[Federal Register Volume 63, Number 157 (Friday, August 14, 1998)]
[Notices]
[Pages 43756-43828]
From the Federal Register Online via the Government Publishing Office [www.gpo.gov]
[FR Doc No: 98-21517]


      

[[Page 43755]]

_______________________________________________________________________

Part II





Environmental Protection Agency





_______________________________________________________________________



Draft Water Quality Criteria Methodology Revisions: Human Health; 
Notice

  Federal Register / Vol. 63, No. 157 / Friday, August 14, 1998 / 
Notices  

[[Page 43756]]



ENVIRONMENTAL PROTECTION AGENCY

[WH-FRL-6141-3]


Draft Water Quality Criteria Methodology Revisions: Human Health

AGENCY: Environmental Protection Agency (EPA).

ACTION: Notice of Draft Revisions to the Methodology for Deriving 
Ambient Water Quality Criteria for the Protection of Human Health.

-----------------------------------------------------------------------

SUMMARY: EPA is announcing the availability for public comment of draft 
revisions to the Methodology for Deriving Ambient Water Quality 
Criteria for the Protection of Human Health (``AWQC Methodology 
Revisions'') published pursuant to Section 304(a)(1) of the Clean Water 
Act (CWA). These AWQC Methodology Revisions, once finalized, will 
supersede the existing Guidelines and Methodology Used in the 
Preparation of Health Effect Assessment Chapters of the Consent Decree 
Water Criteria Documents (``1980 AWQC National Guidelines''), published 
by EPA in November 1980 (45 FR 79347, Appendix C). Today's document is 
intended to satisfy the requirements of Section 304(a)(1) of the CWA 
that EPA periodically revise criteria for water quality to accurately 
reflect the latest scientific knowledge on the kind and extent of all 
identifiable effects on health and welfare that may be expected from 
the presence of pollutants in any body of water, including ground 
water. These AWQC Methodology Revisions are necessitated by the many 
significant scientific advances that have occurred during the past 17 
years in such key areas as cancer and noncancer risk assessments, 
exposure assessments, and bioaccumulation. These revisions are not 
regulations and do not impose legally-binding requirements on EPA, 
States, Territories, Tribes, or the public. Also published as part of 
this document are draft AWQC criteria document summaries for three 
contaminants that reflect the Draft AWQC Methodology Revisions.

AVAILABILITY OF DOCUMENTS: The Draft AWQC Methodology Revisions are 
published below. Copies of the technical support document and the three 
complete criteria documents cited in this document may be obtained from 
the U.S. EPA National Center for Environmental Publications and 
Information (NCEPI), 11029 Kenwood Road, Cincinnati, OH 45242 or (513) 
489-8190. Materials in the public docket will be available for public 
inspection and copying during normal business hours at the Office of 
Water Docket, 401 M St., S.W., Washington, D.C. 20460 by appointment 
only. Appointments may be made by calling (202) 260-3027 and requesting 
item W-97-20. A reasonable fee will be charged for photocopies.
    Selected documents supporting the Draft AWQC Methodology Revisions 
will also be available for viewing by the public at the following 
locations:

I. Region 1 Library, JFK Federal Building, One Congress Street, Boston, 
MA 02203 (617) 565-3300
II. Region 2 Library, 290 Broadway, 16th Floor, New York, NY 10007 
(212) 637-3185
III. Region 3 Library, 841 Chestnut Building, Philadelphia, PA 19107 
(215) 566-5254
IV. Region 4 Library, Atlanta Federal Center, 61 Forsyth St, SW, 9th 
Floor Tower, Atlanta, GA 30303-3104 (404) 347-4216
V. Region 5 Library, 77 West Jackson Boulevard, Chicago, IL 60604-3590 
(312) 353-2022
VI. Region 6 Library, 1445 Ross Avenue, Dallas, TX 75202 (214) 665-6424
VII. Region 7 Information Resource Center, 726 Minnesota Avenue, Kansas 
City, KS 66101-2728 (913) 551-7241
VIII. Region 8 Library, 999 18th Street, Suite 500, Denver, CO 80202-
2466 (303) 312-6746
IX. Region 9 Library, 75 Hawthorne Street, San Francisco, CA 94105 
(415) 744-1517
X. Region 10 Library, 1200 Sixth Avenue, Seattle, WA 98101 (206) 553-
1289

DATES: EPA will accept public comments on the Draft AWQC Methodology 
Revisions on or before December 14, 1998. Comments postmarked after 
this date may not be considered.

ADDRESSES: An original and three copies of all comments and enclosures, 
including references, on the draft AWQC Methodology Revisions should be 
addressed to the W-97-20 Docket Clerk, Water Docket (4101), U.S. EPA, 
401 M St., S.W., Washington, D.C. 20460. Electronic comments must be 
submitted as a WordPerfect 5.1 or WP 6.1 file or as an ASCII file 
avoiding the use of special characters. Comments and data will also be 
accepted on disks in WordPerfect 5.1 or WP 6.1 or ASCII file format. 
Electronic comments on this document may be filed via e-mail at: ow-
[email protected]. Commenters who want EPA to acknowledge receipt 
of their comments should include a self-addressed stamped envelope. No 
facsimiles (faxes) will be accepted.

FOR FURTHER INFORMATION CONTACT: Denis Borum (4304), U.S. EPA, 401 M 
St. S.W., Washington, D.C. 20460 (Telephone: (202) 260-8996).

SUPPLEMENTARY INFORMATION:

                          List of Acronyms Used                         
                                                                        
                                                                        
                                                                        
ADI                                     Acceptable Daily Intake.        
ARAR                                    Applicable or Relevant and      
                                         Appropriate Requirements.      
ASTM                                    American Society of Testing and 
                                         Materials.                     
AWQC                                    Ambient Water Quality Criteria. 
BAF                                     Bioaccumulation Factor.         
BCF                                     Bioconcentration Factor.        
BMD                                     Benchmark Dose.                 
BMR                                     Benchmark Response.             
BSAF                                    Biota-Sediment Accumulation     
                                         Factors.                       
BW                                      Body Weight.                    
C18                                     Carbon-18                       
CDC                                     U.S. Centers for Disease Control
                                         and Prevention.                
CR                                      Consumption Rate.               
CSFII                                   Continuing Survey of Food Intake
                                         by Individuals.                
CTR                                     California Toxics Rule.         
CWA                                     Clean Water Act.                
DI                                      Drinking Water Intake.          
DNA                                     Deoxyribonucleic Acid.          
DOC                                     Dissolved Organic Carbon.       
DT                                      Non-Fish Dietary Intake.        
ED10                                    Dose Associated with a 10       
                                         Percent Extra Risk.            
EMAP                                    Environmental Modeling and      
                                         Assessment Program.            
EPA                                     Environmental Protection Agency.
FCM                                     Food Chain Multiplier.          
FDA                                     Food and Drug Administration.   
FEL                                     Frank Effect Level.             
FI                                      Fish Intake.                    
FIFRA                                   Federal Insecticide, Fungicide, 
                                         and Rodenticide Act.           
FR                                      Federal Register.               
FSTRAC                                  Federal State Toxicology and    
                                         Risk Analysis Committee.       
GI                                      Gastrointestinal.               
GLI                                     Great Lakes Water Quality       
                                         Initiative.                    
IARC                                    International Agency for        
                                         Research on Cancer.            
II                                      Incidental Intake.              
ILSI                                    International Life Sciences     
                                         Institute.                     
IN                                      Inhalation Intake.              
IRIS                                    Integration Risk Information    
                                         System.                        
kg                                      kilogram                        
Kow                                     Octanol-Water Partition         
                                         Coefficient.                   
L                                       Liter.                          
LED10                                   The Lower 95 Percent Confidence 
                                         Limit on a Dose Associated with
                                         a 10 Percent Extra Risk.       

[[Page 43757]]

                                                                        
LMS                                     Linear Multistage Model.        
LOAEL                                   Lowest Observed Adverse Effect  
                                         Level.                         
LR                                      Lifetime Risk.                  
MCL                                     Maximum Contaminant Level.      
MCLG                                    Maximum Contaminant Level Goal. 
MF                                      Modifying Factor.               
mg                                      Milligrams.                     
ml                                      Milliliters.                    
MoA                                     Mode of Action.                 
MoE                                     Margin of Exposure.             
MoS                                     Margin of Safety.               
NCHS                                    National Center for Health      
                                         Statistics.                    
NHANES                                  National Health and Nutrition   
                                         Examination Survey.            
NIEHS                                   National Institute of           
                                         Environmental Health Sciences. 
NOAEL                                   No Observed Adverse Effect      
                                         Level.                         
NOEL                                    No Observed Effect Level.       
NPDES                                   National Pollutant Discharge    
                                         Elimination System.            
NTIS                                    National Technical Information  
                                         Service.                       
NTR                                     National Toxics Rule.           
ODES                                    Ocean Data Evaluation System.   
PAH                                     Polycyclic Aromatic Hydrocarbon.
PBPK                                    Physiologically Based           
                                         Pharmacokinetic.               
PCB                                     Polychlorinated BIPHENYLS.      
PCS                                     Permits Compliance System.      
Pdp                                     Point of Departure.             
POC                                     Particulate Organic Carbon.     
q1*                                     Cancer Potency Factors.         
RDA                                     Recommended Daily Allowance.    
RfC                                     Reference Concentration.        
RfD                                     Reference Dose.                 
RPF                                     Relative Potency Factor.        
RSC                                     Relative Source Contribution.   
RSD                                     Risk Specific Dose.             
SAR                                     Structure-Activity Relationship.
SAB                                     Science Advisory Board.         
SDWA                                    Safe Drinking Water Act.        
SF                                      Safety Factor.                  
STORET                                  Storage Retrieval.              
TCDD-dioxin                             Tetrachlorodibenzo-p-dioxin.    
TEAM                                    Total Exposure Assessment       
                                         Methodology.                   
TEF                                     Toxicity Equivalency Factor.    
TMDL                                    Total Maximum Daily Load.       
TSD                                     Technical Support Document.     
USDA                                    United States Department of     
                                         Agriculture.                   
UF                                      Uncertainty Factor.             
WQBEL                                   Water Quality-Based Effluent    
                                         Limits.                        
                                                                        

Table of Contents

Summary of Today's Action

Appendix I. Background
    A. Water Quality Criteria and Standards
    1. Water Quality Criteria and the Criteria Derivation 
Methodology
    2. Summary of the 1980 AWQC National Guidelines
    3. Water Quality Standards
    B. Need for Revision of the 1980 AWQC National Guidelines
    1. Scientific Advances Since 1980
    2. EPA Human Health Risk Assessment Guidelines Development Since 
1980
    3. Differing Risk Assessment and Risk Management Approaches for 
AWQC and MCLGs
    C. Steps Taken toward Evaluating and Revising the 1980 AWQC 
National Guidelines
    1. September 1992 National Workshop
    2. Science Advisory Board Review
    3. FSTRAC Review
    4. Water Quality Guidance for the Great Lakes System
    D. Overview of AWQC Methodology Revisions, Major Changes, and 
Issues
    E. Risk Characterization Considerations
    1. Background
    2. Additional Guiding Principles
    3. Risk Characterization Applied to the Revised AWQC Methodology
    4. Science, Science Policy, and Risk Management
    5. Discussion of Uncertainty
    (a) Observed Range of Toxicity Versus Range of Environmental 
Exposure
    (b) Continuum of Preferred Data/Use of Defaults
    (c) Significant Figures
Appendix II. Implementation of AWQC Methodology Revisions
    A. Relationship to Other EPA Activities
    B. Status of Existing 304(a) Criteria for Priority Pollutants 
and Methodology
    C. State and Tribal Criteria Development
    D. Process for Developing New or Revised 304(a) Criteria
    E. Development of Future Criteria Documents
    F. Prioritization Scheme for Selecting Chemicals for Updating
    G. Request for Comments
Appendix III. Elements of Methodology Revisions and Issues by 
Technical Area
    A. Cancer Effects
    1. Background on EPA Cancer Assessment Guidelines
    (a) 1980 AWQC National Guidelines
    (b) 1986 EPA Guidelines for Carcinogenic Risk Assessment
    (c) Scientific Issues Associated with the Current Cancer Risk 
Assessment Methodology for the Development of AWQC
    2. Proposed Revisions to EPA's Carcinogen Risk Assessment 
Guidelines
    3. Revised Carcinogen Risk Assessment Methodology for Deriving 
AWQC
    (a) Weight-of-Evidence Narrative
    (b) Dose Estimation
    (1) Determining the Human Equivalent Dose
    (2) Dose Adjustments for Less-than-Lifetime Exposure Periods
    (3) Dose-Response Analysis
    (c) Characterizing Dose-Response Relationships in the Range of 
Observation
    (1) Extrapolation to Low, Environmentally Relevant Doses
    (2) Biologically Based Modeling Approaches
    (3) Default Linear Extrapolation Approach
    (4) Default Nonlinear Approach
    (5) Both Linear and Nonlinear Approaches
    (d) AWQC Calculation
    (e) Risk Characterization
    (f) Use of Toxicity Equivalence Factors (TEF) and Relative 
Potency Estimates
    4. Request for Comments

References for Cancer Effects

    B. Noncancer Effects
    1. 1980 AWQC National Guidelines for Noncancer Effects
    2. Noncancer Risk Assessment Developments Since 1980
    3. Issues and Recommendations Concerning the Derivation of AWQC 
for Noncarcinogens
    (a) Using the Current NOAEL-UF Based RfD Approach or Adopting 
More Quantitative Approaches for Noncancer Risk Assessment
    (1) The Benchmark Dose
    (2) Categorical Regression
    (3) Summary
    (b) Presenting the RfD as a Single Point or as a Range for 
Deriving AWQC
    (c) Guidelines to be Adopted for Derivation of Noncancer Health 
Effects Values
    (d) Treatment of Uncertainty Factors/Severity of Effects During 
the RfD Derivation and Verification Process
    (e) Use of Less-Than-90-Day Studies to Derive RfDs
    (f) Use of Reproductive/Developmental, Immunotoxicity, and 
Neurotoxicity Data as the Basis for Deriving RfDs
    (g) Applicability of Physiologically Based Pharmacokinetic 
(PBPK) Data in Risk Assessment
    (h) Consideration of Linearity (or Lack of a Threshold) for 
Noncarcinogenic Chemicals
    (i) Minimum Data Requirements
    4. SAB Comments
    5. Request for Comments

References for Noncancer Effects

    C. Exposure
    1. Policy Issues
    (a) Identifying the Population Subgroup that the AWQC Should 
Protect
    (b) Appropriateness of Including the Drinking Water Pathway in 
AWQC
    (c) Relationship Between Human Health AWQC and Drinking Water 
Standards
    (d) Setting Separate AWQC for Drinking Water and Fish 
Consumption
    (e) Incidental Ingestion from Ambient Surface Waters
    2. Consideration of Nonwater Sources of Exposure When Setting 
AWQC
    (a) Background
    (b) Exposure Decision Tree Approach
    (c) Quantification of Exposure
    (d) Inclusion of Inhalation and Dermal Exposures From Household 
Drinking Water Uses
    (e) Inclusion of Inhalation Exposures in RSC Analysis

[[Page 43758]]

    (f) Bioavailability of Substances from Different Routes of 
Exposure
    (g) Consideration of Non-water Exposure Procedures for 
Noncarcinogens, Linear Carcinogens, and Nonlinear Carcinogens
    3. Factors Used in the AWQC Computation
    (a) Human Body Weight Values for Dose Calculations
    (1) Rate Protective of Human Health from Chronic Exposure
    (2) Rates Protective of Developmental Human Health Effects
    (3) Rates Based on Combining Intake and Body Weight
    (b) Drinking Water Intake Rates
    (1) Rate Protective of Human Health from Chronic Exposure
    (2) Rates Protective of Developmental Human Health Effects
    (3) Rates Based on Combining Drinking Water Intake and Body 
Weight
    (c) Incidental Ingestion from Ambient Surface Waters
    (d) Fish Intake Rates
    (1) Rates Protective of Human Health from Chronic Exposure
    (2) Rates Protective of Developmental Human Health Effects
    (3) Rates Based on Combining Fish Intake and Body Weight
    4. Request for Comments

References for Exposure

    D. Bioaccumulation
    1. Introduction
    2. Bioaccumulation and Bioconcentration Concepts
    3. Existing EPA Guidance
    4. Definitions
    5. Determining Bioaccumulation Factors for Nonpolar Organic 
Chemicals
    6. Estimating Baseline BAFs
    (a) Field-Measured Baseline BAF
    (b) Baseline BAF Derived from BSAFs
    (c) Calculation of a Baseline BAF from a Laboratory-Measured BCF 
and FCM
    (d) Calculation of a Baseline BAF from a Kow and FCM
    (e) Metabolism
    7. BAFs Used in Deriving AWQC
    8. Inorganic Substances
    9. SAB Comments
    10. Issues for Public Comment

References for Bioaccumulation

    E. Microbiology
    1. Existing Microbiological Criteria
    2. Plans for Future Work
    3. SAB Comments

References for Microbiology

    F. Other Considerations
    1. Minimum Data Considerations
    2. Site-Specific Criterion Calculation
    3. Organoleptic Criteria
    4. Criteria for Chemical Classes
    5. Criteria for Essential Elements
Appendix IV. Summary of Ambient Water Quality Criteria for the 
Protection of Human Health: Acrylonitrile
Appendix V. Summary of Ambient Water Quality Criteria for the 
Protection of Human Health: 1,3-Dichloropropene
Appendix VI. Summary of Ambient Water Quality Criteria for the 
Protection of Human Health: Hexachlorobutadiene

Summary of Today's Action

I. Background

    Section 304(a)(1) of the Clean Water Act requires EPA to develop 
and periodically revise criteria for water quality accurately 
reflecting the latest scientific knowledge. In 1980, EPA published 
ambient water quality criteria (AWQC) for 64 pollutants/pollutant 
classes and provided a methodology for deriving the criteria. The 1980 
AWQC National Guidelines for developing human health AWQC addressed 
three types of endpoints: noncancer, cancer and organoleptic (taste and 
odor) effects. Criteria values for the protection against noncancer and 
cancer effects were estimated by using risk assessment-based 
procedures, including extrapolation from animal toxicity or human 
epidemiological studies. Basic human exposure assumptions were applied 
to the criterion equation, such as: the exposed individual is a 70-
kilogram adult male; the assumed consumption of freshwater and 
estuarine fish and shellfish is 6.5 grams/day; and the assumed 
ingestion rate of drinking water is 2 liters/day. When using cancer as 
the critical risk assessment endpoint, which was assumed not to have a 
threshold, the AWQC were presented for information purposes as a range 
of concentrations associated with specified incremental lifetime risk 
levels (i.e., a range from 10-5 to 10-7). When 
using noncancer effects as the critical endpoint, the AWQC reflected an 
assessment of a ``no-effect'' level, since noncancer effects generally 
exhibit a threshold.

Scientific Advances Since 1980

    Since 1980, EPA risk assessment practices have evolved 
significantly, particularly in the areas of cancer and noncancer risk 
assessments, exposure assessments and bioaccumulation. In cancer risk 
assessment, there have been advances with respect to the use of mode of 
action information to support both the identification of carcinogens 
and the selection of procedures to characterize risk at low, 
environmentally relevant exposure levels. Related to this is the 
development of new procedures to quantify cancer risks at low doses to 
replace the current default use of the linearized multistage (LMS) 
model. In noncancer risk assessment, the Agency is moving toward the 
use of the benchmark dose (BMD) and other dose-response approaches in 
place of the traditional NOAEL approach to estimate a reference dose or 
concentration. In exposure analysis, several new studies have addressed 
water consumption and fish tissue consumption. These exposure studies 
provide a more current and comprehensive description of national, 
regional and special population consumption patterns that EPA has 
reflected in the Draft AWQC Methodology Revisions. In addition, more 
formalized procedures are now available to account for human exposure 
to multiple sources when setting health goals such as AWQC that have 
addressed only one exposure source. With respect to bioaccumulation, 
the Agency has moved toward the use of a bioaccumulation factor (BAF) 
to reflect the uptake of a contaminant from all sources (e.g., 
ingestion, sediment) by fish and shellfish, rather than just from the 
water column as reflected by the use of a bioconcentration factor (BCF) 
as included in the 1980 methodology. The Agency has developed detailed 
procedures and guidelines for estimating BAF values.

EPA Human Health Risk Assessment Guidelines Developed Since 1980

    When the 1980 AWQC National Guidelines were developed, EPA had not 
yet developed formal cancer or noncancer risk assessment guidelines. 
Since then EPA has published several risk assessment guidelines 
documents. In 1996, the Agency published Proposed Guidelines for 
Carcinogen Risk Assessment (61 FR 17960) which when finalized will 
supersede the carcinogenic risk assessment guidelines published in 1986 
(51 FR 33992). In addition, guidelines for mutagenicity assessment were 
also published in 1986 (51 FR 34006). The Agency also issued guidelines 
for assessing the health risks to chemical mixtures in 1986 (51 FR 
34014). With respect to noncancer risk assessment, the Agency published 
guidelines in 1988 for assessing male and female reproductive risk (53 
FR 24834) and in 1991 for assessing developmental toxicity (56 FR 
63798). The guidelines for assessing reproductive toxicity were 
subsequently updated and finalized (61 FR 56274) in 1996. In 1991, the 
Agency also developed an external review draft of revised risk 
assessment guidelines for noncancer health effects. In 1995, EPA also 
proposed guidelines for neurotoxicity risk assessment (60 FR 52032).
    In addition to these risk assessment guidelines, EPA also published 
the ``Exposure Factors Handbook'' in 1989, which presents commonly used 
Agency exposure assumptions and the surveys from which they are 
derived. The Exposure Factors Handbook (EPA/600/P-95/002Fa) was updated 
in 1997. In 1992, EPA published the revised

[[Page 43759]]

Guidelines for Exposure Assessment (57 FR 22888), which describe 
general concepts of exposure assessment, including definitions and 
associated units, and provide guidance on planning and conducting an 
exposure assessment. Also, in the 1980s the Agency published the Total 
Exposure Assessment Methodology (TEAM), which presents a process for 
conducting comprehensive evaluation of human exposures. The Agency has 
recently developed the Relative Source Contribution Policy, which is 
currently undergoing Agency review, for assessing total human exposure 
to a contaminant and allocating the RfD among the media of concern. In 
1997, EPA developed draft Guiding Principles for Monte Carlo analysis.
    Also, in 1986, the Agency made available to the public the 
Integrated Risk Information System (IRIS). IRIS is a data base that 
contains risk information on the cancer and noncancer effects of 
chemicals. The IRIS assessments are peer reviewed and represent EPA 
consensus positions across the Agency's program offices and regional 
offices. In 1995, the Agency initiated an IRIS pilot program to test 
improvements to the internal peer review and consensus processes, and 
to provide more integrated characterizations of cancer and noncancer 
health effects.

Differing Risk Assessment and Risk Management Approaches for AWQC and 
MCLGs

    Another reason for these revisions is the need to bridge the gap 
between the differences in the risk assessment and risk management 
approaches used by EPA's Office of Water for the derivation of AWQC 
under the authority of the CWA and MCLGs (Maximum Contaminant Level 
Goals) under the Safe Drinking Water Act (SDWA). Three notable 
differences are with respect to the treatment of chemicals designated 
as Group C possible human carcinogens--under the 1986 Guidelines for 
Carcinogen Risk Assessment, the consideration of nonwater sources of 
exposure when setting an AWQC or MCLG for a noncarcinogen, and cancer 
risk ranges.
    1. Group C Chemicals. Chemicals have been typically classified as 
Group C--i.e., possible human carcinogens'--under the existing (1986) 
EPA cancer classification scheme for any of the following reasons:
    (1) Carcinogenicity has been documented in only one test species 
and/or only one cancer bioassay and the results do not meet the 
requirements of ``sufficient evidence.''
    (2) Tumor response is of marginal significance due to inadequate 
design or reporting.
    (3) Benign, but not malignant, tumors occur with an agent showing 
no response in a variety of short-term tests for mutagenicity.
    (4) There are responses of marginal statistical significance in a 
tissue known to have a high or variable background rate.
    The 1986 Guidelines for Carcinogen Risk Assessment specifically 
recognized the need for flexibility with respect to quantifying the 
risk of Group C agents. The guidelines noted that agents judged to be 
in Group C, possible human carcinogens, may generally be regarded as 
suitable for quantitative risk assessment, but that case-by-case 
judgments may be made in this regard.
    The EPA Office of Water has historically treated Group C chemicals 
differently under the CWA and the SDWA. It is important to note that 
the 1980 AWQC National Guidelines for setting AWQC under the CWA 
predated EPA's carcinogen classification system, which was proposed in 
1984 (49 FR 46294) and finalized in 1986 (51 FR 33992). The 1980 AWQC 
National Guidelines did not explicitly differentiate among agents with 
respect to the weight-of-evidence for characterizing them as likely to 
be carcinogenic to humans. For all pollutants judged as having adequate 
data for quantifying carcinogenic risk--including those now classified 
as Group C--AWQC were derived based on data on cancer incidence. In the 
November 1980 Federal Register document, EPA emphasized that the AWQC 
for carcinogens should state that the recommended concentration for 
maximum protection of human health is zero. At the same time, the 
criteria published for specific carcinogens presented water 
concentrations for these pollutants corresponding to individual 
lifetime cancer risk levels in the range of 10-7 to 
10-5.
    In the development of national primary drinking water regulations 
under the SDWA, EPA is required to promulgate a health-based MCLG for 
each contaminant. The Agency policy has been to set the MCLG at zero 
for chemicals with strong evidence of carcinogenicity associated with 
exposure from water. For chemicals with limited evidence of 
carcinogenicity, including many Group C agents, the MCLG is usually 
obtained using an RfD based on its noncancer effects with the 
application of an additional uncertainty factor of 1 to 10 to account 
for its possible carcinogenicity. If valid noncancer data for a Group C 
agent are not available to establish an RfD but adequate data are 
available to quantify the cancer risk, then the MCLG is based upon a 
nominal lifetime excess cancer risk calculation in the range of 
10-5 to 10-6 (ranging from one case in a 
population of one hundred thousand to one case in a population of one 
million). Even in those cases where the RfD approach has been used for 
the derivation of the MCLG for a Group C agent, the drinking water 
concentrations associated with excess cancer risks in the range of 
10-5 to 10-6 were also provided for comparison.
    It should also be noted that EPA's pesticides program has applied 
both of the previously described methods for addressing Group C 
chemicals in actions taken under the Federal Insecticide, Fungicide, 
and Rodenticide Act (FIFRA) and finds both methods applicable on a 
case-by-case basis. Unlike the drinking water program, however, the 
pesticides program does not add an extra uncertainty factor to account 
for potential carcinogenicity when using the RfD approach.
    2. Consideration of Nonwater Sources of Exposure. The 1980 AWQC 
National Guidelines for setting AWQC recommended that contributions 
from nonwater sources, namely air and non-fish dietary intake, be 
subtracted from the ADI, thus reducing the amount of the ADI 
``available'' for water-related sources of intake. In practice, 
however, when calculating human health criteria, these other exposures 
were generally not considered because reliable data on these exposure 
pathways were not available. Consequently, the AWQC were usually 
derived such that drinking water and fish ingestion accounted for the 
entire ADI (now called RfD).
    In the drinking water program, a similar ``subtraction'' method was 
used in the derivation of MCLGs proposed and promulgated in drinking 
water regulations through the mid-1980s. More recently, the drinking 
water program has consistently used a ``percentage'' method in the 
derivation of MCLGs for noncarcinogens. In this approach, the 
percentage of total exposure typically accounted for by drinking water, 
referred to as the relative source contribution (RSC), is applied to 
the RfD to determine the maximum amount of the RfD ``allocated'' to 
drinking water reflected by the MCLG value. In using this percentage 
procedure, the drinking water program also applies a ceiling level of 
80 percent of the RfD and a floor level of 20 percent of the RfD. That 
is, the MCLG cannot account for more than 80 percent of the RfD, nor 
less than 20 percent of the RfD.

[[Page 43760]]

    The drinking water program usually takes a conservative public 
health approach of applying an RSC factor of 20 percent to the RfD when 
adequate exposure data do not exist, assuming that the major portion 
(80 percent) of the total exposure comes from other sources, such as 
diet.
    3. Cancer Risk Ranges. In addition to the different risk assessment 
approaches discussed above for deriving AWQC and MCLGs for Group C 
agents, different risk management approaches have arisen between the 
drinking water and ambient surface water programs with respect to using 
lifetime excess risk values when setting health-based criteria for 
carcinogens. As indicated previously, the surface water program 
historically derived AWQC for carcinogens that generally corresponded 
to lifetime excess cancer risk levels of 10-7 to 
10-5. The drinking water program has set MCLGs for Group C 
agents based on a slightly less stringent risk range of 10-6 
to 10-5, while MCLGs for chemicals with strong evidence of 
carcinogenicity (that is, classified as Group A (known) or B (probable) 
human carcinogen) are set at zero.
    It is also important to note that under the drinking water program, 
for those substances having an MCLG of zero, enforceable Maximum 
Contaminant Levels (MCLs) have generally been promulgated to correspond 
with cancer risk levels ranging from 10-6 to 
10-4. Unlike AWQC and MCLGs which are strictly health-based 
criteria, MCLs are developed with consideration given to the costs and 
technological feasibility of reducing contaminant levels in water to 
meet those standards.
Steps Taken Toward Evaluating and Revising the 1980 AWQC National 
Guidelines
    In order to begin developing a ``state-of-the-science'' approach to 
revising the 1980 AWQC National Guidelines, EPA prepared an issues 
paper that described the 1980 methodology, discussed areas that needed 
strengthening, and proposed revisions. This paper was then distributed 
for review and comment to experts at EPA headquarters, regional 
offices, and laboratories; other Federal Agencies, such as the Food and 
Drug Administration (FDA), the National Institute of Environmental 
Health Sciences (NIEHS), and the Centers for Disease Control and 
Prevention (CDC); State health organizations; Canadian health agencies; 
academe; and environmental, industry, and consulting organizations.
    1. September 1992 National Workshop. On September 13-16, 1992, more 
than 100 invited participants discussed the critical issues in a 
workshop convened in Bethesda, Maryland. Based on their expertise, 
attendees were assigned to specific technical work groups. The work 
group topics were cancer risk, noncancer risk, exposure, microbiology, 
minimum data, and bioaccumulation. Each work group member received a 
set of detailed questions that served to focus discussions on critical 
factors in the 1980 AWQC National Guidelines. After the work group 
members deliberated separately on their specific technical areas, all 
workshop participants were given the opportunity to comment on the 
proceedings. After the workshop concluded, the chairperson for each 
technical work group prepared a written summary of that group's 
deliberations and recommendations. Each work group participant was 
given the opportunity to review and comment on the summaries; these 
comments were used to prepare a draft of the proposed revision to the 
methodology.
    2. Science Advisory Board Review. After review of the draft of the 
proposed revisions to the methodology by EPA, the workshop 
participants, and other relevant parties, a summary document was 
submitted for review and comment to the Science Advisory Board (SAB) in 
January 1993 and presented to the Drinking Water Committee of the SAB 
during its meeting on February 8-9, 1993. The SAB presented its 
official comments to EPA on August 12, 1993. The SAB comments have been 
highlighted and addressed in each of the technical areas discussed in 
Appendix III of this document. A complete copy of the document 
submitted to the SAB and SAB's comments are available in the docket 
accompanying this document.
    3. FSTRAC Review. At the Federal State Toxicology and Risk Analysis 
Committee (FSTRAC) meeting on December 1-3, 1993, in Washington, D.C., 
several State representatives presented their opinions on the 
preliminary draft recommendations for revisions to the 1980 AWQC 
National Guidelines. A summary of this meeting is presented in a 
document entitled ``Workshop Summary: State Comments on the Preliminary 
Draft Revisions of the Methodology for Deriving National Ambient Water 
Quality Criteria for the Protection of Human Health.'' This document is 
also available for review in the docket supporting this proposal.
    4. Water Quality Guidance for the Great Lakes System. In March 
1995, EPA published the Final Water Quality Guidance for the Great 
Lakes System (60 FR 15366). The Great Lakes Water Quality Guidance, 
developed under Section 118(c)(2) of the CWA, provides water quality 
criteria for 29 pollutants as well as methodologies, policies, and 
procedures for Great Lakes States and Tribes to establish consistent, 
long-term protection for fish and shellfish in the Great Lakes and 
their tributaries, as well as for the people and wildlife who consume 
them. In developing the methodology to derive human health criteria for 
the waters of the Great Lakes System, the Agency was mindful of the 
need for consistency with the planned changes in the methodology for 
deriving national AWQC for the protection of human health presented in 
today's proposal. Throughout the following text, references are made to 
comparisons of the two methodologies, national and Great Lakes Water 
Quality Guidance, especially whenever differences occur due to regional 
exposure assumptions made for the Great Lakes System.
Major Changes in the Draft AWQC Methodology Revisions
    The proposal presents several changes from the 1980 AWQC National 
Guidelines:
    1. EPA's future role in developing AWQC for the protection of human 
health will include the refinement of the revised methodology, the 
development of revised criteria for chemicals of high priority and 
national importance (including, but not limited to chemicals that 
bioaccumulate, such as PCBs, dioxin, and mercury), and the development 
or revision of AWQC for some additional priority chemicals. EPA does 
not plan to completely revise all of the criteria developed in 1980 or 
those updated as part of the proposed California Toxics Rule (CTR) 62 
FR 42160, August 5, 1997. (This rule proposes for California, numeric 
water quality criteria for priority toxic pollutants necessary to 
fulfill the requirements of Section 303(c)(2)(b) of the CWA.) Further, 
EPA intends to revise 304(a) criteria on the basis of one or more 
components (e.g., BAF, fish intake, toxicological assessment) rather 
than a full set of components. Appendix II of the FR document discusses 
how the Agency is proposing to implement the methodology and revise the 
304(a) criteria. EPA also discusses the role of 304(a) criteria in 
State/Tribal adoption of water quality standards under Section 303(c) 
of the CWA, EPA's responsibilities in reviewing and approving State/
Tribal standards, and EPA's duties in regards to promulgating State/
Tribal standards when necessary.
    2. EPA encourages States and Tribes to use the revised methodology, 
once finalized, to develop or revise AWQC to

[[Page 43761]]

appropriately reflect local conditions. EPA believes that AWQC 
inherently require several risk management decisions that are, in many 
cases, better made at the State and Tribal level (e.g., fish 
consumption rates, target risk levels). EPA will continue to develop 
and update necessary toxicological and exposure data needed in the 
derivation of AWQC that may not be practical for the States or Tribes 
to obtain. EPA encourages States and Tribes to use local or regional 
fish consumption data when available.
    3. The equations for deriving AWQC include toxicological and 
exposure assessment parameters which are derived from scientific 
analysis, science policy, and risk management decisions. For example, 
parameters such as a field-measured BAF or a point of departure from an 
animal study (in the form of a LOAEL/NOAEL/LED10) are 
scientific values which are empirically measured, whereas the decision 
to use animal effects as a surrogate for human effects involves 
judgment on the part of the EPA (and similarly, by other agencies) as 
to the best practice to follow when human data are lacking. Such a 
decision is, therefore, a matter of science policy. On the other hand, 
the choice of default fish consumption rates for protection of a 
certain percentage of the general population, is clearly a risk 
management decision. In many cases, the Agency has selected parameters 
using its best judgment regarding the overall protection afforded by 
the resulting AWQC when all parameters are combined. Appendix I 
discusses in detail the differences between science, science policy, 
and risk management. Appendix I also provides further details with 
regard to risk characterization as related to this methodology, with 
emphasis placed on explaining the uncertainties in the overall risk 
assessment.
    4. The Draft AWQC Methodology Revisions provide an alternative to 
expressing AWQC as a water concentration. AWQC may also be expressed in 
terms of a fish tissue concentration. For some substances, particularly 
those that are expected to exhibit substantial bioaccumulation, the 
AWQC derived using the above equations may have extremely low values, 
possibly below the practical limits for detecting and quantifying the 
substance in the water column. It may, therefore, be more practical and 
meaningful in these cases to focus on the concentration of those 
substances in fish tissue, since fish ingestion would be the 
predominant source of exposure for substances that bioaccumulate.
    5. EPA is proposing an incidental water ingestion exposure rate of 
0.01 L/day to account for long-term incidental recreational ingestion 
(i.e., swimming, boating, fishing) for use in those cases where AWQC 
are developed for recreational waters that are not used as drinking 
water sources.
    6. AWQC for the protection of human health are designed to minimize 
the risk of adverse effects occurring to humans from chronic (lifetime) 
exposure to substances through the ingestion of drinking water and 
consumption of fish obtained from surface waters. The Agency is not 
recommending the development of additional water quality criteria 
similar to the ``drinking water health advisories'' that focus on acute 
or short-term effects, since these are not seen routinely as having a 
meaningful role in the water quality criteria and standards program.
    However, there may be some instances where the consideration of 
short-term toxicity and exposure in the derivation of AWQC is 
warranted. Although the AWQC are based on chronic health effects data 
(both cancer and noncancer effects), the criteria are intended to also 
be protective with respect to adverse effects that may reasonably be 
expected to occur as a result of elevated short-term exposures. That 
is, through the use of conservative assumptions with respect to both 
toxicity and exposure parameters, the resulting AWQC values should 
provide adequate protection not only for the general population over a 
lifetime of exposure, but also for special subpopulations who, because 
of high water- or fish-intake rates, or because of biological 
sensitivities, have an increased risk of receiving a dose that would 
elicit adverse effects from short-term exposures. The Agency 
recognizes, however, that there may be some cases where the AWQC values 
based on chronic toxicity may not provide adequate protection for a 
subpopulation at special risk from such exposures. The Agency 
encourages States, Tribes, and others employing the proposed 
methodology to give consideration to such circumstances in deriving 
criteria to ensure that adequate protection is afforded to all 
identifiable subpopulations. (Appendix III discusses this in greater 
detail.)
    7. For noncarcinogens, risk managers may select another value 
within an RfD range rather than the default point estimate RfD value, 
in criteria development, where a rationale for the range and the value 
selected can be provided. General guidance for the use of values within 
the RfD range is provided based on the overall uncertainty associated 
with the RfD and when adverse health effects in children are not the 
basis for the RfD. For example, if the IRIS RfD is 1 mg/kg/day and the 
uncertainty factor (UF) is 1,000, a log-symmetrical order of magnitude 
around 1 mg/kg/day could be used resulting in a range of 0.3 to 3 mg/
kg/day. If the UF were less than 1,000, the overall range would be 
reduced accordingly (e.g., \1/2\ log for UFs between 100 and 1,000; and 
no range for UFs of 100 or less). However, EPA would select the point 
estimate as a default (the midpoint within the range) when calculating 
a 304(a) criteria value for the purposes of promulgating State or 
Tribal water quality standards.
    8. The Draft AWQC Methodology Revisions reflect EPA's 1996 Proposed 
Guidelines for Carcinogen Risk Assessment. For instance, mode of action 
(MoA) information is used to determine the most appropriate low-dose 
extrapolation approach for carcinogenic agents. The dose-response 
assessment under the new guidelines is a two-step process. In the first 
step, the response data are modeled in the range of empirical 
observation. Modeling in the observed range is done with biologically 
based or appropriate curve-fitting modeling. In the second step, 
extrapolation below the range of observation is accomplished by 
biologically based modeling if there are sufficient data or by a 
default procedure (linear, nonlinear, or both). A point of departure 
for extrapolation is estimated from modeling observed data. The lower 
95 percent confidence limit on a dose associated with 10 percent extra 
risk (i.e., LED10) is proposed as a standard point of 
departure for low-dose extrapolation. If it is determined that the MoA 
understanding supports a nonlinear extrapolation, the AWQC is derived 
using the nonlinear default which is based on a margin of exposure 
(MoE) analysis for the point of departure (e.g., the LED10) 
and applying a safety factor(s) in the risk management. The linear 
default would be considered for those agents that are better supported 
by the assumption of linearity (e.g., direct DNA reactive mutagens) for 
their MoA. A linear approach would also be applied when inadequate or 
no information is available to explain the carcinogenic MoA as a 
science policy choice in the interest of public health. The linear 
default is a straight line extrapolation to the origin (i.e., zero 
dose, zero extra risk) from the point of departure (e.g., 
LED10) identified in the observable response range. There 
may be situations where it is appropriate to apply both the linear and 
nonlinear

[[Page 43762]]

default procedures (e.g., for an agent that is both DNA reactive and 
active as a promoter at higher doses).
    9. For substances that are carcinogenic, particularly those for 
which the mode of action suggests nonlinearity at low doses, the Agency 
recommends that an integrated approach be taken in looking at cancer 
and noncancer effects, and if one pathway does not predominate, AWQC 
values should be determined for both carcinogenic and noncarcinogenic 
effects. The lower of the resulting values should be used for the AWQC.
    10. When deriving AWQC for noncarcinogens and nonlinear 
carcinogens, a factor must be included to account for other nonwater 
exposure sources so that the entire RfD, or [Point of Departure (Pdp) 
divided by a safety factor (SF); (Pdp)/SF)] is not allocated to 
drinking water and fish consumption alone. Guidance is provided in the 
revised methodology for determining the factor, referred to as the 
relative source contribution (RSC), to be used for a particular 
chemical. The Agency is proposing the use of a decision tree procedure 
to support the determination of the appropriate RSC value for a given 
water contaminant. In the absence of data, the Agency will use 20 
percent of the RfD as the default RSC in calculating a 304(a) criteria 
value for the purposes of promulgating State or Tribal water quality 
standards.
    11. When deriving AWQC for linear carcinogens, the Agency 
recommends that risk levels in the range of 10-5 to 
10-6 be used for the protection of the general population. 
States and Tribes can always choose a more stringent risk level, such 
as 10-7. Care should be taken, however, in situations where 
the AWQC includes fish intake levels based on the general population to 
ensure that the risk to more highly exposed subgroups (sportfishers or 
subsistence fishers) does not exceed the 10-4 level.
    12. The default fish consumption values are 17.80 grams/day for the 
general population, which represents the 90th percentile consumption 
rate for the entire population (and approximates the average 
consumption rate for sport anglers, nationally) and 86.30 grams/day for 
subsistence fishers/minority anglers, which represents the 99th 
percentile consumption rate for the general population and is within 
the range of average intakes for subsistence fishers/minority anglers 
(comments are requested on alternatively using 39.04 grams/day for 
subsistence fishers/minority anglers, which is lower in the range of 
averages). These values are derived from the United States Department 
of Agriculture's (USDA) Continuing Survey of Food Intake by Individuals 
(CSFII) from 1989-1991. These rates replace the single default value of 
6.5 grams/day used in the 1980 AWQC National Guidelines. These default 
values are chosen to be protective of the majority of the individuals 
in those groups. However, States and Tribes are urged to use a fish 
intake level derived from local data on fish consumption in place of 
these default values when deriving AWQC, ensuring that the fish intake 
level chosen be protective of highly exposed individuals in the 
population. Consumption rates for women of childbearing age and 
children younger than 14 are also provided to maximize protection in 
those cases where these subpopulations may be at greatest risk.
    13. All criteria should be derived using a BAF rather than a BCF, 
which was used in the 1980 AWQC National Guidelines. The BAF should be 
developed using the EPA methodology or any method consistent with the 
EPA method. EPA's highest preference in developing BAFs are BAFs based 
on field-measured data from local/regional fish.
    14. EPA is neither setting organoleptic criteria nor a default 
methodology for deriving such criteria. Such criteria will necessitate 
case-by-case analysis.
    The attached document includes six major sections: Appendix I, 
which discusses the purpose of the methodology, the background 
associated with the original methodology and the need for revision, and 
the major changes in the revised methodology; Appendix II, which 
addresses implementation issues associated with the methodology; 
Appendix III, which presents the main scientific areas that make up the 
methodology (cancer, noncancer, exposure, and bioaccumulation methods); 
and Appendices IV through VI, which present summaries of the three 
criteria developed for inclusion with the revised methodology. Complete 
versions of the three criteria documents are available on the Internet 
at http://www.epa.gov/OST/Rules/index.html#open.
    This document proposes revisions to EPA's 1980 methodology for the 
development of water quality criteria to protect human health. The 
revisions reflect scientific advancements since 1980 in a number of 
areas, including cancer and noncancer risk assessments, exposure 
assessments and bioaccumulation. When final, the revised methodology 
will provide guidance to States, Tribes, and the public on the approach 
that EPA expects to take in developing recommended human health 
criteria. The revised methodology also will provide guidance to States 
and Tribes that they may use in developing human health criteria as 
part of their water quality standards; States and Tribes use such 
standards in implementing a number of environmental programs, including 
setting discharge limits in NPDES permits. The revised methodology does 
not substitute for the Clean Water Act or EPA's regulations; nor is it 
a regulation itself. Thus, the revised methodology cannot impose 
legally-binding requirements on EPA, States, or the public, and may not 
apply to a particular situation based upon the circumstances. EPA and 
State decisionmakers retain the discretion to use different, 
scientifically defensible, methodologies to develop human health 
criteria. EPA may change the methodology in the future.
    This criteria methodology incorporates scientific advancements made 
over the past two decades. The use of this methodology is an important 
component of the Agency's efforts to improve the quality of the 
Nation's waters. EPA believes the methodology will enhance the overall 
scientific basis of water quality criteria. Further, the methodology 
should help States and Tribes address their unique water quality issues 
and risk management decisions, and afford them greater flexibility in 
developing their water quality programs.

    Dated: August 3, 1998.
J. Charles Fox,
Acting Assistant Administrator for Water.

Appendix I. Background

A. Water Quality Criteria and Standards

1. Water Quality Criteria and the Criteria Derivation Methodology
    EPA published the availability of ambient water quality criteria 
(AWQC) documents for 64 toxic pollutants and pollutant categories 
identified in Section 307(a) of the Clean Water Act (CWA) in the 
Federal Register on November 28, 1980 (45 FR 79318). The November 1980 
Federal Register document also summarized the criteria documents and 
discussed in detail the methods used to derive the AWQC for those 
pollutants. The AWQC for those 64 pollutants and pollutant categories 
were published pursuant to Section 304(a)(1) of the CWA:

    ``The Administrator, * * * shall develop and publish, * * *, 
(and from time to time thereafter revise) criteria for water quality 
accurately reflecting the latest scientific knowledge (A) on the 
kind and extent of all identifiable effects on health and welfare 
including, but not limited to, plankton, fish, shellfish, wildlife, 
plant life, shorelines,

[[Page 43763]]

beaches, esthetics, and recreation which may be expected from the 
presence of pollutants in any body of water, including ground water; 
(B) on the concentration and dispersal of pollutants, or their 
byproducts, through biological, physical, and chemical processes; 
and (C) on the effects of pollutants on the biological community 
diversity, productivity, and stability, including information on the 
factors affecting rates of eutrophication and rates of organic and 
inorganic sedimentation for varying types of receiving waters.''

    The AWQC published in November 1980 provided two essential types of 
information: (1) discussions of available scientific data on the 
effects of the pollutants on public health and welfare, aquatic life, 
and recreation; and (2) quantitative concentrations or qualitative 
assessments of the levels of pollutants in water which, if not 
exceeded, will generally ensure adequate water quality for a specified 
water use. Water quality criteria developed under Section 304(a) are 
based solely on data and scientific judgments on the relationship 
between pollutant concentrations and environmental and human health 
effects. The 304(a) criteria do not reflect consideration of economic 
impacts or the technological feasibility of meeting the chemical 
concentrations in ambient water. As discussed below, 304(a) criteria 
may be used as guidance by States and Tribes to establish water quality 
standards, which ultimately provide a basis for controlling discharges 
or releases of pollutants.
    The 1980 AWQC were derived using guidelines and methodologies 
developed by the Agency for calculating the impact of waterborne 
pollutants on aquatic organisms and on human health. Those guidelines 
and methodologies consisted of systematic procedures for assessing 
valid and appropriate data concerning a pollutant's acute and chronic 
adverse effects on aquatic organisms, nonhuman mammals, and humans. The 
guidelines and methodologies were fully described in Appendix B (for 
protection of aquatic life and its uses) and Appendix C (for protection 
of human health) of the November 1980 Federal Register document.
    This revised methodology addresses the development of AWQC to 
protect human health; a similar process to revise the methodology for 
deriving AWQC for the protection of aquatic life is currently underway 
at the Agency. When finalized, the Agency intends to use the revised 
AWQC human health methodology to both develop new AWQC for additional 
chemicals and to revise existing AWQC. Appendices IV-VI are summaries 
of criteria developed using the revised methodology. These AWQC were 
developed to demonstrate the different risk assessment and exposure 
approaches presented in the revised methodology. The complete criteria 
documents are available from NTIS or on EPA's Internet web site. In 
addition, EPA intends to derive AWQC for the protection of human health 
for several chemicals of high priority, including but not limited to, 
PCBs, lead, mercury, arsenic, and dioxin, within the next several 
years. EPA anticipates that the focus of 304(a) criteria development 
will be criteria for bioaccumulative chemicals and chemicals considered 
highest priority by the Agency. The Draft AWQC Methodology Revisions 
presented here are also intended to provide States and Tribes 
flexibility in setting water quality standards by providing 
scientifically valid options for developing their own water quality 
criteria that consider local conditions. States and Tribes are 
encouraged to use the methodology once it is finalized to derive their 
own AWQC. However, the revised methodology also defines the default 
factors EPA intends to use in evaluating and determining consistency of 
State water quality standards with the requirements of the CWA. The 
Agency intends to use these default factors to calculate water quality 
criteria when promulgating water quality standards for a State or Tribe 
under Section 303(c) of the Act.
2. Summary of the 1980 AWQC National Guidelines
    The 1980 AWQC National Guidelines for developing AWQC for the 
protection of human health addressed three types of endpoints: 
noncancer, cancer, and organoleptic (taste and odor) effects. Criteria 
values for protection against noncancer and cancer effects were 
estimated by using risk assessment-based procedures, including 
extrapolation from animal toxicity or human epidemiological studies. 
Basic human exposure assumptions were applied, such as: the exposed 
individual is a 70-kilogram adult male; the assumed consumption of 
freshwater and estuarine fish and shellfish is 6.5 grams per day; and 
the assumed ingestion rate of drinking water is 2 liters per day.
    When using cancer as the critical risk assessment endpoint, which 
has been assumed not to have a threshold, the AWQC were presented as a 
range of concentrations associated with specified incremental lifetime 
risk levels 1 (i.e., a range from 10-5 to 
10-7). When using noncancer effects as the endpoint, the 
AWQC reflected an assessment of a ``no-effect'' level, since noncancer 
effects generally exhibit a threshold. The risk assessment-based 
procedures used to derive the AWQC to protect human health were 
specific to whether the endpoint was cancer or noncancer. The key 
features of each procedure are described briefly in the following 
sections.
---------------------------------------------------------------------------

    \1\ Throughout this document, the term ``risk level'' regarding 
a cancer assessment endpoint specifically refers to an upper-bound 
estimate of excess lifetime cancer risk.
---------------------------------------------------------------------------

    Cancer effects. If human or animal studies on a contaminant 
indicated that it induced a statistically significant carcinogenic 
response, the 1980 AWQC National Guidelines treated the contaminant as 
a carcinogen and derived a low-dose cancer potency factor from 
available animal data using the linearized multistage model (LMS). The 
LMS, which uses a linear, nonthreshold assumption for low-dose risk, 
was used by the Agency as a science policy choice in protecting public 
health, and represents the most plausible upper limit for low-dose 
risk. The cancer potency factor, which expresses incremental, lifetime 
risk as a function of the rate of intake of the contaminant, was then 
combined with exposure assumptions to express that risk in terms of an 
ambient water concentration. In the 1980 AWQC National Guidelines, the 
Agency presented a range of contaminant concentrations corresponding to 
incremental cancer risks of 10-7 to 10-5 (that 
is, a risk of one additional case of cancer in a population of ten 
million to one additional cancer case in a population of one hundred 
thousand, respectively). The risk range was presented for information 
purposes and did not represent an Agency judgment on ``acceptable'' 
risk level. The Agency stated in 1980 that: ``for the maximum 
protection of human health from the potential carcinogenic effects due 
to exposure of Chemical X through ingestion of contaminated water and 
aquatic organisms, the ambient water concentration should be zero based 
on the nonthreshold assumption for this chemical. However, zero level 
may not be attainable at the present time. Therefore, the levels which 
may result in incremental cancer risk over the lifetime are estimated 
at 10-5, 10-6, and 10-7.''
    Noncancer effects. If the pollutant was not considered to have the 
potential for causing cancer in humans (this was later defined as a 
known, probable, or possible human carcinogen by the 1986 Guidelines 
for Cancer Risk), the 1980 AWQC National Guidelines treated the 
contaminant as a noncarcinogen, and a criterion was derived using a 
threshold

[[Page 43764]]

concentration for noncancer adverse effects. The criteria derived from 
noncancer data were based on the Acceptable Daily Intake (ADI) (now 
termed the reference dose [RfD]). ADI values were generally derived 
using no-observed- adverse-effect level (NOAEL) data from animal 
studies, although human data were used whenever available. The ADI was 
calculated by dividing the NOAEL by an uncertainty factor to account 
for uncertainties inherent in extrapolating toxicological data from 
animal studies to humans. In accordance with the National Research 
Council recommendations of 1977, safety factors (later termed 
uncertainty factors) of 10, 100, or 1,000 were used, depending on the 
quality and quantity of the data.
    Organoleptic effects. Organoleptic characteristics were also used 
in developing criteria for some contaminants to control undesirable 
taste and/or odor imparted by them to ambient water. In some cases, a 
water quality criterion based on organoleptic effects would be more 
stringent than a criterion based on toxicologic endpoints. The 1980 
AWQC National Guidelines emphasized that criteria derived for 
organoleptic endpoints are not based on toxicologic information, have 
no direct relationship to adverse human health effects and, therefore, 
do not necessarily represent approximations of acceptable risk levels 
for humans.
3. Water Quality Standards
    Under Section 303 of the CWA, States have the primary 
responsibility to establish water quality standards, defined under the 
Act as designated beneficial uses of a water segment and the water 
quality criteria necessary to support those uses. Additionally, Native 
American Tribes authorized to administer the water quality standards 
program under 40 CFR 131.8 establish water quality standards for waters 
within their jurisdictions. This statutory framework allows States and 
Tribes to work with local communities to establish appropriate 
designated uses, and adopt criteria to protect those designated uses. 
Section 303 provides for EPA review of Water Quality Standards and for 
promulgation of a superseding Federal rule in cases where State or 
Tribal standards are not consistent with the applicable requirements of 
the CWA, or in situations where the Agency determines Federal standards 
are necessary to meet the requirements of the Act. Section 303(c)(2)(B) 
specifically requires States and Tribes to adopt AWQC for toxics for 
which EPA has published criteria under Section 304(a), and for which 
the discharge or presence could reasonably be expected to interfere 
with the designated use adopted by the State or Tribe. In adopting such 
criteria, States and Tribes must establish numerical values based on 
one of the following: (1) 304(a) criteria; (2) 304(a) criteria modified 
to reflect site-specific conditions; or, (3) other scientifically 
defensible methods.
    In order to avoid confusion, it must be recognized that the Act 
uses the term ``criteria'' in two separate ways. In Section 303(c), the 
term is part of the definition of a water quality standard. That is, a 
water quality standard is composed of designated uses and the criteria 
necessary to protect those uses. Thus, States and Tribes are required 
to adopt regulations which contain legally enforceable criteria. 
However, in Section 304(a) the term criteria is used to describe the 
scientific information that EPA develops to be used as guidance in the 
State, Tribal, or Federal adoption of water quality standards pursuant 
to 303(c). Thus, two distinct purposes are served by the 
304(a)criteria. The first is as guidance to the States and Tribes in 
the development and adoption of water quality criteria which will 
protect designated uses, and the second is as the basis for 
promulgation of a superseding Federal rule when such action is 
necessary.

B. Need for Revision of the 1980 AWQC National Guidelines

l. Scientific Advances Since 1980
    Since 1980, EPA risk assessment practices have evolved 
significantly, particularly in the areas of cancer and noncancer risk 
assessments, exposure assessments, and bioaccumulation. In cancer risk 
assessment, there have been advances with respect to the use of mode of 
action information to support both the identification of carcinogens 
and the selection of procedures to characterize risk at low, 
environmentally relevant exposure levels. Related to this is the 
development of new procedures to quantify cancer risk at low doses to 
replace the current default use of the LMS model. (See discussion in 
Appendix III, Section A.) In noncancer risk assessment, the Agency is 
moving toward the use of the benchmark dose (BMD) and other dose-
response approaches in place of the traditional NOAEL approach to 
estimate a reference dose or concentration. A BMD is calculated by 
fitting a mathematical dose-response model to data using appropriate 
statistical procedures. (See discussion in Appendix III, Section B.)
    In exposure analysis, several new studies have addressed water 
consumption and fish-tissue consumption. These studies provide a more 
current and comprehensive description of national, regional, and 
special-population consumption patterns that EPA has reflected in the 
Draft AWQC Methodology Revisions presented today. In addition, more 
formalized procedures are now available to account for human exposure 
from multiple sources when setting health goals such as AWQC that 
address only one exposure source. (See discussion in Appendix III, 
Section C.)
    With respect to bioaccumulation, the Agency has moved toward the 
use of a bioaccumulation factor (BAF) to reflect the uptake of a 
contaminant from all sources (e.g., ingestion, sediment) by fish and 
shellfish, rather than just from the water column as reflected by the 
use of a bioconcentration factor (BCF) as included in the 1980 
methodology. The Agency has also developed detailed procedures and 
guidelines for estimating BAF values. (See discussion in Appendix III, 
Section D.)
2. EPA Human Health Risk Assessment Guidelines Development Since 1980
    When the 1980 AWQC methodology was developed, EPA had not yet 
developed formal cancer or noncancer risk assessment guidelines. Since 
then EPA has published several risk assessment guidelines documents. In 
1996, the Agency proposed revised guidelines for carcinogenic risk 
assessment (61 FR 17960) which when finalized will supersede the 
carcinogenic risk assessment guidelines published in 1986 (51 FR 
33992). In addition, guidelines for mutagenicity assessment were also 
published in 1986 (51 FR 34006). The Agency also issued guidelines for 
assessing the health risks to chemical mixtures in 1986 (51 FR 34014). 
With respect to noncancer risk assessment, the Agency published 
guidelines in 1988 for assessing male and female reproductive risk (53 
FR 24834) and in 1991 for assessing developmental toxicity (56 FR 
63798). The guidelines for assessing reproductive toxicity were 
subsequently updated and finalized (61 FR 56274) in 1996. In 1991, the 
Agency also developed an external review draft of revised risk 
assessment guidelines for noncancer health effects. In 1995, EPA also 
proposed guidelines for neurotoxicity risk assessment (60 FR 52032).
    In addition to these risk assessment guidelines, EPA also published 
the

[[Page 43765]]

``Exposure Factors Handbook'' in 1989, which presents commonly used 
Agency exposure assumptions and the surveys from which they are 
derived. The Exposure Factors Handbook (EPA/600/P-95/002Fa) was updated 
in 1997. In 1992 EPA published the revised Guidelines for Exposure 
Assessment (57 FR 22888), which describe general concepts of exposure 
assessment, including definitions and associated units, and provide 
guidance on planning and conducting an exposure assessment. Also, in 
the 1980s the Agency published the Total Exposure Assessment 
Methodology (TEAM), which presents a process for conducting 
comprehensive evaluation of human exposures. The Agency has recently 
developed the Relative Source Contribution Policy, which is currently 
undergoing Agency review, for assessing total human exposure to a 
contaminant and allocating the RfD among the media of concern. In 1997, 
EPA developed draft Guiding Principles for Monte Carlo analysis.
    Also, in 1986, the Agency made available to the public the 
Integrated Risk Information System (IRIS). IRIS is a data base that 
contains risk information on the cancer and noncancer effects of 
chemicals. The IRIS assessments are peer reviewed and represent EPA 
consensus positions across the Agency's program and regional offices. 
In 1995, the Agency initiated an IRIS pilot program to test 
improvements to the internal peer review and consensus processes, and 
to provide more integrated characterizations of cancer and noncancer 
health effects.
3. Differing Risk Assessment and Risk Management Approaches for AWQC 
and MCLGs
    There are some differences in the risk assessment and risk 
management approaches used by EPA's Office of Water for the derivation 
of AWQC under the authority of the CWA and MCLGs (Maximum Contaminant 
Level Goals) under the Safe Drinking Water Act (SDWA). Two notable 
differences are with respect to the treatment of chemicals designated 
as Group C possible human carcinogens under the 1986 Guidelines for 
Carcinogen Risk Assessment and the consideration of nonwater sources of 
exposure when setting an AWQC or MCLG for a noncarcinogen.
    Group C Chemicals. Chemicals have been typically classified as 
Group C--i.e., possible human carcinogens--under the existing (1986) 
EPA cancer classification scheme for any of the following reasons:
    1. Carcinogenicity has been documented in only one test species 
and/or only one cancer bioassay and the results do not meet the 
requirements of ``sufficient evidence.''
    2. Tumor response is of marginal significance due to inadequate 
design or reporting.
    3. Benign, but not malignant, tumors occur with an agent showing no 
response in a variety of short-term tests for mutagenicity.
    4. There are responses of marginal statistical significance in a 
tissue known to have a high or variable background rate.
    The 1986 Guidelines for Carcinogen Risk Assessment specifically 
recognized the need for flexibility with respect to quantifying the 
risk of Group C agents. The guidelines noted that agents judged to be 
in Group C, possible human carcinogens, may generally be regarded as 
suitable for quantitative risk assessment, but that case-by-case 
judgments may be made in this regard.
    The EPA Office of Water has historically treated Group C chemicals 
differently under the CWA and the SDWA. It is important to note that 
the 1980 AWQC National Guidelines for setting AWQC under the CWA 
predated EPA's carcinogen classification system, which was proposed in 
1984 (49 FR 46294) and finalized in 1986 (51 FR 33992). The 1980 AWQC 
National Guidelines did not explicitly differentiate among agents with 
respect to the weight-of-evidence for characterizing them as likely to 
be carcinogenic to humans. For all pollutants judged as having adequate 
data for quantifying carcinogenic risk--including those now classified 
as Group C--AWQC were derived based on data on cancer incidence. In the 
November 1980 Federal Register document, EPA emphasized that the AWQC 
for carcinogens should state that the recommended concentration for 
maximum protection of human health is zero. At the same time, the 
criteria published for specific carcinogens presented water 
concentrations for these pollutants corresponding to individual 
lifetime cancer risk levels in the range of 10-7 to 
10-5.
    In the development of national primary drinking water regulations 
under the SDWA, EPA is required to promulgate a health-based MCLG for 
each contaminant. The Agency policy has been to set the MCLG at zero 
for chemicals with strong evidence of carcinogenicity associated with 
exposure from water. For chemicals with limited evidence of 
carcinogenicity, including many Group C agents, the MCLG is usually 
obtained using an RfD based on its noncancer effects with the 
application of an additional uncertainty factor of 1 to 10 to account 
for its possible carcinogenicity. If valid noncancer data for a Group C 
agent are not available to establish an RfD but adequate data are 
available to quantify the cancer risk, then the MCLG is based upon a 
nominal lifetime excess cancer risk calculation in the range of 
10-5 to 10-6 (ranging from one case in a 
population of one hundred thousand to one case in a population of one 
million). Even in those cases where the RfD approach has been used for 
the derivation of the MCLG for a Group C agent, the drinking water 
concentrations associated with excess cancer risks in the range of 
10-5 to 10-6 were also provided for comparison.
    It should also be noted that EPA's pesticides program has applied 
both of the previously described methods for addressing Group C 
chemicals in actions taken under the Federal Insecticide, Fungicide, 
and Rodenticide Act (FIFRA) and finds both methods applicable on a 
case-by-case basis. Unlike the drinking water program, however, the 
pesticides program does not add an extra uncertainty factor to account 
for potential carcinogenicity when using the RfD approach.
    Consideration of Nonwater Sources of Exposure. The 1980 AWQC 
National Guidelines for setting AWQC recommended the use of the 
following equation to derive the criterion:
[GRAPHIC] [TIFF OMITTED] TN14AU98.000

where:

C=The criterion value
ADI=Acceptable daily intake (mg/kg-day)
DT=Non-fish dietary intake (mg/kg-day)
IN=Inhalation intake (mg/kg-day)
2=Assumed daily water intake (L/day)
0.0065=Assumed daily fish consumption (kg)
R=Bioconcentration factor (L/kg)

    As implied by this equation, the contributions from nonwater 
sources, namely air and non-fish dietary intake, were to be subtracted 
from the ADI, thus reducing the amount of the ADI ``available'' for 
water-related sources of intake. In practice, however, when calculating 
human health criteria, these other exposures were generally not 
considered because reliable data on these exposure pathways were not 
available. Consequently, the AWQC were usually derived such that 
drinking water and fish ingestion accounted for the entire ADI (now 
called RfD).
    In the drinking water program, a similar ``subtraction'' method was 
used

[[Page 43766]]

in the derivation of MCLGs proposed and promulgated in drinking water 
regulations through the mid-1980s. More recently, the drinking water 
program has consistently used a ``percentage'' method in the derivation 
of MCLGs for noncarcinogens. In this approach, the percentage of total 
exposure typically accounted for by drinking water, referred to as the 
relative source contribution (RSC), is applied to the RfD to determine 
the maximum amount of the RfD ``allocated'' to drinking water reflected 
by the MCLG value. In using this percentage procedure, the drinking 
water program also applies a ceiling level of 80 percent of the RfD and 
a floor level of 20 percent of the RfD. That is, the MCLG cannot 
account for more than 80 percent of the RfD, nor less than 20 percent 
of the RfD.
    The drinking water program usually takes a conservative public 
health approach of applying an RSC factor of 20 percent to the RfD when 
adequate exposure data do not exist, assuming that the major portion 
(80 percent) of the total exposure comes from other sources, such as 
diet.
    Cancer Risk Ranges. In addition to the different risk assessment 
approaches discussed above for deriving AWQC and MCLGs for Group C 
agents, different risk management approaches have arisen between the 
drinking water and ambient surface water programs with respect to using 
lifetime excess risk values when setting health-based criteria for 
carcinogens. As indicated previously, the surface water program has 
derived AWQC for carcinogens that generally correspond to lifetime 
excess cancer risk levels of 10-7 to 10-5. The 
drinking water program has set MCLGs for Group C agents based on a 
slightly less stringent risk range of 10-6 to 
10-5, while MCLGs for chemicals with strong evidence of 
carcinogenicity (that is, classified as Group A, known, or B probable, 
human carcinogen) are set at zero.
    It is also important to note that under the drinking water program, 
for those substances having an MCLG of zero, enforceable Maximum 
Contaminant Levels (MCLs) have generally been promulgated to correspond 
with cancer risk levels ranging from 10-6 to 
10-4. Unlike AWQC and MCLGs which are strictly health-based 
criteria, MCLs are developed with consideration given to the costs and 
technological feasibility of reducing contaminant levels in water to 
meet those standards.

C. Steps Taken Toward Evaluating and Revising the 1980 AWQC National 
Guidelines

    In order to begin developing a ``state-of-the-science'' approach to 
revising the 1980 AWQC National Guidelines, EPA prepared an issues 
paper that described the 1980 methodology, discussed areas that needed 
strengthening, and proposed revisions. This paper was then distributed 
for review and comment to experts at EPA headquarters, regional 
offices, and laboratories; other Federal Agencies, such as the Food and 
Drug Administration (FDA), the National Institute of Environmental 
Health Sciences (NIEHS), and the Centers for Disease Control and 
Prevention (CDC); State health organizations; Canadian health agencies; 
academe; and environmental, industry, and consulting organizations.
1. September 1992 National Workshop
    On September 13-16, 1992, more than 100 invited participants 
discussed the critical issues in a workshop convened in Bethesda, 
Maryland. Based on their expertise, attendees were assigned to specific 
technical work groups. The work group topics were cancer risk, 
noncancer risk, exposure, microbiology, minimum data, and 
bioaccumulation. Each work group member received a set of detailed 
questions that served to focus discussions on critical factors in the 
1980 AWQC National Guidelines. After the work group members deliberated 
separately on their specific technical areas, all workshop participants 
were given the opportunity to comment on the proceedings. After the 
workshop concluded, the chairperson for each technical work group 
prepared a written summary of that group's deliberations and 
recommendations. Each work group participant was given the opportunity 
to review and comment on the summaries; these comments were used to 
prepare an initial draft of the revised methodology.
2. Science Advisory Board Review
    After review of the initial draft of the revisions to the 
methodology by EPA, the workshop participants, and other relevant 
parties, a summary document was submitted for review and comment to the 
Science Advisory Board (SAB) in January 1993 and presented to the 
Drinking Water Committee of the SAB during its meeting on February 8-9, 
1993. The SAB presented its official comments to EPA on August 12, 
1993. The SAB comments have been highlighted and addressed in each of 
the technical areas discussed in Appendix III of this document. A 
complete copy of the document submitted to the SAB and SAB's comments 
are available in the docket supporting this Notice.
3. FSTRAC Review
    At the Federal State Toxicology and Risk Analysis Committee 
(FSTRAC) meeting on December 1-3, 1993, in Washington, D.C., several 
State representatives presented their opinions on the initial draft 
revised methodology and the SAB's comments. A summary of this meeting 
is presented in a document entitled ``Summary Report: State Comments on 
the Proposed Revision of the Methodology for Deriving National Ambient 
Water Quality Criteria for the Protection of Human Health.'' This 
document is also available for review in the docket supporting this 
Notice.
4. Water Quality Guidance for the Great Lakes System
    In March 1995, EPA published the Final Water Quality Guidance for 
the Great Lakes System (60 FR 15366). The Great Lakes Water Quality 
Guidance, developed under Section 118(c)(2) of the CWA, provides water 
quality criteria for 29 pollutants as well as methodologies, policies, 
and procedures for Great Lakes States and Tribes to establish 
consistent, long-term protection for fish and shellfish in the Great 
Lakes and their tributaries, as well as for the people and wildlife who 
consume them. In developing the methodology to derive human health 
criteria for the waters of the Great Lakes System, the Agency was 
mindful of the need for consistency with the planned changes in the 
methodology for deriving national AWQC for the protection of human 
health presented today. Throughout the following text, references are 
made to comparisons of the two methodologies, national and Great Lakes 
Water Quality Guidance, especially whenever differences occur due to 
regional exposure assumptions made for the Great Lakes System.

D. Overview of AWQC Methodology Revisions, Major Changes, and Issues

    Following is a summary of the major revisions to the 1980 AWQC 
National Guidelines:
    1. EPA's future role in developing AWQC for the protection of human 
health will include the refinement of the revised methodology, the 
development of revised criteria for chemicals of high priority and 
national importance (including, but not limited to chemicals that 
bioaccumulate, such as PCBs, TCDD-dioxin, and mercury), and the 
development or revision of AWQC for some additional priority chemicals. 
EPA does not plan to completely revise all of

[[Page 43767]]

the criteria developed in 1980 or those updated as part of either the 
1992 National Toxics Rule (NTR) or the 1997 proposed California Toxics 
Rule (CTR). Partial updates of all criteria may be plausible. (Appendix 
II discusses how the Agency is proposing to implement the methodology 
and update or revise the 304(a) criteria.)
    2. EPA encourages States and Tribes to use the revised methodology, 
once finalized, to develop or revise AWQC to appropriately reflect 
local conditions. EPA believes that AWQC inherently require several 
risk management decisions that are, in many cases, better made at the 
State, Tribal, and local level (e.g., fish consumption rates, target 
risk levels). EPA will continue to develop and update necessary 
toxicological and exposure data needed to use in the derivation of AWQC 
that may not be practical to obtain at the State, Tribal, or local 
level. EPA encourages States and Tribes to use local or regional fish 
consumption data when available.
    3. The following equations for deriving AWQC include toxicological 
and exposure assessment parameters which are derived from scientific 
analysis, science policy, and risk management decisions. For example, 
parameters such as a field-measured BAF or a point of departure from an 
animal study (in the form of a LOAEL/NOAEL/LED10) are 
scientific values which are empirically measured, whereas the decision 
to use animal effects as a surrogate for human effects involves 
judgment on the part of the EPA (and similarly, by other agencies) as 
to the best practice to follow when human data are lacking. Such a 
decision is, therefore, a matter of science policy. On the other hand, 
the choice of default fish consumption rates for protection of a 
certain percentage (in this case, 90 percent and 95 percent 
respectively) of the general population, is clearly a risk management 
decision. In many cases, the Agency has selected parameters using its 
best judgment regarding the overall protection afforded by the 
resulting AWQC when all parameters are combined. For a longer 
discussion of the differences between science, science policy, and risk 
management, please refer to Section E. Section E also provides further 
details with regard to risk characterization as related to this 
methodology, with emphasis placed on explaining the uncertainties in 
the overall risk assessment.
    The generalized equations for deriving AWQC based on noncancer 
effects are: 2
---------------------------------------------------------------------------

    \2\ The fish intake (FI) and bioaccumulation factor (BAF) 
parameters are presented here in simplified form. It is preferable 
to calculate criteria by splitting these out by trophic level since 
bioaccumulation may vary significantly from one level to another. 
This is discussed further in the bioaccumulation section and 
specific guidance is given in the Technical Support Document for 
this methodology. Also, the proposed example criteria that accompany 
these proposed revisions use trophic level breakouts for these 
parameters.
---------------------------------------------------------------------------

Noncancer Effects 3
[GRAPHIC] [TIFF OMITTED] TN14AU98.001

Nonlinear Cancer Effects
[GRAPHIC] [TIFF OMITTED] TN14AU98.002

Linear Cancer Effects
[GRAPHIC] [TIFF OMITTED] TN14AU98.003

where:

    \3\ Although appearing in this equation as a factor to be 
multiplied, the RSC can also be an amount subtracted. Refer to the 
explanation key below the equations.
---------------------------------------------------------------------------

AWQC=Ambient Water Quality Criterion (mg/L)
RfD=Reference dose for noncancer effects (mg/kg-day)
Pdp=Point of departure for nonlinear carcinogens (mg/kg-day), usually a 
LOAEL, NOAEL, or LED10
SF=Safety Factor for nonlinear carcinogens (unitless)
RSD=Risk-specific dose for linear carcinogens (mg/kg-day) (Dose 
associated with a target risk, such as 10-6)
RSC=Relative source contribution factor to account for nonwater sources 
of exposure. (Not used for linear carcinogens.) May be either a 
percentage (multiplied) or amount subtracted, depending on whether 
multiple criteria are relevant to the chemical.
BW=Human body weight (proposed default=70 kg for adults)
DI=Drinking water intake (proposed default=2 L/day for adults)
FI=Fish intake (proposed defaults=0.01780 kg/day for general adult 
population and sport anglers, and 0.08630 kg/day for subsistence 
fishers)
BAF=Bioaccumulation factor, lipid normalized (L/kg)
    4. As an alternative to expressing AWQC as a water concentration as 
provided in the above equations, AWQC may also be expressed in terms of 
a fish tissue concentration. For some substances, particularly those 
that are expected to exhibit substantial bioaccumulation, the AWQC 
derived using the above equations may have extremely low values, 
possibly below the practical limits for detecting and quantifying the 
substance in the water column. It may, therefore, be more practical and 
meaningful in these cases to focus on the concentration of those 
substances in fish tissue, since fish ingestion would be the 
predominant source of exposure for substances that bioaccumulate. Fish 
tissue criteria that correspond to an AWQC expressed as a water 
concentration obtained from one of the above equations is computed as 
(note, the BAF used should be the same one that was used to calculate 
the AWQC):

[[Page 43768]]

[GRAPHIC] [TIFF OMITTED] TN14AU98.004


    5. EPA is recommending an incidental water ingestion exposure rate 
of 0.01 L/day to account for long-term incidental recreational 
ingestion (i.e., swimming, boating, fishing) for use in those cases 
where AWQC are developed for recreational waters that are not used as 
drinking water sources.
    6. AWQC for the protection of human health are designed to minimize 
the risk of adverse effects occurring to humans from chronic (lifetime) 
exposure to substances through the ingestion of drinking water and 
consumption of fish obtained from surface waters. The Agency is not 
recommending the development of additional water quality criteria 
similar to the ``drinking water health advisories'' that focus on acute 
or short-term effects, since these are not seen routinely as having a 
meaningful role in the water quality criteria and standards program. 
However, as discussed below, there may be some instances where the 
consideration of acute or short-term toxicity and exposure in the 
derivation of AWQC is warranted.
    Although the AWQC are based on chronic health effects data (both 
cancer and noncancer effects), the criteria are intended to also be 
protective with respect to adverse effects that may reasonably be 
expected to occur as a result of elevated acute or short-term 
exposures. That is, through the use of conservative assumptions with 
respect to both toxicity and exposure parameters, the resulting AWQC 
values should provide adequate protection not only for the general 
population over a lifetime of exposure, but also for special 
subpopulations who, because of high water- or fish-intake rates, or 
because of biological sensitivities, have an increased risk of 
receiving a dose that would elicit adverse effects. The Agency 
recognizes, however, that there may be some cases where the AWQC values 
based on chronic toxicity may not provide adequate protection for a 
subpopulation at special risk from shorter-term exposures. The Agency 
encourages States, Tribes, and others employing the revised methodology 
to give consideration to such circumstances in deriving criteria to 
ensure that adequate protection is afforded to all identifiable 
subpopulations. (See Appendix III, Section C.3 for additional 
discussion of these subpopulations.)
    7. For noncarcinogens, risk managers may select an RfD range rather 
than a single RfD value, in criteria development, where a rationale for 
the range and the value selected can be provided. General guidance for 
the use of values within the RfD range is provided based on the overall 
uncertainty associated with the RfD. For example, if the IRIS RfD is 1 
mg/kg/day and the uncertainty factor (UF) is 1,000, a log-symmetrical 
order of magnitude (i.e., 10-fold) around 1 mg/kg/day could be used 
resulting in a range of 0.3 to 3 mg/kg/day. If the UF were less than 
1,000, the overall range would be reduced accordingly (i.e., \1/2\ log 
(3-fold) for UFs between 100 and 1,000, resulting in a range of 0.67 to 
1.5 mg/kg/day; and no range for UFs of 100 or less). However, EPA 
intends to select the point estimate as a default (the midpoint within 
the range) when calculating a 304(a) criteria value for the purposes of 
promulgating State or Tribal water quality standards. Furthermore, an 
RfD range should not be used when children are identified as the 
exposed population of concern.
    8. As explained in EPA's 1996 Proposed Guidelines for Carcinogen 
Risk Assessment, mode of action (MoA) information is used to determine 
the most appropriate low-dose extrapolation approach for carcinogenic 
agents. The dose-response assessment under the new guidelines is a two- 
step process. In the first step, the response data are modeled in the 
range of empirical observation. Modeling in the observed range is done 
with biologically based or appropriate curve-fitting modeling. In the 
second step, extrapolation below the range of observation is 
accomplished by biologically based modeling if there are sufficient 
data or by a default procedure (linear, nonlinear, or both). A point of 
departure for extrapolation is estimated from modeling observed data. 
The lower 95 percent confidence limit on a dose associated with 10 
percent extra risk (LED10) is proposed as a standard point 
of departure for low-dose extrapolation. If it is determined that the 
MoA understanding supports a nonlinear extrapolation, the AWQC is 
derived using the nonlinear default which is based on a margin of 
exposure (MoE) analysis for the point of departure (LED10) 
and applying a margin of safety (MoS) in the risk management. The 
linear default would be considered for those agents that are better 
supported by the assumption of linearity (e.g., direct DNA reactive 
mutagens) for their MoA. A linear approach would also be applied when 
inadequate or no information is available to explain the carcinogenic 
MoA as a science policy choice in the interest of public health. The 
linear default is a straight line extrapolation to the origin (i.e., 
zero dose, zero extra risk) from the point of departure 
(LED10) identified in the observable response range. There 
may be situations where it is appropriate to apply both the linear and 
nonlinear default procedures (e.g., for an agent that is both DNA 
reactive and active as a promoter at higher doses).
    9. For substances that are carcinogenic, particularly those for 
which the mode of action suggests nonlinearity at low doses, the Agency 
recommends that an integrated approach be taken in looking at cancer 
and noncancer effects, and if one pathway does not predominate, AWQC 
values should be determined for both carcinogenic and noncarcinogenic 
effects. The lower of the resulting values should be used for the AWQC.
    10. When deriving AWQC for noncarcinogens and nonlinear 
carcinogens, a factor must be included to account for other nonwater 
exposure sources so that the entire RfD, or [Point of Departure (Pdp) 
divided by a safety factor (SF) (Pdp)/SF)] is not allocated to drinking 
water and fish consumption alone. Guidance is provided in the revised 
methodology for determining the factor, referred to as the RSC, to be 
used for a particular chemical. The Agency is recommending the use of a 
decision tree procedure to support the determination of the appropriate 
RSC value for a given water contaminant. In the absence of data, the 
Agency intends to use 20 percent of the RfD as the default RSC in 
calculating a 304(a) criteria value for the purposes of promulgating 
State or Tribal water quality standards.
    11. For AWQC derived for linear carcinogens, the Agency recommends 
that risk levels in the range of 10-5 to 10-6 be 
used. (See RSD factor in Equation ID-3, above.) States and Tribes can 
always choose a more stringent risk level, such as 10-7. 
Care should be taken, however, in situations where the AWQC includes 
fish intake levels based on the general population to ensure that the 
risk to more highly exposed subgroups (sportfishers or subsistence 
fishers) does not exceed the 10-4 level.
    12. The default fish consumption values in the revised methodology 
are 17.80 grams/day for the general adult population, which represents 
the 90th percentile consumption rate for the entire adult population 
(and approximates the average consumption

[[Page 43769]]

rate for sport anglers, nationally); and 86.30 grams/day for 
subsistence fishers/minority anglers, which represents the 99th 
percentile consumption rate for the general population and falls within 
the range of averages for subsistence/minority anglers. Public comments 
are requested on alternatively using 39.04 grams/day, which represents 
the 95th percentile (and is also within the range of averages), and 
which of these two values (i.e., 39.04 or 86.30 grams/day) is more 
representative of fresh/estuarine fish consumption among subsistence 
fishers/minority anglers. These values are derived from the United 
States Department of Agriculture's (USDA) Continuing Survey of Food 
Intake by Individuals (CSFII) from 1989-1991. These rates replace the 
single default value of 6.5 grams/day used in the 1980 AWQC National 
Guidelines. These default values are chosen to be protective of the 
majority of the individuals in those groups. However, States and Tribes 
are urged to use a fish intake level derived from local data on fish 
consumption in place of these default values when deriving AWQC, 
ensuring that the fish intake level chosen be protective of highly 
exposed individuals in the population. Consumption rates for women of 
childbearing age and children younger than 14 are also provided to 
maximize protection in those cases where these subpopulations may be at 
greatest risk.
    13. In the revised methodology, criteria are derived using a BAF 
rather than a BCF, which was used in the 1980 AWQC National Guidelines. 
To derive the BAF, States and Tribes may use EPA's methodology or any 
method consistent with the EPA method. EPA's highest preference in 
developing BAFs are BAFs based on field-measured data from local/
regional fish.
    14. EPA is neither setting organoleptic criteria nor recommending a 
default methodology for deriving such criteria. Such criteria will 
necessitate case-by-case analysis.

E. Risk Characterization Considerations

1. Background
    On March 21, 1995, the EPA Administrator, Carol Browner, issued the 
EPA Risk Characterization Policy and Guidance. This policy and guidance 
is intended to ensure that characterization information from each stage 
of a risk assessment is used in forming conclusions about risk and that 
this information is communicated from risk assessors to risk managers, 
and from EPA to the public. The policy also provides the basis for 
greater clarity, transparency, reasonableness, and consistency in risk 
assessments across EPA programs. The fundamental principles which form 
the basis for a risk characterization are as follows:
     Risk assessments should be transparent, in that the 
conclusions drawn from the science are identified separately from 
policy judgments, and the use of default values or methods and the use 
of assumptions in the risk assessment are clearly articulated.
     Risk characterizations should include a summary of the key 
issues and conclusions of each of the other components of the risk 
assessments, as well as describe the likelihood of harm. The summary 
should include a description of the overall strengths and limitations 
(including uncertainties) of the assessment and conclusions.
     Risk characterizations should be consistent in general 
format, but recognize the unique characteristics of each specific 
situation.
     Risk characterizations should include, at least in a 
qualitative sense, a discussion of how a specific risk and its context 
compares with similar risks. This may be accomplished by comparisons 
with other chemicals or situations on which the Agency has decided to 
act, or other situations with which the public may be familiar. The 
discussion should highlight the limitations of such comparisons.
     Risk characterization is a key component of risk 
communication, which is an interactive process involving exchange of 
information and expert opinion among individuals, groups, and 
institutions.
2. Additional Guiding Principles
     The risk characterization integrates the information from the 
hazard identification, dose-response, and exposure assessments, using a 
combination of qualitative information, quantitative information, and 
information regarding uncertainties.
     The risk characterization includes a discussion of 
uncertainty and variability.
     Well-balanced risk characterizations present conclusions and 
information regarding the strengths and limitations of the assessment 
for other risk assessors, EPA decision- makers, and the public.
3. Risk Characterization Applied to the Revised AWQC Methodology
    In developing the methodology presented today, the EPA has closely 
followed the risk characterization guiding principles listed above. As 
States and Tribes develop criteria using the revised methodology, they 
are strongly encouraged to follow EPA's risk characterization guidance. 
There are a number of areas within the methodology and criteria 
development process where risk characterization principles apply:
     Integration of cancer and noncancer assessments with exposure 
assessments, including bioaccumulation potential determinations, in 
essence, weighing the strengths and weaknesses of the risk assessment 
as a whole when developing a criterion.
     Selecting a fish consumption rate, locally derived or default 
value, within the context of a target population (e.g., sensitive 
subpopulations) as compared to the general population.
     Presenting cancer and/or noncancer risk assessment options.
     Describing the uncertainty and variability in both the hazard 
identification, the dose-response and the exposure assessment.
    Health Risks to Children.
    In recognition that children have a special vulnerability to many 
toxic substances, Administrator Carol Browner directed EPA in 1995 to 
explicitly and consistently take into account environmental health 
risks to infants and children in all risk assessments, risk 
characterizations and public health standards set for the United 
States. In April 1997, President Clinton signed Executive Order 13045 
on the protection of children from environmental health risks, which 
assigned a high priority to addressing risks to children. In May 1997, 
EPA established the Office of Children's Health Protection to ensure 
the implementation of the President's Executive Order. Circumstances 
where risks to children should be considered in the context of the AWQC 
Methodology, along with specific recommendations, are discussed in 
relevant sections throughout this proposal.
    Details on risk characterization and the guiding principles stated 
above are included in the March 21, 1995 policy statement and the 
discussion of risk characterization which accompanies the Proposed 
Guidelines for Carcinogen Risk Assessment 61 FR 17960 (April 23, 1996) 
and the Reproductive and Toxicity Risk Assessment Guidelines also of 
1996 (61 FR 56274).
4. Science, Science Policy, and Risk Management
    An important part of risk characterization, as described at the 
beginning of this Section, is to make risk assessments transparent. 
This means that conclusions drawn from the science are identified 
separately from policy judgments and risk management decisions, and 
that the use of default

[[Page 43770]]

values or methods, as well as the use of assumptions in risk 
assessments, are clearly articulated. For the purposes of this revised 
methodology, EPA will attempt to separate out scientific analysis from 
science policy and risk management decisions. This will ultimately 
allow the States and Tribes, and specifically users of this 
methodology, such as scientists, policy setters, and risk managers, to 
understand the elements of the methodology accurately and clearly, and 
to easily separate out the scientific decisions from the science policy 
and risk management decisions. This is important so that when questions 
are asked regarding the scientific merit, validity, or apparent 
stringency or leniency of AWQC, the implementer of the criteria can 
clearly explain what judgments were made to develop the criterion in 
question and to what degree these judgments were based on science, 
science policy, or risk management. To some extent this process will 
also be displayed in future AWQC documents.
    When EPA speaks of science or scientific analysis, we are referring 
to the extraction of data from either toxicological or exposure studies 
and surveys with a minimum of judgment being used to make inferences 
from the available evidence. For example, if we are describing a point 
of departure from an animal study (e.g., a lowest-observed-adverse-
effect level, or LOAEL), this is usually determined as a lowest dose 
which produces an observable adverse effect. This would constitute a 
scientific determination. Judgments applying science policy, however, 
may enter this determination. For example, several scientists may 
differ in their opinion of what is adverse, and this in turn can 
influence the selection of a LOAEL in a given study. The use of an 
animal study to predict effects in a human in the absence of human data 
is an inherent science policy decision. The selection of specific 
uncertainty factors when developing a reference dose is another example 
of science policy. In any risk assessment, a number of decision points 
occur where risk to humans can only be inferred from the available 
evidence. Both scientific judgments and policy choices may be involved 
in selecting from among several possible inferential bridges when 
conducting a risk assessment.
    Risk management is the process of weighing policy alternatives and 
selecting the most appropriate regulatory action, integrating the 
results of risk assessment with engineering data and with social, 
economic, and political concerns to reach a decision. In this 
methodology, the choice of a default fish consumption rate which is 
protective of 90 percent of the general population is a risk management 
decision. The choice of an acceptable cancer risk by a State or Tribe 
is a risk management decision.
    Many of the parameters in the revised methodology are an amalgam of 
science, science policy, and/or risk management. For example, most of 
the defaults chosen by EPA are based on the examination of scientific 
data and the application of either science policy or risk management. 
This includes the default assumptions of 2 liters a day of drinking 
water; the assumption of 70 kilograms for an adult body weight; the use 
of default percent lipid and particulate organic carbon/dissolved 
organic carbon (POC/DOC) for developing national BAFs; the default fish 
consumption rates for the general population and sport and subsistence 
anglers; the choice of a default cancer risk level. Some decisions are 
more heavily steeped in science and science policy, such as the choice 
of default BAFs, and others are more obviously risk management 
decisions, such as the determination of default fish consumption rates 
and cancer risk levels. Throughout the revised methodology, EPA has 
identified just what kind of decision was necessary to develop defaults 
and what the basis for the decision was. More details on the concepts 
of science analysis, science policy, risk management and how they are 
introduced into risk assessments are included in Risk Assessment in the 
Federal Government: Managing the Process, National Academy Press. 1983.
5. Discussion of Uncertainty
    (a) Observed Range of Toxicity Versus Range of Environmental 
Exposure. When characterizing a risk assessment, an important 
distinction to make is between the observed range of adverse effects 
(from an epidemiology or animal study) and the environmentally observed 
range of exposure (or anticipated human exposure) to the contaminant. 
In many cases, EPA intends to apply a number of default factors to 
account for uncertainties or incomplete knowledge in developing RfDs or 
nonlinear cancer risk assessments to provide a margin of protection. In 
reality, the actual effect level and the environmental exposure levels 
may be separated by several orders of magnitude. The difference between 
some observed response and the anticipated human exposure should be 
described by risk assessors and managers, especially when comparing 
criteria to environmental levels of a contaminant.
    (b) Continuum of Preferred Data/Use of Defaults. In both 
toxicological and exposure assessments, EPA has defined a continuum of 
preferred data ranging from a highest preference of chronic human data 
for toxicological assessments (e.g., studies that examine a long-term 
exposure of humans to a chemical, usually from occupational and/or 
residential exposure); and actual field data for many of the exposure 
decisions that need to be made (e.g., locally derived fish consumption 
rates, waterbody-specific bioaccumulation rates); to default values 
which are at the lower end of the preference continuum. EPA has 
supplied default values for all of the risk assessment parameters in 
the revised methodology; however, it is important to note that when 
default values are used, the uncertainty in the final risk assessment 
is usually higher, and the final resulting criterion may not be as 
applicable to local conditions, than is a risk assessment derived from 
human/field data. Using defaults assumes generalized conditions and may 
not capture the actual variability in the population (e.g., sensitive 
subpopulations/high-end consumers). If defaults are chosen as the basis 
for criteria, these inherent uncertainties should be communicated to 
the risk manager and the public. While this continuum is an expression 
of preference on the part of EPA, it does not imply in any way that any 
of the choices are unacceptable or scientifically indefensible.
    (c) Significant Figures. The number of significant figures in a 
numeric value is the number of certain digits plus one estimated digit. 
Digits should not be confused with decimal places. For example, 15.1, 
.0151, and .0150 all have 3 significant figures. Decimal places may 
have been used to maintain the correct number of significant figures, 
but in themselves they do not indicate significant figures (Brinker, 
1984). Since the number of significant figures must include only one 
estimated digit, the sources of input parameters (e.g., fish 
consumption and water consumption rates) should be checked to determine 
the number of significant figures associated with data they provide. 
However, the original measured values may not be available to determine 
the number of significant figures in the input parameters. In these 
situations, EPA recommends utilizing the data as presented.
    When developing criteria, EPA recommends rounding the number of 
significant figures at the end of the criterion calculation to the same 
number of significant figures in the least precise parameter. This is a 
generally accepted

[[Page 43771]]

practice which can be found described in greater detail in APHA, 1992 
and Brinker, 1984. The general rule is that for multiplication or 
division, the resulting value should not possess any more significant 
figures than is associated with the factor in the calculation with the 
least precision. When numbers are added or subtracted, the number that 
has the fewest decimal places, not necessarily the fewest significant 
figures, puts the limit on the number of places that justifiably may be 
carried in the sum or difference. Rounding off a number is the process 
of dropping one or more digits so that the value contains only those 
digits that are significant or necessary in subsequent computations 
(Brinker, 1984). The following rounding procedures are recommended: (1) 
if the digit 6, 7, 8, or 9 is dropped, increase the preceding digit by 
one unit; (2) if the digit 0, 1, 2, 3, or 4 is dropped, do not alter 
the preceding digit; and (3) if the digit 5 is dropped, round off the 
preceding digit to the nearest even number (e.g., 2.25 becomes 2.2 and 
2.35 becomes 2.4) (APHA, 1992 and Brinker, 1984).
    EPA recommends that calculations of water quality criteria be 
performed without rounding of intermediate step values. The resulting 
criterion may be rounded to a manageable number of decimal places. 
However, in no case should the number of digits presented exceed the 
number of significant figures implied in the data and calculations 
performed on them. The term ``intermediate step values'' refers to 
values of the parameters in Equations ID-1 through ID-3. The final step 
is considered the resulting AWQC. Although AWQC are, in turn, used for 
purposes of establishing WQBELs in NPDES permits, calculating TMDLs, 
and with Superfund ARARs, they are considered the final step of this 
methodology and, for the purpose of this discussion, where the rounding 
should occur.
    The determination of appropriate significant figures inevitably 
involves some judgment regarding the fact that some of the equation 
parameters are adopted default exposure values. Specifically, the 
default drinking water intake rate of 2 L/day is a value adopted to 
represent a majority of the population over the course of a lifetime. 
Although supported by drinking water consumption survey data, this 
value was adopted as a policy decision and, as such, does not have to 
be considered in determining the parameter with the least precision. 
That is, the resulting AWQC need not always be reduced to one 
significant digit. Similarly, the 70-kg adult body weight has been 
adopted Agency-wide and represents a default policy decision.
    The following example illustrates the rule described above. The 
example is for hexachlorobutadiene (HCBD), the revised criterion 
summarized in Appendix VI. The parameters that were calculated (i.e., 
not policy adopted values) include values with significant figures of 
two (the Pdp and RSC), three (the SF), and four (the FI and BAF). Based 
on the revised methodology, the final criterion should be rounded to 
two significant figures. The bold numbers in parentheses indicate the 
number of significant figures and those with asterisks also indicate 
Agency adopted policy values.
[GRAPHIC] [TIFF OMITTED] TN14AU98.005

    Example (refer to HCBD document for details on the data):
    [GRAPHIC] [TIFF OMITTED] TN14AU98.006
    
    * represents Agency adopted policy value.

    A number of the values used in the equation may result in 
intermediate step values that have more than four figures past the 
decimal place and may be carried throughout the equation. However, 
carrying more than four figures past the decimal place (equivalent to 
the most precise parameter) is unnecessary as it has no effect on the 
resulting criterion calculation.

References

APHA. American Public Health Association. 1992. Standard Methods: 
For the Examination of Water and Wastewater. 18th Edition. Prepared 
and published jointly by: American Public Health Association, 
American Water Works Association, and Water Environment Federation. 
Washington, D.C.
Brinker, R.C. 1984. Elementary Surveying. 7th Edition. Cliff 
Robichaud and Robert Greiner, Eds. Harper and Row Publishers, Inc. 
New York, NY.

Appendix II. Implementation of AWQC Methodology Revisions

    Today's Draft AWQC Methodology Revisions raise several important 
implementation issues. These include the following: (1) the 
relationship of the 304(a) criteria revisions to other EPA water 
quality standards activities; (2) the status of existing 304(a) 
criteria once any revisions to the criteria and the associated 
methodologies are finalized; (3) the role of States and Tribes in 
developing the criteria; (4) the appropriateness of EPA revising 304(a) 
criteria on the basis of a change in one, or fewer than all, 
parameters; (5) the process EPA will utilize in developing new criteria 
for additional chemicals and revising existing criteria; and (6) the 
development of a priority setting process for selecting appropriate 
304(a) criteria for revising. Each of these areas is discussed below.

A. Relationship to Other EPA Activities

    New information leads to new insights as to how a chemical induces 
a toxic effect. In response to such new information, EPA continually 
updates

[[Page 43772]]

RfDs and dose-response information in IRIS. Toxicity information and 
exposure assumptions change as additional data become available. This 
ongoing evolution effects two important and interrelated 
responsibilities of the Agency, which are carried out concurrently. 
First, from time to time EPA recalculates the 304(a) water quality 
criteria to reflect the latest data. These recalculations have been 
compiled in a series of guidance documents: the Green Book in 1968, the 
Blue Book in 1972, the Red Book in 1976, and the Gold Book in 1986. The 
second responsibility pertains to the requirements of Section 303(c).
    As part of the water quality standards triennial review process 
defined in Section 303(c)(1), the States and Tribes are responsible for 
maintaining and revising water quality standards. Section 303(c)(1) 
requires States and Tribes to review, and modify if appropriate, their 
water quality standards at least once every three years. When a State 
or Tribe fails to revise or adopt water quality standards consistent 
with the requirements of the CWA, Section 303(c)(4) authorizes EPA to 
promulgate replacement water quality standards for them. From time to 
time, EPA has undertaken such promulgations and calculated numeric 
water quality criteria for the purposes of the Act. In doing so, EPA 
utilizes the most current available scientific information, such as 
toxicity data and exposure assumptions.
    With the promulgation of Federal criteria under 303(c)(4) and the 
publication of new or revised 304(a) criteria, the criteria in an early 
Federal action may differ from the criteria in a subsequent Federal 
action. Some confusion has arisen among the public with regard to what 
EPA's current recommended 304(a) water quality criteria are for a given 
chemical at any given time.
    The most recent Federal action establishes the Agency's current 
water quality criteria. To date, the most recent Federal recalculation 
of 304(a) criteria occurred in the CTR, not withstanding the fact the 
CTR was proposed pursuant to Section 303(c)(4) of the Act. (See 
discussion below.) Again, EPA views the criteria program as constantly 
evolving. When the AWQC Methodology Revisions are final, any chemical-
specific 304(a) criteria published using the revised methodology will 
be considered the Agency's most current 304(a) criteria. EPA notes 
revisions of existing 304(a) criteria prior to the finalization of the 
revised methodology may be undertaken and are not precluded.
    As discussed in Appendix I, Section B.3., States and Tribes have 
three options when adopting water quality criteria for which EPA has 
published 304(a) criteria. They can establish numerical values based on 
304(a) criteria, 304(a) criteria modified to reflect site specific 
conditions, or other scientifically defensible methods. When States or 
Tribes revise their water quality criteria to correct deficiencies 
identified in a Federal promulgation, EPA will assess the scientific 
defensibility of the criteria in terms of the Agency's most recent 
recommended water quality criteria. Thus, there may be cases where 
applicable policies and science have evolved such that EPA would be 
evaluating the scientific defensibility of State or Tribal criteria, 
adopted using one of the three options discussed above, on the basis of 
new information. Furthermore, EPA views Federal 303(c)(4) promulgations 
as temporary corrections of deficiencies in State and Tribal water 
quality standards. The triennial review process provides States and 
Tribes with a process for addressing these deficiencies. Since CWA 
Section 303(c)(1) requires States and Tribes to review and modify their 
water quality standards at least once every three years, EPA does not 
expect or intend to assume the State and Tribal responsibility of 
periodically reviewing and revising water quality standards, including 
water quality criteria, through federal promulgations.
    EPA developed and published final Water Quality Guidance for the 
Great Lakes System (the Guidance), codified at 40 CFR part 132, in 
March 1995 (58 FR 15366). The Guidance consists of water quality 
criteria for 29 pollutants to protect aquatic life, wildlife, and human 
health, and detailed methodologies to develop criteria for additional 
pollutants, implementation procedures, and antidegradation policies and 
procedures tailored to the Great Lakes system. The Guidance was 
developed using the best available science, and reflects the unique 
nature of the Great Lakes ecosystem. Great Lakes States and Tribes are 
to use the water quality criteria, methodologies, policies and 
procedures in the Guidance to establish consistent, enforceable, long-
term protection for the waters of the Great Lakes system. Under the 
CWA, the Great Lakes States are to adopt provisions into their water 
quality standards and National Pollutant Discharge Elimination System 
(NPDES) permit programs by March 1997 that are consistent with the 
Guidance. The Guidance promotes consistency in standards and 
implementation procedures while allowing appropriate flexibility to 
States and Tribes to develop equitable strategies to control pollution 
sources and to promote pollution prevention practices. Today's Draft 
AWQC Methodology Revisions are being undertaken pursuant to Section 304 
of the CWA, is independent of, and does not supersede, the Guidance.
    Although consistency in State water quality standards programs is 
an important goal for EPA, EPA also recognizes it is necessary to 
provide appropriate flexibility to States and Tribes, both Great Lakes 
States and non-Great Lakes States, in the development and 
implementation of place-based water quality programs. In overseeing 
States' implementation of the CWA, EPA has found that reasonable 
flexibility is not only necessary to accommodate site-specific 
conditions and unforseen circumstances, but also to enable innovations 
and improvements as new approaches and information become available. 
Recognition of a general need for flexibility is not incompatible with 
the requirements for the Great Lakes States and Tribes established at 
Section 118(c)(2). Once States and Tribes have adopted provisions 
consistent with the Guidance, EPA intends to extend to them flexibility 
in utilizing new data and information in developing and updating water 
quality criteria using the Great Lakes Water Quality Guidance 
methodologies. In the event a Great Lakes State or Tribe fails to adopt 
provisions consistent with the Guidance, EPA will promulgate provisions 
consistent with 40 CFR part 132 that will apply to waters and 
discharges within that jurisdiction.
    In the Draft AWQC Methodology Revisions, EPA is presenting the 
acceptable lifetime cancer risk for the general population in the range 
of 10-5 to 10-6 as opposed to the previous range 
of 10-5 to 10-7. The Draft AWQC Methodology also 
provides that States and Tribes should ensure the most highly exposed 
populations do not exceed a 10-4 risk level. EPA emphasizes 
selection of a risk level is a component used in the derivation of 
water quality criteria, and is thus subject to EPA review under Section 
303(c) of the CWA. These proposed revisions are consistent with current 
program office guidance and Agency regulatory actions.
    The three criteria summary documents in Appendices IV through VI 
were derived using a 10-6 risk level, which the Agency 
believes reflects an appropriate risk for the general population. This 
risk level is already used by many States and Tribes. EPA

[[Page 43773]]

intends to continue to derive 304(a) criteria at the 10-6 
risk level, applying a risk management policy which ensures protection 
for all exposed population groups. EPA acknowledges that at any given 
risk level for the general population, those segments of the population 
that are more highly exposed face a higher relative risk. For example, 
if fish are contaminated at a level permitted by criteria derived on 
the basis of a risk level of 10-6, individuals consuming up 
to 10 times the assumed fish consumption rate would still be protected 
at a 10-5 risk level. States and Tribes have the flexibility 
to adopt water quality criteria that result in a higher risk level 
(e.g., 10-5). EPA expects to approve such criteria if the 
State or Tribe has identified the most highly exposed subpopulation 
within the State or Tribe, demonstrates the chosen risk level is 
adequately protective of the most highly exposed subpopulation and has 
completed all necessary public participation. EPA notes that concerns 
regarding highly exposed subpopulations make it unlikely EPA would 
approve a State-wide 10-4 risk level, unless it was 
demonstrated that the potentially highly exposed subpopulations are, in 
fact, not experiencing higher exposures than the general population. In 
effect, risk for such subpopulations would not exceed a 10-4 
risk level. EPA further notes that risk levels and criteria need to be 
protective of tribal rights under federal law (e.g., fishing, hunting, 
or gathering rights) that are related to water quality. Such rights may 
raise unique issues and will need to be evaluated on a case-by-case 
basis.

B. Status of Existing 304(a) Criteria for Priority Pollutants and 
Methodology

    In November 1980, EPA published criteria development guidelines for 
the protection of human health, along with criteria for 64 toxic 
pollutants and pollutant classes (45 FR 79318). The total number of 
human health criteria published in 1980 was 105. Subsequently, three 
volatile chemicals (dichlorodifluoromethane, trichlorofluoromethane, 
and bis-(chloromethyl)-ether) were removed from the priority list. In 
1984, the criteria for dioxin were published; this resulted in a total 
of 103 criteria. In 1986, EPA summarized the available criteria 
information in Quality Criteria for Water 1986 (1986 ``Gold Book''). 
The 103 human health criteria for the protection of human health were 
included in the proposed NTR in November 1991 (56 FR 58420). At that 
time, 83 of the 103 criteria were revised to reflect the contemporary 
IRIS values. The final NTR (codified at 40 CFR 131.36(b)(1)) included 
91 human health 304(a) criteria. Nine previously published criteria 
were not included in the NTR for the purposes of promulgating federal 
water quality under 303(c), but remain in effect as published 304(a) 
criteria. Previously published criteria for seven pollutants were 
withdrawn in the NTR. The NTR directed permit authorities to 
specifically address five other pollutants in NPDES permit actions 
using the States' existing narrative ``free from toxicity'' criteria. 
In August, 1997, EPA included revised human health criteria for 22 
pollutants in the CTR (62 FR 42160). These 22 criteria, plus the 
previously published 78 criteria, are the Agency's recommended human 
health criteria. As such, they will continue to be used as the basis 
for Agency decisions, both regulatory and nonregulatory, until EPA 
revises and reissues chemical-specific criteria. For example, EPA 
intends to use these criteria: (1) as guidance to States and Tribes for 
use in establishing water quality standards; (2) as the basis for EPA 
promulgation of water quality standards; (3) in establishing NPDES 
water quality-based permit limits, where the criteria have been adopted 
by a State or Tribe or promulgated by EPA; and (4) for all other 
purposes of Section 304(a) criteria under the Act. It is important to 
emphasize again two distinct purposes which are served by the 
304(a)criteria. The first is as guidance to the States and Tribes in 
the development and adoption of water quality criteria which will 
protect designated uses, and the second is as the basis for 
promulgation of a superseding Federal rule when such action is 
necessary.
    As stated above, until such time as EPA re-evaluates a chemical, 
subjects the criteria to appropriate peer review, and subsequently 
publishes a revised chemical-specific 304(a) criteria, the existing 
304(a) criteria remain in effect. While the Draft AWQC Methodology 
Revisions represent improvements to the 1980 methodology, EPA believes 
the 1980 human health 304(a) criteria methodology and the resulting 
criteria are fundamentally sound from a scientific standpoint. In the 
Draft AWQC Methodology Revisions, EPA is presenting for public review 
and comment the latest advancements in risk and exposure assessment and 
the application of the most recent data available. In this manner, the 
Agency will continue to strengthen the scientific and technical 
foundations of the Agency's human health 304(a) criteria and provide an 
incremental improvement in the level of protection afforded to the 
public.
    EPA has long supported this position. For example, while 
undertaking reassessments of dioxin, PCBs, and other chemicals, EPA has 
consistently upheld the use of the current 304(a) criteria for these 
chemicals and has maintained their scientific acceptability on the 
grounds that until such time as a reassessment is completed, the 
existing 304(a) criteria represent EPA's best assessment for that 
particular chemical.

C. State and Tribal Criteria Development

    In keeping with their primary responsibility in establishing water 
quality standards, EPA encourages States and Tribes to develop and 
adopt water quality criteria which reflect local and regional 
conditions by using the options discussed above. States and Tribes will 
have access to EPA regional, laboratory, and headquarters staff when 
help is needed for interpretation of the methodology revisions, and for 
making critical risk assessment decisions. However, when establishing a 
numerical value based on 304(a) criteria modified to reflect site 
specific conditions, or on other scientifically defensible methods, EPA 
strongly cautions States and Tribes not to selectively apply data in 
order to ensure a water quality criteria which is less stringent than 
EPA's 304(a) criteria. Such an approach would inaccurately characterize 
risk in particular.
    Once revisions to the human health methodology are finalized, EPA 
intends to continue to update a limited number of 304(a) criteria per 
year, developing the toxicological and exposure data needed to conduct 
risk assessments associated with many of the toxic pollutants covered 
by the current universe of 304(a) criteria. As discussed below in 
Section D, updating the exposure factors used in deriving a criterion 
is not as time- and resource-intensive as completing the toxicological 
evaluation. EPA intends to update a limited number of 304(a) criteria 
each year over the next several years using new national default 
exposure assumptions, national default BAFs, and updated toxicological 
values (i.e., new or revised RfDs, cancer dose-response assessments). 
In establishing water quality criteria, States and Tribes are urged to 
continue to use the IRIS noncancer and cancer risk assessments, but to 
adjust the exposure assumptions (e.g., fish consumption and relative 
source contribution) to account for local and regional conditions. If a 
State- or

[[Page 43774]]

waterbody-specific exposure analysis cannot be conducted, States and 
Tribes should rely on EPA national defaults.
    Generally, EPA has sought to conduct re-evaluations of all of the 
components of each of the 304(a) criteria before revising the criteria. 
However in recent years, in recognition of both time and resource 
limitations, EPA has revised existing 304(a) criteria on the basis of a 
limited number of components for which there are new data or improved 
science is a reasonable and efficient means to: (1) implement the 
latest advances in scientific information and Agency policy for 
exposure analysis; and (2) publish revised 304(a) criteria on a more 
frequent basis. This approach promotes up-to-date and robust 304(a) 
criteria.
    Once new or revised 304(a) criteria are published by EPA, the 
Agency expects States and Tribes to adopt new or revised water quality 
criteria into their water quality standards consistent with the three 
options discussed above. EPA believes State and Tribal adoption of up-
to-date water quality criteria for all pollutants for which EPA has 
published 304(a) criteria is important for ensuring full and complete 
protection of human health. EPA emphasizes it will be reviewing State 
and Tribal water quality standards to assess the need for new or 
revised water quality criteria. EPA believes five years from the date 
of publication of new or revised 304(a) criteria is a reasonable time 
frame by which States and Tribes should take action. This period is 
intended to accommodate those States and Tribes which have begun a 
triennial review and wish to complete the actions they have underway, 
deferring initiating adoption of new or revised water quality criteria 
until the next triennial review.

D. Process for Developing New or Revised 304(a) Criteria

    Section 304(a)(1) directs the Agency to ``develop and publish * * * 
and from time to time * * * revise criteria for water quality 
accurately reflecting the latest scientific knowledge.'' Recent changes 
in Agency policies and procedures, as well as potential future changes, 
have implications for 304(a) criteria. These include IRIS updates, the 
proposed revisions to the cancer risk assessment guidelines, and 
revisions to the human health criteria methodology such as those in 
today's document. Additionally, when supported by additional scientific 
information, EPA has approved site-specific and chemical-specific 
decisions which differ from the 304(a) criteria published in the Gold 
Book. This situation, as well as the need for Federal promulgations of 
water quality standards under Section 303(c)(4) discussed above, has 
led to confusion among States, Tribes, and the public as to the process 
for developing 304(a) criteria.
    Several steps need to occur before a new 304(a) criterion for a 
chemical is developed or an existing 304(a) criterion is revised. 
First, new data must be evaluated by appropriate EPA Offices, 
calculations of a new criterion or any revisions to existing criteria 
must be completed, and any implications to other EPA programs must be 
determined. EPA estimates the time to conduct risk assessment ranges 
from a few months to a year or more. For exposure analyses, EPA 
estimates the time to be much shorter, ranging from a few weeks to a 
few months. EPA's experience is that toxicological evaluations take 
longer to complete than exposure assessments due the degree and 
complexity of the analysis. EPA will utilize new, relevant data in 
calculating a revised criterion value without regard to whether the 
revised criterion is more or less stringent. As noted above, EPA may 
revise 304(a) criteria on the basis of one or more components (e.g., 
BAF, fish intake, toxicity assessment), rather than a full set of 
components. This approach is in keeping with the Agency's ongoing 
efforts to strengthen the scientific and technical foundations of the 
304(a) criteria.
    Second, EPA policy is to subject derivations of new criteria or 
revisions of existing criteria to appropriate peer review. Agency peer 
review consists of a documented critical review by qualified 
individuals or organizations who are independent of those who 
originally performed the work, but who are collectively equivalent in 
technical expertise to them. Conducting peer review will help ensure 
the criteria are technically adequate, appropriately derived, properly 
documented and satisfy quality requirements. In addition, EPA will 
accept data and information from interested members of the public 
during the peer review process. Through peer review of 304(a) criteria, 
EPA will provide a sound basis for its decisions, enhancing both the 
credibility and acceptance of the 304(a) criteria.
    Finally, EPA publishes criteria and announces their availability in 
the Federal Register. While the process for developing a new 304(a) 
criterion is basically the same as for revising an existing criterion, 
the time and resources for developing the necessary data bases for new 
criteria are significantly greater. However, the criteria development 
process described above is essentially the same whether undertaken 
pursuant to 304(a) or 303(c)(4).
    In an effort to keep the States, Tribes, and public apprised of the 
most current Agency information, EPA intends to publish on a regular 
basis the current recommended 304(a) criteria, and the individual 
component values used in their derivation, for guidance to States and 
Tribes in adopting water quality standards under Section 303. 
Traditionally, EPA has published criteria documents or summaries of 
these documents (e.g., the Gold Book) as the process for incorporating 
the latest scientific knowledge and updating 304(a) criteria. Under 
this new approach, EPA expects to publish annually in the Federal 
Register a table, similar to the one EPA publishes for the drinking 
water MCLs and Health Advisories, entitled Drinking Water Regulations 
and Health Advisories (EPA 822-B-96-002). The drinking water matrix 
includes information on the existing MCLs, MCLGs, health advisories 
including the RfD, and the cancer assessment for the chemical. The AWQC 
table will contain all current recommended human health and aquatic 
life 304(a) criteria values. This table will only include water quality 
criteria of general national applicability. Water quality criteria 
derived to address a site specific or watershed situation will not be 
included. Water quality criteria from proposed or promulgated Federal 
water quality standards or new or revised 304(a) criteria documents 
will be regularly incorporated into the table. Additionally, for easier 
public access, EPA intends to maintain this repository of current EPA 
304(a) criteria and supporting information on the Internet on EPA's 
home pages on the World Wide Web (www.epa.gov).

E. Development of Future Criteria Documents

    The Agency intends to implement a streamlined approach to 
developing criteria documents which focuses on critical toxicological 
and exposure related studies. This is a departure from the past format 
in which all existing toxicological and exposure studies were presented 
in the 1980 criteria documents, with equal emphasis placed on exposure, 
pharmacokinetics, toxicological effects, and criterion formulation. Due 
to limited resources and a need to revise and update criteria more 
frequently, future criteria documents will be more abbreviated, with an 
emphasis on using current risk assessments (on IRIS or other EPA health 
assessment documents) where available and focusing to a greater

[[Page 43775]]

extent on critical exposure and toxicological studies which may 
influence the development of a 304(a) criterion (e.g., critical effects 
studies which form the basis of RfD development or cancer assessment). 
EPA will still review the literature for the latest studies, but does 
not intend to provide an exhaustive amount of information for those 
areas which are deemed less significant in the criterion development 
process. Where there is a significant amount of literature on an area 
of study (for instance, pharmacokinetics), EPA expects to reference the 
information or cite existing IRIS support documents which discuss the 
information in greater detail.
    The overall objective of this change in approach is to allow EPA to 
revise and update 304(a) criteria more frequently, while still 
maintaining the scientific rigor which EPA requires. With this new 
format, EPA estimates it can revise several criteria for the same cost 
as revising a single criterion under the old format.
    In Appendices IV through VI of today's document, EPA is publishing 
summaries of revised criteria for three chemicals using the Draft AWQC 
Methodology Revisions; the full criteria documents are available on 
EPA's Internet web site at: http://www.epa.gov/OST/Rules. The three 
chemicals for which criteria have been developed are: acrylonitrile, 
1,3-dichloropropene, and hexachlorobutadiene.
1. Acrylonitrile
    The revised criterion for protection of human health from the 
consumption of drinking water and organisms is 0.055 g/L. The 
criterion for the protection of human health from the consumption of 
organisms and incidental ingestion of water is 4.0 g/L. These 
values are based on an assumed risk level of 1 x 10-6. For 
more details on assumed parameters in this calculation, see the summary 
in Appendix IV of this document. The complete criteria document is 
available through NTIS or on EPA's Internet web site.
2. 1,3-Dichloropropene
    The revised criterion for protection of human health from the 
consumption of drinking water and organisms is 0.34 g/L. The 
criterion for the protection of human health from the consumption of 
organisms and incidental ingestion of water is 14 g/L. These 
values are based on an assumed risk level of 1 x 10-6. For 
more details on assumed parameters in this calculation, see the summary 
in Appendix V of this document. The complete criteria document is 
available through NTIS or EPA's Internet web site.
3. Hexachlorobutadiene
    The revised criteria were derived using a nonlinear (MOE) approach. 
However, both linear and nonlinear approaches are demonstrated for this 
chemical. Using the linear approach, the criterion for protection of 
human health from the consumption of drinking water and organisms is 
0.046 g/L (assumed risk level of 1 x 10-6); and the 
criterion for the protection of human health from the consumption of 
organisms and incidental ingestion of water is 0.049 g/L. 
Using the nonlinear approach, the criterion for protection of human 
health from the consumption of drinking water and organisms is 0.11 
g/L; and the criterion for the protection of human health from 
the consumption of organisms and incidental ingestion of water is 
0.12g/L. Again, EPA recommends the nonlinear approach based on 
the fact that in this specific case, there is too much uncertainty and 
not enough confidence using the tumor data (only one data point at a 
very high dose where the MTD has been exceeded and toxicity is severe) 
to do a linear high to low dose extrapolation for the estimation of 
human risk. Moreover, since data from both rats and mice support the 
same NOAEL value, there is greater confidence in the data base for a 
nonlinear approach. For more details on assumed parameters in this 
calculation, see the summary in Appendix VI of this document. The 
complete criteria document is available through NTIS or on EPA's 
Internet web site.

F. Prioritization Scheme for Selecting Chemicals for Updating

    As discussed above, the Agency does not have the resources to 
immediately develop human health criteria, either new or revised, for 
all the contaminants found in surface water. Because of this, EPA is 
soliciting comment on how to prioritize chemicals for future 
recommended 304(a) criteria using the revised human health methodology. 
One approach for prioritizing chemicals is for EPA to publish on an 
annual basis in the Federal Register a list of substances for which EPA 
plans to initiate criterion development or updating. The Federal 
Register document would provide the status of any ongoing criteria 
updates or developments of new criteria. EPA would also ask the public 
for candidates for new or updated recommended AWQC and would ask for 
scientific data (either toxicological or exposure related) or a 
compelling reason(s) to revise a current criterion or develop a new 
AWQC. This process would be similar to that used by EPA to announce its 
lists of agents for which cancer hazard and dose-response assessments 
will be initiated on an annual basis (61 FR 32799). Using the 
information submitted from the public and other data, the Agency would 
establish a list of chemicals for which it will initiate work, on an 
annual basis. EPA intends to maintain an open docket on the Internet 
which would allow the public and/or interested parties to review 
external submissions to the Agency for given chemicals and would also 
allow an exchange of pertinent information between the public and the 
Agency.
    To initiate this process for prioritization, EPA evaluated 
chemicals to generate a preliminary list of candidates for revision. 
Focusing on chemicals that pose the greatest potential risk to human 
health, the initial universe considered by EPA included the 126 
priority pollutants designated as toxic under Section 307(a) of the 
Act, plus seven additional pollutants included because of their 
bioaccumulation potential. (EPA was required to publish criteria 
documents for 65 pollutants and pollutant classes which Congress, in 
the 1977 amendments to the Clean Water Act, designated as toxic under 
Section 307(a)(1). The 65 pollutants and pollutant classes were, in 
total, 129 chemicals which became known as the list of 129 priority 
pollutants. The final number became 126 when 3 priority pollutants were 
subsequently deleted.) After careful consideration, EPA identified 98 
chemicals as possible candidates for new or revised 304(a) criteria. 
The 98 chemicals were selected based on the following factors:
     The NTR promulgated 304(a) human health criteria for 91 
chemicals. EPA considers these 91 chemicals as a good representation of 
the priority pollutants for which sufficient data exist to revise 
304(a) criteria. (The NTR did not include human health criteria for 35 
priority pollutants for the reasons discussed in the final NTR.)
     Seven chemicals for which human health criteria were not 
developed in the NTR but which have a high potential for 
bioaccumulation, based on information contained in the recently 
promulgated Great Lakes Water Quality Guidance (hexachlorocyclohexane, 
mirex, octachlorostyrene, pentachlorobenzene, photomirex, 1,2,3,4-
tetrachlorobenzene, 1,2,3,5-tetrachlorobenzene).
    In prioritizing the 98 chemicals discussed above, EPA considered 
four factors: (1) toxicity data from IRIS; (2)

[[Page 43776]]

data on occurrence in fish tissue from The Incidence and Severity of 
Sediment Contamination in Surface Waters of the United States (EPA-823-
R-97-006); (3) data on the occurrence in sediments from The Incidence 
and Severity of Sediment Contamination in Surface Waters of the United 
States; and (4) data on BAFs for trophic level 4 from either the 
proposed or final Great Lakes Water Quality Initiative Guidance (GLWQI 
or GLI). Of these four factors, EPA selected the potential for 
bioaccumulation (i.e., BAFs and Log Kow) along with toxicity 
(i.e., cancer slope factor or RfD) as the most indicative of potential 
risk to human health. Taking these two factors into consideration, EPA 
chose 29 chemicals from the list of 98 originally considered. This list 
provides the initial basis for criteria revision decisions, along with 
other Agency chemical ranking lists and input from States and Tribes. 
Furthermore, EPA intends to use these two factors for ranking 
contaminants in the future. EPA would review these priorities in light 
of Agency resources and programmatic commitments when making decisions 
to develop and/or revise 304(a) criteria in the future. New criterion 
updates and starts would be presented in an annual Federal Register 
document, as described in Section D. PCBs, mercury, and dioxin are not 
on the priority list because EPA is already committed to developing 
updated AWQC for these chemicals. The 29 highest ranked chemicals out 
of the 98 considered (not in order of priority) are the following:

Benz(a)-Anthracene
Benzo(a)-Pyrene
4-Bromo-phenyl Phenyl-Ether
4-Chloro-phenyl Phenyl Ether
Dibenzo(a,h)Anthracene
Di-n-Butyl Phthalate
Hexachloro-benzene
Hexachloro-butadiene
Aldrin
Hexachlorocyclohexane
alpha-BHC
beta-BHC
gamma-BHC
delta-BHC
Chlordane
4,4'-DDT
4,4'-DDE
4,4'-DDD
Dieldrin
Endrin
Heptachlor
Heptachlor Epoxide
Mirex/dechlorane
Octachlorostyrene
Pentachlorobenzene
Photomirex
1,2,3,4-Tetrachlorobenzene
1,2,3,5-Tetrachlorobenzene
Toxaphene

    EPA is also planning to review other prioritization efforts within 
the Agency to consider possible non-bioaccumulative contaminants found 
in surface water. Specifically, EPA will evaluate the Safe Drinking 
Water Contaminant List and risk analyses from the Office of Pesticide 
Programs.

G. Request for Comments

    EPA requests comment on all aspects of the implementation strategy 
and specifically requests comment on the following areas.
    1. Because, as a general matter, EPA uses the cancer risk range of 
10-4 to 10-6 when setting criteria and standards, 
the Agency recommends a consistent approach here (i.e., 10-5 
to 10-6 for the general population, while ensuring that the 
most highly exposed population does not exceed a risk level of 
10-4). EPA requests comment on this recommendation and its 
intention to derive 304(a) criteria at the 10-6 level. Are 
there other issues that the Agency should consider regarding this 
policy?
    2. Should EPA revise existing 304(a) criteria on the basis of a 
partially updated data set (e.g., update exposure factors to be used in 
calculating 304(a) criteria)?
    3. With what frequency should new criteria be developed or existing 
criteria updated? Is annually sufficient?
    4. Does the streamlined approach to developing criteria documents 
appropriately characterize the derivation of criteria using the 
proposed methodology? Readers are directed to the three criteria 
documents available through NTIS and EPA's Internet site as examples of 
this new approach.
    5. Is the list of 29 chemicals which EPA selected for 
prioritization appropriate? What other chemicals should be added to the 
list, and why should they be added to the list?

Appendix III. Elements of Methodology Revisions and Issues by 
Technical Area

A. Cancer Effects

1. Background on EPA Cancer Assessment Guidelines
    (a) 1980 AWQC National Guidelines. When EPA published the 1980 AWQC 
National Guideline (USEPA, 1980), formal Agency guidelines for 
assessing carcinogenic risk from exposure to chemicals had not yet been 
adopted. The methodology for assessing carcinogenic risk used by EPA in 
the 1980 AWQC National Guidelines is based primarily on the Interim 
Procedures and Guidelines for Health Risks and Economic Impact 
Assessment of Suspected Carcinogens published by EPA in 1976 (USEPA, 
1976). Although the 1980 AWQC National Guidelines recommended the use 
of both human epidemiological and animal studies to identify 
carcinogens, potential human carcinogens were primarily identified as 
those substances causing a statistically significant carcinogenic 
response in animals. It was also assumed for risk assessment purposes 
that any dose of the carcinogen results in some possibility of a tumor 
(i.e., a nonthreshold phenomenon).
    Under the 1980 guidelines, two types of data are used for 
quantitative estimates: (1) lifetime animal studies; and (2) human 
studies where excess cancer risk is associated with exposure to the 
agent. (Human data with sufficient quantification to carry out risk 
assessment are generally not available for most agents because there is 
a lack of exposure data, especially for confounders.) The scaling of 
doses from animals to humans uses a conversion factor of body weight to 
the \2/3\ power (BW2/3) to approximate the expression of 
dose in terms of surface area of the target organ (represented as a 
perfect sphere), with exposure defined in mg of contaminant/(body 
weight)2/3/day 4. This approach is based on the 
assumption that equivalent doses between animal species can be 
expressed in terms of mg/surface area/day (Mantel and Schneiderman, 
1975). This assumption is more appropriate at low applied-dose 
concentrations where sources of nonlinearity, such as saturation or 
induction of enzyme activity, are less likely to occur.
---------------------------------------------------------------------------

    \43\ The specific equation for converting an animal dose to a 
human equivalent dose using the BW2/3 scaling factor is:
    Human Equivalent Dose (mg/kg-day) = Animal Dose (mg/kg-day)  x  
Animal BW  Animal BW2/3  x  Human 
BW2/3  Human BW
    that is equivalent to
    Animal Dose Animal BW  Human BW1/3
---------------------------------------------------------------------------

    The estimation of cancer risk to humans typically used animal 
bioassay data extrapolated to low doses approximating human exposure 
using the LMS. The LMS model was fit to tumor data using a computer 
program (e.g., GLOBAL 86) that calculated the 95th percentile upper 
confidence limit on the linear slope in the low-dose range. The slope 
that is obtained is referred to as the q1*, and was used as 
an estimate of cancer potency. When animal data are used for these 
calculations, the body weights are scaled using BW2/3, as 
discussed above. The q1* values obtained using the LMS model 
and slope factors derived from other models were expressed in the form 
of x (mg/kg-day) -1 and are often used to estimate the upper 
bound of the

[[Page 43777]]

lifetime cancer risk for long-term low- level exposure to agents.
    Upper-bound risk assessments carried out with the low-dose linear 
model were generally considered conservative, representing the most 
plausible 95th percentile upper bound for risk. The ``true risk'' was 
considered unlikely to exceed the risk estimate derived by this 
procedure, and could be as low as zero at low doses. The use of low-
dose linear extrapolation with a default to LMS was endorsed by four 
agencies in the Interagency Regulatory Liaison Group and was 
characterized as less likely to underestimate risk at the low doses 
typical of environmental exposure than other models and approaches that 
were available. Because of the uncertainties associated with 
extrapolation from high to low dose and from animals to humans, assumed 
water and fish exposure, and the serious public health consequences 
that could result if risk were underestimated, EPA believed that it was 
prudent to use the LMS to estimate cancer risk for the AWQC. In 
deriving water quality criteria, the slope factors are currently 
estimated using the LMS model under most circumstances.
    Basic assumptions that are used to calculate the AWQC include a 
daily consumption rate of 2 liters of water per day (from all sources), 
a daily fish consumption rate of 6.5 grams per day, and a body weight 
of 70 kilograms (kg) (154 pounds). The maximum lifetime cancer risk 
generated by waterborne exposure to the agent is targeted in the range 
of one in one hundred thousand to one in ten million (10-5 
to 10-7). The formula for deriving the AWQC in mg/L for 
carcinogens presented in the 1980 AWQC National Guidelines is:
where:

10-6=target cancer risk level; the 1980 AWQC National 
Guidelines recommended risk levels in the range of 10-5 to 
10-7
[GRAPHIC] [TIFF OMITTED] TN14AU98.007

70=assumed body weight of an adult human being (kg)
q1*=carcinogenic potency factor for humans derived from LMS 
model (mg/kg-day)-1
2=assumed daily water consumption of an adult human (L/day)
0.0065=assumed daily consumption of fish (kg)
R=bioconcentration factor (L/kg) from water to food (e.g., fish, birds)

    (b) 1986 EPA Guidelines for Carcinogenic Risk Assessment. Since 
1980, EPA risk assessment practices have evolved significantly. In 
September 1986, EPA published its Guidelines for Carcinogen Risk 
Assessment (referred to subsequently in this document as the 1986 
Cancer Guidelines) in the Federal Register (51 FR 33992) (USEPA, 1986). 
The 1986 Cancer Guidelines were based on the publication by the Office 
of Science and Technology Policy (OSTP, 1985) that provided a summary 
of the state of knowledge in the field of carcinogenesis and a 
statement of broad scientific principles of carcinogen risk assessment 
on behalf of the Federal government. The 1986 Cancer Guidelines 
categorize chemicals into alpha-numerical groups: A (known human 
carcinogen; sufficient evidence from epidemiological studies or other 
human studies); B (probable human carcinogen; sufficient evidence in 
animals and limited or inadequate evidence in humans); C (possible 
human carcinogen; limited evidence of carcinogenicity in animals in the 
absence of human data); D (not classifiable; inadequate or no animal 
evidence of carcinogenicity); and E (no evidence of carcinogenicity in 
at least two adequate species or in both epidemiological and animal 
studies). Within Group B there are two subgroups, Groups B1 and B2. 
Group B1 is reserved for agents for which there is limited evidence of 
carcinogenicity from epidemiological studies. It is reasonable, for 
practical purposes, to regard an agent for which there is 
``sufficient'' evidence of carcinogenicity in animals as if it 
presented a carcinogenic risk to humans. Therefore, agents for which 
there is ``sufficient evidence'' from animal studies and for which 
there is ``inadequate evidence'' or ``no data'' from epidemiological 
studies would usually be categorized under Group B2 (USEPA, 1986). The 
system was similar to that used by the International Agency for 
Research on Cancer (IARC).
    The 1986 Cancer Guidelines include guidance on what constitutes 
sufficient, limited, or inadequate evidence. In epidemiological 
studies, sufficient evidence indicates a causal relationship between 
the agent and human cancer; limited evidence indicates that a causal 
relationship is credible, but that alternative explanations, such as 
chance, bias, or confounding, could not adequately be excluded; 
inadequate evidence indicates either lack of pertinent data, or a 
causal interpretation is not credible. In animal studies, sufficient 
evidence includes an increased incidence of malignant tumors or 
combined malignant and benign tumors:
    (a) In multiple species or strains;
    (b) In multiple experiments (e.g., with different routes of 
administration or using different dose levels);
    (c) To an unusual degree in a single experiment with regard to high 
incidence, unusual site or type of tumor, or early age at onset;
    (d) Additional data on dose-response; short-term tests or 
structural activity relationship.
    Limited evidence includes studies involving a single species, 
strain, or experiment which do not meet criteria for sufficient 
evidence; experiments restricted by inadequate dosage levels, 
inadequate duration of exposure, inadequate period of follow-up, poor 
survival, too few animals, or inadequate reporting; an increase in 
benign but not malignant tumors with an agent showing no response in a 
variety of short-term tests for mutagenicity; or responses of marginal 
statistical significance in a tissue known to have a high or variable 
background rate.
    In the 1986 Cancer Guidelines, hazard identification and the 
weight-of-evidence process focus on tumor findings. The human 
carcinogenic potential of agents is characterized by a six-category 
alphanumeric classification system. The weight-of-evidence approach for 
making judgment about cancer hazard analyzes human and animal tumor 
data separately, then combines them to make the overall conclusion 
about potential human carcinogenicity. The next step of the hazard 
analysis is an evaluation of supporting evidence (e.g., mutagenicity, 
cell transformation) to determine whether the overall weight-of-
evidence conclusion should be modified.
    For cancer risk quantification, the 1986 Cancer Guidelines 
recommend the use of LMS as the only default approach. The 1986 Cancer 
Guidelines also mention that a low-dose extrapolation model other than 
the LMS might be considered more appropriate based on biological 
grounds. However, no guidance was given in choosing

[[Page 43778]]

other approaches. The 1986 Cancer Guidelines continued to recommend the 
use of (BW) 2/3 as a dose scaling factor between species.
    (c) Scientific Issues Associated with the Current Cancer Risk 
Assessment Methodology for the Development of AWQC. In reviewing the 
current approach for the development of Water Quality Criteria for 
Human Health, EPA feels that the alphanumeric classification scheme for 
carcinogens adopted in 1986 was too rigid and relied too heavily on 
tumor findings and the full use of all relevant information, an 
understanding of how the agent induces tumors, and the relevance of the 
mode of action to humans was not promoted. Because guidance was not 
provided in the 1986 Cancer Guidelines for developing a mode of action 
understanding about how the agent induces tumors, dose-response 
assessments have been traditionally based on the modeling of tumor data 
with the LMS approach. There is an increasing number of examples of 
where the use of linear extrapolation may not be appropriate (e.g., 
nonmutagenic carcinogens causing a hormonal imbalance and thyroid gland 
neoplasia, or inducing bladder tumors secondary to bladder calculi-
induced hyperplasia). Additionally, the circumstances or conditions 
under which a particular hazard is expressed (e.g., route, duration, 
pattern, or magnitude of exposure) are not conveyed with the 1986 
letter classification system.
    The Office of Water has also reviewed the guidance provided by the 
1992 National Workshop on Revision of the Methods for Deriving National 
Ambient Water Quality Criteria for the Protection of Human Health 
(USEPA, 1993) and EPA's SAB review of the 1992 National Workshop report 
on cancer-related issues.5 As recommended by these two 
groups, the Office of Water is revising the cancer risk assessment 
methodology for the development of AWQC by incorporating principles 
consistent with the Proposed Guidelines for Carcinogenic Risk 
Assessment dated April 23, 1996 (USEPA, 1996).
---------------------------------------------------------------------------

    \5\ The 1992 National Workshop on Revision of the Methods for 
Deriving National Ambient Water Quality Criteria for the Protection 
of Human Health (USEPA, 1993) and EPA's Scientific Advisory Board 
(SAB) review of the workshop identified several issues on cancer. 
EPA was encouraged by both groups to incorporate new approaches into 
the AWQC methodology. Further, the SAB recommended against the 
interim adoption of the 1986 Cancer Guidelines into the AWQC 
methodology, indicating that it might create considerable confusion 
in the future, once new Cancer Guidelines are formally proposed and 
implemented.
---------------------------------------------------------------------------

2. Proposed Revisions to EPA's Carcinogen Risk Assessment Guidelines 
    EPA has recently published Proposed Guidelines for Carcinogen Risk 
Assessment (USEPA, 1996), that revise the 1986 Cancer Guidelines. These 
revisions are designed to ensure that the Agency's cancer risk 
assessment methods reflect the most current scientific 
information.6 Although many fundamental aspects of the 
current cancer risk assessment approach have been retained, there are a 
number of key changes proposed, some of which address the specific 
problems mentioned in the preceding section. Proposed changes to the 
cancer guidelines are discussed here because many of the changes that 
are proposed are incorporated into the AWQC methodology in this 
document.
---------------------------------------------------------------------------

    \6\ They are referred to hereafter as the Proposed Cancer 
Guidelines.
---------------------------------------------------------------------------

    The key changes in the Proposed Cancer Guidelines include:
    (a) Hazard assessment promotes the analysis of all biological 
information rather than just tumor findings.
    (b) An agent's mode of action in causing tumors is emphasized to 
reduce the uncertainty in describing the likelihood of harm and in 
determining the dose-response approach(es).
    (c) Increased emphasis on hazard characterization to integrate the 
data analysis of all relevant studies into a weight-of-evidence 
conclusion of hazard, to develop a working conclusion regarding the 
agent's mode of action in leading to tumor development, and to describe 
the conditions under which the hazard may be expressed (e.g., route, 
pattern, duration and magnitude of exposure).
    (d) A weight-of-evidence narrative with accompanying descriptors 
(listed in Section 3 below) replaces the current alphanumeric 
classification system. The narrative is intended for the risk manager 
and lays out a summary of the key evidence, describes the agent's mode 
of action, characterizes the conditions of hazard expression, and 
recommends appropriate dose-response approach(es). Significant 
strengths, weaknesses, and uncertainties of contributing evidence are 
highlighted. The overall conclusion as to the likelihood of human 
carcinogenicity is given by route of exposure.
    (e) Biologically based extrapolation models are the preferred 
approach for quantifying risk. It is anticipated, however that the 
necessary data for the parameters used in such models will not be 
available for most chemicals. The new guidelines allow for alternative 
quantitative methods, including several default approaches.
    (f) Dose-response assessment is a two-step process. In the first 
step, response data are modeled in the range of observation, and in the 
second step, a determination of the point of departure or range of 
extrapolation below the range of observation is made. In addition to 
modeling tumor data, the new guidelines call for the use and modeling 
of other kinds of responses if they are considered to be more informed 
measures of carcinogenic risk.
    (g) Three default approaches are provided--linear, nonlinear, or 
both. Curve fitting in the observed range would be used to determine a 
point of departure. A standard point of departure is proposed as the 
effective dose corresponding to the lower 95 percent limit on a dose 
associated with 10 percent extra risk (LED10).7 
The linear default is a straight line extrapolation from the response 
at LED10 to the origin (zero dose, zero extra risk). The 
nonlinear default begins with the identified point of departure and 
provides an MoE analysis rather than estimating the probability of 
effects at low doses. The MoE analysis is used to determine the 
appropriate margin between the Pdp and the projected exposure level 
(i.e., the AWQC). The key objective of the MoE analysis is to describe 
for the risk manager how rapidly responses may decline with dose. Other 
factors are also considered in the MoE analysis (nature of the 
response, human variation, species differences, biopersistence).
---------------------------------------------------------------------------

    \7\ Use of the LED10 as the point of departure is 
recommended with this methodology, as it is with the Proposed Cancer 
Guidelines. Public comments were requested on the use of the 
LED10, ED10, or other points. EPA is currently 
evaluating these comments and any changes in the Cancer Guidelines 
will be reflected in the Final AWQC Methodology.
---------------------------------------------------------------------------

    (h) Refining the approach used to calculate oral human equivalent 
dose when assessments are based on animal bioassays including a change 
in the default assumption for interspecies dose scaling (using body 
weight raised to the \3/4\ power).
    With recent proposals to emphasize mode of action understanding in 
risk assessment and to model response data in the observable range to 
derive points of departure or BMDs for both cancer and noncancer 
endpoints, EPA health risk assessment practices are beginning to come 
together. The modeling of observed response data to identify points of 
departure in a standard way will help to harmonize cancer and noncancer 
dose-response approaches

[[Page 43779]]

and permit comparisons of cancer and noncancer risk estimates.
    The Notice, 61 FR 17960 April 23, 1996, and its supporting 
administrative record should be consulted for detailed information 
(USEPA, 1996).
3. Revised Carcinogen Risk Assessment Methodology for Deriving AWQC 
8
---------------------------------------------------------------------------

    \8\ Additional information regarding the revised methodology may 
be found in Ambient Water Quality Criteria Derivation Methodology--
Human Health. Technical Support Document. (USEPA, 1998).
---------------------------------------------------------------------------

    The revised methodology for deriving numerical AWQC for carcinogens 
incorporates the principles consistent with the Proposed Cancer 
Guidelines. This discussion of the revised methodology for carcinogens 
focuses primarily on the quantitative aspects of deriving numerical 
AWQC values. It is important to note that the cancer risk assessment 
process outlined in the Proposed Cancer Guidelines is not limited just 
to the quantitative aspects. A numerical AWQC value derived for a 
carcinogen is to be accompanied by appropriate hazard assessment and 
risk characterization information.
    This Section contains a discussion of the weight-of-evidence 
narrative, that describes all information relevant to a cancer risk 
evaluation, followed by a discussion of the quantitative aspects of 
deriving numerical AWQC values for carcinogens. It is assumed that data 
from an appropriately conducted animal bioassay provide the underlying 
basis for deriving the AWQC value. The discussion focuses on the 
following: (1) dose estimation; (2) characterizing dose-response 
relationships in the range of observation and at low, environmentally 
relevant doses; (3) calculating the AWQC value; (4) risk 
characterization; and (5) use of toxicity equivalent factors (TEF) and 
Relative Potency Estimates. The first three listed topics encompass the 
quantitative aspects of deriving AWQC for carcinogens.
    (a) Weight-of-Evidence Narrative.9 As stated in the EPA 
Proposed Cancer Guidelines, the new method includes a weight-of-
evidence narrative that is based on an overall weight-of-evidence of 
biological and chemical/physical considerations. Hazard assessment 
information accompanying an AWQC value for a carcinogen is provided in 
the form of a weight-of-evidence narrative as described in the 
footnote. Of particular importance is that the weight-of-evidence 
narrative explicitly provides adequate support based on human studies, 
animal bioassays, and other key evidence for the conclusion that the 
substance is a ``known or likely'' human carcinogen from exposures 
through drinking water and/or fish ingestion. The Agency emphasizes the 
importance of providing an explicit discussion of the mode of action 
for the substance in the weight-of-evidence narrative, including a 
discussion that relates the mode of action to the quantitative 
procedures used in the derivation of the AWQC.
---------------------------------------------------------------------------

    \9\ The weight-of-evidence narrative is intended for the risk 
manager, and thus explains in nontechnical language the key data and 
conclusions, as well as the conditions for hazard expression. 
Conclusions about potential human carcinogenicity are presented by 
route of exposure. Contained within this narrative are simple 
likelihood descriptors that essentially distinguish whether there is 
enough evidence to make a projection about human hazard (i.e., known 
human carcinogen, likely to be a human carcinogen, or not likely to 
be a human carcinogen) or whether there is insufficient evidence to 
make a projection (i.e., the cancer potential cannot be determined 
because evidence is lacking, conflicting, inadequate, or because 
there is some evidence but it is not sufficient to make a projection 
to humans). Because one encounters a variety of data sets on agents, 
these descriptors are not meant to stand alone; rather, the context 
of the weight-of-evidence narrative is intended to provide a 
transparent explanation of the biological evidence and how the 
conclusions were derived. Moreover, these descriptors should not be 
viewed as classification categories (like the alphameric system), 
which often obscure key scientific differences among chemicals. The 
new weight-of-evidence narrative also presents conclusions about how 
the agent induces tumors and the relevance of the mode of action to 
humans, and recommends a dose-response approach based on the mode-
of-action understanding (USEPA, 1996).
---------------------------------------------------------------------------

    (b) Dose Estimation.
    (1) Determining the Human Equivalent Dose. An important objective 
in the dose-response assessment is to use a measure of internal or 
delivered dose at the target site where possible. This is particularly 
important in those cases where the carcinogenic response information is 
being extrapolated to humans from animal studies. Generally, the 
measure of dose provided in the underlying human studies and animal 
bioassays is the applied dose, typically given in terms of unit mass 
per unit body weight per unit time, (e.g., mg/kg-day). When animal 
bioassay data are used, it is necessary to make adjustments to the 
applied dose values to account for differences in pharmacokinetics 
between animals and humans that affect the relationship between applied 
dose and delivered dose at the target organ.
    In the estimation of a human equivalent dose, the Proposed Cancer 
Guidelines recommend that when adequate data are available, the doses 
used in animal studies can be adjusted to equivalent human doses using 
toxicokinetic information on the particular agent. However, in most 
cases, there are insufficient data available to compare doses between 
species. In these cases, the estimate of a human equivalent dose is 
based on science policy default assumptions. To derive an equivalent 
human oral dose from animal data, the new default procedure is to scale 
daily applied oral doses experienced for lifetime in proportion to body 
weight raised to the \3/4\ power. The adjustment factor is used because 
metabolic rates, as well as most rates of physiological processes that 
determine the disposition of dose, scale this way. Thus, the rationale 
for this factor rests on the empirical observation that rates of 
physiological processes consistently tend to maintain proportionality 
with body weight raised to \3/4\ power (USEPA, 1996).

Human Equivalent Dose=(Animal Dose)[(Animal BW)/(Human 
BW)]1/4

    The use of body weight raised to \3/4\ power (BW3/4) is 
a departure from the scaling factor of BW2/3, which was 
based on surface area adjustment and was included in the 1980 AWQC 
National Guidelines as well as the 1986 Cancer Guidelines. A more 
extensive discussion of the rationale and data supporting the Agency's 
adoption of this scaling factor is in USEPA (1992) and the Proposed 
Cancer Guidelines.
    (2) Dose Adjustments for Less-than-Lifetime Exposure Periods. In 
the 1980 AWQC National Guidelines, two other dose-related adjustments 
were discussed. The first addressed situations where the experimental 
dosing period (le) is less than the duration of the 
experiment (Le). In these cases, the average daily dose is 
adjusted downward by multiplying by the ratio (le/
Le) to obtain an equivalent average daily dose for the full 
experimental period. This adjustment would also be used in situations 
where animals are dosed fewer than 7 days per week. If, for example, 
``daily'' dosing is done only 5 days each week, the lifetime daily dose 
would be calculated as \5/7\ of the actual dose given on each of the 5 
days.
    The second dose adjustment addresses situations where the 
experimental duration (Le) is substantially less than the 
natural lifespan (L) of the test animal. For example, for mice and rats 
the natural lifespans are defined as 90 weeks and 104 weeks 
respectively. If the study duration is less than 78 weeks for mice, or 
less than 90 weeks for rats, applied doses are adjusted by dividing by 
a factor of (L/Le)\3\. (Alternatively, the cancer potency 
factor obtained from the study could be adjusted upward by multiplying 
by the factor of (L/Le)\3\.)
    This adjustment is considered necessary because a shortened 
experimental duration does not permit

[[Page 43780]]

the full expression of cancer incidence that would be expressed during 
a lifetime study. In addition, most carcinogenic responses are manifest 
in humans and animals at higher rates later in life. Age-specific rates 
of cancer increase as a constant function of the background cancer rate 
(Anderson, 1983) by the 2nd or higher power of age (Doll, 1971). In the 
adjustment recommended here, it is assumed that the cumulative tumor 
rate will increase by at least the 3rd power of age. It is important to 
note that although both dose adjustments discussed in this Section were 
included in the 1980 AWQC National Guidelines, the second adjustment 
has not been commonly used in practice.
    (3) Dose-Response Analysis. If data on the agent are sufficient to 
support the parameters of a biologically based or case-specific model 
and the purpose of the assessment is such as to justify investing 
resources supporting use, this is the first choice for both the 
observed tumor and related response data and for extrapolation below 
the range of observed data in either animal or human studies.
    (c) Characterizing Dose-Response Relationships in the Range of 
Observation. The first quantitative component in the derivation of AWQC 
for carcinogens is the dose-response assessment in the range of 
observation. For most agents, in the absence of adequate data to 
generate a biologically based model or case-specific model, dose-
response relationships in the observed range can be addressed through 
curve-fitting procedures for response data. It should be noted that the 
1996 proposed guidelines call for modeling of not only tumor data in 
the observable range, but also other responses thought to be important 
events proceeding tumor development (e.g., DNA adducts, cellular 
proliferation, receptor binding, hormonal changes). The modeling of 
these data are intended to better inform the dose-response assessment 
by providing insights into the relationships of exposure (or dose) and 
tumor response below the observable range. These nontumor response data 
can only play a role in the dose-response assessment if the agent's 
carcinogenic mode of action is reasonably understood, as well as, the 
role of that precursor event.
    The Proposed Cancer Guidelines recommend calculating the lower 95 
percent confidence limit on a dose associated with an estimated 10 
percent increased tumor or relevant nontumor response 
(LED10) for quantitative modeling of dose-response 
relationships in the observed range. The estimate of the 
LED10 is used as the point of departure for low-dose 
extrapolations discussed below. The LED10, the lower 95 
percent confidence limit on a dose associated with 10 percent extra 
risk, a standard point of departure, is adopted as a matter of science 
policy to remain as consistent and comparable from case to case as 
possible. It is also a convenient comparison point for noncancer 
endpoints. The rationale supporting use of the LED10 is that 
a 10 percent response is at or just below the limit of sensitivity of 
discerning a significant difference in most long-term rodent studies. 
The lower confidence limit on dose is used to appropriately account for 
experimental uncertainty (Barnes et al., 1995); it does not provide 
information about human variability. The estimate of the 
LED10 involves considerable judgment in dealing with 
uncertainties related to such factors as selection of approach, number 
and spacing of doses, sample sizes, the precision and accuracy of dose 
measurements, and the accuracy of pathological findings.
    For some data sets, a choice of the point of departure other than 
the LED10 may be appropriate. The objective is to determine 
the lowest reliable part of the dose-response curve for the beginning 
of the second step of the dose-response assessment--determine the 
extrapolation range. Therefore, if the observed response is below the 
LED10, then a lower point may be a better choice (e.g., 
LED5). Moreover, some forms of data may not be amenable to 
curve-fitting estimation, but to estimation of a LOAEL or NOAEL 
instead, e.g., certain continuous data.
    Analysis of human studies in the observed range is designed on a 
case-by-case basis depending on the type of study and how dose and 
response are measured in the study.
    (1) Extrapolation to Low, Environmentally Relevant Doses. In most 
cases, the derivation of an AWQC will require an evaluation of 
carcinogenic risk at environmental exposure levels substantially lower 
than those used in the underlying bioassay. Various approaches are used 
to extrapolate risk outside the range of observed experimental data. In 
the Proposed Cancer Guidelines, the choice of extrapolation method is 
largely dependent on the mode of action. The Proposed Guidelines also 
indicate that the choice of extrapolation procedure follows the 
conclusions developed in the hazard assessment about the agent's 
carcinogenic mode of action, and it is this mode of action 
understanding that guides the selection of the most appropriate dose-
response extrapolation procedure. It should be noted that the term 
``mode of action'' is deliberately chosen in the new guidelines in lieu 
of the term ``mechanism'' to indicate using knowledge that is 
sufficient to draw a reasonable working conclusion without having to 
know the processes in detail as the term mechanism might imply. The 
proposed guidelines preferred the choice of a biologically based model, 
if the parameters of such models can be calculated from data sources 
independent of tumor data. It is anticipated that the necessary data 
for such parameters will not be available for most chemicals. Thus, the 
new guidelines allow for several default extrapolation approaches (low-
dose linear, nonlinear, or both).
    (2) Biologically Based Modeling Approaches. If a biologically based 
or case-specific modeling approach has been used to characterize the 
dose-response relationships in the observed range, and the confidence 
in the model is high, it may be used to extrapolate the dose-response 
relationship to environmentally relevant doses. For the purposes of 
risk management derivation of AWQC, the environmentally relevant dose 
would be the RSD associated with incremental lifetime cancer risks in 
the 10-4 to 10-6 range for carcinogens on which a 
linear extrapolation approach is applied.10 The use of the 
RSD and the Pdp/SF to compute the AWQC is presented in Appendix II, 
Section A.3(d), below. Although biologically based and case-specific 
approaches are appropriate both for characterizing observed dose-
response relationships and extrapolating to environmentally relevant 
doses, it is not expected that adequate data will be available to 
support the use of such approaches for most substances. In the absence 
of such data, the default linear approach, the nonlinear (margin of 
exposure) approach, or both linear and nonlinear approaches will be 
used.
---------------------------------------------------------------------------

    \10\ For discussion of the cancer risk range, see Appendix II, 
Section A and Appendix III, Section C.1(a).
---------------------------------------------------------------------------

    (3) Default Linear Extrapolation Approach. The default linear 
approach proposed here is a replacement of the LMS approach that has 
served as the default approach for EPA cancer risk assessments. This 
new approach is used in the derivation of AWQC for (1) agents with a 
mode of action of gene mutation due to DNA reactivity; (2) agents with 
evidence that supports a mode of action other than DNA reactivity that 
are better supported by the assumption of low-dose linearity; and (3) 
carcinogenic agents lacking information on the mode

[[Page 43781]]

of action. The proposed default linear approach is considered generally 
conservative regarding the protection of public health. Evidence of 
effects on cell growth control via direct interaction with DNA 
constitutes an expectation of a linear dose-response relationship in 
the low dose range, unless there is other information to the contrary.
    The procedures for implementing the default linear approach begin 
with the estimation of a point of departure as described above. The 
point of departure, LED10, reflects the interspecies 
conversion to the human equivalent dose and the other adjustments for 
less-than-lifetime experimental duration. In most cases, the 
extrapolation for estimating response rates at low, environmentally 
relevant exposures is accomplished by drawing a straight line between 
the response at the point of departure and the origin (i.e., zero dose, 
zero extra risk). This is mathematically represented as:
[GRAPHIC] [TIFF OMITTED] TN14AU98.008

Where:

[GRAPHIC] [TIFF OMITTED] TN14AU98.009

    The slope of the line, ``m'' (the estimated cancer potency factor 
at low doses), is computed as:
[GRAPHIC] [TIFF OMITTED] TN14AU98.010

    The RSD is then calculated for a specific incremental targeted 
lifetime cancer risk (in the range of 10-4 to 
10-6) as:
[GRAPHIC] [TIFF OMITTED] TN14AU98.011

Where:

RSD=Risk-specific dose (mg/kg-day)
Target Incremental Cancer Risk 11=Value in the range of 
10-4 to 10-6
---------------------------------------------------------------------------

    \11\ In 1980, the target lifetime cancer risk range was set at 
10-7 to 10-5. However, both the expert panel 
for the AWQC workshop (1992) and SAB recommended that EPA change the 
risk range to 10-6 to 10-4, to be consistent 
with drinking water. See Appendix I, Section D for more details.
---------------------------------------------------------------------------

m=Cancer potency factor (mg/kg-day)-1

    The use of the RSD to compute the AWQC is described in Section D 
below.
    (4) Default Nonlinear Approach. As discussed in the Proposed Cancer 
Guidelines, the use of a nonlinear approach for risk assessment is 
recommended where there is no evidence for linearity and there is 
sufficient evidence to support an assumption of nonlinearity.
    The nonlinear approach is indicated for agents having a mode of 
action that may lead to a dose-response relationship that is nonlinear, 
with response falling much more quickly than linearly with dose, or 
being most influenced by individual differences in sensitivity. The 
mode of action may theoretically be nonlinear because of a threshold 
(e.g., the carcinogenic response may be a secondary effect of toxicity 
or of an induced physiological change that is itself a threshold 
phenomenon).
    Mode of action data are used for all cases. The nonlinear approach 
may be used, for instance, in the case of an organophosphate, where the 
chemical is not mutagenic and causes only stone formation in male rat 
bladders at high doses. This dynamic leads to tumor formation only (at 
the high doses). Stone and subsequent tumor formation are not expected 
to occur at doses lower than those that induce the physiological 
changes that lead to stone formation. (More detail on this chemical is 
provided in the cancer section of the Technical Support Document). EPA 
does not generally try to distinguish between modes of action that 
might imply a ``true threshold'' from others with a nonlinear dose-
response relationship, because there is usually not sufficient 
information to distinguish between these empirically.
    The nonlinear margin of exposure (MoE) approach in the Proposed 
Cancer Guidelines compares an observed response rate such as the 
LED10, NOAEL, or LOAEL with actual environmental exposures 
of interest by computing the ratio between the two. In the context of 
deriving AWQC, the environmentally relevant exposures are targets 
rather than actual exposures.
    If the evidence for an agent indicates a nonlinearity (e.g., when 
carcinogenicity is secondary to another toxicity for which there is a 
threshold), the MoE analysis for the toxicity is similar to what is 
done for a noncancer endpoint, and an RfD or RfC for that toxicity may 
also be estimated and considered in the cancer assessment. However, a 
threshold of carcinogenic response is not necessarily assumed. It 
should be noted that for cancer assessment, the margin of exposure 
analysis begins from a point of departure that is adjusted for 
toxicokinetic differences between species to give a human equivalent 
dose.
    To support the use of the MoE approach, information is provided in 
the risk assessment about the current understanding of the phenomena 
that may be occurring as dose (exposure) decreases substantially below 
the observed data. This provides information about the risk reduction 
that is expected to accompany a lowering of exposure. Information 
regarding the various factors that influence the selection of the SF in 
an MoE approach are included in the discussion.
    There are two main steps in the MoE approach. The first step is the 
selection of a point of departure (Pdp). The Pdp may be the 
LED10 for tumor incidence, or in some cases, it may also be 
appropriate to use a NOAEL or LOAEL value from a response that is a 
precursor to tumors. When animal data are used, the Pdp is a human 
equivalent dose or concentration arrived at by interspecies dose 
adjustment (as discussed previously in this Notice) or toxicokinetic 
analysis.
    The second step in using MoE analysis to establish AWQC is the 
selection of an appropriate margin or SF to apply to the Pdp. This is 
supported by analyses in the MoE discussion in the risk assessment. The 
following issues should be considered when establishing the overall SF 
for the derivation of AWQC using the MoE approach (others may be found 
appropriate in specific cases):
     The slope of the observed dose-response relationship at 
the point of departure and its uncertainties and implications for risk 
reduction associated with exposure reduction.
    (A steeper slope implies a greater reduction in risk as exposure 
decreases. This may support a smaller margin);
     Variation in sensitivity to the phenomenon involved, among 
members of the human population;
     Variation in sensitivity between humans and the animal study 
population;
     The nature of the response used for the dose-response 
assessment, for instance, a precursor effect, or tumor response. The 
latter may support a greater margin of exposure; and

[[Page 43782]]

     Persistence of the agent in the body. This is particularly 
relevant when precursor data from less-than-lifetime studies are the 
response data being assessed.
    As a default assumption for two of these points, the Proposed 
Cancer Guidelines recommend a factor of no less than 10-fold each be 
employed to account for human variability and for interspecies 
differences in sensitivity when humans may be more sensitive than 
animals. When data indicate that humans are less sensitive than 
animals, a default factor of no smaller than \1/10\ fraction may be 
employed to account for this. If information about human variability or 
interspecies differences is available, it is used.
    After considering all the issues together, the risk manager decides 
on the margin of safety (MoS). The size of the MoS is a matter of 
policy and is selected on a case-by-case basis, considering the weight-
of-evidence and the margin of exposure analysis provided in the risk 
assessment.12
    (5) Both Linear and Nonlinear Approaches. In some cases both linear 
and nonlinear procedures may be used. When data indicate that there may 
be more than one operant mode of action for cancer induction at 
different tumor sites, an appropriate procedure is used for each site 
(USEPA, 1996). The use of both the default linear approach and the 
nonlinear approach may be appropriate to discuss implications of 
complex dose-response relationships. For example, if it is apparent 
that an agent is both DNA reactive and is highly active as a promoter 
at high doses, and there are insufficient data for modeling, both 
linear and nonlinear default procedures may be needed to decouple and 
consider the contribution of both phenomena (USEPA, 1996). For further 
discussion on making risk assessment decisions between these 
approaches, refer to the Proposed Cancer Guidelines (USEPA, 1996).
    (d) AWQC Calculation.

Linear Approach

    The following equation is used for the calculation of the AWQC for 
carcinogens where an RSD is obtained from the default linear approach:
[GRAPHIC] [TIFF OMITTED] TN14AU98.012

Nonlinear Approach

    In those cases where the nonlinear, MoE approach is used, a similar 
equation is used to calculate the AWQC 13
[GRAPHIC] [TIFF OMITTED] TN14AU98.013

Where:
AWQC=Ambient water quality criterion (mg/L)
RSD=Risk-specific dose (mg/kg-day)
Pdp=Point of departure (mg/kg-day)
SF=Safety factor (unitless)
BW=Human body weight (kg)
DI=Drinking water intake (L/day)
FI=Fish intake (kg/day)
BAF=Bioaccumulation factor (L/kg)
RSC=Relative source contribution (percentage or subtraction)

    A difference between the AWQC values obtained using the linear and 
nonlinear approaches should be noted. First, the AWQC value obtained 
using the default linear approach corresponds to a specific estimated 
incremental lifetime cancer risk level in the range of 10-4 
to 10-6. In contrast, the AWQC obtained using the nonlinear 
approach does not describe a specific cancer risk.
    The AWQC calculations shown above are appropriate for waterbodies 
that are used as sources of drinking water. If the waterbodies are not 
used as drinking water sources, the approach is modified. The drinking 
water value (DI in the equations above) is substituted with an 
incidental ingestion value (II) of 0.01 L/day. The incidental intake is 
assumed to occur from swimming and other activities. The fish intake 
value is assumed to remain the same.
    The actual AWQC chosen for the protection of human health is based 
on a review of all relevant information, including cancer and noncancer 
data. The AWQC may, or may not, utilize the value obtained from the 
cancer analysis in the final AWQC value. The endpoint selected for the 
AWQC will be based on consideration of the weight-of-evidence and a 
complete analysis of all toxicity endpoints.
    (e) Risk Characterization. Risk assessment is an integrative 
process that culminates ultimately into a risk characterization 
summary. Risk characterization is the final step of the risk assessment 
process in which all preceding analyses (i.e., hazard, dose-response, 
and exposure assessments) are tied together to convey the overall 
conclusions about potential human risk. This component of the risk 
assessment process characterizes the data in nontechnical terms, 
explaining the extent and weight-of-evidence, major points of 
interpretation and rationale, strengths and weaknesses of the evidence, 
and discusses alternative approaches, conclusions, and uncertainties 
that deserve serious consideration.
---------------------------------------------------------------------------

    \12\ Guidance on selecting appropriate safety factors is 
provided in the Proposed Guidelines for Carcinogenic Risk Assessment 
(USEPA, 1996).
---------------------------------------------------------------------------

    Risk characterization information is included with the numerical 
AWQC value and addresses the major strengths and weaknesses of the 
assessment arising from the availability of data and the current limits 
of understanding of the process of cancer causation. Key issues 
relating to the confidence in the hazard assessment and the dose-
response analysis (including the low-dose extrapolation procedure used) 
are discussed. Whenever more than one interpretation of the weight-of-
evidence for carcinogenicity or the dose-response characterization can 
be supported, and when choosing among them is difficult, the 
alternative views are provided along with the rationale for the 
interpretation chosen in the derivation of the AWQC value. Where 
possible, quantitative uncertainty analyses of the data are provided; 
at a minimum, a qualitative discussion of the important uncertainties 
is presented.
---------------------------------------------------------------------------

    \13\ Although appearing in this equation as a factor to be 
multiplied, the RSC can also be an amount subtracted.
---------------------------------------------------------------------------

    (f) Use of Toxicity Equivalence Factors (TEF) and Relative Potency 
Estimates. The 1996 Proposed Guidelines for Carcinogen Risk Assessment 
(USEPA, 1991; 1996) state: ``A toxicity equivalence factor (TEF) 
procedure is one used to derive

[[Page 43783]]

quantitative dose-response estimates for agents that are members of a 
category or class of agents. TEFs are based on shared characteristics 
that can be used to order the class members by carcinogenic potency 
when cancer bioassay data are inadequate for this purpose. The ordering 
is by reference to the characteristics and potency of a well-studied 
member or members of the class. Other class members are indexed to the 
reference agent(s) by one or more shared characteristics to generate 
their TEFs.'' In addition, the Proposed Cancer Guidelines state that 
TEFs are generated and used for the limited purpose of assessment of 
agents or mixtures of agents in environmental media when better data 
are not available. When better data become available for an agent, its 
TEF should be replaced or revised. To date, according to the Proposed 
Cancer Guidelines, adequate data to support use of TEFs has been found 
in only one class of compounds (dioxins) (USEPA, 1989; 1996).
    The uncertainties associated with TEFs are explained when this 
approach is used. This is a default approach to be used when tumor data 
are not available for individual components in a mixture. Relative 
potency factors (RPFs) can be similarly derived and used for agents 
with carcinogenicity or other supporting data. These are conceptually 
similar to TEFs, but are less firmly based on science and do not have 
the same levels of data to support them. TEFs and relative potencies 
are used only when there is no better alternative. When they are used, 
uncertainties associated with them are discussed. As of today, there 
are only three classes of compounds for which relative potency 
approaches have been examined by EPA: dioxins, polychlorinated 
biphenyls (PCBs), and polycyclic aromatic hydrocarbons (PAHs).
4. Request for Comments
    EPA's Office of Water requests comments on the revised methodology 
in this Notice. Topics on which comment is particularly sought are 
indicated below. Comments on the Proposed Cancer Guidelines are not 
solicited here; the comment period on the Proposed Cancer Guidelines 
ended in August 1996. EPA will reflect changes in the final Cancer 
Guidelines in the final Human Health methodology. Comments on the 
application of the concepts and principles of the revised AWQC 
methodology are relevant and solicited here.
    The Agency requests comment on the new approaches to dose-response 
assessment and modeling described in this Section.

References for Cancer Effects

Anderson, E.L. 1983. Quantitative Approaches in Use to Assess Cancer 
Risk. Risk Analysis. 3(4):227-295.
Barnes, D.G., G.P Daston, J.S. Evans, A.M. Jarabek, R.J. Kavlock, 
C.A. Kimmel, C. Park, and H.L. Spitzer. 1995. Benchmark Dose 
Workshop: Criteria for Use of a Benchmark Dose to Estimate a 
Reference Dose. Regul. Toxicol. Pharmacol. 21:296-306.
Doll, R. 1971. Weibull Distribution of Cancer: Implications for 
Models of Carcinogenesis. J. Roy. Stat. Soc. A. 13: 133-166.
Mantel, N. and M.A. Schneiderman. 1975. Estimating ``Safe Levels,'' 
a Hazardous Undertaking. Cancer Res. 35: 1379.
Office of Science and Technology Policy (OSTP). 1985. Chemical 
Carcinogens: Review of the Science and its Associated Principles. 
Federal Register 50: 10372-10442.
USEPA. 1976. Interim Procedures and Guidelines for Health Risks and 
Economic Impact Assessment of Suspected Carcinogens. Federal 
Register 41:21402-21405.
USEPA. 1980. Water Quality Criteria Documents. Federal Register. 
45(231): 79318-79379.
USEPA. 1986. Guidelines for Carcinogen Risk Assessment. Federal 
Register 51:33992-34003.
USEPA. 1989. Interim Procedures for Estimating Risks Associated with 
Exposures to Mixtures of Chlorinated Dibenzo-p-dioxins and -
Dibenzofurans (CDDs and CDFs) and 1989 Update. Risk Assessment 
Forum. Washington, DC. EPA/625/3-89/016.
USEPA. 1991. Workshop Report on Toxicity Equivalency Factors for 
Polychlorinated Biphenyl Congeners. Risk Assessment Forum. 
Washington, DC. EPA/625/3-91/020.
USEPA. 1992. Draft Report: A Cross-Species Scaling Factor for 
Carcinogen Risk Assessment Based on Equivalence of mg/kg 
3/4/day. Federal Register 57: 24152-24173.
USEPA. 1993. Report of the National Workshop and Preliminary 
Recommendations for Revision. Submitted to the Science Advisory 
Board. Washington, DC. January 8.
USEPA. 1996. Proposed Guidelines for Carcinogen Risk Assessment 
Federal Register April 23.
USEPA. 1998. Ambient Water Quality Criteria Derivation Methodology--
Human Health. Technical Support Document. Final Draft. EPA 822-B-98-
005. Office of Water. Washington, DC. July.

B. Noncancer Effects

1. 1980 AWQC National Guidelines for Noncancer Effects
    In the 1980 AWQC National Guidelines, the Agency evaluated 
noncancer human health effects from exposure to chemical contaminants 
using ADI levels. ADIs were calculated by dividing NOAELs by SFs to 
obtain estimates of doses of chemicals that would not be expected to 
cause adverse effects over a lifetime of exposure. In accordance with 
the National Research Council report of 1977 (NAS, 1977), EPA used SFs 
of 10, 100, or 1,000, depending on the quality and quantity of the 
overall data base. In general, a factor of 10 was suggested when good-
quality data identifying a NOAEL from human studies were available. A 
factor of 100 was suggested if no human data were available but the 
data base contained valid chronic animal data. For chemicals with no 
human data and scant animal data, a factor of 1,000 was recommended. 
Intermediate SFs could also be used for data bases that fell between 
these categories.
    AWQC were then calculated using the ADI levels together with 
standard exposure assumptions about the rates of human ingestion of 
water and fish, and also accounting for intake from other sources (see 
Equation IB-1 in the Introduction). Surface water concentrations at or 
below the calculated criteria concentrations would be expected to 
result in human exposure levels at or below the ADI. Inherent in these 
calculations is the assumption that, generally, noncarcinogens exhibit 
a threshold.
2. Noncancer Risk Assessment Developments Since 1980
    Since 1980, the risk assessment of noncarcinogenic chemicals has 
changed. To remove the value judgments implied by the words 
``acceptable'' and ``safety,'' the ADI and SF terms have been replaced 
with the terms RfD and UF/modifying factor (MF), respectively.
    For the risk assessment of general systemic toxicity, the Agency 
currently uses the guidelines contained in the IRIS Background Document 
entitled Reference Dose (RfD): Description and Use in Health Risk 
Assessments. That document defines an RfD as ``an estimate (with 
uncertainty spanning approximately an order of magnitude) of a daily 
exposure to the human population (including sensitive subgroups) that 
is likely to be without appreciable risk of deleterious effects over a 
lifetime'' (USEPA, 1993a). The most common approach for deriving the 
RfD does not involve dose-response modelling. Instead, an RfD for a 
given chemical is usually derived by first identifying the NOAEL for 
the most sensitive known toxicity endpoint, that is, the toxic effect 
that occurs at the lowest dose. This effect is called the critical 
effect. Factors such as the study methodology, the species of 
experimental animal, the nature of the toxicity endpoint assessed and 
its

[[Page 43784]]

relevance to human effects, the route of exposure, and exposure 
duration are critically evaluated in order to select the most 
appropriate NOAEL from among all available studies in the chemical's 
data base. If no appropriate NOAEL can be identified from any study, 
then the LOAEL for the critical effect endpoint is used and an 
uncertainty factor for LOAEL to NOAEL extrapolation is applied. Using 
this approach, the RfD is equal to the NOAEL (or LOAEL) divided by the 
product of uncertainty factors and, occasionally, a modifying factor:
[GRAPHIC] [TIFF OMITTED] TN14AU98.014

The definitions and guidance for use of the uncertainty factors and the 
modifying factor are provided in the IRIS Background Document and are 
repeated in Table IIIB-1.
    The IRIS Background Document on the Reference Dose (USEPA, 1993a) 
provides guidance for critically assessing noncarcinogenic effects of 
chemicals and for deriving the RfD. Another reference on this topic is 
Dourson (1994). Furthermore, the Agency has also published separate 
guidelines for assessing specific toxic endpoints, such as 
developmental toxicity (USEPA, 1991a); reproductive toxicity (USEPA, 
1996a); and neurotoxicity risk assessment (USEPA, 1995a). These 
endpoint-specific guidelines will be used for their respective areas in 
the hazard assessment step and will complement the overall 
toxicological assessment. It should be noted, however, that an RfD, 
derived using the most sensitive known endpoint, is considered 
protective against all noncarcinogenic effects.

                           Table IIIB-1.--Uncertainty Factors and the Modifying Factor                          
----------------------------------------------------------------------------------------------------------------
          Uncertainty Factor                                           Definition                               
----------------------------------------------------------------------------------------------------------------
UFH...................................  Use a 1, 3, or 10-fold factor when extrapolating from valid data in     
                                         studies using long-term exposure to average healthy humans. This factor
                                         is intended to account for the variation in sensitivity (intraspecies  
                                         variation) among the members of the human population.                  
UFA...................................  Use an additional factor of 1, 3, or 10 when extrapolating from valid   
                                         results of long-term studies on experimental animals when results of   
                                         studies of human exposure are not available or are inadequate. This    
                                         factor is intended to account for the uncertainty involved in          
                                         extrapolating from animal data to humans (interspecies variation).     
UFS...................................  Use an additional factor of 1, 3, or 10 when extrapolating from less-   
                                         than-chronic results on experimental animals when there are no useful  
                                         long-term human data. This factor is intended to account for the       
                                         uncertainty involved in extrapolating from less-than-chronic NOAELs to 
                                         chronic NOAELs.                                                        
UFL...................................  Use an additional factor of 1, 3, or 10 when deriving an RfD from a     
                                         LOAEL, instead of a NOAEL. This factor is intended to account for the  
                                         uncertainty involved in extrapolating from LOAELs to NOAELs.           
UFD...................................  Use an additional 3- or 10-fold factor when deriving an RfD from an     
                                         ``incomplete'' data base. This factor is meant to account for the      
                                         inability of any single type of study to consider all toxic endpoints. 
                                         The intermediate factor of 3 (approximately \1/2\ log10 unit, i.e., the
                                         square root of 10) is often used when there is a single data gap       
                                         exclusive of chronic data. It is often designated as UFD.              
----------------------------------------------------------------------------------------------------------------
                                                Modifying Factor                                                
----------------------------------------------------------------------------------------------------------------
                                        Use professional judgment to determine the MF, which is an additional   
                                         uncertainty factor that is greater than zero and less than or equal to 
                                         10. The magnitude of the MF depends upon the professional assessment of
                                         scientific uncertainties of the study and data base not explicitly     
                                         treated above (e.g., the number of species tested). The default value  
                                         for the MF is 1.                                                       
----------------------------------------------------------------------------------------------------------------
Note: With each UF or MF assignment, it is recognized that professional scientific judgment must be used. The   
  total product of the uncertainty factors and modifying factor should not exceed 3,000.                        

    Similar to the procedure used in the 1980 AWQC National Guidelines, 
the revised derivation of AWQC values for noncarcinogens uses the RfD 
together with various assumptions concerning intake of the contaminant 
from both water and nonwater sources of exposure. The objective of the 
AWQC value for noncarcinogens is to ensure that human exposure to a 
substance related to its presence in surface water, combined with 
exposure from other sources, does not exceed the RfD. The algorithm for 
deriving AWQC for noncarcinogens using the RfD is presented as Equation 
ID-1 in the Introduction and discussed further in Appendix II, Section 
C in this Notice.
3. Issues and Recommendations Concerning the Derivation of AWQC for 
Noncarcinogens
    During a review of the 1980 AWQC National Guidelines (USEPA, 
1993b), the Agency identified several issues that must be resolved in 
order to develop a final revised methodology for deriving AWQC based on 
noncancer effects. These issues, as discussed below, mainly concern the 
derivation of the RfD as the basis for such an AWQC value. Foremost 
among these issues is whether the Agency should revise the present 
method or adopt entirely new procedures that use quantitative dose-
response modelling for the derivation of the RfD. Other issues include 
the following:
     Presenting the RfD as a single point value or as a range to 
reflect the inherent imprecision of the RfD;
     Selecting specific guidance documents for derivation of 
noncancer health effect levels;
     Considering severity of effect in the development of the RfD;
     Using less-than-90-day studies as the basis for RfDs;
     Integrating reproductive/developmental, immunotoxicity, and 
neurotoxicity data into the RfD calculation;
     Applying pharmacokinetic data in risk assessments; and

[[Page 43785]]

     Considering the possibility that some noncarcinogenic effects 
do not exhibit a threshold.
    (a) Using the Current NOAEL-UF Based RfD Approach or Adopting More 
Quantitative Approaches for Noncancer Risk Assessment. The current 
NOAEL-UF-based RfD methodology, or its predecessor ADI/SF methodology, 
have been used since 1980. This approach assumes that there exists a 
threshold exposure below which adverse noncancer health effects are not 
expected to occur. Exposures above this threshold are believed to pose 
some risk to exposed individuals; however, the current approach does 
not address the nature and magnitude of the risk above the threshold 
level (i.e., the shape of the dose-response curve above the threshold). 
The NOAEL-UF-based RfD approach is intended primarily to ensure that 
the RfD value derived from the available data falls below the 
population effects threshold. However, the NOAEL-UF-based RfD procedure 
has limitations. In particular, this method requires that one of the 
actual experimental doses used by the researchers in the critical study 
be selected as the NOAEL or LOAEL value. The determination that a dose 
is a NOAEL or LOAEL will depend on the biological endpoints used and 
the statistical significance of the data. Statistical significance will 
depend on the number and spacing of dose groups and the numbers of 
animals used in each dose group. Studies using a small number of 
animals can limit the ability to distinguish statistically significant 
differences between measurable responses seen in dose groups and 
control groups. Furthermore, the determination of the NOAEL or LOAEL 
also depends on the dose spacing of the study. Doses are often widely 
spaced, typically differing by factors of three to ten. A study can 
identify a NOAEL and a LOAEL from among the doses studied, but the 
``true'' NOAEL cannot be determined from those results. The study size 
and dose spacing limitations also limit the ability to characterize the 
nature of the expected response to exposures between the observed NOAEL 
and the LOAEL values.
    The limitations of the NOAEL-UF approach have prompted development 
of alternative approaches that incorporate more quantitative dose-
response information. The traditional NOAEL approach for noncancer risk 
assessment has often been a source of controversy and has been 
criticized in several ways. For example, experiments involving fewer 
animals tend to produce higher NOAELs and, as a consequence, may 
produce higher RfDs. The reverse would seem more appropriate in a 
regulatory context because larger sample sizes should provide greater 
experimental sensitivity. The focus of the NOAEL approach is only on 
the dose that is the NOAEL, and the NOAEL must be one of the 
experimental doses. It also ignores the shape of the dose-response 
curve. Thus, the slope of the dose-response plays little role in 
determining acceptable exposures for human beings. Therefore, in 
addition to the NOAEL-UF-based RfD approach described above, EPA is 
considering using other approaches that incorporate more quantitative 
dose-response information in appropriate situations for the evaluation 
of noncancer effects and the derivation of RfDs. However, the Agency 
wishes to emphasize that it still believes the NOAEL-UF RfD methodology 
is valid and can continue to be used to develop RfDs.
    Two alternative approaches that may have relevance in assisting in 
the derivation of the RfD for a chemical are the BMD and the 
Categorical Regression approaches. These alternative approaches may 
overcome some of the inherent limitations in the NOAEL-UF approach. For 
example, the BMD analyses for developmental effects show that NOAELs 
from studies correlate well with a 5 percent response level (Allen et 
al., 1994). The BMD and the Categorical Regression approaches usually 
have greater data requirements than the RfD approach. Thus, it is 
unlikely that any one approach will apply to every circumstance; in 
some cases, different approaches may be needed to accommodate the 
varying data bases for the range of chemicals for which water quality 
criteria must be developed. Acceptable approaches will satisfy the 
following criteria: (1) Meet the appropriate risk assessment goal; (2) 
adequately describe the toxicity data base and its quality; (3) 
characterize the endpoints properly; (4) provide a measure of the 
quality of the ``fit'' of the model when a model is used for dose-
response analysis; and (5) describe the key assumptions and 
uncertainties.
    (1) The Benchmark Dose. The BMD is defined as the statistical lower 
confidence limit on the dose estimated to produce a predetermined level 
of change in response (the Benchmark Response, or BMR) relative to 
control. In the derivation of an RfD, the BMD is used as the dose to 
which uncertainty factors are applied instead of the NOAEL. The BMD 
approach first models a dose-response curve for the critical effect(s) 
using available experimental data. Several functional forms can be used 
to model the dose- response curve, such as polynomial or Weibull 
functions. To define a BMD from the modeled curve for quantal data, the 
assessor first selects the BMR. The choice of the BMR is critical. For 
quantal endpoints, a particular level of response is chosen (e.g., 1 
percent, 5 percent, or 10 percent). For continuous endpoints, the BMR 
is the degree of change from controls and is based on what is 
considered a biologically significant change. The BMD is derived from 
the BMR dose by applying the desired confidence limit calculation. The 
RfD is obtained by dividing the BMD by one or more uncertainty factors, 
similar to the NOAEL approach. Because the BMD is used like the NOAEL 
to obtain the RfD, the BMR should be selected at or near the low end of 
the range of increased risks that can be detected in a study of typical 
size. Generally, this falls in the range between the ED01 
and the ED10.
    The Agency is considering the use of a BMD approach to derive RfDs 
for those agents for which there is an adequate data base. There are a 
number of technical decisions associated with the application of the 
BMD technique. These include the following:
     Selection of response data to model;
     The form of the data used (continuous versus quantal);
     The definition of an adverse response;
     The choice of mathematical model (including use of 
nonstandard models for unusual data sets);
     The choice of the measures of increased risk (extra risk 
versus additional risk);
     The selection of the BMR;
     Methods for calculating the confidence interval;
     Selection of the appropriate BMD as the basis for the RfD 
(when multiple endpoints are modeled from a single study, when multiple 
models are applied to a single response, and when multiple BMDs are 
calculated from different studies); and
     The use of uncertainty factors with the BMD approach.
    These topics are discussed in detail in Crump et al. (1995) and the 
TSD that accompanies this Notice. The use of the BMD approach has been 
discussed in general terms by several authors (Gaylor, 1983; Crump, 
1984; Dourson et al., 1985; Kimmel and Gaylor, 1988; Brown and 
Erdreich, 1989; Kimmel, 1990). The International Life Sciences 
Institute (ILSI) also held a major workshop on the BMD in September 
1993; the workshop proceedings are summarized in ILSI (1993) and in 
Barnes et al. (1995). For further information on these technical 
issues,

[[Page 43786]]

the reader is referred to these publications.
    The BMD approach addresses several of the quantitative or 
statistical criticisms of the NOAEL approach. These are discussed at 
greater length in Crump et al. (1995) and are summarized here. First, 
the BMD approach uses information on variability in the selected study 
rather than just a single data point, such as the NOAEL or LOAEL. By 
using response data from all of the dose groups to model a dose-
response curve, the BMD approach allows for consideration of the 
steepness of the slope of the curve when estimating the 
ED10. The use of the full data set also makes the BMD 
approach less sensitive to small changes in data than the NOAEL 
approach, which relies on the statistical comparison of individual dose 
groups. The BMD approach also allows consistency in the consideration 
of the level of effect (e.g., a 10 percent response rate) across 
endpoints.
    The BMD approach accounts more appropriately for the size of each 
dose group than the NOAEL approach. Laboratory tests with fewer animals 
per dose group tend to yield higher NOAELs, and thus higher RfDs, 
because statistically significant differences in response rates are 
harder to detect. Therefore, in the NOAEL approach, dose groups with 
fewer animals lead to a higher (less conservative) RfD. In contrast, 
with the BMD approach, smaller dose groups will tend to have the effect 
of extending the confidence interval around the ED10; 
therefore, the lower confidence limit on the ED10 (the BMD) 
will be lower. With the BMD approach, greater uncertainty (smaller test 
groups) leads to a lower (more conservative) RfD.
    There are some issues to be resolved before the BMD approach is 
used routinely. These were identified in a 1996 Peer Consultation 
Workshop (USEPA, 1996b). Methods for routine use of the BMD are 
currently under development by EPA. Several RfCs and RfDs based on the 
BMD approach are included in EPA's IRIS data base. These include that 
for methyl mercury based on delayed postnatal development in humans; 
carbon disulfide based on neurotoxicity; 1,1,1,2-tetrafluoroethane 
based on testicular effects in rats; and antimony trioxide based on 
chronic pulmonary interstitial inflammation in female rats.
    Various mathematical approaches have been proposed for modeling 
developmental toxicity data (e.g., Crump, 1984; Kimmel and Gaylor, 
1988; Rai and Van Ryzin, 1985; Faustman et al., 1989), which could be 
used to calculate a BMD. Similar methods can be used to model other 
types of toxicity data, such as neurotoxicity data (Gaylor and Slikker, 
1990, 1992; Glowa and MacPhail, 1995). The choice of the mathematical 
model may not be critical, as long as estimation is within the observed 
dose range. Since the model is used only to fit the observed data, the 
assumptions in a particular model regarding the existence or absence of 
a threshold for the effect may not be pertinent (USEPA, 1997). Thus, 
any model that suitably fits the empirical data is likely to provide a 
reasonable estimate of a BMD. However, research has shown that flexible 
models that are nonsymmetric (e.g., the Weibull) are superior to 
symmetric models (e.g., the probit) in estimating the BMD because the 
data points at the higher doses have less influence on the shape of the 
curve than at low doses. In addition, models should incorporate 
fundamental biological factors where such factors are known (e.g., 
intralitter correlation for developmental toxicity data) in order to 
account for as much variability in the data as possible. The Agency is 
currently supporting research studies to evaluate the application of 
several models to data sets for calculating the BMD.
    (2) Categorical Regression. Categorical Regression is an emerging 
technique that may have relevance for the derivation of RfDs or for 
estimating risk above the RfD (Dourson et al., 1997; Guth et al., 
1997). The Categorical Regression approach, like the BMD approach, can 
be used to estimate a dose that corresponds to a given probability of 
adverse effects. This dose would then be divided by uncertainty factors 
to establish a reference dose. However, unlike the BMD approach, the 
Categorical Regression approach can incorporate information on 
different health endpoints in a single dose-response analysis. For 
those health effects for which studies exist, responses to the 
substance in question are grouped into severity categories; for example 
(1) no effect, (2) no adverse effect, (3) mild-to-moderate adverse 
effect, and (4) frank effect. These categories correspond to the dose 
categories currently used in setting the RfD, namely, the no-observed-
effect level (NOEL), NOAEL, LOAEL, and frank-effect level (FEL), 
respectively. Logistic transform or other applicable mathematical 
operations are used to model the probability of experiencing effects in 
a certain category as a function of dose (Harrell, 1986; Hertzberg, 
1989). The ``acceptability'' of the fit of the model to the data can be 
judged using several statistical measures, including the 
2 statistic, correlation coefficients, and the 
statistical significance of its model parameter estimates.
    The resulting function can be used to find a dose (or the lower 
confidence bound on the dose) at which the probability of experiencing 
adverse effects does not exceed a selected level, e.g., 10 percent. 
This dose (like the NOAEL or BMD) would then be divided by relevant 
uncertainty factors to calculate a RfD. For more detail on how to 
employ the categorical regression approach, see the discussion in the 
TSD.
    As with the BMD approach, the Categorical Regression approach has 
the advantage of using more of the available dose-response data to 
account for response variability as well as accounting for uncertainty 
due to sample size through the use of confidence intervals. Additional 
advantages of categorical regression include the combining of data sets 
prior to modeling, thus allowing the calculation of the slope of a 
dose-response curve for multiple adverse effects rather than only one 
effect at a time, and the ability to estimate risks for different 
levels of severity from exposures above the RfD.
    On the other hand, as with BMD, opinions differ over the amount and 
adequacy of data necessary to implement the method. The Categorical 
Regression approach also requires judgments regarding combining data 
sets, judging goodness-of-fit, and assigning severity to a particular 
effect. Furthermore, this approach is still in the developmental stage. 
It is not recommended for routine use, but may be used when data are 
available and justify the extensive analyses required.
    (3) Summary. Whether a NOAEL-based methodology, a BMD, a 
Categorical Regression model, or other approach is used to develop the 
RfD, the dose-response-evaluation step of a risk assessment process 
should include additional discussion about the nature of the toxicity 
data and its applicability to human exposure and toxicity. The 
discussion should present the range of doses that are effective in 
producing toxicity for a given agent; the route, timing, and duration 
of exposure; species specificity of effects; and any pharmacokinetic or 
other considerations relevant to extrapolation from the toxicity data 
to human-health-based AWQC. This information should always accompany 
the characterization of the adequacy of the data.
    (b) Presenting the RfD as a Single Point or as a Range for Deriving 
AWQC. Although the RfD has traditionally been presented and used as a 
single point, its

[[Page 43787]]

definition contains the phrase ``. . . an estimate (with uncertainty 
spanning perhaps an order of magnitude) . . .'' (USEPA, 1993a). 
Underlying this concept is the reasoning that the selection of the 
critical effect and the total uncertainty factor used in the derivation 
of the RfD is based on the ``best'' scientific judgment, and that 
competent scientists examining the same data base could derive RfDs 
which varied within an order of magnitude.
    In one case, the RfD was presented as a point value within an 
accompanying range. EPA derived a single number as the RfD for arsenic 
(0.3 g/kg-day), but added that ``strong scientific arguments 
can be made for various values within a factor of 2 or 3 of the 
currently recommended RfD value, i.e., 0.1 to 0.8 g/kg/day'' 
(USEPA, 1993c). EPA noted that regulatory managers should be aware of 
the flexibility afforded them through this action.
    In today's Notice, EPA discusses situations where the risk manager 
can consider a range around the point estimate. As explained further 
below, the Agency is recommending that sometimes considering the use of 
a range for the RfD is more appropriate in characterizing risk than 
only the use of the point estimate. The selection of an appropriate 
range must be determined for each individual situation, since several 
factors affect the magnitude of the range associated with the RfD. For 
example, the completeness of the data base plays a major role. 
Observing similar effects in several animal species, including humans, 
can increase confidence in the selection of the critical effect and 
thereby narrow the range of uncertainty. Other factors that can affect 
the precision are: the slope of the dose-response curve, seriousness of 
the observed effect, dose spacing, and possibly the route of the 
experimental doses. For example, a steep dose-response curve indicates 
that relatively large differences in response occur with a small change 
in dose. For chemicals that elicit a serious effect near the LOAEL, an 
additional uncertainty factor is often used in the RfD derivation to 
protect against less serious but still observable adverse effects that 
could occur at lower doses, thus increasing the range of uncertainty 
for the RfD. Dose spacing and the number of animals in the study groups 
used in the experiment can also affect the confidence in the RfD.
    To derive the AWQC, the point estimate of the RfD is the default. 
Based on considerations of available data, the use of another number 
within the range defined by the UF could be justified in a specific 
case. This means that there are risk considerations which indicate that 
some value in the range other than the point estimate may be more 
appropriate than the point estimate, based on human health or 
environmental fate considerations.
    Because the uncertainty around the dose-response relationship 
increases as extrapolation below the observed data increases, the use 
of a point within the RfD range may be more appropriate in 
characterizing the risk than the use of the point estimate. Therefore, 
as a matter of risk management policy, it is proposed that if the 
product of the UFs and MF used to derive the RfD is 100 or less, there 
would be no consideration of a range because there is great confidence 
in the hazard and dose-response characterization. If greater than 100 
and less than 1,000, the maximum range that could be considered would 
be one half of a log10 (3-fold) or a number ranging from the 
point estimate divided by 1.5 to the point estimate multiplied by 1.5. 
At 1,000 and above, the maximum range would be a log10 (10-
fold) or a number ranging from the point estimate divided by 3 to the 
point estimate multiplied by 3. Use of any point other than the RfD 
must be justified.
    The following examples illustrate situations where EPA believes the 
use of a range is not appropriate. The RfD for zinc (USEPA, 1992) is 
based on consideration of nutritional data, a minimal LOAEL, and a UF 
of 3. If a factor of 3 were used to bound the RfD for zinc, then the 
upper-bound level would approach the minimal LOAEL. This situation must 
be avoided, since it is unacceptable to set a standard at levels that 
may cause an adverse effect. Another case in point is nitrate. Since 
the RfD for nitrate was based on the lack of effects in human infants 
and was assigned a UF of 1 (USEPA, 1991b), it would be difficult if not 
impossible to justify the use of an RfD range for infants exposed to 
nitrate. Table IIIB-2 gives examples of factors to consider when 
determining whether to use the point estimate of the RfD, or a value 
higher or lower than the point estimate (see the TSD for additional 
detail on this topic).

                   Table IIIB-2.--Some Scientific Factors To Consider When Using the RfD Range                  
----------------------------------------------------------------------------------------------------------------
                                                                                                                
----------------------------------------------------------------------------------------------------------------
Use point estimate RfD.................  --Default position                                                     
                                         --Total uncertainty factor, modifying factor product 100 or less       
                                         --Essential nutrient                                                   
Use lower range of RfD.................  --Increased bioavailability from medium                                
                                         --The seriousness of the effect and whether or not it is reversible    
                                         --A shallow dose-response curve in the range of observation            
                                         --Exposed group contains a sensitive population (e.g., children or     
                                          fetuses)                                                              
Use upper range of RfD.................  --Decreased bioavailability with humans                                
                                         --RfD based on minimal LOAEL and large uncertainty factor              
                                         --A steep dose-response curve in the range of observation              
                                         --No sensitive populations identified                                  
----------------------------------------------------------------------------------------------------------------

    The risk-characterization step of the risk assessment provides a 
mechanism for communicating such issues. The risk manager must be 
informed of those specific cases when it is not scientifically correct 
to estimate a RfD range. In addition, the risk characterization should 
provide risk managers with guidelines (see Table IIIB-2) on the 
scientific basis for using a value within the range as the RfD.
    (c) Guidelines to be Adopted for Derivation of Noncancer Health 
Effects Values. The Agency is currently using IRIS Background Document 
1A entitled Reference Dose (RfD): Description and Use in Health Risk 
Assessments as the general basis for the risk assessment of 
noncarcinogenic effects of chemicals (USEPA, 1993a). EPA recommends 
continued use of this document for this purpose. However, it should be 
noted that the process for evaluating chemicals for inclusion in IRIS 
is undergoing revision. The Agency is currently conducting a pilot 
program for the continued development of the IRIS

[[Page 43788]]

assessment process. Under this program, a more integrated assessment 
for cancer and noncancer effects is being developed for 11 chemicals: 
arsenic, bentazon, beryllium, chlordane, chromium compounds, cumene, 
methyl methacrylate, methylene diphenyl isocyanate, napthalene, 
tributyltin oxide and vinyl chloride (USEPA, 1996c). The results for 
these 11 are expected to be in IRIS soon. A second set of chemical 
assessments have also been initiated and are expected to be complete by 
the end of 1998. The second set includes the following eight chemicals: 
acetonitrile; barium; benzene; 1,3-butadiene; cadmium; chloroethane; 
diesel emissions; and ethylene glycol butyl ether (USEPA 1998). A third 
set of chemicals is planned for completion by the end of 1999, which 
includes boron; bromate; chloral hydrate; chloroform; dichloroacetic 
acid; 1,3-dichloropropene; formaldehyde; lindane; nitrobenzene; 
pentachlorophenol; PCBs (noncancer endpoints); styrene; 
tetrachloroethylene; tetrahydrofuran; toxaphene; trichloroethylene; and 
vinyl acetate (USEPA, 1998).
    (d) Treatment of Uncertainty Factors/Severity of Effects During the 
RfD Derivation and Verification Process. During the RfD derivation and 
review process, EPA considers the uncertainty of extrapolations between 
animal species and within individuals of a species, as well as specific 
uncertainties associated with the completeness of the data base, as 
described in Table IIIB-1.
    The Agency's RfD Work Group has always considered the severity of 
the observed effects induced by the chemical under review when choosing 
the value of the UF with a LOAEL. For example, during the derivation 
and verification of the RfD for zinc (USEPA, 1992), an uncertainty 
factor less than the standard factor of 10 (UF of 3) was assigned to 
the relatively mild adverse effects seen in experimental studies in 
humans, namely, a decrease in erythrocyte superoxide dismutase 
activity. EPA recommends that an assessment of the severity of the 
critical effect be determined when deriving an RfD and that risk 
managers be made aware of the severity of the effect and the weight 
placed on this attribute of the effect when the RfD was derived.
    (e) Use of Less-Than-90-Day Studies to Derive RfDs. Generally, 
less-than-90-day experimental studies are not used to derive an RfD. 
This is based on the rationale that studies lasting for less than 90 
days may be too short to detect various toxic effects. However, EPA, 
has in certain circumstances, derived an RfD based on a less-than-90-
day study. For example, the RfD for nonradioactive effects of uranium 
is based on a 30-day rabbit study (USEPA, 1989). The short-term 
exposure period was used since it was adequate for determining doses 
that cause chronic toxicity. In other cases, it may be appropriate to 
use a less-than-90-day study because the critical effect is expressed 
in less than 90 days. For example, the RfD for nitrate was derived and 
verified using studies that were less than 3-months duration (USEPA, 
1991b). The reason for this decision was that the critical effect, 
methemoglobinemia in infants, occurs in less than 90 days. When it can 
be demonstrated from other data in the toxicological data base that the 
critical adverse effect is expressed within the study period and that a 
longer exposure duration would not exacerbate the observed effect or 
cause the appearance of some other adverse effect, the Agency may 
choose to use less-than-90-day studies as the basis of the RfD. Such 
values would have to be used with care because of the uncertainty in 
determining if other effects might be expressed if exposure was of 
greater duration than 90 days.
    (f) Use of Reproductive/Developmental, Immunotoxicity, and 
Neurotoxicity Data as the Basis for Deriving RfDs. All relevant 
toxicity data have some bearing on the RfD derivation and verification 
and are considered by EPA. The ``critical'' effect is the adverse 
effect most relevant to humans or, in the absence of an effect known to 
be relevant to humans, the adverse effect that occurs at the lowest 
dose in animal studies. For example, if the critical effect is 
neurotoxicity, EPA may use this specific toxicity data as the basis for 
the derivation and verification of an RfD, as it did for the RfD for 
acrylamide. Moreover, the Agency is continually revising its procedures 
for noncancer risk assessment. For example, EPA has recently released 
guidelines for deriving developmental RfDs (RfDDT, USEPA, 
1991a), for using reproductive toxicity (USEPA, 1996a), and 
neurotoxicity (USEPA, 1995) data in risk assessments. The Agency is 
currently working on guidelines for using immunotoxicity to derive 
RfDs. In addition, the Agency is proceeding with the process of 
generating acceptable emergency health levels for hazardous substances 
in acute exposure situations based on established guidelines (NRC, 
1993).
    (g) Applicability of Physiologically Based Pharmacokinetic (PBPK) 
Data in Risk Assessment. EPA believes that all pertinent data should be 
used in the risk assessment process, including PBPK data. In fact, the 
Agency has used PBPK data in deriving the RfD for cadmium and other 
compounds. In addition, the Agency is currently using PBPK data to 
better characterize human inhalation exposures from animal inhalation 
experiments during derivation/verification of RfCs. In analogy to the 
RfD, the RfC is considered to be an estimate of a level in the air that 
is not anticipated to cause adverse effects over a lifetime of 
inhalation exposure (Jarabek et al., 1990). With RfCs, a kinetic 
adjustment is made for the differences between animals and humans in 
respiration and deposition. This procedure results in calculation of a 
``human equivalent concentration.'' Based on the use of these 
procedures, an interspecies UF of 3 (i.e., approximately 
100.5), instead of the standard factor of 10, is used in the 
RfC derivation.
    The rationale for the use of PBPK models is that the 
pharmacokinetics and pharmacodynamics of a chemical each contribute to 
a chemical's observed toxicity, and specifically, to observed 
differences among species in sensitivity. Pharmacokinetics describes 
the absorption, distribution, metabolism, and elimination of chemicals 
in the body, while pharmacodynamics describes the toxic interaction of 
the agent with the target cell. In the absence of specific data on 
their relative contributions to the toxic effects observed in species, 
each is considered to account for approximately one-half of the 
variability in observed effects, as is assumed in the development of 
RfCs and RfDs. The implication of this assumption is that an 
interspecies uncertainty factor of 3 rather than 10 could be used for 
deriving an RfD when valid pharmacokinetic data and models can be 
applied to obtain an oral ``human equivalent applied dose'' (Jarabek et 
al., 1990). If specific data exist on the relative contribution of 
either element to observed effects, that proportion will be used.
    (h) Consideration of Linearity (or Lack of a Threshold) for 
Noncarcinogenic Chemicals. It is quite possible that there are 
chemicals with noncarcinogenic endpoints that have no threshold 
exposure level. For example, it appears that, after skin sensitization 
occurs from exposure to nickel, there is no apparent threshold in 
subpopulations of hypersensitive individuals for subsequent dermal 
effects of the chemical. Other examples could include genotoxic 
teratogens and germline mutagens. Genotoxic teratogens act by causing 
mutational events during organogenesis, histogenesis, or other stages 
of development. Germline mutagens interact with germ cells to produce 
mutations which may be transmitted to the zygote and expressed

[[Page 43789]]

during one or more stages of development. However, there are few 
chemicals which currently have sufficient mechanistic information about 
these possible modes of action. It should be recognized that although a 
mode of action consistent with linearity is possible (especially for 
agents known to be mutagenic), this has yet to be reasonably 
demonstrated for most toxic endpoints other than cancer.
    EPA has recognized the potential for nonthreshold noncarcinogenic 
endpoints and discussed this issue in the Guidelines for Developmental 
Toxicity Risk Assessment (USEPA, 1991a) and in the 1986 Guidelines for 
Mutagenicity Risk Assessment (USEPA, 1986). An awareness of the 
potential for such teratogenic/mutagenic effects should be established 
in order to deal with such data. However, without adequate data to 
support a genetic or mutational basis for developmental or reproductive 
effects, the default becomes an uncertainty factor or mechanism of 
action approach, which are procedures utilized for noncarcinogens 
assumed to have a threshold. Therefore, genotoxic teratogens and 
germline mutagens should be considered an exception while the 
traditional uncertainty factor approach is the general rule for 
calculating criteria or values for chemicals demonstrating 
developmental/reproductive effects. For the exceptional cases, since 
there is no well-established mechanism for calculating criteria 
protective of human health from the effects of these agents, criteria 
will be established on a case-by-case basis. Other types of 
nonthreshold noncarcinogens must also be handled on a case-by-case 
basis.
    (i) Minimum Data Requirements. For details on minimum data 
requirements related to RfD development, see the TSD.
4. SAB Comments
    The SAB commented that the BMD approach, and other approaches, have 
strengths and weaknesses. As described previously, these approaches 
permit use of more of the entire data base, derive a number that is 
independent of dose spacing, and can be applied in a manner that 
reflects the quality of the data. The SAB counseled against using a low 
BMD (e.g., ED01) that is outside the dose range able to be 
detected by current toxicological methodology. The SAB further 
mentioned that the ``threshold'' for a noncancer effect must be 
considered when using these approaches. EPA does not disagree with the 
SAB comments on the BMD and other new approaches for dose-response 
evaluation. The AWQC Methodology allows for using the benchmark, 
categorical regression or traditional approach (i.e., NOAEL/LOAEL) in 
deriving an RfD. This allows for flexibility in choosing the approach 
that best suits the data. In most cases, the concept of a threshold 
will be intrinsic to the risk characterization for noncarcinogens. 
However, as pointed out in Section B.3(h), there are some toxins (such 
as lead) that appear to have no threshold.
    The SAB has expressed the opinion that few data demonstrate that 
the precision of the RfD derivation process is ``an order of 
magnitude'' and mentioned that the precision of each RfD is specific 
for that RfD. The SAB also questions the application of the term 
``precision'' in this case, because of the difficulty in evaluating the 
precision of a particular RfD. In responding to comments, EPA attempted 
to remove terminology that implied that there was an order of magnitude 
in the precision of the RfD but still allowed for choosing a value 
other than the point estimate of the RfD in establishing the AWQC. The 
acceptable range around the RfD has been tied to the uncertainty in the 
data, rather than any assessment of the analytical precision or 
accuracy of the calculation. The word precision is still used in the 
text, but, hopefully, in a context that implies a general rather than 
analytical meaning.
    The SAB concurs that the severity of effect should be considered 
during the RfD derivation and verification process. However, the SAB 
has expressed concern about the type of scale that would be used to 
rate the level of severity. SAB suggests that a severity scale could be 
based on whether the effect is reversible or if it is irreversible and 
cumulative. Another possible construct could consider whether the 
effect is an overt pathology, functional deficit, adverse biochemical 
change, or a biochemical change of unknown consequence. Finally, a 
severity scale could be developed based on consideration of target 
organ affected. The SAB commented that the second type of scale is 
likely to have greatest applicability to noncancer effects, and would 
require that biochemical effects be specifically related to functional 
changes and/or to overt pathology. The SAB expressed skepticism about 
scales based on relative value given to target organ systems. EPA 
agrees that it is difficult to develop a simple scale for expressing 
the severity of an effect. Such a judgment is best left to experienced 
toxicologists. References for guidelines to consider in evaluating the 
seriousness of effects are included in the TSD as resource information 
for the reader.
    The SAB has expressed the opinion that, as a rule, less-than-90-day 
studies are not adequate for RfD derivation, and cited the danger of 
false-negative studies. It believes that RfDs derived in this manner 
should be labeled as ``temporary'' or ``interim.'' However, as 
demonstrated above, each case must be considered individually. The AWQC 
guidelines are in agreement with SAB regarding the use of data from 
studies of less than 90-day duration, but point out that there are 
circumstances (such as occurrence of a critical acute effect or a 
developmental RfD) where data from durations of less than 90-days are 
used.
    The SAB believes that PBPK modeling is useful for RfD derivation 
but needs to be based on understanding the mechanisms of toxicity. EPA 
is in general agreement with the SAB's opinions about the limitations 
on the use of PBPK data, and require that pharmacokinetic models be 
verified and understood before they are used. This implies that there 
is an understanding of the pharmacodynamic interactions of the toxic 
agent with a target cell.
5. Request for Comments
    1. EPA requests comment on the application of the NOAEL-UF, BMD, 
Categorical Regression, and other approaches to derive RfDs in support 
of the derivation of AWQC for the protection of human health.
    2. EPA requests comment on the issue of permitting the use of a 
point within the RfD range for deriving the AWQC, rather than a single 
point estimate. It must be emphasized that appropriate scientific 
justification must be given when using any number other than the point 
estimate RfD. EPA requests comment on how to develop the RfD range and 
how to determine which point estimate in the range is appropriate.
    3. EPA requests comment on approaches to incorporate severity of 
effect in deriving the RfD.
    4. EPA requests comment on the use of less-than-90-day studies to 
derive RfDs.
    5. EPA requests comment on the use of reproductive/developmental, 
immunotoxicity, and neurotoxicity data as the basis for deriving RfDs.
    6. EPA requests comment on the use of PBPK data in deriving an RfD.
    7. EPA requests comment on allowing, on a case-by-case basis, 
consideration of a nonthreshold mode of action for certain chemicals 
that cause noncancer effects when deriving RfDs.

[[Page 43790]]

References for Noncancer Effects

Allen, B.C., R.T. Kavlock, C.A. Kimmel, and E.M. Faustman. 1994. 
Dose-response Assessment for Developmental Toxicity. Fund. Appl. 
Toxicol. 23: 496-509.
Barnes, D.G., G.P Daston, J.S. Evans, A.M. Jarabek, R.J. Kavlock, 
C.A. Kimmel, C. Park, and H.L. Spitzer. 1995. Benchmark Dose 
Workshop: Criteria for Use of a Benchmark Dose to Estimate a 
Reference Dose. In: Regulatory Toxicology and Pharmacology, 21: 296-
306.
Brown, K.G. and L.S. Erdreich. 1989. Statistical Uncertainty in the 
No-observed-adverse-effect Level. Fund. Appl. Toxicol. 13(2): 235-
244.
Crump, K.S., B. Allen, and E. Faustman. 1995. The Use of the 
Benchmark Dose Approach in Health Risk Assessment. Prepared for 
USEPA Risk Assessment Forum. EPA/630/R-94-007.
Crump, K.S. 1984. A New Method for Determining Acceptable Daily 
Intakes. Fund. Appl. Toxicol. 4: 854-871.
Dews, P.B. 1986. On the Assessment of Risk. In: Developmental 
Behavioral Pharmacology. N. Krasnegor, J. Gray and T. Thompson 
(eds.). Hillsdale, NJ: Lawrence Erlbaun Assoc. 53-65.
Dourson, M.L. 1994. Methodology for Establishing Oral Reference 
Doses (RfDs). In: Risk Assessment of Essential Elements. W. Mertz, 
C.O. Abernathy and S.S. Olin (eds.) Washington, DC: ILSI Press. pp. 
51-61.
Dourson, M.L., R.C. Hertzberg, R. Hartung and K. Blackburn. 1985. 
Novel Approaches for the Estimation of Acceptable Daily Intake. 
Toxicol. Ind. Health 1(4): 23-41.
Dourson, M.L., L.K. Teuschler, P.R. Durkin, and W.M. Stiteler. 1997. 
Categorical Regression of Toxicity Data, A Case Study Using 
Aldicarb. Regul. Toxicol. Pharmacol. 25: 121-129.
Faustman, E.M., D.G. Wellington, W.P. Smith and C.A. Kimmel. 1989. 
Characterization of a Developmental Toxicity Dose-response Model. 
Environ. Health Perspect. 79: 229-241.
Gaylor, D.W. 1983. The Use of Safety Factors for Controlling Risk. 
J. Toxicol. Environ. Health 11: 329-336.
Gaylor, D.W. and W. Slikker. 1990. Risk Assessment for Neurotoxic 
Effects. Neurotoxicology 11:211-218.
Gaylor, D.W. and W. Slikker. 1992. Risk Assessment for 
Neurotoxicants. In: Neurotoxicology. H. Tilson and C. Mitchel (eds). 
New York: Raven Press. 331-343.
Glowa, J.R. and R.C. MacPhail. 1995. Quantitative Approaches to Risk 
Assessment in Neurotoxicology. In: Neurotoxicology: Approaches and 
Methods. Academic Press. 777-787.
Guth, D.J., R.J. Carroll, D.G. Simpson, and H. Zhou. 1997. 
Categorical Regression Analysis of Acute Exposure to 
Tetrachloroethylene. Risk Anal. 17(3): 321-332.
Harrell, F. 1986. The Logist Procedure. SUGI Supplemental Library 
Users Guide, Ver. 5th ed. Cary, NC: SAS Institute.
Hertzberg, R.C. 1989. Fitting a Model to Categorical Response Data 
with Application to Species Extrapolation of Toxicity. Health 
Physics 57: 405-409.
International Life Sciences Institute (ILSI). 1993. Report of the 
Benchmark Dose Workshop. Washington, DC: International Life Sciences 
Institute, Risk Science Institute.
Jarabek, A.M., M.G. Menache, J.H. Overton, M.L. Dourson, and F.J. 
Miller. 1990. The U.S. Environmental Protection Agency's Inhalation 
RfD Methodology. Risk Assessment for Air Toxics. Toxicol. Indust. 
Hlth. 6: 279-301.
Kimmel, C.A. 1990. Quantitative Approaches to Human Risk Assessment 
for Noncancer Health Effects. Neurotoxicology 11: 189-198.
Kimmel, C.A. and D.W. Gaylor. 1988. Issues in Qualitative and 
Quantitative Risk Analysis for Developmental Toxicity. Risk Anal. 
8(1): 15-20.
National Academy of Sciences. 1977. Decision Making in the 
Environmental Protection Agency. Vol. 2. Washington, DC: National 
Academy of Sciences. 32-33 and 241-242.
National Research Council. 1993. Guidelines for Developing Emergency 
Exposure Levels for Hazardous Substances. Subcommittee on Guidelines 
for Developing Community Emergency Exposure Levels (CEELs) for 
Hazardous Substances. Committee on Toxicology, NRC. Washington, DC: 
National Academy Press.
Rai, K. and J. Van Ryzin. 1985. A Dose-response Model for 
Teratological Experiments Involving Quantal Responses. Biometrics 
41: 1-10.
USEPA. 1986. Guidelines for Mutagenicity Assessment. Federal 
Register 51: 34006-34012. September 24.
USEPA. 1989. Reference Dose (RfD) for Oral Exposure for Uranium 
(Soluble Salts). Integrated Risk Information System (IRIS). Online. 
(Verification date 10/1/89). Office of Health and Environmental 
Assessment, Environmental Criteria and Assessment Office. 
Cincinnati, OH.
USEPA. 1991a. Final Guidelines for Developmental Toxicity Risk 
Assessment. Federal Register 56: 63798-63826. December 5.
USEPA. 1991b. Reference Dose (RfD) for Oral Exposure for Nitrate. 
Integrated Risk Information System (IRIS). Online. (Verification 
date 10/01/91). Office of Health and Environmental Assessment, 
Environmental Criteria and Assessment Office. Cincinnati, OH.
USEPA. 1992. Reference Dose (RfD) for Oral Exposure for Inorganic 
Zinc. Integrated Risk Information System (IRIS). Online. 
(Verification date 10/1/92). Office of Health and Environmental 
Assessment, Environmental Criteria and Assessment Office. 
Cincinnati, OH.
USEPA. 1993a. Reference Dose (RfD): Description and Use in Health 
Risk Assessments. Integrated Risk Information System (IRIS). Online. 
Intra-Agency Reference Dose (RfD) Work Group, Office of Health and 
Environmental Assessment, Environmental Criteria and Assessment 
Office. Cincinnati, OH. March 15.
USEPA. 1993b. Revision of Methodology for Deriving National Ambient 
Water Quality Criteria for the Protection of Human Health: Report of 
Workshop and EPA's Preliminary Recommendations for Revision. 
Submitted to the EPA Science Advisory Board by the Human Health Risk 
Assessment Branch, Health and Ecological Criteria Division, Office 
of Science and Technology, Office of Water. Washington, DC. January 
8.
USEPA. 1993c. Reference Dose (RfD) for Oral Exposure for Inorganic 
Arsenic. Integrated Risk Information System (IRIS). Online. 
(Verification date 02/01/93). Office of Health and Environmental 
Assessment, Environmental Criteria and Assessment Office. 
Cincinnati, OH.
USEPA. 1995. Proposed Guidelines for Neurotoxicity Risk Assessment. 
Federal Register 60: 52032-52056. October 4.
USEPA. 1996a. Reproductive Toxicity Risk Assessment Guidelines. 
Federal Register 61: 56274-56322. October 31.
USEPA. 1996b. Report on the Benchmark Dose Peer Consultation 
Workshop. Risk Assessment Forum. Washington, DC: EPA/630/R-96/011.
USEPA. 1996c. Integrated Risk Information System (IRIS); 
Announcement of Pilot Program; Request for Information. Federal 
Register. 61: 14570. April 2.
USEPA. 1997. Mercury Study: Report to Congress. Volume 5: Health 
Effects of Mercury and Mercury Compounds. Office of Air Quality 
Planning and Standards, and Office of Research and Development. EPA-
452/R-97-007.
USEPA. 1998. Integrated Risk Information System (IRIS). Announcement 
of 1998 Program; Request for Information. Federal Register 63(1): 
75-77. January 2.

C. Exposure

    As discussed in the Introduction, the derivation of AWQC for the 
protection of human health requires information about both the 
toxicological endpoints of concern for water pollutants and the 
pathways of human exposure to those pollutants. Historically, two 
primary pathways of human exposure to pollutants present in a 
particular ambient waterbody have been considered in deriving AWQC: 
direct ingestion and other exposure from household uses (e.g., 
showering) of drinking water obtained from that waterbody, and the 
consumption of fish/shellfish indigenous to that waterbody. A third 
pathway that has also been of concern in some circumstances is 
incidental ingestion of ambient water in conjunction with recreational 
uses. The derivation of an ambient water quality criterion for a 
pollutant entails the calculation of the maximum water concentration of 
that pollutant which ensures that drinking water exposures and/or fish 
consumption, as well as incidental ingestion, do not result in human 
intake of that pollutant in amounts that exceed a specified level based 
upon the toxicological endpoint of concern.
    There are many exposure topics and issues involved in the 
derivation of

[[Page 43791]]

AWQC. The first category includes several broad policy issues 
concerning the major objectives that the Agency believes should be met 
in setting AWQC. These issues include the following:
     Specifying which sources of exposure associated with ambient 
water should be explicitly included in the derivation of AWQC (e.g., 
Should drinking water be included in AWQC given that there may be 
separate national drinking water standards? Should AWQC be separate for 
drinking water exposure and fish consumption, or should they reflect 
combined exposure potential? Should there be an AWQC based on 
incidental water ingestion?)
     Identifying which segment or subgroup of the population AWQC 
should be designed to protect (e.g., Should the derivation be based on 
providing protection for individuals having average or ``typical'' 
exposures? Should it be based on protecting highly exposed individuals, 
or most sensitive individuals?)
    The second category includes determining whether nonwater sources 
of exposure (e.g., dietary, inhalation) should also be explicitly 
considered in the derivation of AWQC. (i.e., Should they be included 
when setting AWQC based on carcinogenicity as the toxicological 
endpoint? Should they be considered when setting AWQC based on an RfD 
for a noncarcinogenic endpoint? What specific procedures should be 
followed to account for the nonwater sources?)
    The third category of issues involves those that mainly address the 
selection of specific values for the exposure factors included in the 
AWQC derivation algorithms and which (for the most part) involve 
considerations independent of the particular method or procedure 
selected for deriving the criterion. These include such considerations 
as drinking water consumption rates, fish ingestion rates, and human 
body weight.
    The following sections present exposure issues relevant to the 
Draft AWQC Methodology Revisions, organized according to the three 
topics introduced above: policy issues are presented first, followed by 
the consideration of nonwater sources of exposure, and finally the 
factors used in AWQC computation. In relevant sections, comments 
provided from the SAB in its August 1993 review of the AWQC methodology 
are presented and discussed.
    The TSD presents suggested sources of contaminant concentration and 
exposure intake information, in addition to some suggestions of survey 
methods for obtaining and analyzing exposure data, necessary for 
setting AWQC. The following topics are also addressed in the TSD 
accompanying this Notice regarding exposure assessments for the AWQC: 
evaluating available exposure data; describing highly exposed 
subpopulations; distinguishing between major and minor exposure 
sources; comparing exposures to RfD values; addressing uncertainty and 
variability of the estimate; the question of current and future uses of 
the chemical; considering chemical and physical properties; and 
addressing unquantifiable exposures via an allocation ceiling.
1. Policy Issues
    The following discussions are qualitative in nature and are 
discussed in greater detail in Section C.3., Factors Used in the AWQC 
Computation.
    (a) Identifying the Population Subgroup that the AWQC Should 
Protect. The AWQC criteria are derived to establish ambient 
concentrations of chemicals which, if not exceeded, will protect the 
general population from adverse health impacts from that chemical due 
to consumption of aquatic organisms and water, including incidental 
water consumption related to recreational activities. For each 
chemical, chronic criteria are derived to reflect long-term consumption 
of food and water. An important decision to make when setting AWQC is 
the choice of the particular population to protect. For instance, the 
criteria might be set to protect those individuals who have average or 
``typical'' exposures, or the criteria could be set so that they offer 
greater protection to those individuals who are more highly exposed 
(e.g., subsistence fishers). EPA has selected default assumptions that 
are representative of the defined populations being addressed. These 
defined populations are: adults in the general population; sport 
(recreational) fishers; subsistence fishers; women of childbearing age 
(defined as ages 15-44); and children. In deciding on default 
assumptions, EPA is aware that multiple assumptions are used in 
combination (e.g., intake rate and body weight). In the section on the 
exposure factors used in the AWQC computations, EPA describes the 
populations that are represented by the different exposure intake 
assumptions. EPA recommends that priority should be given to 
identifying and adequately protecting the most highly exposed 
population. In carrying out regulatory actions under its statutory 
authorities, including the CWA, EPA's risk management goal is to 
establish criteria that are protective of human health and generally 
views that an upper-bound incremental cancer risk in the range of 
10-5 to 10-6 achieves this goal. EPA also 
considers that the goal is satisfied if the population as a whole will 
be adequately protected by human health criteria when the criteria are 
met in ambient water. As stated previously in Appendix II, Section A, 
EPA is proposing criteria at the 10-6 risk level. However, 
States and Tribes should have the flexibility to develop criteria, on a 
site-specific basis, that provides additional protection appropriate 
for highly exposed populations. EPA understands that highly exposed 
populations may be widely distributed geographically throughout a given 
State and Tribal area. Thus, if the State or Tribe determines that a 
highly exposed population would not be adequately protected by criteria 
based on the general population, EPA recommends that the State/Tribe 
adopt more stringent criteria. Furthermore, EPA recommends that States 
and Tribes ensure that the most highly exposed populations not exceed a 
risk level of 10-4. EPA acknowledges that at any given risk 
level for the general population, those segments of the population that 
are more highly exposed face a higher relative risk. For example, if 
fish are contaminated at a level permitted by criteria that are derived 
based on a risk level of 10-6, individuals consuming up to 
10 times the assumed fish consumption rate would still be protected at 
a 10-5 risk level.
    For RfD-based chemicals, EPA's policy is that, in general, the RfD 
should not be exceeded (see discussion in Section B.3.b on the RfD 
range) and that the exposure assumptions used should reflect the 
population of concern. It is recommended that when setting waterbody-
specific AWQC, States and Tribes should consider the populations most 
exposed via water and fish.
    (b) Appropriateness of Including the Drinking Water Pathway in 
AWQC. Under the 1980 AWQC National Guidelines, the derivation of AWQC 
for the protection of human health accounted for potential human 
exposure via both consumption of drinking water and ingestion of fish. 
During the 1992 Workshop, there was discussion regarding the need to 
include drinking water consumption as a factor in calculating AWQC for 
surface waters. The principal argument presented against the explicit 
inclusion of drinking water consumption is that most drinking water, 
and almost all drinking water obtained from surface water sources, is 
treated prior to its distribution to consumers. That is, the

[[Page 43792]]

direct ingestion of untreated ambient water is extremely rare and, 
therefore, direct ingestion of water should only be taken into account 
in setting AWQC when it is a significant route of exposure for a 
population of concern. However, the majority opinion from the 1992 
workshop was that direct ingestion is relevant to the AWQC (for the 
reasons stated below).
    EPA recommends continuing to include the drinking water exposure 
pathway explicitly in deriving AWQC for the protection of human health 
where drinking water is a designated use, for the following reasons: 
(1) drinking water is a designated use for surface waters under the CWA 
and, therefore, criteria are needed to assure that this designated use 
can be maintained; (2) although rare, there are some public water 
supplies that provide drinking water from surface water sources without 
treatment; (3) even among the majority of water supplies that do treat 
surface waters, existing treatments may not necessarily be effective 
for reducing levels of particular contaminants; (4) in consideration of 
the Agency's goals of pollution prevention, ambient waters should not 
be contaminated to a level where the burden of achieving health 
objectives is shifted away from those responsible for pollutant 
discharges and placed on downstream users to bear the costs of upgraded 
or supplemental water treatment.
    (c) Relationship Between Human Health AWQC and Drinking Water 
Standards. In conjunction with the preceding issue, EPA has also given 
consideration to whether there should be an equivalency between the 
drinking water component of AWQC and either MCLGs or MCLs promulgated 
under the SDWA.
    Under the SDWA, MCLGs are established as health-based goals without 
explicit consideration of either the costs or technological feasibility 
of achieving those goals. MCLs are then set as close to the MCLGs as 
possible, taking costs of the drinking water treatment technologies and 
the availability of analytical methodologies into account. Because MCLs 
are based in part on cost and technology considerations, they are not 
considered counterparts to AWQC for the protection of human health. As 
strictly health-based goals, however, MCLGs and AWQC for the protection 
of human health are highly analogous. There are some states that have 
utilized MCLGs as human health water quality criteria under the CWA.
    The application of the health goals set under the SDWA is quite 
different from the application of goals set under the CWA. Under the 
SDWA, the MCLGs (and MCLs) apply to the chemical concentration in 
distributed tap water, whereas under the CWA, AWQC are used to develop 
State or Tribal standards, which are then used with water transport 
models to derive permit limits for point source discharges. Because the 
water transport model uses protective assumptions which provide a 
margin of safety (such as 30-year, low-flow rates), it is generally 
unlikely that the water column concentration will be as high as the 
AWQC concentration limit for an extended period of time.
    In some cases, MCLs or MCLGs are more stringent than AWQC. In other 
cases, AWQC are more stringent than the drinking water MCLs or MCLGs. 
The reason is that the methodology used for deriving drinking water 
levels is different than the methodology used for deriving AWQC. 
Although both methods predominantly use the same reference dose or 
cancer risk assessment, and both methods assume a 70 kg adult and 
consumption of 2 liters of water per day, there are several important 
risk management differences. One difference is that MCLGs for chemicals 
that are known or likely carcinogens have usually been set equal to 
zero, while AWQC for carcinogens are based on an incremental cancer 
risk level. For chemicals with limited evidence of carcinogenicity 
(classified as C, possible carcinogen, under the 1986 Cancer 
Guidelines), the MCLG is usually based on the chemical's reference dose 
for noncancer effects with the application of an additional uncertainty 
factor of 1 to 10 to account for its possible carcinogenicity. The 1980 
AWQC guidelines do not differentiate among carcinogens with respect to 
the weight-of-evidence grouping; all were derived based on lifetime 
carcinogenic risk levels. Another difference is that a single 
determined risk value (i.e., within the range of 10-4 to 
10-6) is selected in setting risk-based MCLs, while AWQC 
have been derived by providing incremental risk levels spanning 
10-5 to 10-7 (i.e., three values were presented). 
Different numerical values between the two may also be due to the 
information that each criterion is based on at the time of development. 
That is, criteria developed at different times for the same chemical 
may be based on different exposure data and, perhaps, different 
toxicity studies. However, the principal difference is in the approach 
to accounting for exposure sources, including the fact that AWQC are 
based on a prediction of exposure from fish and shellfish using a 
bioaccumulation factor for the individual chemical and a fish/shellfish 
consumption rate. With the current MCLG methodology, bioaccumulation 
factors have not been used in the exposure estimates and fish/shellfish 
consumption rates have not been fully accounted for. Additionally, 
MCLGs for RfD-based chemicals developed under the SDWA follow a 
relative source contribution (RSC) approach in which the percentage of 
exposure that is attributed to drinking water is determined relative to 
the total exposure from all sources (e.g., drinking water, food, air). 
The rationale for this approach is to ensure that an individual's total 
exposure to a chemical does not exceed the RfD. Although the 1980 AWQC 
guidelines recommended taking non-fish dietary sources and inhalation 
into account, data on these other sources were generally not available. 
Therefore, it was typically assumed that an individual's total exposure 
to a chemical came solely from drinking water from the water body and 
consumption of fish and shellfish living in the water body. Lastly, as 
stated previously, when an MCL is adjusted based on cost or 
availability of treatment technology or analytical methods, then the 
MCL may become much less stringent than the AWQC, regardless of the 
exposure assumptions or toxicological basis.
    The SAB, in its 1993 review of EPA's preliminary recommendations, 
commented that there would be difficulties in using the concept of 
drinking water MCLGs for setting AWQC. The SAB was concerned about the 
possible introduction of the zero MCLG concept into the methodology for 
deriving AWQC. The SAB was also concerned that AWQC are considerably 
different from MCLGs, and that developing AWQC that are different from 
MCLGs may be reasonable in certain specific cases (e.g., for 
disinfectant byproducts). EPA's proposed methodology addresses the 
specific concerns that the SAB has raised regarding the incorporation 
of the zero MCLG concept.
    The Agency believes that for a given pollutant, the drinking water 
component of an AWQC should be consistent with the MCLG that has been 
established for that substance (if one has been developed) and, 
therefore, proposes to use similar assessment methodologies for 
deriving AWQC and MCLGs. EPA stated its policy on the use of Section 
304(a) human health criteria (i.e., the AWQC) versus MCLs in 45 FR 
79318, November 28, 1980. Additionally, a memorandum from R. Hanmer to 
the Regional Water

[[Page 43793]]

Management Division Directors dated December 12, 1988, provided 
detailed guidance with regard to this policy. Specifically, for the 
protection of public water supplies, EPA encouraged the use of MCLs. 
When fish ingestion is considered an important activity, EPA 
recommended the use of AWQC to protect human health. In all cases, if 
an AWQC did not exist for a chemical, an MCL was deemed a suitable 
level of protection. EPA is now recommending a slightly different 
approach. Although the use of MCLs is acceptable in the absence of 
304(a) criteria, EPA is recommending that MCLs only be used when they 
are numerically the same as the MCLG and only when the sole concern is 
the protection of public water supply sources and not the protection of 
the CWA section 101(a) goal regarding fish consumption (e.g., where the 
chemically toxic form in water is not the form found in fish tissue 
and, therefore, fish ingestion exposure is not an issue of concern). 
Where consideration of available treatment technology, costs, or 
availability of analytical methodologies has resulted in MCLs that are 
less protective than MCLGs or AWQC, States and Tribes should consider 
using MCLGs and/or health-based AWQC to protect water uses. Where fish 
consumption is an existing or potential activity, States and Tribes 
should ensure that their adopted human health criteria adequately 
address this exposure route. When fish consumption is a use, EPA 
recommends development of AWQC due to the fact that fish consumption 
and bioaccumulation are explicitly addressed. In all cases, AWQC should 
be set to ensure that all routes of exposure have been considered. EPA 
believes if water monitored at existing drinking water intakes has 
concentrations at or below MCLGs, then the water could be considered to 
meet a designated use under the CWA as a drinking water supply. In 
situations where a 304(a) criterion was less protective than an MCL, it 
is advisable to use the MCL as the criterion for segments designated as 
drinking water supplies. For carcinogens where the MCLG is equal to 
zero, States are encouraged to base an AWQC at the drinking water 
intake on an acceptable cancer risk level (i.e., a level within the 
range of 10-4 to 10-6), to promote pollution 
prevention and anti-degradation.
    (d) Setting Separate AWQC for Drinking Water and Fish Consumption. 
In conjunction with the issue of the appropriateness of including the 
drinking water pathway explicitly in the derivation of AWQC for the 
protection of human health, there has been discussion of whether these 
AWQC should be single values that account for potential exposure from 
drinking water and fish consumption together, or whether it is more 
appropriate to calculate separate AWQC explicitly for each pathway. One 
of the factors considered has been that setting separate criteria could 
provide a more straightforward means of developing AWQC for the 
drinking water pathway that would be consistent with MCLG development.
    The 1980 AWQC National Guidelines used the approach of setting a 
single AWQC accounting for both drinking water and fish consumption, as 
well as a separate AWQC based on ingestion of aquatic organisms alone. 
This latter criterion was intended to apply in those cases where the 
designated uses of a waterbody include supporting fish or shellfish for 
human consumption, but not as a drinking water supply source (e.g., 
non-potable estuarine waters).
    Although the SAB recommended the use of separate criteria based on 
fish intake and water consumption, in the revised methodology, the 
Agency is recommending continuing the practice of setting AWQC that 
account for combined drinking water and fish consumption, as well as a 
separate criterion for fish/shellfish consumption alone. The reason for 
this is because most State and Tribal programs designate their waters 
to cover both uses.
    (e) Incidental Ingestion from Ambient Surface Waters. The 1980 AWQC 
National Guidelines did not include criteria to address incidental 
ingestion from recreational uses. As noted previously, there are cases 
where AWQC for the protection of human health do not include 
consideration of the waterbody as a source of potable water (e.g., 
estuaries). In these cases, criteria based only on fish ingestion (or 
aquatic life criteria) may not adequately protect recreational users 
from health effects resulting from incidental ingestion. In order to 
protect recreational users, EPA recommends including exposure resulting 
from incidental ingestion of water in those cases where the waterbody 
is not used for potable water. However, it should be noted that the SAB 
felt there was not a great need for incidental ingestion criteria for 
recreational uses where drinking water criteria are inapplicable (e.g., 
estuaries). The exposure factors section of this document (Appendix II, 
Section C.3.(c)) discusses incidental ingestion estimates for 
calculating both chronic and acute ingestion rates.
2. Consideration of Nonwater Sources of Exposure When Setting AWQC
    (a) Background. In the 1980 AWQC National Guidelines, different 
approaches for addressing nonwater exposure pathways were used in 
setting AWQC for the protection of human health depending upon the 
toxicological endpoint of concern. For those substances for which the 
appropriate toxic endpoint was linear carcinogenicity, only the two 
water sources (i.e., drinking water consumption and fish ingestion) 
were considered in the derivation of the AWQC. Nonwater sources were 
not considered explicitly. In the case of linear carcinogens, the AWQC 
is being determined with respect to the incremental lifetime risk posed 
by a substance's presence in water, and is not being set with regard to 
an individual's total risk from all sources of exposure.
    In the case of substances for which the AWQC is set on the basis of 
a nonlinear carcinogen or a noncancer endpoint where a threshold is 
assumed to exist, nonwater exposures were to be considered when 
deriving the AWQC under the 1980 AWQC National Guidelines. In effect, 
the 1980 AWQC National Guidelines specified that the AWQC be calculated 
to account for no more than that portion of the ADI that remains after 
contributions from other expected sources of exposure have been 
subtracted out. The ADI is equivalent to the RfD, which is discussed in 
Appendix II, Section B of this Notice. The rationale for this approach 
has been that for pollutants exhibiting threshold effects, the 
objective of the AWQC is to ensure that an individual's total exposure 
does not exceed that threshold level.
    It is useful to note that while the 1980 AWQC National Guidelines 
recommended taking non-fish dietary sources and inhalation into account 
in setting the AWQC for threshold contaminants, in practice the data on 
these other sources were generally not available and, therefore, the 
AWQC usually were derived such that they accounted for all of the ADI 
(RfD). When the 1980 AWQC National Guidelines were published, EPA noted 
that the inability to estimate intake from non-fish dietary sources and 
inhalation, as well as the wide variability that may exist in such 
exposures, would add to the uncertainty in the criteria derivation. EPA 
also noted in the 1980 AWQC National Guidelines that in terms of 
scientific validity, the accurate estimate of the ADI (RfD) is the 
major

[[Page 43794]]

factor in the satisfactory derivation of AWQC.

    Note: In the drinking water MCLG methodology, noncarcinogenic 
criteria follow an RSC approach in which the percentage of exposure 
that is attributed to drinking water is determined relative to the 
total exposure from all sources (e.g., drinking water, food, air, 
soil). The rationale for this approach is to ensure that an 
individual's total exposure to a chemical does not exceed the 
reference dose.

    Given the inability to reasonably predict future changes in 
exposure patterns, the uncertainties in the exposure estimates due to 
both data inadequacy and possible unknown sources of exposure, as well 
as the potential for some populations to experience greater exposures 
than indicated by the available data, EPA believes that utilizing the 
entire RfD (or Pdp/SF) may not be adequately protective. Additionally, 
the uncertainties associated with the derivation of the RfD (or Pdp/SF) 
(e.g., limitations in the toxicity study, extrapolation from the study 
species to humans) are independent of the exposure assessment and the 
associated intake sources and intake uncertainties.
    If the AWQC are set so that the RfD or Pdp/SF (or some ceiling 
value less than either of these) is not exceeded after taking other 
sources of exposure into account, a procedure to consider the nonwater 
sources in the derivation of AWQC must be adopted.
    As discussed above, the 1980 AWQC National Guidelines did not 
account for nonwater sources when setting AWQC for those chemicals that 
were evaluated as carcinogens. The formula for setting the criterion 
for carcinogens was:
[GRAPHIC] [TIFF OMITTED] TN14AU98.015

Where:
C=The AWQC (mg/L)
70=human body weight (kg)
LR=lifetime cancer risk factor being used to set the criterion, 
generally in the range of 10-5 to 10-7
q\1\*=cancer slope factor in (mg/kg-day)-1
2=drinking water consumption (L/day)
0.0065=fish ingestion (kg/day)
R=bioconcentration factor (L/kg)

    As indicated by the above equation, if the lifetime risk value (LR) 
in the above equation is 10-6, then the value computed for C 
is the water concentration that would be expected to increase an 
individual's lifetime risk of carcinogenicity from exposure to the 
particular pollutant by no more than one chance in one million, 
regardless of the additional lifetime cancer risk due to exposure, if 
any, to that particular substance from other sources.
    For noncarcinogens for which nonwater exposures were to be 
considered, however, the 1980 methodology included the following 
general formula for setting the criterion:
[GRAPHIC] [TIFF OMITTED] TN14AU98.016

Where:

C=The criterion (mg/L)
ADI=Acceptable daily intake (mg), developed as a dose specifically for 
a 70 kg adult (replaced by the use of Reference Dose (RfD) in units of 
mg/kg-day, as discussed in Appendix II, Section B of this document)
DT=Non-fish dietary intake (mg/kg-day)
IN=Inhalation intake (mg/kg-day)

    The other elements are the same as for the cancer-based formula, 
above. As indicated by the above equation, the 1980 AWQC National 
Guidelines used a ``subtraction'' approach to account for nonwater 
exposure sources when calculating AWQC for noncarcinogenic, threshold 
pollutants. That is, the amount of the ADI (RfD) ``available'' for 
water sources was determined by first subtracting out contributions 
from nonwater sources. A similar subtraction approach was used, albeit 
inconsistently, in the derivation of drinking water MCLG values in the 
early and mid-1980's; along with a percentage method. More recently, 
the approach used in the drinking water program has been to determine 
the MCLGs exclusively by the percentage method. To foster meeting the 
objective noted earlier of establishing consistency in deriving MCLGs 
and the drinking water component of AWQC, EPA would like to use the 
same approach for both MCLGs and AWQC.
    There has been some discussion of whether it is, in fact, necessary 
in most cases to explicitly account for other sources of exposure when 
computing the AWQC for pollutants exhibiting threshold effects. It has 
been argued that because of the conservative assumptions generally 
incorporated in the calculation of reference doses used as the basis 
for the AWQC derivation, total exposures slightly exceeding the RfD are 
unlikely to produce adverse effects. It could be argued, therefore, 
that reducing AWQC by accounting for other exposure sources relative to 
what they would be if they were derived from the full RfD value 
provides little or no actual additional risk reduction.
    In its report, SAB's Drinking Water Committee did not feel that it 
is appropriate to develop AWQC geared to ensure that the sum of all 
theoretically possible exposures never exceeds the RfD by even a small 
amount. The Committee rejected the routine use of the percentage or 
subtraction methods for the allocation of the RfD, and the use of 
default values in the absence of reliable exposure data. They also 
expressed concern that EPA could ``focus intense regulatory attention 
on insignificant problems, thus wasting scarce resources'' if 
``compensat[ion] for other routes of exposure'' was attempted. (For the 
complete discussion, refer to SAB, 1993.)
    Instead, the Committee endorsed the recommendation from the AWQC 
Workshop held by the Agency in 1992 which calls for bringing together 
knowledgeable individuals from all the appropriate offices or agencies 
for discussions when significant contributions to exposure are expected 
from multiple sources, and the total of those contributions exceeds the 
RfD. For certain chemicals (e.g., dioxin, mercury), EPA has coordinated 
efforts throughout the Agency. However, such extensively coordinated 
efforts may prove to be impractical on a routine basis. It is 
reasonable that the initially developed assessments and proposed 
criteria, including proposals for RfD allocation, could be circulated 
for

[[Page 43795]]

comments and input from staff of the appropriate offices or agencies.
    However, the SAB also stated that apportionment can be attempted 
when data are available. When total exposures are below the RfD, SAB 
suggested that EPA's goal should be to develop criteria ``to ensure 
that a problem does not develop in the future.'' Yet, they made no 
specific suggestions on how to achieve this goal. For situations when 
exposures may exceed the RfD, the SAB stated that ``it is unlikely that 
exposure of any populations to doses slightly over the RfD (even up to 
twice the RfD) would produce significant health effects.'' However, 
they seem to contradict this by advising that ``if total exposures are 
at or higher than the RfD, then remedial actions may need to be 
considered.'' EPA disagrees with the idea that the conservative way in 
which the RfD is calculated automatically makes it unlikely that 
populations would experience ``significant health effects'' from 
exposures greater than the RfD. RfDs are not all equivalent in their 
derivation, and EPA believes multiple route exposures may be 
particularly important when the uncertainty factors associated with the 
RfD are small. Furthermore, the opinion that unless ``total exposures 
[are] significantly in excess of the RfD, exposure from other routes 
should be neglected in calculations of AWQC'' is counter to strong 
Agency directives to routinely consider and account for all routes of 
exposure when setting health-based criteria and with consideration to 
other regulatory activities. Despite arguments raised by SAB, EPA is 
recommending that only a portion of the RfD (or Pdp/SF) be used in 
setting AWQC in order to account for other sources of exposure. EPA is 
also considering whether toxicity information (such as uncertainty 
factors, severity of effects, essentiality, possible additive/
synergistic effects) should be considered in allocating the RfD or Pdp/
SF. While combined exposures above the RfD or Pdp/SF may or may not be 
an actual health risk, a combination of health criteria exceeding the 
RfD or Pdp/SF may not be sufficiently protective. Therefore, EPA 
recommends routinely accounting for all sources and routes of non-
occupational exposure when setting AWQC. EPA believes that maintaining 
total exposure below the RfD (Pdp/SF) is a reasonable health goal and 
that there are circumstances where health-based criteria for a chemical 
should not exceed the RfD (Pdp/SF), either alone (if only one criterion 
is relevant, along with other intake sources considered as background 
exposures) or in combination.
    EPA has considered several alternative approaches to account for 
nonwater sources and to resolve past inconsistencies in setting 
criteria. Specifically, the Agency's Relative Source Contribution 
Policy Workgroup has considered six alternatives:
     Exposure Decision Tree Approach;
     Subtraction Approach;
     Percentage Approach;
     Tiered Approach;
     Safety Factor Approach; and
     Margin of Safety Approach.14
---------------------------------------------------------------------------

    \14\ This term refers to a method for accounting for nonwater 
sources of exposure and should not be confused with the nonlinear 
cancer assessment approach known as Margin of Exposure.
---------------------------------------------------------------------------

    The Workgroup discussed, during the series of meetings, the various 
approaches to evaluating human exposure for regulatory and other risk 
assessment activities. Each approach has advantages and disadvantages 
that were discussed at length during these meetings, as do the basic 
concepts surrounding the subtraction and percentage methods of 
accounting for relevant exposures when allocating an RfD (Pdp/SF). The 
other four approaches are variations on the fundamental concepts of the 
subtraction or the percentage approaches.
    Each of these six approaches is discussed in detail in a separate 
document contained in the public docket for this proposal (Borum, 
unpublished). The Agency recommends the Exposure Decision Tree Approach 
as described below. More detailed discussion and an example of how the 
Exposure Decision Tree is implemented are presented in the TSD.
    As will become clear when reading the Exposure Decision Tree 
Approach, a typical evaluation will likely involve multiple sources/
pathways of exposure and may involve more than one health-based 
criterion (either existing or in consideration for development). The 
current EPA policy discussions include the potential for applying this 
approach to other program offices to the extent practicable when 
conducting exposure assessments. As such, the broader goals are to 
ensure more comprehensive evaluations of exposure Agencywide and 
consistent allocations of the RfD (Pdp/SF) for criteria-setting 
purposes when appropriate.
    (b) Exposure Decision Tree Approach. The Exposure Decision Tree 
approach allows flexibility in the RfD (Pdp/SF) allocation among 
sources of exposure. When adequate data are available they are used to 
make accurate exposure predictions for the population(s) of concern. 
When this is not possible, a series of qualitative alternatives is 
proposed using less adequate data or default assumptions that allow for 
the inadequacies of the data while protecting human health. The 
decision tree allows for use of both subtraction and percentage methods 
of accounting for other exposures, depending on whether one or more 
health criterion is relevant for the chemical in question. The 
subtraction method is considered acceptable when only one criterion is 
relevant for a particular chemical. In these cases, other sources of 
exposure can be considered ``background'' and can be subtracted from 
the RfD (Pdp/SF). When more than one criterion is relevant to a 
particular chemical, apportioning the RfD (Pdp/SF) via the percentage 
method is considered appropriate to ensure that the combination of 
criteria, and thus the potential for resulting exposures, do not exceed 
the RfD (Pdp/SF). The decision tree (with numbered boxes) is shown in 
Figure IIIC-1. The underlying objective is to maintain total exposure 
below the RfD (Pdp/SF) while avoiding an extremely low limit in a 
single medium that represents just a fraction of the total exposure. To 
meet this objective, all proposed numeric limits lie between 80 percent 
and 20 percent of the RfD (Pdp/SF). EPA recommends use of the decision 
tree approach but also recognizes that departures from the approach may 
be appropriate in certain cases. The Agency endorses such action as 
long as reasons are given as to why it is not appropriate to follow the 
decision tree approach as long as the steps taken to evaluate the 
potential sources and levels of exposure are clearly indicated.
    The first step in the decision process, problem formulation, is to 
identify the population(s) of concern (Box 1) and identify the relevant 
exposure sources and pathways (Box 2). The second step is to identify 
what data are available and whether they are adequate for calculating 
exposure estimates (Box 3). The term ``data,'' as used here and 
discussed throughout the document, refers to ambient sampling data 
(from Federal, regional, State or area-specific studies) and not 
internal human exposure measurements. The adequacy of data is a 
professional judgment for each individual chemical of concern, but EPA 
recommends that the minimum acceptable data for Box 3 are exposure 
distributions that can be used to determine, with an acceptable 95 
percent confidence interval, the central tendency and high-end exposure 
levels for each source. Once the two initial steps are complete, the 
next step

[[Page 43796]]

depends on the type and quantity of data available.

BILLING CODE 6560-50-P

[[Page 43797]]

[GRAPHIC] [TIFF OMITTED] TN14AU98.017



BILLING CODE 6560-50-C

[[Page 43798]]

    If adequate data are available to describe the central tendencies 
and high-end levels from each exposure source/pathway, the levels of 
exposure are compared to the RfD or Pdp/SF (Box 11). If the levels of 
exposure for the chemical in question are not near (currently defined 
as greater than 80 percent), at, or in excess of the RfD (Pdp/SF), then 
a determination is made (Box 13) as to whether there is more than one 
regulatory action relevant for the given chemical (i.e., more than one 
criterion, standard or other guidance being planned, performed or in 
existence for the chemical).
    If the action under consideration is the sole action (i.e., 
multiple criteria, etc. are not relevant), then the recommended method 
for setting a health-based criterion is to use a subtraction 
calculation (Box 14). The criterion is the result after the appropriate 
intake levels from all other sources have been subtracted from the RfD 
(Pdp/SF). In addition, there is a ceiling on the amount of the RfD 
(Pdp/SF) available for allocation. This ceiling, 80 percent of the RfD 
(Pdp/SF), is to provide adequate protection for individuals whose total 
exposure to a contaminant is, due to any of the exposure sources, 
higher than currently indicated by the available data. This also 
increases the margin of safety to account for possible unknown sources 
of exposure. There is also a floor of 20 percent to prevent a de 
minimis exposure allocation in a particular medium.
    If more than one regulatory action is relevant (as described 
above), then the recommended method for setting health-based criteria 
is to allocate the RfD (Pdp/SF) among those sources for which health-
based criteria are being set (Box 15). Two main options for allocating 
the RfD (Pdp/SF) are presented in this Box. Option 1 for allocation is 
the percentage approach (with a ceiling and floor). This option simply 
refers to the percentage of overall exposure contributed by an 
individual exposure source. That is, if for a particular chemical, 
drinking water were to represent half of total exposure and diet were 
to represent the other half, then the drinking water contribution 
(known as the ``relative source contribution'' or RSC) would be 50 
percent. The health-based criterion would, in turn, be set at 50 
percent of the RfD (Pdp/SF).
    This option also uses an appropriate combination of intake values 
for each exposure source based on the variability in occurrence levels 
and determined on a case-by-case basis. Option 2 would involve 
subtracting from the RfD (Pdp/SF) the exposure levels from all sources 
of exposure and apportioning the free space among those sources for 
which health-based criteria are being set. There are several ways to do 
this: (1) Divide the free space among the sources with preference given 
to the source likely to need the most increase (e.g., because of 
intentional uses or because of physical/chemical properties like 
solubility in water, etc.); (2) Divide the free space in proportion to 
the ``base'' amount used (e.g., the source accounting for 60 percent of 
exposure gets 60 percent of the free space--this is identical to the 
percentage method; the outcome is the same); and (3) Divide the free 
space based on current variability of exposure from each source (i.e., 
such that more free space is allocated to the source that varies the 
most). The resulting criterion would then be equal to the amount of 
free space allocated plus the amount subtracted for that source.
    If the levels of exposure for the chemical in question are near 
(again, currently defined as greater than 80 percent), at, or in excess 
of the RfD (Pdp/SF), then the estimates of exposures and related 
uncertainties, potential allocations, toxicity-related information, 
control issues, and other information will be presented to managers for 
a decision (Box 12). The high levels referred to in Box 11 may be due 
to a single dominant source or to a combination of sources. The 
estimates of exposure performed in these instances and any allocations 
made would be done as described above for Boxes 13, 14, and 15. 
However, because exposures that approach or exceed the RfD (Pdp/SF) and 
the feasibility of controlling different sources of exposure are 
complicated issues, risk managers will need to be directly involved in 
formulating any allocation decisions.
    If the data fail the adequacy test (Box 3), any limited data that 
are available are evaluated (Box 4). This includes information about 
the chemical/physical properties, uses, environmental fate and 
transformation, limited sampling data that did not fulfill the 
requirements of Box 3, as well as any other information that would 
characterize the likelihood of exposure from various media for the 
chemical and aid in making a qualitative determination regarding the 
relation of one exposure source to another. Because these data are less 
certain (i.e., include information that does not directly measure 
exposure, or very limited data), criteria based on this information 
should be more conservative as shown in the remainder of the decision 
tree.
    If there are not sufficient data/information to give any 
characterization of exposure, then it may be best to defer action on 
the chemical until better information becomes available (Boxes 5 & 6). 
If this is not possible, then the ``default'' assumption of 20 percent 
of the RfD or Pdp/SF (Box 7) should be used, which has been used in 
past Agency water program regulations.
    If there are sufficient data to give a characterization of 
exposure, the RfD (Pdp/SF) allocation depends on whether there are 
other known or potential uses or sources of concern (Box 8). If the 
source of concern is the sole source then EPA recommends an allocation 
of 50 percent of the RfD or Pdp/SF (Box 9). If there are multiple 
sources of concern and some information is available on each (Box 10A), 
the procedure, as shown in Box 10C, is the same as that in Box 14 or 
Box 15 depending on whether one or more criterion is relevant, but with 
a 50 percent ceiling to account for uncertainties from the limited 
amount of data (compared to Box 3). As with Box 11, if a determination 
is made in Box 10A (i.e., if information is available) that exposures 
are near, at or above the RfD (or Pdp/SF) based on the available 
information, the allocations made need to be presented to risk managers 
for decision. If information is lacking on some of the multiple 
exposure sources then EPA would use an allocation of 20 percent of the 
RfD or Pdp/SF (Box 10B).
    (c) Quantification of Exposure. When selecting contaminant 
concentration values in environmental media and exposure intake values 
for the Relative Source Contribution (RSC) analysis, it is important to 
realize that each value selected (including those intakes recommended 
as default assumptions in the AWQC equation) is associated with a 
distribution of values for that parameter. Determining how various 
subgroups fall within the distributions of overall exposure and how the 
combination of exposure variables defines what population is being 
protected is a complicated and, perhaps, unmanageable task, depending 
on the amount of information available on each exposure factor 
included. Many times, the default assumptions used in EPA risk 
assessments are derived from the evaluation of numerous studies and are 
generally considered to represent a particular population group or some 
national average. Therefore, describing with certainty the exact 
percentile of a particular population that is protected with a 
resulting criteria is often not possible.
    General recommendations for selecting values to be used in exposure 
assessments for both individual and population exposures are discussed 
in

[[Page 43799]]

EPA's Guidelines for Exposure Assessment (USEPA 1992). The ultimate 
choice of the contaminant concentration values used in the RSC estimate 
and the exposure intake rates requires the use of professional 
judgment. This is discussed in greater detail in the TSD (Section 
2.3.3).
    (d) Inclusion of Inhalation and Dermal Exposures From Household 
Drinking Water Uses. A number of drinking water contaminants are 
volatile and thus diffuse from water into the air where they may be 
inhaled. In addition, drinking water is used for bathing and, thus, 
there is at least the possibility that some contaminants in water may 
be dermally absorbed.
    Volatilization may increase exposure via inhalation and decrease 
exposure via ingestion and dermal absorption. The net effect of 
volatilization and dermal absorption upon total exposure to volatile 
drinking water contaminants is unclear. Although several approaches can 
be found in the literature, including various models that have been 
used by EPA, the Agency currently does not have a recommended 
methodology for explicitly incorporating inhalation (i.e., from 
volatilization) and dermal absorption exposures from household water 
uses in the derivation of health-based criteria. However, the Agency is 
supporting research in this area.
    (e) Inclusion of Inhalation Exposures in RSC Analysis. The type and 
magnitude of toxicity produced may differ between routes; that is, the 
route of exposure can impact the effective concentration of a chemical 
and can also change the toxicity. For example, an inhaled chemical such 
as hydrogen fluoride may produce local effects upon the lung that are 
not observed (or only observed at much higher doses) when the chemical 
is administered orally. Also, the active form of a chemical (and 
principal toxicity) can be the parent compound and/or one or more 
metabolites. With this Methodology, EPA recommends that differences in 
absorption and toxicity by different routes of exposure be determined 
and converted to reflect the differences in bioavailability and applied 
to the exposure assessment. EPA acknowledges that the issue of whether 
the doses received from inhalation and ingestion exposures are 
cumulative (i.e., toward the same threshold of toxicity) is 
complicated. Such a determination involves evaluating the chemical's 
physical characteristics, speciation and reactivity. A chemical may 
also exhibit different metabolism by inhalation versus oral exposure 
and may not typically be metabolized by all tissues. In addition, a 
metabolite may be much more or much less toxic than the parent 
compound. Certainly with a systemic effect, if the chemical enters the 
bloodstream, then there is some likelihood to contact the same target 
organ. Attention also needs to be given to the fact that both the RfD 
and RfC are derived based on the administered level. Toxicologists 
generally believe that the effective concentration of the active form 
of a chemical(s) at the site(s) of action determines the toxicity. If 
specific differences between routes of exposure are not known, it may 
be reasonable to assume that the internal concentration at the site 
from any route contributes as much to the same effect as any other 
route. A default of assuming equal absorption has often been used. 
However, for many of the chemicals that the Agency has reviewed, there 
is a substantial amount of information already known to determine 
differences in rates of absorption. For example, absorption, in part, 
is a function of blood solubility (i.e., Henry's Constant) and better 
estimations than the default can be made.
    The RSC analyses that accompany these proposed Methodology 
revisions include consideration of inhalation exposures. Comment is 
requested on whether this is a reasonable approach to accounting for 
exposures for setting AWQC. Even if different target organs are 
involved between different routes of exposure, a conservative policy 
may be appropriate to keep all exposures below a certain level. One 
suggestion is to set allowable levels (via an equation) such that the 
total of ingestion exposures over the ingestion RfD in addition to the 
total of inhalation exposures over the inhalation RfC is not greater 
than 1 (Note: the RfD is typically presented in mg/kg-day and the RfC 
is in mg/m\3\).
    (f) Bioavailability of Substances from Different Routes of 
Exposure. For many chemicals, the rate of absorption can differ 
substantially from ingestion compared to inhalation. There is also 
available information for some chemicals which demonstrates appreciable 
differences in gastrointestinal absorption depending on whether the 
chemical is ingested from water, soil, or food. For some contaminants, 
plant and animal food products may also have appreciably different 
absorption rates. Regardless of the allocation approach used, EPA 
recommends using existing data on differences in bioavailability 
between water, air, soils, and different foods when estimating total 
exposure for use in allocating the RfD or Pdp/SF. The Agency has 
developed such exposure estimates for cadmium (USEPA, 1994). In the 
absence of data, EPA will assume equal rates of absorption from 
different routes and sources of exposure.
    (g) Consideration of Non-water Exposure Procedures for 
Noncarcinogens, Linear Carcinogens, and Nonlinear Carcinogens. In the 
revised methodology, EPA recommends continuing to use the incremental 
risk approach that does not consider other exposure sources explicitly 
when setting AWQC for linear carcinogens. EPA recommends continuing to 
consider other exposure sources in setting AWQC for threshold 
toxicants, including both noncarcinogens and nonlinear carcinogens. 
Nonlinear carcinogens are discussed in detail in Appendix II, Section 
A.
3. Factors Used in the AWQC Computation
    This section presents values for several exposure factors that are 
currently used in the derivation of AWQC. A new factor being considered 
by EPA, incidental ingestion from surface water, is also discussed in 
this Section, with a suggested default value.
    When choosing exposure factors to include in the derivation of a 
criterion for a given pollutant, EPA recommends considering exposure 
factors relevant to populations that are most susceptible to that 
pollutant. In addition, highly exposed individuals should be considered 
when setting criteria. In general, exposure factors specific to adults 
and relevant to lifetime exposures are the most appropriate exposure 
factors to consider when determining criteria to protect against 
effects from long-term exposure. However, infants and children have a 
higher rate of water and food consumption per body weight compared to 
adults and also may be more susceptible to some pollutants than adults 
(USEPA, 1997c). In addition, exposure by pregnant women to certain 
toxic chemicals may cause developmental effects in the fetus (USEPA, 
1997c). Exposures resulting in developmental effects may be of concern 
for some contaminants and should be considered along with data 
applicable to long-term health effects when setting AWQC. (See Section 
B for further discussion of this issue.) Short-term exposure may 
include multiple or continuous exposures occurring over a week or so. 
Exposure factors relevant for considering chronic toxicity as well as 
exposure factors relevant for short-term developmental exposure 
concerns that could result in adverse health effects are discussed in 
the Sections below. States and Tribes may choose to develop criteria 
for developmental health effects

[[Page 43800]]

based on exposure factors specific to children or to women of 
childbearing age.
    EPA believes that the recommended exposure factor default intakes 
for adults with chronic exposure situations are adequately protective 
of the population over a lifetime. In providing additional exposure 
intake factors for women of childbearing age and children, EPA is 
providing flexibility for States and Tribes to establish criteria 
specifically targeted to provide additional protection to sensitive 
subpopulations (e.g., pregnant/nursing women, infants, children) or 
highly exposed subpopulations (e.g., sport anglers, subsistence 
fishers) using adjusted values for exposure parameters for body weight, 
drinking water intake, and fish consumption.
    Each of the following Sections recommends exposure parameters for 
use in developing AWQC. These are based on both science policy 
decisions that consider the best available data, as well as risk 
management judgments regarding the overall protection afforded by their 
choice in the derivation of AWQC.
    (a) Human Body Weight Values for Dose Calculations.
    (1) Rate Protective of Human Health from Chronic Exposure. The 1980 
AWQC National Guidelines assumed a body weight of 70 kg for derivation 
of AWQC. EPA recommends maintaining the default body weight of 70 kg 
for calculating AWQC as a representative average value for both male 
and female adults. As stated above, exposure factors specific to adults 
are recommended to protect against effects from long-term exposure. 
This value is based on the following information. In an analysis of the 
NHANES II (the second National Health and Nutrition Examination Survey) 
data base, the 10th, 25th, and 50th percentile values for female adults 
18-74 years old are 50.3, 55.4, and 62.4 kg, respectively (adapted from 
NCHS, 1987). For males in the same age range the comparable percentile 
values are 62.3, 68.7, and 76.9 kg, respectively. The mean body weight 
value for men and women ages 18 to 75 years old from this survey is 
71.8 kg (adapted from NCHS, 1987). The mean value for body weight for 
adults ages 20-64 years old from another survey which primarily 
measured drinking water intake is 70.5 kg (Ershow and Cantor, 1989). 
The revised EPA Exposure Factors Handbook (USEPA 1997a) recommends 71.8 
kg for adults, based on the NHANES II data. However, the Handbook also 
acknowledges the 70 kg value commonly used in EPA risk assessments and 
cautions assessors on the use of values other than 70 kg. Specifically, 
the point is made that the 70 kg value is used in the derivation of 
cancer slope factors and unit risks that appear in IRIS. Consistency is 
advocated between the dose-response relationship and exposure factors 
assumed.
    (2) Rates Protective of Developmental Human Health Effects. As 
noted above, pregnant women may represent a more appropriate population 
for which to assess exposure from chemicals in ambient waters in some 
cases, because of the potential for developmental effects in fetuses. 
In these cases, body weights representative of women of childbearing 
age may be appropriate to adequately protect offspring from such health 
effects. To determine a mean body weight value appropriate to this 
population, separate body weight values for women in individual age 
groups within the range of 15-44 years old, taken from NHANES II (NCHS, 
1987), were combined and weighted by current population percentages 
(U.S. Bureau of the Census, 1996) to obtain a value applicable to the 
current population. The resulting mean body weight value is 63.8 kg. 
Ershow and Cantor (1989) present body weight values specifically for 
pregnant women included in the survey; mean and median weights are 65.8 
and 64.4 kilograms, respectively. Ershow and Cantor (1989), however, do 
not indicate the ages of these pregnant women. Based on this 
information for women of childbearing age and pregnant women, States 
may wish to use the mean body weight value of 65 kg in cases where 
pregnant women are the specific population of concern and the chemical 
of concern exhibits reproductive and/or developmental effects (i.e., 
the critical effect upon which the RfD or Pdp/SF is based). Using the 
65 kg assumption would result in lower (more protective) criteria than 
criteria based on 70 kg.
    As discussed earlier, because infants and children have a higher 
rate of water and food consumption per body weight compared to adults, 
a higher intake rate per body weight factor may be needed when 
comparing estimated exposure doses with critical doses when RfDs are 
based on health effects in children. To calculate these intake rates 
relevant to such effects, the body weight of children should be used. 
As with the default body weight for pregnant women, EPA is not 
recommending the development of additional AWQC (i.e., similar to 
drinking water health advisories) that focus on acute or short-term 
effects since these are not seen routinely as having a meaningful role 
in the water quality criteria and standards program. However, there may 
be circumstances where the consideration of exposures for these groups 
is warranted. Although the AWQC are generally based on chronic health 
effects data, they are intended to also be protective with respect to 
adverse effects that may reasonably be expected to occur as a result of 
elevated shorter-term exposures. EPA acknowledges this as a potential 
course of action and is, therefore, recommending these default values 
for States and Tribes to utilize in such situations.
    EPA is recommending an assumption of 28 kg as a default body weight 
to calculate AWQC to provide additional protection for children when 
the chemical of concern indicates health effects in children are of 
predominant concern (i.e., test results show children are more 
susceptible due to less developed immune systems, neurological systems, 
and/or lower body weights). The value is based on the mean body weight 
value of 28 kilograms for children ages 0-14 years old, which combines 
body weight values for individual age groups within this larger group. 
The mean value is based on body weight information from NHANES II 
(NCHS, 1987) for individual-year age groups between 6 months and 14 
years old, and weights the values for these different ages by current 
population percentages (from U.S. Bureau of the Census, 1996) to 
represent a body weight value applicable to the current population of 
children aged 0-14 years. The same mean body weight of 28 kilograms is 
also obtained using body weight values from Ershow and Cantor (1989) 
for five age groups within this range of 0-14 years, and applying the 
above weighting method. The 28 kg assumption is also consistent with 
the estimated fish intake rates proposed for children in the same age 
range. Unfortunately, fish intake rates for finer age group divisions 
are not possible due to the limited sampling base from the fish intake 
survey; there is limited confidence in calculated values (e.g., the 
mean) for such fine age groups. Given this limitation, the broad age 
category of body weight for children is suitable for use with the 
default fish intake assumption.
    Given the hierarchy of preferences regarding the use of fish intake 
information [see Section C.3.(d)], States may have more comprehensive 
data and prefer to target a more narrow, younger age group. If States 
choose to specifically evaluate infants and toddlers, EPA would 
recommend 10 kg as a default body weight assumption for water intake 
for children ages 1-3 years old, as has been used in other EPA water 
programs. The 10th, 25th, and

[[Page 43801]]

50th percentile values of body weight for children 1-3 years old are 
10.4, 11.8, and 13.6 kg, respectively, with a mean value of 14.1 kg 
(Ershow and Cantor 1989). Based on an analysis of the NHANES II data 
base reported in the EPA's Exposure Factors Handbook, the 10th, 25th, 
and 50th percentile values for children less than 3 years old are 8.5, 
9.6, and 11.3 kg for females, and 9.1, 10.3, and 11.8 kg for males, 
respectively (USEPA, 1989). The mean for both sexes from NHANES II is 
11.6 kg. The 10 kg body weight assumption is representative of the 
majority of children under the age of 3. As with the 28 kg assumption, 
EPA recommends a more protective body weight assumption than the median 
value because of the increased susceptibility of infants and toddlers 
to acute effects from water-based formula intake.
    Body weight values for individual ages within the larger range of 
0-14 years are listed in the TSD for this Notice for those States and 
Tribes who wish to use body weight values for these individual groups. 
States and Tribes may wish to consider certain general developmental 
ages (e.g., infants, pre-adolescents, etc.), or certain specific 
developmental landmarks (e.g., neurological development in the first 
four years), depending on the chemical of concern. EPA encourages 
States and Tribes to choose a body weight intake from the tables 
presented in the TSD, if they believe a particular age subgroup is more 
appropriate.
    (3) Rates Based on Combining Intake and Body Weight. As discussed 
below, EPA is also soliciting comments on whether intake assumptions 
should be given on a per kg body weight basis. Under this alternate 
approach, default body weight assumptions of 10, 28, 65, or 70 kg are 
not needed because the approach involves dividing individual 
respondents' intake rates (determined in surveys of drinking water or 
fish intake) by their own seif-reported body weights.
    (b) Drinking Water Intake Rates. (1) Rate Protective of Human 
Health from Chronic Exposure. The 1980 AWQC National Guidelines assumed 
a water intake rate of 2 L/day. There is comparatively little 
variability in water intake within the population, compared to fish 
intake (i.e., drinking water intake varies, by and large, by about a 
three-fold range, whereas fish intake can vary by 100-fold). The 50th, 
75th, and 90th percentile values for adults 20-64 years old are 1.3, 
1.7, and 2.3 L/day, respectively (Ershow and Cantor, 1989). The 2 L/day 
value represents the 84th percentile for adults from the Ershow and 
Cantor study. EPA recommends maintaining the default tap water intake 
rate of 2 L/day. Individuals who work or exercise in hot climates could 
have water consumption rates significantly above 2 L/day, and EPA 
believes that States and Tribes should consider regional or 
occupational variations in water consumption. EPA believes that the 2 
L/day assumption is representative of a majority of the population over 
the course of a lifetime. This assumption was used with the 1980 
methodology and has also been used in EPA's drinking water program. 
Although a policy decision, 2 L/day is a reasonable and protective 
determination that represents the intake of most water consumers in the 
general population according to available drinking water studies, as 
summarized above and described in greater detail in the TSD. EPA 
believes that this assumption continues to represent an appropriate 
risk management decision.15 Based on the study data, EPA 
also recommends 2 L/day for women of childbearing age.
---------------------------------------------------------------------------

    \ 15\ EPA is currently conducting an analysis to generate 
estimates of water intake based on recent data from the USDA's 
CSFII. Estimates will be generated by population demographics 
including, age, gender, race, socioeconomic status and geographical 
region. Results of this analysis may be considered in the future 
with this methodology.
---------------------------------------------------------------------------

    (2) Rates Protective of Developmental Human Health Effects. As 
noted above, because infants and children have a higher water 
consumption per body weight compared to adults, a water consumption 
rate indicative of children is proposed for use when RfDs are based on 
health effects in children. Use of this water consumption rate should 
result in adequate protection for infants and children when setting 
criteria based on health effects for this target population. Estimating 
a mean drinking water intake for children ages 0-14 years old, 
combining drinking water intake for five age groups within the larger 
age group of 0-14 years from Ershow and Cantor (1989) and weighting by 
current population estimates (from U.S. Bureau of the Census, 1996) 
results in a drinking water intake of approximately 750 ml. As a 
slightly more protective measure than using 750 ml, EPA recommends a 
drinking water intake of 1 L/day to, again, represent a majority of the 
population in this age group. This value is equivalent to about the 
75th percentile value, which is 960 ml, for children ages 1-10 years 
old (Ershow and Cantor, 1989). The 50th, 75th, and 90th percentile 
values for children 1-3 years old are 0.6, 0.8, and 1.2 L/day, 
respectively (Ershow and Cantor, 1989).
    (3) Rates Based on Combining Drinking Water Intake and Body Weight. 
As an alternative to considering body weight and drinking water intake 
rates separately, EPA is considering using the actual intake per body 
weight data that is available in the Ershow and Cantor (1989) report. 
This approach has the advantage of using self-reported body weights of 
survey respondents, instead of converting to the 70 kg or 10 kg default 
assumptions. These alternate values are presented in Ershow and Cantor 
(1989) or can be determined from Ershow and Cantor (1989) and U.S. 
Bureau of the Census (1996) using the methods described above to 
determine a weighted mean. For example, the mean, 50th, 75th, and 90th 
percentile values of tap water intake for adults 20-64 years old are 
19.9, 18.2, 25.3, and 33.7 ml/kg body weight, respectively. Using 
information from Ershow and Cantor (1989) for fine age categories, the 
weighted mean intake for children ages 0-14 years old is 32.6 ml/kg, 
and using the same weighting procedure, the approximate 50th, 75th, and 
90th percentiles for this age group are 28.6, 42.3, and 59.3 ml/kg. The 
50th, 75th, and 90th percentile values of tap water intake for children 
1-3 years old are 41.4, 60.4, and 82.1 ml/kg body weight, respectively. 
It should be noted that in their 1993 review, SAB felt that using 
drinking water intake rate assumptions on a per body weight basis would 
be more accurate, but did not believe this change would appreciably 
affect the criteria values.
    (c) Incidental Ingestion from Ambient Surface Waters. To prevent 
potential health risks from incidental recreational ingestion, an 
incidental intake rate is necessary. EPA recommends using 10 ml/day as 
the chronic incidental ingestion rate. The value would be divided by 
the adult body weight of 70 kg. This chronic intake is based on 
information about the amount of water that may be ingested in a given 
hour of recreational exposure to water (30 ml) multiplied by the number 
of hours of recreational water use throughout a year and averaged over 
the year to obtain an average intake per day. (Refer to the TSD for 
further explanation.) As stated earlier, this intake would only be used 
in those cases where the waterbody is not used for potable water (e.g., 
estuaries) and criteria are based solely on fish ingestion. When 
developing criteria for waterbodies that are potential drinking water 
sources, the assumption of 2 L/day of direct ingestion is likely to 
account for the additional possible ingestion via recreational 
activities and, therefore, this incidental rate will not be added.
    (d) Fish Intake Rates. (1) Rates Protective of Human Health from 
Chronic Exposure. When deriving AWQC, EPA strives to provide adequate

[[Page 43802]]

protection [as described earlier in Section C.1.(a)(1), Policy Issues] 
from adverse health effects to highly exposed populations such as 
recreational and subsistence fishers as well as the general population. 
Based on available studies that characterize consumers of fish, 
recreational fishers and subsistence fishers appear to be two distinct 
groups whose intake rates are greater than the general population. It 
is, therefore, EPA's decision to discuss intakes for these two groups, 
in addition to the general population. Because the level of fish intake 
in highly exposed populations varies by geographical location, EPA 
suggests a four preference hierarchy for deriving consumption rates 
that encourages use of the best local, State, or regional data 
available but provides a default rate based on national statistics if 
there are no other data. A thorough discussion of the development of 
this policy method and relevant data sources is contained in the TSD. 
The four preference hierarchy is: (1) use of local data; (2) use of 
data reflecting similar geography/population groups; (3) use of data 
from national surveys; and (4) use of proposed default intake rates.
    The recommended four preference hierarchy is intended for use in 
evaluating fish intake from fresh and estuarine species only. 
Therefore, to protect humans who additionally consume marine species of 
fish, the marine portion should be considered as part of the ``other 
sources of exposure'' when calculating an RSC or dietary value (DT in 
the 1980 methodology equation). Refer to the TSD for further 
discussion. States and Tribes need to ensure that when evaluating 
overall exposure to a contaminant, marine fish intake is not double-
counted with the other dietary intake estimate used. Coastal States and 
Tribes that believe accounting for total fish consumption (i.e., fresh/
estuarine and marine species) is more appropriate for protecting the 
population of concern may do so, provided that the marine intake 
component is not double-counted with the RSC estimate. Throughout this 
Section, the terms ``fish intake'' or ``fish consumption'' are used. 
They generally refer to the consumption of finfish and shellfish, and 
the national survey described in this section includes both. States and 
Tribes should ensure that when selecting local or regionally-specific 
studies, both types are included when the population exposed are 
consumers of both types.
    EPA's first preference is that States and Tribes use the results 
from fish intake surveys of local watersheds within the State to 
establish fish intake assumptions that are representative of the 
defined populations being addressed for the particular waterbody. 
Again, EPA recommends that data indicative of fresh/estuarine species 
only be used which is, by and large, most appropriate for developing 
AWQC. EPA also recommends the use of cooked weight intake values which 
is discussed in greater detail with the fourth preference. States and 
Tribes may use either high-end values (such as the 90th or 95th 
percentile values) or central tendency values (mean or medians) for an 
identified population that they plan to protect (e.g., subsistence 
fishers or sport fishers). The mean or median value should be the 
lowest value considered by States or Tribes when choosing intake rates 
for use in criteria derivation. Furthermore, when considering median 
values from fish consumption studies, States and Tribes need to ensure 
that the distribution is based on survey respondents who reported 
consuming fish because surveys based on both consumers and nonconsumers 
typically result in median values of zero. If a State or Tribe chooses 
values (whether the central tendency or high-end values) from studies 
that particularly target high-end consumers, these values should be 
compared to high-end fish intake rates for the general population to 
make sure that the high-end consumers within the general population 
would be protected by the chosen intake rates. EPA believes this is a 
reasonable procedure and is also consistent with recent water quality 
guidance established for the Great Lakes. (See 60 FR 15366, Thursday, 
March 23, 1995). States and Tribes may wish to conduct their own 
surveys of fish intake, and EPA guidance is available on methods to 
conduct such studies in Guidance for Conducting Fish and Wildlife 
Consumption Surveys (USEPA, 1997b). Results from broader geographic 
regions in which the State or Tribe is located can also be used, but 
may not be as applicable as results from local watersheds. Since such 
studies would ultimately form the basis of a State or Tribe's AWQC, EPA 
would review any surveys of fish intake for consistency with the 
principles of EPA's guidance, as part of the Agency's review under 
303(c).
    If surveys conducted in the geographic area of the State or Tribe 
are not available, EPA's second preference is that States and Tribes 
consider results from existing fish intake surveys that reflect similar 
geography and population groups (e.g., from a neighboring State or 
Tribe or a similar watershed type), and follow the method described 
above regarding target values to derive a fish intake rate. Again, EPA 
recommends the use of cooked weight intake values and the use of fresh/
estuarine species data only. Results of existing local and regional 
surveys are discussed in greater detail in the TSD.
    If applicable consumption rates are not available from local, 
State, or regional surveys, EPA's third preference is that States and 
Tribes select intake rate assumptions for different population groups 
from national food consumption surveys. EPA has analyzed one such 
national survey, the combined 1989, 1990, and 1991 Continuing Survey of 
Food Intake by Individuals (CSFII). The CSFII, conducted annually by 
the USDA, collects food consumption information from a probability 
sample of the population of the 48 conterminous states. Respondents to 
the survey provide three days of dietary recall data. A detailed 
description of the combined 1989-1991 CSFII survey, the statistical 
methodology, and the results and uncertainties of the EPA analyses are 
provided in USEPA (1998). The TSD for this Notice presents selected 
results from this report including point and interval estimates of 
combined finfish and shellfish consumption for the mean, 50th (median), 
90th, 95th, and 99th percentiles. The estimated fish consumption rates 
are by fish habitat (i.e., freshwater/estuarine, marine and all 
habitats) for the following population groups: (1) All individuals; (2) 
individuals age 18 and over; (3) women ages 15-44; and (4) children age 
14 and under. Three kinds of estimated fish consumption rates are 
provided: (1) per capita rates [i.e., rates based on consumers and 
nonconsumers of fish (from the survey period. Refer to the TSD for 
further discussion)]; (2) acute consumption rates (i.e., rates based on 
respondents who reported consuming finfish or shellfish during the 
three-day reporting period); and (3) per capita consumption by body 
weight (i.e., per capita rates reported as milligrams of fish per 
kilogram of body weight per day).
    In addition, the TSD presents estimated per capita finfish and 
shellfish consumption rates for nine geographical regions of the U.S. 
based on the 1989-1991 CSFII. States and Tribes may wish to use these 
regional values if they do not have significant tier one or tier two 
data but do have limited regional data, and if they believe that the 
consumption rates of the particular population of concern differ from 
the national rates. The TSD also discusses precautions regarding their 
use due to limitations in the data set.

[[Page 43803]]

Similarly, if a State or Tribe has not identified a separate well-
defined population of high-end consumers and believes that the national 
data from the CSFII are representative, they may choose these rates.
    EPA's fourth preference is that States and Tribes use as fish 
intake assumptions the following default rates, based on the 1989-1991 
CSFII data, that EPA believes are representative of fish intake for 
different population groups: 17.80 g/day for the general adult 
population and sport fishers, and 86.30 g/day for subsistence fishers. 
These are risk management decisions that EPA has made after evaluating 
numerous fish intake surveys. These values represent the intake of 
freshwater/estuarine finfish and shellfish as consumed. As with the 
other preferences, EPA requests that States and Tribes routinely 
consider whether there is a substantial population of sport fishers or 
subsistence fishers when developing site-specific estimates, rather 
than automatically basing them on the typical individual. Because the 
combined 1989-1991 CSFII survey is national in scope, EPA proposes that 
the results from this survey be used to estimate fish intake for 
deriving national criteria. EPA has recognized the data gaps and 
uncertainties associated with the analysis of the CSFII in the process 
of making its default recommendations. The estimated mean of freshwater 
and estuarine fish ingestion for adults is 5.6 g/day, and the median is 
0 g/day. The estimated 90th percentile is 17.80 g/day; the estimated 
95th percentile is 39.04 g/day; and the estimated 99th percentile is 
86.30 g/day. The median value of 0 g/day may reflect the portion of 
individuals in the population who never eat fish as well as the limited 
reporting period (3 days) over which intake was measured. By applying 
as a default 17.8 g/day for the general adult population, EPA intends 
to select an intake rate that is protective of a majority of the 
population (again, the 90th percentile of consumers and nonconsumers 
according to the CSFII survey data). EPA further considers this rate to 
be indicative of the average consumption among sport fishers based on 
averages in the studies reviewed, which are presented in the TSD. 
Similarly, EPA believes that the assumption of 86.30 g/day is within 
the range of average consumption estimates for subsistence fishers 
based on the studies reviewed. The 95th percentile value, 39.04 g/day, 
is also within the range of average consumption for subsistence 
fishers, although on the low end according to the studies reviewed. The 
1992 National Workshop experts acknowledged that the high-end values 
are representative of rates for highly exposed groups such as 
subsistence fishermen, specific ethnic groups, or other high-risk 
people. EPA is aware that some local and regional studies indicate 
greater consumption among Native American, Pacific Asian American, and 
other subsistence consumers and recommends the use of those studies in 
appropriate cases, as indicated by the first and second preferences.
    The estimated values derived from the combined 1989-1991 CSFII 
survey can be compared with the default values in the 1980 AWQC 
National Guidelines. The 1980 AWQC National Guidelines recommended a 
fish intake rate of 6.5 g/day. This value was based on the mean per 
capita consumption rate of freshwater and estuarine finfish and 
shellfish from 30-day diary results that were reported in the 1973-1974 
National Purchase Diary Survey. It is generally believed that the 
consumption of fish has increased somewhat in recent years due to 
nutritional and other preferential choices. When comparing the old 
default rate of 6.5 g/day with the new arithmetic mean indicated above 
(5.6 g/day), the use of cooked weights and the redesignation of certain 
species (as described in the TSD) must be kept in mind.
    As indicated above, the default intake values proposed, as well as 
the rest of the CSFII values presented in the TSD tables, are based on 
the cooked weights of the fish analyzed, which was the basis of the 
survey design. There has been some question regarding whether to use 
cooked or uncooked weights of fish intake for deriving the AWQC. 
Studies show that, typically, with a filet or steak of fish, the weight 
loss in cooking is about 20 percent; that is, the uncooked weight is 
approximately 20 percent higher (Jacobs et al., 1998). This obviously 
means that using cooked weights results in a slightly lower intake rate 
and slightly less stringent AWQC. In researching consumption surveys 
for this proposal, EPA has found that some surveys have reported rates 
for cooked fish, others have reported uncooked rates, and many more are 
unclear as to whether cooked or uncooked rates are used.
    There are several issues regarding whether to use cooked or 
uncooked weights when estimating fish consumption rates. The first 
issue concerns the effect of the cooking process on the concentration 
of the toxicant in the fish tissue. For example, if in the cooking 
process, the mass of a toxicant in the fish tissue remains constant, 
then the concentration in the fish tissue will increase (the weight of 
the fish tissue decreases). This appears to be the case with a chemical 
such as mercury because it binds strongly to proteins and, thus, 
concentrates in the muscle tissue (Minnesota Department of Health, 
1992). However, as has been seen with numerous organic chemicals (e.g., 
PCBs), some cooking processes tend to decrease the mass of toxicant, 
thus reducing the concentration in the fish tissue (Zabik, et al., 
1993). Of importance here is that the mass of the contaminant in the 
fish tissue stays constant or is reduced. Unfortunately, there are 
rather few chemicals for which measurements are available. This issue 
is complicated further by the fact that different chemicals accumulate 
in different parts of the fish; that is, some chemicals accumulate in 
the muscle tissue, some in the gills, some in the viscera, etc. 
Therefore, the method of preparation (i.e., cleaning and trimming) can 
greatly affect the potential intake of the contaminant, as can the 
cooking method and the considerable variation in both of these factors 
between species of fish. In addition, there is the relatively 
unexplored area of how the cooking process affects the nature of the 
chemical. Specifically, the cooking process may change the ``parent'' 
compound to a by-product, or form a different compound altogether.
    Nevertheless, the cooked weight values are consistent with the 
recent Great Lakes guidance (which was specifically based on studies 
describing consumption rates of cooked fish) and, by and large, cooked 
fish is what people consume. This is also consistent with non-fish 
dietary estimates made by both EPA's pesticide program and FDA's Total 
Diet Study program. That is, their analyses are based on prepared 
foods, not raw commodities. However, EPA's Guidance For Assessing 
Chemical Contaminant Data For Use In Fish Advisories recommends 
analysis and advisories based on uncooked fish (USEPA, 1997c). States 
and Tribes should have the flexibility to consider raw fish consumption 
if they believe that the population they are targeting are consumers of 
raw fish. It should be noted that any raw shellfish consumed by 
respondents in the CSFII survey is included in the ``as consumed'' 
values. EPA cautions States and Tribes that the as consumed weights 
provided are not to be used for developing fish advisories, which is a 
substantially different program than the water quality criteria 
program.
    Therefore, EPA recommends using cooked weight intake rates, as they 
better reflect the potential exposure from fish consumption versus 
using the

[[Page 43804]]

uncooked weights. If States and Tribes find that, when using site-
specific or regional data, they are limited to data for uncooked 
weights only, they may choose to use these data in their calculations, 
provided that they adjust for the weight loss in cooking (i.e., by 
reducing the value by 20 percent). If a State or Tribe believes that 
the population of concern is preparing fish in such a manner that the 
amount normally lost is actually consumed as well, then they may 
consider using the uncooked weight. In addition, EPA recommends 
assuming no change in contaminant concentration from cooking as a 
default. If information on chemical change from cooking is available, 
then States are encouraged to use this information. If a State or Tribe 
has information on chemical change from cooking, they may consider 
using a cooking loss factor to adjust the BAF accordingly.
    It should be noted that there has been a redesignation of several 
species from how they were classified in the 1973-74 National Purchase 
Diary Fish Consumption Survey. Most significantly, salmon has been 
reclassified from a freshwater/estuarine species to a marine species. 
As marine harvested salmon represents approximately 99 percent of 
salmon consumption, removal reduces the overall fresh/estuarine fish 
consumption rate by 13 percent. Although they represent a very small 
percentage of freshwater/estuarine intake, land-locked and farm-raised 
salmon are still included. The basis for this decision is that the 
majority of the life span of all species of salmon (except land-locked 
and farm-raised populations) is spent in marine waters. This includes 
most of the species' growth phase, including the pre-spawning food 
gorging that the fish undertake. For the actual spawning event, most 
salmon fast, thus spending their energy making the trip to their 
spawning destination. This rationale is explained more fully, with 
citations, in the TSD. All of the species apportionments are indicated 
in Appendix A of the TSD (Tables A.31 through A.34) in parenthesis by 
the species name. The 13 percent reduction described above for salmon 
can be calculated via these tables.
    (2) Rates Protective of Developmental Human Health Effects. 
Exposures resulting in health effects in children or developmental 
effects in fetuses may be of primary concern. As discussed at the 
beginning of Section C.3, depending on the type of exposure or effect, 
States and Tribes may wish to use exposure factors for children or 
women of childbearing age in these situations. As stated previously, 
EPA is not recommending the development of additional AWQC but is 
acknowledging that basing a criterion on these population groups is a 
potential course of action and is, therefore, proposing the following 
default intake rates for States and Tribes to utilize in such 
situations.
    Since children have a higher fish consumption per body weight 
compared to adults, using a higher fish consumption rate per body 
weight may be needed for setting AWQC to assure adequate protection for 
children. EPA's preferences for States and Tribes in selecting 
assumptions for intake rates relevant for children is the same as that 
discussed above for establishing assumptions for average daily 
consumption rates for chronic effects, i.e., in order of decreasing 
preference, results from fish intake surveys of local watersheds, 
results from existing fish intake surveys that reflect similar 
geography and population groups, the distribution of intake rates from 
nationally based surveys (e.g., the CSFII), or finally, the default 
rate that EPA recommends below that is representative of a selected 
population group. The TSD for this Notice will present some 
distributional values related to the intake values relevant for 
assessing exposure when health effects to children are of concern. When 
an RfD is based on health effects in children, EPA recommends a default 
intake rate of 108.36 g/day for assessing those contaminants that 
exhibit adverse effects. This is equivalent to about the 90th 
percentile consumption rate for actual consumers of freshwater/
estuarine finfish and shellfish for children ages 14 and under using 
the combined 1989-1991 results from the CSFII survey. The value was 
calculated based on data for only those children who ate any fish 
during the 3-day survey period, and the intake was averaged over the 
number of days during which fish was actually consumed. EPA believes 
that by selecting the data for consumers only, the 90th percentile is a 
reasonable intake rate to use in assessments for effects where children 
are of primary concern. As discussed previously, EPA is recommending a 
default body weight of 28 kg to address such potential effects from 
fish consumption by children. EPA is providing these intake assumption 
values for States and Tribes that choose to provide additional 
protection when developing criteria that they believe should be based 
on health effects in children. This is consistent with the rationale in 
the recent guidance established for the Great Lakes (as already cited) 
and is an approach that EPA believes is reasonable.
    There are also cases in which pregnant women may be the population 
of most concern, due to the possibility of developmental effects that 
may result from exposures of the mother to toxicants. In these cases, 
fish intake rates specific to females of childbearing age are most 
appropriate when assessing exposures to developmental toxicants. When 
an RfD is based on developmental toxicity, EPA proposes a default 
intake rate of 148.83 g/day for assessing exposures for women of 
childbearing age from contaminants that cause developmental effects. 
This is equivalent to about the 90th percentile consumption rate for 
actual consumers of freshwater/estuarine finfish and shellfish for 
women ages 15-44 using the combined 1989-1991 results from the CSFII 
survey. As with the rate for children, this value represents only those 
women who ate fish during the 3-day survey period. As discussed 
previously, EPA is recommending a default body weight of 65 kg for 
women of childbearing age.
    (3) Rates Based on Combining Fish Intake and Body Weight. As an 
alternative to looking at fish intake values separately from body 
weight, EPA is considering using the actual intake per body weight 
data. This approach has the advantage of using actual body weights of 
survey respondents, instead of converting to the 70 kg, 65 kg, 28 kg, 
or 10 kg default assumptions. In its 1993 review, SAB felt that using 
fish intake rate assumptions on a per body weight basis would be more 
accurate, but did not believe this change would appreciably affect the 
criteria values.
4. Request for Comments
    1. EPA requests comment on the choice of population to protect and 
on the adequacy of their assumptions in protecting this population.
    2. EPA requests comment on the Agency's recommendation to include 
the drinking water pathway explicitly in deriving the AWQC for the 
protection of human health where drinking water is a designated use.
    3. EPA requests comment on the Agency's recommendation to continue 
the practice of setting AWQC that account for combined drinking water 
and fish consumption, as well as a separate criterion for fish/
shellfish consumption alone.
    4. EPA requests comment on whether AWQC based only on fish 
ingestion (or aquatic life criteria) adequately protect recreational 
users from health effects resulting from incidental ingestion from

[[Page 43805]]

water bodies not considered sources of potable water (e.g., estuaries).
    5. EPA requests comment on the Agency's recommendation to include 
incidental ingestion in the calculation of AWQC in those cases where 
the water body is not used for potable water.
    6. EPA requests comment on the Agency's recommendation that only a 
portion of the RfD be used in setting AWQC in order to account for 
other sources of exposure.
    7. The Agency also requests comment on whether toxicity information 
(such as uncertainty factors, severity of effects, essentiality, and 
possible additive/synergistic effects) should be considered in 
allocating the RfD.
    8. EPA requests comment on the choice of the Exposure Decision Tree 
approach and the choice of the 80 percent ceiling and 20 percent floor 
as bounding levels for the RfD allocation. The Agency also requests 
comment on the use of the subtraction approach and the percentage 
approach within the decision tree.
    9. EPA requests comment on how inhalation and dermal absorption 
exposures from water should be estimated and included in calculating 
health-based criteria.
    10. EPA requests comment on the appropriateness of including 
inhalation exposures when accounting for other sources of exposure in 
setting AWQC.
    11. EPA requests comment on the Agency's recommendation to use 
existing data on differences in bioavailability between water, air, 
soils, and different foods when estimating total exposure for use in 
allocating the RfD. In the absence of such data, EPA will assume equal 
rates of absorption from different routes and sources of exposure. EPA 
requests comment on this assumption.
    12. EPA requests comment on the Agency's recommendation to continue 
using the incremental risk approach that does not consider other 
exposure sources explicitly when setting AWQC for linear carcinogens, 
and to continue using other exposure sources in setting AWQC for 
threshold toxicants including noncarcinogens and nonlinear carcinogens.
    13. EPA requests comment on whether a default body weight of 65 kg 
should be used in cases where pregnant women constitute the target 
population.
    14. EPA requests comment on the Agency's proposal to use 28 kg as 
the default body weight to calculate AWQC which protects against 
adverse effects in children when the chemical of concern has an RfD 
based on health effects in children.
    15. EPA requests comment on whether 10 kg or a different body 
weight should be used as the default assumption to calculate AWQC for 
children's health effects from water intake for children 1-3 years old, 
as has been used in other EPA water programs.
    16. EPA requests comment on whether additional default body weights 
should be developed for finer age categories due to the consideration 
of different developmental stages.
    17. EPA requests comment on whether to use separate tap water 
intake and body weight assumptions (e.g., 2 L/day, 70 kg body weight) 
or assumptions that combine tap water intake and body weight (e.g., 30 
ml tap water/kg body weight), and what values should be used.
    18. Although EPA is not recommending an incidental ingestion rate 
for derivation of criteria based on short-term health effects at this 
time, the Agency requests comment on the use of an intake of 30 ml/hour 
in cases where shorter-term effects may be considered in the derivation 
of criteria. (EPA assumes that this 30 ml incidental rate may be 
ingested by children, and thus for RfDs based on health effects in 
children, this value may be divided by the lower body weights of 
children to adequately protect them from health effects resulting from 
incidental ingestion.)
    19. EPA requests comment on (1) the use of the CSFII survey results 
in setting national criteria given the known limitations (i.e., the 3-
day reporting period); (2) whether EPA should select default rates for 
different population groups, including 17.80 g/day for sportfishers and 
86.30 g/day for subsistence fishers in addition to the value of 17.80 
g/day for the typical adult individual (EPA also requests comment on 
alternatively using 39.04 g/day for subsistence fishers); and (3) which 
default intake rate(s) should be used in setting criteria. With regard 
to the default alternative for subsistence fishers, EPA requests 
comment on which is more indicative of fresh/estuarine consumption 
rates among the population group.
    20. EPA requests comment on the use of cooked versus uncooked fish 
intake weights, the concepts of mass and concentration of a toxicant in 
fish tissue and the potential changes from cooking, as well as the 
potential changes in the structure of the toxicant.
    21. EPA requests comments on the rationale for redesignating salmon 
as a marine species, as well as the rationale for the other species 
designations.
    22. EPA requests comments on the use of the default rate of 108.36 
g/day of fish intake for children when assessing effects from 
contaminants that are based on health effects in children. EPA 
similarly requests comments on the use of the default intake rate of 
148.83 g/day for women of childbearing age when assessing exposures 
from contaminants that cause developmental effects.
    23. EPA requests comments on whether to use separate fish intake 
and body weight assumptions (e.g., 17.80 g/day, 70 kg body weight) or 
assumptions that combine fish intake and body weight (e.g., 254.3 mg 
fish/kg body weight), and what values should be used.

References for Exposure

Borum, D.R. Unpublished. Approaches to Allocating the RfD for 
Setting Health-Based Criteria. Available in the Public Docket for 
the Proposed Revisions to the Ambient Water Quality Criteria Human 
Health Methodology.
Ershow A.G. and K.P. Cantor. 1989. Total Water and Tap Water Intake 
in the United States: Population-based Estimates of Quantities and 
Sources. Bethesda, MD: National Cancer Institute. Order #263-MD-
810264.
Jacobs, H.L., H.D. Kahn, K.A. Stralka, and D.B. Phan. 1998. 
Estimates of Per Capita Fish Consumption in the U.S. Based on the 
Continuing Survey of Food Intake by Individuals (CSFII). Risk 
Analysis: An International Journal 18(3).
Minnesota Department of Health. 1992. Minnesota Fish Consumption 
Advisory. Minneapolis, MN. May. Cited in USEPA, 1998.
National Center for Health Statistics (NCHS). 1987. Anthropometric 
Reference Data and Prevalence of Overweight, United States, 1976-
1980. Data from the National Health and Nutrition Examination 
Survey, Series 11, No. 238. Hyattsville, MD: U.S. Department of 
Health and Human Services, National Center for Health Statistics. 
DHHS Publication No. PHS 87-1688.
SAB. 1993. Review of the Methodology for Developing Ambient Water 
Quality Criteria for the Protection of Human Health. Prepared by the 
Drinking Water Committee of the Science Advisory Board. EPA-SAB-DWC-
93-016.
U.S. Bureau of the Census. 1996. Personal Communication Between Jean 
Dee, U.S. Bureau of the Census and Amy Benson, Abt Associates. May 
10.
USEPA. 1989. Exposure Factors Handbook. Office of Health and 
Environmental Assessment. Washington, DC. EPA 600/8-89-043.
USEPA. 1994. Integrated Risk Information System (IRIS). Reference 
Dose (RfD) for Oral Exposure for Cadmium. Online. (Verification date 
02/01/94.) Office of Health and Environmental Assessment, 
Environmental Criteria and Assessment Office, Cincinnati, OH.

[[Page 43806]]

USEPA. 1997a. Exposure Factors Handbook. National Center for 
Environmental Assessment, Office of Research and Development. 
Washington, DC. EPA/600/P-95/002Fa. August.
USEPA. 1997b. Guidance for Conducting Fish and Wildlife Consumption 
Surveys. September.
USEPA. 1997c. Guidance for Assessing Chemical Contaminant Data for 
Use in Fish Advisories. Volume II: Risk Assessment and Fish 
Consumption Limits. Second Edition. Office of Water. Washington DC. 
EPA 823-B-97-009.
USEPA. 1998. Daily Average Per Capita Fish Consumption Estimates 
Based on the Combined USDA 1989, 1990, 1991 Continuing Survey of 
Food Intakes by Individuals (CSFII). Volume I: Uncooked Fish 
Consumption National Estimates; Volume II: As Consumed Fish 
Consumption National Estimates. Prepared by SAIC under Contract #68-
C4-0046. March.
Zabik, M.E., et al. 1993. Assessment of Contaminants in Five Species 
of Great Lakes Fish at the Dinner Table. Final Report to the Great 
Lakes Protection Fund. March. Cited in USEPA. 1998.

D. Bioaccumulation

1. Introduction
    Aquatic organisms can accumulate certain types of chemicals in 
their bodies when exposed to these chemicals in water, food, and other 
sources. This process is called bioaccumulation. For some chemicals, 
uptake through the food chain is the most important route of exposure. 
As lower trophic level organisms are consumed by higher trophic level 
organisms, the tissue concentrations of these chemicals may increase 
with each trophic level so that chemical residues in top carnivores may 
be many orders of magnitude greater than the concentration of the 
chemical in the environment. Although ambient concentrations of certain 
chemicals in the environment may be too low to affect the lowest level 
organisms, this biomagnification process can result in concentrations 
which may pose severe health risks to the consumers of top trophic 
level aquatic organisms.
    In order to properly account for potential human exposure to 
waterborne contaminants, human health ambient water quality criteria 
should be developed based on principles of bioaccumulation. The degree 
to which chemicals bioaccumulate can vary widely (spanning several 
orders of magnitude) for different chemicals. Thus, if two chemicals 
are equal in every respect except for the extent to which they 
bioaccumulate, the chemical with the higher bioaccumulation factor (a 
measure of bioaccumulation) will have the lower water quality 
criterion. Prior to deriving a human health water quality criterion, 
the extent of bioaccumulation for the chemical of interest must be 
established.
2. Bioaccumulation and Bioconcentration Concepts
    Bioaccumulation reflects the uptake and retention of a chemical by 
an aquatic organism from all surrounding media (e.g., water, food, 
sediment). Bioconcentration refers to the uptake and retention of a 
chemical by an aquatic organism from water only. Both bioaccumulation 
and bioconcentration can be viewed simply as the result of competing 
rates of chemical uptake and depuration (chemical loss) by an aquatic 
organism. However, the rates of uptake and depuration can be affected 
by numerous factors including the physical and chemical properties of 
the chemical, the physiology and biology of the organism, environmental 
conditions, ecological factors such as food web structure, and the 
amount and source of the chemical. When the rates of chemical uptake 
and depuration are equal, the distribution of the chemical between the 
organism and its source(s) is said to be at equilibrium or at steady-
state. For a constant chemical exposure, the time required to achieve 
steady-state conditions varies according to the properties of the 
chemical and other factors. For example, some chemicals require a long 
time to reach steady-state conditions between environmental 
compartments (e.g., many months for certain highly hydrophobic 
chemicals) while others reach steady-state relatively quickly (e.g., 
hours to days for certain hydrophilic chemicals).
    The concept of steady-state or equilibrium conditions is very 
important when assessing or evaluating bioaccumulation and applying 
these principles in real world situations, such as the derivation of 
AWQC. For some chemicals and organisms that require relatively long 
time periods to reach steady-state, changes in water column chemical 
concentrations may occur on a much more rapid time scale compared to 
the corresponding changes in an organism's tissue concentrations. Thus, 
if the system departs substantially from steady-state conditions, the 
ratio of the tissue concentration to a water concentration which is not 
averaged over a sufficient time period may have little resemblance to 
the steady-state ratio and have little predictive value of long-term 
bioaccumulation potential. For highly bioaccumulative pollutants in 
dynamic systems, reliable BAFs can be determined only if, among other 
factors, water column concentrations are averaged over a sufficient 
period of time (e.g., a duration approximating the amount of time 
predicted for the pollutant to reach steady-state). In addition, 
adequate spatial averaging of both tissue and water column 
concentrations is required to develop reliable BAFs for use in deriving 
human health ambient water quality criteria.
    For this reason, a bioaccumulation factor (BAF) is defined in this 
Notice as representing the ratio (in L/kg) of a concentration of a 
substance in tissue to its concentration in the surrounding water in 
situations where the organism and its food are exposed and the ratio 
does not change substantially over time. A bioconcentration factor is 
considered to represent the uptake and retention of a substance by an 
aquatic organism from the surrounding water only, through gill 
membranes or other external body surfaces, in situations where the 
tissue-to-water ratio does not change substantially over time.
3. Existing EPA Guidance
    In developing criteria to protect humans from the consumption of 
contaminated aquatic organisms, EPA has relied upon the BCF and 
occasionally BAF to relate water concentrations to the amount of a 
contaminant that is ingested.
    BCFs are determined either by measuring bioconcentration in 
laboratory tests (comparing fish tissue residues to chemical 
concentrations in test waters), or by predicting the BCF from a 
chemical's octanol-water partition coefficient (Kow or P). 
The log of the octanol-water partition coefficient (log Kow 
or log P) has been shown to be empirically related to the log of the 
BCFs (e.g., Mackay, 1982; Connell, 1988; Veith et al., 1979), as 
described further by the equations below.
    The 1980 AWQC National Guidelines for deriving human health 
criteria allowed for the use of laboratory-measured or predicted BCFs 
when the preferred field-measured BCFs (equivalent to field-measured 
bioaccumulation factors (BAFs) described below) were not available. In 
those cases where an appropriate laboratory-measured BCF was not 
available, the equation ``log BCF = (0.85 log Kow) - 0.70'' 
was used (Veith et al., 1979) to estimate the BCF for aquatic 
organisms.
    In 1991, EPA issued the final ``Technical Support Document for 
Water Quality-Based Toxics Control'' (EPA 505/2-90-001) and a draft 
document entitled ``Assessment and Control of Bioconcentratable 
Contaminants in Surface Waters'' for notice and comment (56 FR 13150). 
These documents, relying on additional research into the relationship 
between BCF and log Kow,

[[Page 43807]]

recommend that a slightly different equation be used to derive BCFs in 
the absence of laboratory-measured BCFs (Veith and Kosian, 1983; log 
BCF = 0.79 log Kow-0.40).
    EPA's 1991 National guidance documents, the ``Technical Support 
Document for Water Quality-Based Toxics Control'' and draft 
``Assessment and Control of Bioconcentratable Contaminants in Surface 
Waters,'' recommend a methodology for estimating the BAF where there is 
an absence of a field-measured BAF. This methodology multiplies the 
laboratory-measured or predicted BCF by a factor which accounts for the 
biomagnification of a pollutant through trophic levels in a food chain. 
As larger predatory aquatic organisms (e.g., salmon) consume other fish 
and aquatic organisms, the amount of some contaminants in the consumed 
fish is concentrated in the predator. The factor which accounts for 
this biomagnification through the food chain is called the food chain 
multiplier (FCM) in these 1991 National guidance documents. EPA 
calculated the FCMs using a model of the step-wise increase in the 
concentration of an organic chemical from phytoplankton (trophic level 
1) through the top predatory fish level of a food chain (trophic level 
4) (Thomann, 1989).
    The FCMs were determined by first running Thomann's model to 
generate BCFs and BAFs for trophic level 2, and BAFs for trophic levels 
3 and 4. This was done for a range of log Kow values from 
3.5 to 6.5, at intervals of a tenth of log Kow value. 
Second, the FCMs for each log Kow value in this range were 
calculated using the following equations:
    For trophic level 2 (zooplankton):
    [GRAPHIC] [TIFF OMITTED] TN14AU98.018
    
    For trophic level 3 (small fish):
    [GRAPHIC] [TIFF OMITTED] TN14AU98.019
    
    For trophic level 4 (top predator fish):
    [GRAPHIC] [TIFF OMITTED] TN14AU98.020
    
    Where BCF2 is the BCF for trophic level 2 organisms, and BAF2, 
BAF3, and BAF4 are the BAFs for trophic levels 2, 3, and 4, 
respectively.
    On March 23, 1995 (60 FR 15366), EPA promulgated the Great Lakes 
Water Quality Initiative (GLWQI or GLI) guidance. The GLWQI guidance 
incorporated BAFs in the derivation of criteria to protect human health 
because it is believed that BAFs are better predictors of chemical 
concentrations in fish tissue than BCFs since BAFs include 
consideration of contaminant uptake from all routes of exposure (i.e., 
which occurs in field situations). The final GLWQI guidance established 
a hierarchy of four methods for deriving BAFs for nonpolar organic 
chemicals: (1) Field-measured BAFs; (2) predicted BAFs derived using a 
field-measured biota-sediment accumulation factor (BSAF); (3) predicted 
BAFs derived by multiplying a laboratory-measured BCF by a food chain 
multiplier; and (4) predicted BAFs derived by multiplying a BCF 
calculated from the Kow by a food-chain multiplier (U.S. 
EPA, 1995a). The GLI incorporated several improvements in the 
methodology for deriving BAFs. For example, the GLI used the Gobas 
model (Gobas, 1993) for estimating FCMs that accounted for both the 
benthic and pelagic food webs. The Thomann model described above only 
accounted for the pelagic food web. Other improvements included the use 
of the BSAF method for estimating BAFs. The BSAF method allows for the 
estimation of BAFs for those chemicals that are difficult to measure in 
the ambient water due to their extremely high hydrophobicity, such as 
the polychlorinated dibenzo-p-dioxins.
    The revised methodology in this Notice for deriving human health 
AWQC explicitly addresses various attributes of how bioaccumulative 
chemicals behave and accumulate in aquatic ecosystems. For certain 
chemicals where uptake from exposure to multiple media is important, 
EPA is emphasizing the assessment of bioaccumulation (i.e., uptake from 
water, food, sediments) over bioconcentration (i.e., uptake from 
water). Consistent with the final GLI, the revisions to EPA's national 
AWQC methodology establishes the same four-method hierarchy of 
procedures for deriving BAFs for nonpolar organic chemicals.
    For inorganic chemicals, EPA proposes that the AWQC be based on (in 
order of preference): (1) An appropriately determined field-measured 
BAF; (2) a laboratory-measured BCF multiplied by a field-measured FCM; 
or (3) a laboratory-measured BCF. Because inorganic substances do not 
predominantly partition to lipids, the BAF for metals do not need to be 
normalized by lipid content.
4. Definitions
    Baseline BAF (BAFlf~~d). For organic 
chemicals, a BAF (in L/kg-lipid) that is based on the concentration of 
freely dissolved chemical in the ambient water and the lipid normalized 
concentration in tissue; for inorganic chemicals, a BAF that is based 
on the wet weight of the tissue.
    Baseline BCF (BCFlf~~d). For organic 
chemicals, a BCF (in L/kg-lipid) that is based on the concentration of 
freely dissolved chemical in the ambient water and the lipid normalized 
concentration in tissue; for inorganic chemicals, a BCF that is based 
on the wet weight of the tissue.
    Bioaccumulation. The net accumulation of a substance by an organism 
as a result of uptake from all environmental sources.
    Bioaccumulation Factor (BAF). The ratio (in L/kg-tissue) of the 
concentration of a substance in tissue to its concentration in the 
ambient water, in situations where both the organism and its food are 
exposed and the ratio does not change substantially over time. The BAF 
is calculated as:
[GRAPHIC] [TIFF OMITTED] TN14AU98.021

where:

Ct = Concentration of the chemical in the wet tissue (either 
whole organism or specified tissue)
Cw = Concentration of chemical in water


[[Page 43808]]


    Bioconcentration. The net accumulation of a substance by an aquatic 
organism as a result of uptake directly from the ambient water, through 
gill membranes or other external body surfaces.
    Bioconcentration Factor (BCF). The ratio (in L/kg-tissue) of the 
concentration of a substance in tissue of an aquatic organism to its 
concentration in the ambient water, in situations where the organism is 
exposed through the water only and the ratio does not change 
substantially over time. The BCF is calculated as:
[GRAPHIC] [TIFF OMITTED] TN14AU98.022

where:

Ct = Concentration of the chemical in the wet tissue (either 
whole organism or specified tissue)
Cw = Concentration of chemical in water

    Biota-Sediment Accumulation Factor (BSAF). The ratio (kg of 
sediment organic carbon per kg of lipid) of the lipid-normalized 
concentration of a substance in tissue of an aquatic organism to its 
organic carbon-normalized concentration in surface sediment, in 
situations where the ratio does not change substantially over time, 
both the organism and its food are exposed, and the surface sediment is 
representative of average surface sediment in the vicinity of the 
organism. The BSAF is defined as:
[GRAPHIC] [TIFF OMITTED] TN14AU98.023

Where:

Cl = The lipid-normalized concentration of the chemical in 
tissues of the biota (g/g lipid)
Csoc = The organic carbon-normalized concentration of the 
chemical in the surface sediment (g/g sediment organic carbon)

    Biomagnification. The increase in tissue concentration of poorly 
depurated materials in organisms along a series of predator-prey 
associations, primarily through the mechanism of dietary accumulation.
    Biomagnification Factor (BMF). The ratio (unitless) of the tissue 
concentration of a predator organism at a particular trophic level to 
the tissue concentration in its prey organism at the next lowest 
trophic level, for a given waterbody and chemical exposure. For organic 
chemicals, a BMF can be calculated using lipid-normalized 
concentrations in the tissue of organisms at two successive trophic 
levels as:
[GRAPHIC] [TIFF OMITTED] TN14AU98.024

Where:

Cl(TL, n) = Lipid-normalized concentration in appropriate 
tissue of predator organism at trophic level ``n''
Cl(TL, n-l) = Lipid-normalized concentration in appropriate 
tissue of prey organism at the next lowest trophic level from the 
predator.

    For inorganic chemicals, a BMF can be calculated using chemical 
concentrations in the tissue of organisms at two successive trophic 
levels as:
[GRAPHIC] [TIFF OMITTED] TN14AU98.025

Where:

Ct(TL, n) = Concentration in appropriate tissue of predator 
organism at trophic level ``n'' (may be either wet weight or dry weight 
concentration so long as both the predator and prey concentrations are 
expressed in the same manner)
Ct (TL, n-1) = Concentration in appropriate tissue of prey 
organism at the next lowest trophic level from the predator (may be 
either wet weight or dry weight concentration so long as both the 
predator and prey concentrations are expressed in the same manner)

    As explained in the TSD, BMFs can also be related to (and 
calculated from) FCMs and baseline BAFs.
    Depuration. The loss of a substance from an organism as a result of 
any active or passive process.
    Food-Chain Multiplier (FCM). The ratio of a baseline BAF for an 
organism of a particular trophic level to the baseline BCF (usually 
determined for organisms in trophic level one).
    Freely Dissolved Concentration. For hydrophobic organic chemicals, 
the concentration of the chemical that is dissolved in ambient water, 
excluding the portion sorbed onto particulate or dissolved organic 
carbon. The freely dissolved concentration is considered to represent 
the most bioavailable form of an organic chemical in water and, thus, 
is the form that best predicts bioaccumulation. The freely dissolved 
concentration can be determined as:
[GRAPHIC] [TIFF OMITTED] TN14AU98.026

Where:

Cwf d = Freely dissolved concentration of the 
organic chemical in ambient water
Cwt = Total concentration of the organic chemical 
in ambient water
ffd = Fraction of the total chemical in ambient water that 
is freely dissolved

    Lipid-normalized Bioaccumulation Factor (BAFl). The 
ratio (in L/kg- lipid) of a substance's lipid-normalized concentration 
in tissue to its concentration in the ambient water, in situations 
where both the organism and its food are exposed and the ratio does not 
change substantially over time. The lipid-normalized BAF is calculated 
as:
[GRAPHIC] [TIFF OMITTED] TN14AU98.027

Where:
Cl = Lipid-normalized concentration of the chemical in whole 
organism or specified tissue
Cw = Concentration of chemical in water

    Lipid-normalized Bioconcentration Factor (BCFl). The 
ratio (in L/kg- lipid) of a substance's lipid-normalized concentration 
in tissue of an aquatic organism to its concentration in the ambient 
water, in situations where the organism is exposed through the water 
only and the ratio does not change substantially over time. The lipid-
normalized BCF is calculated as:
[GRAPHIC] [TIFF OMITTED] TN14AU98.028

Where:

Cl = Lipid-normalized concentration of the chemical in whole 
organism or specified tissue
Cw = Concentration of chemical in water

    Lipid-normalized Concentration (Cl). The total 
concentration of a contaminant in a tissue or whole organism divided by 
the lipid fraction in that tissue or whole organism. The lipid-
normalized concentration can be calculated as:
[GRAPHIC] [TIFF OMITTED] TN14AU98.029

where:

Ct = Concentration of the chemical in the wet tissue (either 
whole organism or specified tissue)
fl = Fraction lipid content in the organism or specified 
tissue

    Octanol-water Partition Coefficient (Kow). The ratio of 
the concentration of a substance in the n-octanol phase to its 
concentration in the aqueous phase in an equilibrated two-phase 
octanol-water system. For log Kow, the log of the octanol-
water partition coefficient is a base 10 logarithm.
    Organic Carbon-normalized Concentration (Csoc). For 
sediments, the total concentration of a contaminant in

[[Page 43809]]

sediment divided by the fraction of organic carbon in sediment. The 
organic carbon-normalized concentration can be calculated as:
[GRAPHIC] [TIFF OMITTED] TN14AU98.030

where:

Cs = Concentration of chemical in sediment
foc = Fraction organic carbon in sediment

    Uptake. Acquisition by an organism of a substance from the 
environment as a result of any active or passive process.
5. Determining Bioaccumulation Factors for Nonpolar Organic Chemicals
    The calculation of a BAF for a nonpolar organic chemical (chemicals 
that do not readily dissolve in water) used in the derivation of AWQC 
is a two-step process. The first step is to calculate a baseline BAF 
for the chemical of interest using information from the field site or 
laboratory where the original data were collected (e.g., the lipid 
content of the species collected and the freely dissolved fraction of 
the chemical in water at the site where the data were collected). If 
information used to estimate fish consumption rates indicates that 
organisms are being consumed from different trophic levels, then 
baseline BAFs need to be determined for each of the relevant trophic 
levels (see Section 6 for determining baseline BAFs).
    The second step is to calculate a BAF (or BAFs) for the chemical 
that will be used in the derivation of AWQC using information from the 
location where the aquatic species of interest are consumed (e.g., the 
lipid content of the aquatic species consumed by humans and the freely 
dissolved fraction of the chemical in water at the site where the 
aquatic species are being consumed). The difference in a baseline BAF 
and a BAF used in the derivation of AWQC is that baseline BAFs can be 
used for extrapolating from one species to another and from one water 
body to another. This is the case because baseline BAFs are lipid-
normalized which enables extrapolation for organic chemicals from one 
species to another and are based on the freely dissolved concentration 
of organic chemicals which enables extrapolation from one water body to 
another (the importance of these concepts is discussed below). Baseline 
BAFs, however, cannot be used directly in the derivation of AWQC 
because they may not reflect the conditions in the area of interest 
(e.g., the lipid content of the aquatic species consumed in the area of 
interest and the freely dissolved fraction of the chemical in the area 
of concern).
    Depending on the type of information available for a given 
chemical, different procedures may be used to determine the baseline 
BAF. The most preferred baseline BAFs are those derived using 
appropriate field data. Field-measured BAFs, however, have not been 
determined for all chemicals. Thus, EPA recommends a hierarchy of 
procedures to determine BAF values. The data preference for derivation 
of baseline BAFs for nonpolar organic chemicals is as follows (in order 
of priority):
    1. A field-measured baseline BAF derived from a field study of 
acceptable quality;
    2. A predicted baseline BAF derived from a field-measured BSAFs of 
acceptable quality;
    3. A predicted baseline BAF derived from a laboratory-measured BCF 
of acceptable quality and a food-chain multiplier (FCM); or
    4. A predicted baseline BAF derived from an acceptable 
Kow and a food-chain multiplier.
    While EPA recommends the above hierarchy for determining final 
baseline BAF values, for comparative purposes, baseline BAFs should be 
determined for each chemical by as many of the four methods as 
available data allow. Comparing baseline BAFs derived using the 
different methods recommended above can provide insight for identifying 
and evaluating any discrepancies in the BAF determinations that might 
occur. The information needed to derive an acceptable baseline BAF 
using each of the four methods is discussed in Section D.6. Section D.7 
discusses the information needed to derive an acceptable BAF for use in 
the calculation of AWQC.
6. Estimating Baseline BAFs
    All the baseline BAFs for nonpolar organic chemicals should be 
expressed on a freely dissolved and lipid-normalized basis. In 
addition, because bioaccumulation can be strongly influenced by the 
trophic level of aquatic organisms, baseline BAFs need to be determined 
on a trophic level-specific basis. The procedures for adjusting a 
field-measured BAF or field-measured BSAF or laboratory-measured BCF to 
a freely dissolved and lipid-normalized basis are discussed below.
    (a) Field-Measured Baseline BAF. Appropriately derived field-
measured BAFs are considered first in the data preference hierarchy for 
calculating baseline BAFs because they directly reflect any chemical 
metabolism that may occur and site-specific differences in the aquatic 
food web that may affect bioaccumulation. The calculation of a field-
measured baseline BAF expressed on a freely dissolved and lipid- 
normalized basis requires information on: (1) A field-measured BAF 
based on the total concentration of a chemical in the tissue of the 
aquatic organism sampled and the total concentration of the chemical in 
the ambient water; (2) the fraction of tissue that is lipid in the 
aquatic organism of interest; and (3) either the measured or estimated 
freely dissolved fraction of the total chemical in the ambient water 
where the aquatic species were collected (to estimate the freely 
dissolved fraction for a chemical requires information on the 
particulate and dissolved organic carbon content in the ambient water 
and the Kow of the chemical of interest). The equation for 
deriving a field-measured baseline BAF expressed on a freely dissolved 
and lipid-normalized basis is:
[GRAPHIC] [TIFF OMITTED] TN14AU98.031

where:

Baseline BAFl/fd = BAF expressed on a freely 
dissolved and lipid-normalized basis
Measured BAFT/t = BAF based on total 
concentration in tissue and water
fl = Fraction of the tissue that is lipid
ffd = Fraction of the total chemical that is freely 
dissolved in the ambient water

    For each trophic level, a species mean baseline BAF is calculated 
as the geometric mean if more than one acceptable, measured baseline 
BAF is available for a given species. For each trophic level, a trophic 
level-specific BAF is calculated as the geometric mean of the species 
mean measured baseline BAFs. Each of the three components for deriving 
the baseline BAF are described in further detail below.

[[Page 43810]]

    Measured BAFtT. To estimate a measured 
BAFtT, information is needed on the total 
concentration of the pollutant in the tissue of the organism and the 
total concentration of the chemical in ambient water at the site of 
sampling. The equation to derive a measured BAFtT 
is:
[GRAPHIC] [TIFF OMITTED] TN14AU98.032

    Application of data quality assurance procedures when measuring, 
estimating, and applying field-measured BAFs is of primary importance. 
The following general procedural and quality assurance requirements are 
important to be met for field-measured BAFs:
    1. The field studies used should be limited to those that include 
fish at or near the top of the aquatic food chain (i.e., in trophic 
levels 3 and/or 4). In situations where consumption of lower trophic 
level organisms represents an important exposure route, such as certain 
types of shellfish at trophic level 2, the field study should also 
include appropriate target species at this trophic level.
    2. The trophic level of the fish species should be determined 
taking into account the life stage(s) consumed and food web structure 
at the location(s) of interest.
    3. Collection of bioaccumulation field data at a specific site for 
which criteria are to be applied and with the species of concern are 
preferred.
    4. If data cannot be collected from every site for which criteria 
are to be applied, the site of the field study should not be so unique 
that the BAF cannot be extrapolated to other locations where the 
criteria and values will apply.
    5. Samples of the appropriate resident species and the water in 
which they reside should be collected and analyzed using appropriate, 
sensitive, accurate, and precise methods to determine the 
concentrations of bioaccumulative chemicals present in the tissues and 
water samples.
    6. For organic chemicals, the percent lipid should be either 
measured or reliably estimated for the tissue used in the determination 
of the BAF to permit the measured concentration of chemical in the 
organism's edible tissues to be lipid-normalized.
    7. The concentration of the chemical in the water should be 
measured in a way that can be related to particulate organic carbon 
(POC) and/or dissolved organic carbon (DOC).
    8. For organic chemicals with log Kow greater than four, 
the concentrations of POC and DOC in the ambient water should be either 
measured or reliably estimated.
    9. For inorganic chemicals where lipid normalization does not 
apply, BAFs should be used only if they are expressed on a wet weight 
basis; BAFs reported on a dry weight basis can be used only if they are 
converted to a wet weight basis using a conversion factor that is 
measured or reliably estimated for the tissue used in the determination 
of the BAF.
    EPA is currently developing guidance for determining field-measured 
BAFs, including recommendations for minimum data base requirements. A 
more detailed discussion of the factors which need to be considered 
when determining field-measured BAFs is provided in the TSD.
    Fraction Freely Dissolved (ffd). Nonpolar organic 
chemicals can exist in water in several different forms including 
freely dissolved chemicals in the water column, chemicals bound to 
particulate matter, or chemicals bound to dissolved organic matter in 
the water. The form of the chemical has been shown to affect 
bioaccumulation, with the freely dissolved fraction of a chemical 
considered to be the best expression of the bioavailable form to 
aquatic organisms. Because the amount of chemical that is freely 
dissolved may differ among water bodies due to differences in the total 
organic carbon in the water, bioaccumulation factors which are based on 
the concentration of freely dissolved chemical in the water will 
provide the most universal bioaccumulation factor for organic chemicals 
when averaging bioaccumulation factors from different studies (i.e., 
BAFs based on the freely dissolved chemical are most predictable 
between sites). However, BAFs based on the total concentration of the 
chemical in water (i.e., the freely dissolved plus that sorbed to 
particulate organic carbon and dissolved organic carbon) can often be 
measured more accurately than BAFs based on freely dissolved 
concentrations in water. Thus, if only BAFs based on total water 
concentrations are reported in a given BAF study, they can be used with 
information on the organic carbon content of water (from the BAF study, 
if available) to predict freely dissolved concentrations.
    To estimate the freely dissolved concentration, the fraction freely 
dissolved (ffd) in the above equation must be estimated, 
using information on the chemical's Kow and both dissolved 
and particulate organic carbon contents of the water. The equation used 
to estimate ffd is as follows:
[GRAPHIC] [TIFF OMITTED] TN14AU98.033

Where:

POC = concentration of particulate organic carbon (kg/L)
DOC = concentration of dissolved organic carbon (kg/L)
Kow = n-octanol water partition coefficient for the chemical

    Additional information on the derivation of Equation IIID-15 is 
provided in the TSD.
    POC/DOC Values. As noted above, when converting from the total 
concentration of a chemical to a freely dissolved concentration, the 
POC and DOC should be obtained from the original study that reports 
BAFs based on total concentrations of a chemical in water. However, if 
the POC and DOC concentrations are not reported in the BAF study, then 
reliable estimates of POC and DOC might be obtained from other studies 
of the same site used in the BAF study or closely related site(s) 
within the same water body. When using POC/DOC data from other studies 
of the same water body, care should be taken to ensure that 
environmental conditions that may affect POC or DOC concentrations are 
reasonably similar to those in the BAF study. Additional

[[Page 43811]]

guidance on selection of POC and DOC values is provided in the TSD.
    Kow Values. The Kow is the octanol-water 
partition coefficient of a chemical and is defined as the ratio of the 
concentration of a substance in the n-octanol phase to its 
concentration in the aqueous phase. Numerous investigations have 
demonstrated a linear relationship between the logarithm of the BCF and 
the logarithm of the octanol-water partition coefficient 
(Kow) for organic chemicals for fish and other aquatic 
organisms. Isnard and Lambert (1988) list various regression equations 
that illustrate this linear relationship. The underlying assumption for 
the linear relationship between the BCF and Kow is that the 
bioconcentration process can be viewed as a partitioning of a chemical 
between the lipid of the aquatic organisms and water and that the 
Kow is an useful surrogate for this partitioning process 
(Mackay, 1982).
    Several of the BAF procedures, including the BSAF method, use of 
the food chain model, and conversion of total chemical concentrations 
in water to freely dissolved chemical concentrations, rely on the 
Kow for chemicals. Because the Kow is used in 
calculating BAFs, it is important that the most accurate and reliable 
Kow measurements for a chemical are used. A variety of 
techniques are available to estimate or predict Kow values, 
some of which are more or less reliable depending on the Kow 
of the chemical.
    In this Notice, EPA discusses two options on how to select a 
reliable Kow value. The first option is EPA's existing 
guidance published in the Great Lakes Water Quality Initiative (60 FR 
15366 (March 23, 1995). A second option is more detailed, draft 
guidance on selecting Kow values which EPA has developed and 
is undergoing external peer review. The salient features of both the 
GLWQI Kow selection guidance (option one) and EPA's new, 
draft guidance (option two) are presented below. Additional details of 
both approaches are provided in the TSD.
    Guidance on selecting reliable values of Kow based on 
the GLWQI approach (option 1) is as follows.
    For chemicals with log Kow <4:

------------------------------------------------------------------------
             Priority                            Technique              
------------------------------------------------------------------------
1................................  Slow-stir.                           
                                   Shake-flask.                         
                                   Generator column.                    
2................................  Measured value from the CLOGP        
                                    program.                            
3................................  Reverse-phase liquid chromatography  
                                    on C18 with extrapolation to zero   
                                    percent solvent.                    
4................................  Reverse-phase liquid chromatography  
                                    on C18 without extrapolation to zero
                                    percent solvent.                    
5................................  Calculated by the CLOGP program.     
------------------------------------------------------------------------

    For chemicals with log Kow 4:

------------------------------------------------------------------------
             Priority                            Technique              
------------------------------------------------------------------------
1................................  Slow-stir,                           
                                   Generator-column.                    
2................................  Reverse-phase liquid chromatography  
                                    on C18 with extrapolation to zero   
                                    percent solvent.                    
3................................  Reverse-phase liquid chromatography  
                                    on C18 without extrapolation to zero
                                    percent solvent.                    
4................................  Shake-flask.                         
5................................  Measured value from the CLOGP        
                                    program.                            
6................................  Calculated by the CLOGP program.     
------------------------------------------------------------------------

    If no measured Kow is available, then the Kow 
must be estimated using the CLOGP program.
    Several general points should be kept in mind when using 
Kow values. Values should be used only if they were obtained 
from the original authors or from a critical review that supplied 
sufficient information. If more than one Kow value is 
available for a chemical using the highest priority method, then the 
arithmetic mean of the available log Kows or the geometric 
mean of the available Kows may be used. Because of potential 
interference due to radioactivity associated with impurities, values 
determined by measuring radioactivity in water and/or octanol should be 
considered less reliable than values determined by a Kow 
method of the same priority that employ nonradioactive techniques. The 
values determined using radioactive methods should be moved down one 
step in the priority below the values determined using the 
nonradioactive technique. Because the Kow is an intermediate 
value in the derivation of a BAF, the value used for the Kow 
of a chemical should not be rounded to less than three significant 
digits. Kow values that are outliers compared with other 
values for a chemical should not be used.
    The salient features of EPA's new draft methodology (option 2) for 
selecting reliable values of Kow is described below.
    I. Assemble/evaluate experimental and calculated data (e.g., CLOGP, 
LOGKOW, SPARC).
    II. If calculated log Kow is >8,
    A. Develop independent estimates of Kow using:

1. Liquid Chromatography (LC) methods with ``appropriate'' standards. 
(See TSD for guidelines for LC application)
2. Structure Activity Relationship (SAR) estimates extrapolated from 
similar chemicals where ``high quality'' measurements are available. 
``High quality'' SARs are defined in the TSD
3. Property Reactivity Correlation (PRC) estimates based on other 
measured properties (solubility, etc.)

    B. If calculated data are in reasonable agreement and are supported 
by independent estimates described above, report the average calculated 
value. Guidance on determining whether Kow values are in 
``reasonable agreement'' are presented in the TSD.
    C. If calculated/estimated data do not agree, use professional 
judgment to evaluate/blend/weight the calculated and estimated data to 
assign Kow value.
    D. Document rationale including relevant statistics.
    III. If calculated log Kow ranges from 6-8,
    A. Look for ``high quality'' measurements. These will generally be 
slow stir measurements, the exception being certain classes of 
compounds where micro emulsions tend to be less of a problem (i.e., 
PNA's, shake flask measurements are good to log Kow of 6.5).
    B. If measured data are available and are in reasonable agreement 
(both measurements and calculations), report average measured value.
    C. If measured data are in reasonable agreement, but differ from 
calculated values, develop independent estimates and apply professional 
judgment to evaluate/blend/weight the measured, calculated and 
estimated data to assign Kow value.
    D. If measured data are not in reasonable agreement (or if only one 
measurement is available), use II A, B, and C to produce a ``best 
estimate''; use this value to evaluate/screen the measured 
Kow data. Report the average value of screened data. If no 
measurements reasonably agree with ``best estimate'', apply 
professional judgment to evaluate/blend/weight the measured, calculated 
and estimated data to assign Kow.
    E. If measured data are unavailable, proceed through II A, B, C and 
report the ``best estimate''.
    F. Document rationale including relevant statistics.
    IV. If calculated log Kow is <6,
    A. Proceed as in III. Slow stir is the preferred method but shake 
flask data can be considered for all chemicals if sufficient attention 
has been given to emulsion problems in the measurement.

[[Page 43812]]

    The general operational guidelines for EPA's new draft methodology 
for selecting Kow values are as follows:
    1. For chemicals with log Kow >5, it is highly unlikely 
to find multiple ``high quality'' measurements. (Note: ``high quality'' 
is data judged to be reliable based on the guidelines presented in the 
TSD).
    2. ``High Quality'' measured data are preferred over estimates, but 
due to the scarcity of ``high quality'' data, the use of estimates is 
important in assigning Kow's.
    3. Kow measurements by slow stir are extendable to 
10\8\. Shake flask Kow measurements are extendable to 10 \6\ 
with sufficient attention to micro emulsion effects; for classes of 
chemicals that are not highly sensitive to emulsion effects (i.e., 
PNA's) this range may extend to 106.5.
    4. What is to be considered reasonable agreement in log 
Kow data (measured or estimated) depends primarily on the 
log Kow magnitude. The following standards for data 
agreement have been set for this guidance: 0.5 for log Kow 
>7; 0.4 for 6 log Kow 7; 0.3 for log 
Kow <6.
    5. Statistical methods should be applied to data as appropriate but 
application is limited due to the scarcity of data, and the 
determinate/methodic nature of most measurement error(s).
    The various techniques for measuring or calculating Kow 
that are referenced in both approaches above are summarized as follows:
     The slow-stir method requires adding the test chemical to a 
reaction flask which contains a water and octanol phase. The chemical 
partitions to these two phases under conditions of slow stirring the 
flask. After the phases are allowed to separate, the concentration of 
the test chemical in each phase is determined (Brooke et al., 1986).
     The shake-flask method also involves adding the chemical to a 
reaction flask with a mixture of octanol and water. In this method, 
however, the flask is shaken to obtain partitioning of the chemical 
between the octanol and water phases.
     The generator-column method involves filling a column with an 
inert material (silanized Chromosorb W or glass beads) that is coated 
with water-saturated octanol and contains the test chemical. Pumping 
water through the column results in an aqueous solution in equilibrium 
with the octanol phase. The water that leaves the column is extracted 
with specifically either an organic solvent or a C18 column 
that is then eluted with hexane or methanol (DeVoe et al., 1981; 
Woodburn et al., 1984; Miller et al., 1984).
     The reverse-phase liquid chromatography method involves 
adding the test chemical in a polar mobile phase (such as water or 
water-methanol) to a hydrophobic porous stationary phase (the 
C18 n-alkanes covalently bound to a silica support). The 
chemical partitions between the column and the polar aqueous phase. 
Kow values are estimated from linear equations between the 
Kow and retention indices that are derived for reference 
chemicals (Konemann et al., 1979; Veith et al., 1979; McDuffie, 1981; 
Garst and Wilson, 1984).
     The CLOGP Program is a computer program that contains 
measured Kow values for some chemicals and can calculate 
Kow values for additional chemicals based on similarities in 
their chemical structure with measured Kow values. The 
method used to calculate the Kow values is described in 
Hansch and Leo (1979).
     LOGKOW is essentially an expanded CLOGP with more recent 
training data and additional fragment constants. The developers were 
Philip Howard, William Meylan and co-workers at Syracuse Research 
Corporation. (See Meylan and Howard, 1994, for model details and 
performance information.)
     SPARC (SPARC Performs Automated Reasoning in Chemistry) is a 
mechanistic model developed at the Ecosystems Research Division of the 
National Exposure Research Laboratory of the Office of Research and 
Development of the U.S. Environmental Protection Agency by Sam 
Karickhoff, Lionel Carreira, and co-workers.
    In some situations, available data may require determination of a 
single Kow value for a class of chemicals or a mixture of 
closely related chemicals (e.g., when toxicity data are class- or 
mixture-specific). However, it is not possible to determine 
experimentally a valid Kow for a substance that is a mixture 
of chemicals (e.g., PCBs, toxaphene, chlordane). For calculating the 
composite freely dissolved fraction used to adjust a composite total 
BAF to a composite baseline BAF, a composite Kow value for 
the mixture can be calculated based on the sum of the total 
concentrations of the mixture components in water (e.g., individual 
congeners for PCBs), the sum of the dissolved concentrations of the 
mixture components in water, and the DOC and POC from the site for 
which the BAF was measured. An example of determining a composite 
Kow for deriving BAFs and AWQC for PCBs under the Great 
Lakes Water Quality Initiative is provided in 62 FR 117250 (March 12, 
1997). Additional details on this methodology are also provided in the 
TSD.
    Fraction lipid (fl)--lipid normalization of data. For 
lipophilic nonpolar organic chemicals, BAFs and BCFs are assumed to be 
directly proportional to the percent lipid in the edible tissue or 
whole body of the organism of interest. For example, an organism with 
two percent lipid content would be expected to accumulate twice the 
amount of a chemical as an organism with one percent lipid content, all 
else being equal. The proportionality of accumulation with lipid 
content for nonpolar organic chemicals has been extensively evaluated 
in the literature (Mackay, 1982; Connell, 1988; Barron, 1990) and is 
generally accepted. Different aquatic organisms, however, have 
different lipid contents thus making it difficult to compare BAFs and 
BCFs. BAFs and BCFs that have been measured in aquatic organisms that 
have different lipid contents can be compared by normalizing the lipids 
between organisms. The lipid values can be normalized by dividing the 
BAF or BCF by the mean lipid fraction in the tissue of the aquatic 
organism sampled. For example, if the BAF for a given chemical and 
tissue of an aquatic organism was determined to be 5,000 L/kg and the 
percent lipid in this tissue was 5 percent, the lipid-normalized BAF 
would be 100,000 L/kg-lipid (i.e., 5,000/0.05).
    Since lipid content is known to vary from one tissue to another and 
from one aquatic species to another, EPA recommends the percent lipid 
used to normalize the BAF or BCF (whole body or edible tissue) be 
obtained from the BAF or BCF study. Unless comparability can be 
determined across organisms, the fraction lipid should be determined in 
the test organism.
    (b) Baseline BAF Derived from BSAFs. When acceptable field-measured 
values of the BAF are not available for a nonpolar organic chemical, 
EPA recommends the use of the BSAF methodology to predict the BAF as 
the second method in the BAF data preference hierarchy. Although BSAFs 
may be used for measuring and predicting bioaccumulation directly from 
concentrations of chemicals in surface sediment, they may also be used 
to estimate BAFs (USEPA, 1993), as described below. Since BSAFs are 
based on field data and incorporate effects of metabolism, 
biomagnification, growth, and other factors, BAFs estimated from BSAFs 
will incorporate the net effect of all these factors. The BSAF approach 
is particularly beneficial for developing water quality criteria for 
chemicals

[[Page 43813]]

which are detectable in fish tissues and sediments, but are difficult 
to measure in the water column and have reduced bioaccumulation 
potential due to metabolism.
    In previously promulgated guidance, ratios of BSAFs of 
polychlorinated dibenzodioxins and polychlorinated dibenzofurans to a 
BSAF for 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) were used for 
evaluation of TCDD toxic equivalency associated with complex mixtures 
of these chemicals (i.e., bioaccumulation equivalency factors, see 60 
FR 15366). This approach is applicable to calculation of BAFs from 
BSAFs for other organic chemicals. The approach of estimating BAFs from 
BSAFs requires data from a steady-state (or near steady-state 
condition) between sediment and water for both a reference chemical 
``r'' with a measured BAF and other chemicals ``n=i'' for which BAFs 
are to be determined. The baseline BAF derived from a BSAF for a 
chemical ``i'' can be calculated using the following equation:
[GRAPHIC] [TIFF OMITTED] TN14AU98.034

Where:

(Baseline BAFl fd)i=BAF expressed on a 
freely dissolved and lipid-normalized basis for chemical of interest 
``i''
(Baseline BAFl fd)r=BAF expressed on a 
freely dissolved and lipid-normalized basis for reference chemical 
``r''
(BSAF)i=Biota-sediment accumulation factor for chemical of 
interest ``i''
(BSAF)r=Biota-sediment accumulation factor for the reference 
chemical ``r''
(Kow)i=octanol-water partition coefficient for 
chemical of interest ``i''
(Kow)r=octanol-water partition coefficient for 
the reference chemical ``r''

    Field-measured BSAFs. As shown in the following equation, BSAFs are 
determined by relating lipid-normalized concentrations of chemicals in 
an organism (Cl) to organic carbon-normalized concentrations 
of the chemicals in surface sediment samples associated with the 
average exposure environment of the organism (Csoc).
[GRAPHIC] [TIFF OMITTED] TN14AU98.035

    The lipid-normalized concentration of a chemical in an organism is 
determined by:
[GRAPHIC] [TIFF OMITTED] TN14AU98.036

where:

Ct=Concentration of the chemical in the wet tissue (either 
whole organism or specified tissue) (g/g)
fl = Fraction lipid content in the organism
    The organic carbon-normalized concentration of a chemical in 
sediment is determined by:
[GRAPHIC] [TIFF OMITTED] TN14AU98.037

where:

Cs=Concentration of chemical in sediment (g/g 
sediment)
foc=Fraction organic carbon in sediment

    Differences between BSAFs for different organic chemicals are good 
measures of the relative bioaccumulation potentials of the chemicals. 
When calculated from a common organism-sediment sample set, chemical-
specific differences in BSAFs primarily reflect the net effect of 
biomagnification, metabolism, bioenergetics, and bioavailability 
factors on each chemical's disequilibrium ratio between biota and 
sediment (i.e., the ratio of the freely dissolved concentration 
associated with water in the tissue to the freely dissolved 
concentration associated with the pore water in the sediment). At 
equilibrium, the disequilibrium (fugacity) ratio between biota and 
sediment is expected to be 1.0. However, deviations from 1.0 
(reflecting disequilibrium) are common and can reflect 
biomagnification, conditions where surface sediment has not reached 
equilibrium, kinetic limitations for chemical transfer, or biological 
processes such as growth or biotransformation. BSAFs are most useful 
(i.e., most predictable from one site to another) when measured under 
steady-state conditions. BSAFs measured for systems with new chemical 
loadings or rapid increases in loadings may be unreliable due to 
underestimation of steady-state Csocs.
    The trophic level to which the baseline BAF applies is the same as 
the trophic level of the organisms used in the determination of the 
BSAF. For each trophic level, a species mean baseline BAF is calculated 
as the geometric mean if more than one acceptable baseline BAF is 
predicted from BSAFs for a given species. For each trophic level, a 
trophic level-specific BAF is calculated as the geometric mean of the 
acceptable species mean baseline BAFs derived using BSAFs.
    The following procedural and quality assurance requirements should 
be met for field-measured BSAFs:
    1. The field studies used should be limited to those conducted with 
fish at or near the top of the aquatic food chain (i.e., in trophic 
levels 3 and/or 4). In situations where consumption of lower trophic 
level organisms represents an important exposure route, such as certain 
types of shellfish at trophic level 2, the field study should also 
include appropriate target species at this trophic level.
    2. Samples of surface sediments (0-1 cm is ideal) should be from 
locations in which sediment is regularly deposited and is 
representative of average surface sediment in the vicinity of the 
organism.
    3. The Kows used should be of acceptable quality as 
described in Section D.6 above.
    4. The site of the field study should not be so unique that the 
resulting BAF cannot be extrapolated to other locations where the 
criteria and values will apply.
    5. The percent lipid should be either measured or reliably 
estimated for the tissue used in the determination of the BAF.
    Further details on these requirements for predicting BAFs from BSAF 
measurements and the data supporting this approach are provided in the 
TSD.
    (c) Calculation of a Baseline BAF from a Laboratory-Measured BCF 
and FCM. As the third tier in the data preference hierarchy for 
nonpolar organic chemicals, EPA recommends the use of a predicted BAF 
derived from a technically defensible, laboratory measurement of the 
BCF and an appropriate FCM. Laboratory-measured BCFs are preferred over 
predicted BCFs because laboratory-measured BCFs inherently account for 
the effects of any metabolism of the chemical on the BCF. The equation 
for deriving a baseline BAF expressed on a freely dissolved and lipid-
normalized basis using this method is:

[[Page 43814]]

[GRAPHIC] [TIFF OMITTED] TN14AU98.038


Where:

Baseline BAFfdl = BAF expressed on a freely 
dissolved and lipid-normalized basis for a given trophic level
Measured BCFtT = BCF based on total concentration 
in tissue and water
fl = Fraction of the tissue that is lipid
ffd = Fraction of the total chemical in the test water that 
is freely dissolved
FCM = The food-chain multiplier either obtained from Tables IIID-1, 
IIID-2, or IIID-3 by linear interpolation for the appropriate trophic 
level, or from appropriate field data

    For each trophic level, the species mean baseline BAF is calculated 
as the geometric mean if more than one acceptable baseline BAF is 
predicted from laboratory-measured BCFs for a given species. For each 
trophic level, the trophic level-specific BAF is calculated as the 
geometric mean of the species mean baseline BAFs based on laboratory-
measured BCFs.
    Measured BCF t T. To estimate a measured 
BCFtT, information is needed on the total 
concentration of the chemical in the tissue of the organism and the 
total concentration of the chemical in the laboratory test waters. The 
equation to derive a measured BCFtT is:
[GRAPHIC] [TIFF OMITTED] TN14AU98.039

    A BCF derived from results of a laboratory exposure study is 
acceptable if the study has met certain specific technical criteria. 
These criteria include, but are not limited to:
    1. The test organism should not be diseased, unhealthy, or 
adversely affected by the concentration of the chemical because these 
attributes may alter accumulation of chemicals by otherwise healthy 
organisms.
    2. The total concentration of the chemical in the water should be 
measured and should be relatively constant during the steady-state time 
period.
    3. The organisms should be exposed to the chemical using a flow-
through or renewal procedure.
    4. For organic chemicals, the percent lipid should be either 
measured or reliably estimated for the tissue used in the determination 
of the BCF.
    5. For organic chemicals with log Kow greater than four, 
the concentrations of POC and DOC in the test solution should be either 
measured or reliably estimated. For organic chemicals with log 
Kow less than four, virtually all of the chemical is 
predicted to be freely dissolved, except in water with extremely high 
DOC and POC concentrations, which is not characteristic of laboratory 
dilution water used in BCF determinations.
    6. Laboratory-measured BCFs should be determined using fish 
species, but BCFs determined with molluscs and other invertebrates may 
be used with caution. For example, because invertebrates metabolize 
some chemicals less efficiently than vertebrates, a baseline BCF 
determined for such a chemical using invertebrates is expected to be 
higher than a comparable baseline BCF determined using fish.
    7. If laboratory-measured BCFs increase or decrease as the 
concentration of the chemical increases in the test solutions in a 
bioconcentration test, the BCF measured at the lowest test 
concentration that is above concentrations existing in the control 
water should be used (i.e., a BCF should not be calculated from a 
control treatment). The concentrations of an inorganic chemical in a 
bioconcentration test should be greater than normal background levels 
and greater than levels required for normal nutrition of the test 
species if the chemical is a micronutrient, but below levels that 
adversely affect the species. Bioaccumulation of an inorganic chemical 
might be overestimated if concentrations are at or below normal 
background levels due to, for example, nutritional requirements of the 
test organisms.
    8. For inorganic chemicals, BCFs should be used only if they are 
expressed on a wet weight basis. BCFs reported on a dry weight basis 
cannot be converted to wet weight unless a conversion factor is 
measured or reliably estimated for the tissue used in the determination 
of the BAF.
    9. BCFs for organic chemicals may be based on measurement of 
radioactivity only when the BCF is intended to include metabolites, 
when there is confidence that there is no interference due to 
metabolites, or when studies are conducted to determine the extent of 
metabolism, thus allowing for a proper correction.
    10. The calculation of the BCF must appropriately address growth 
dilution, which can be particularly important in affecting BCF 
determinations for poorly depurated chemicals.
    11. Other aspects of the methodology used should be similar to 
those described by the American Society of Testing and Materials (ASTM, 
1990).
    In addition, the magnitude of the octanol-water partition 
coefficient (Kow) and the availability of corroborating BCF 
data should be considered. For example, some chemicals with high log 
Kows may require longer than 28 days to obtain steady state 
conditions between the organism and the water column.
    FCMs. The FCM reflects a chemical's tendency to biomagnify in the 
aquatic food web. Food chain multipliers in Tables IIID-1, IIID-2 and 
IIID-3 have been calculated as the ratio of the baseline BAFs for 
various trophic levels to the baseline BCF using the model of Gobas 
(1993). Values of FCMs greater than 1.0 indicate biomagnification and 
typically apply to organic chemicals with log Kow values 
between 4.0 and 9.0. For a given chemical, FCMs tend to be greater at 
higher trophic levels, although FCMs for trophic level three can be 
higher than those for trophic level four. The final GLI established 
FCMs using the food chain model by Gobas (1993) for a range of log 
Kow values from 2.0 to 9.0 at intervals of a tenth of a log 
Kow value.
    EPA recommends using the biomagnification model by Gobas (1993) to 
derive FCMs for nonpolar organic chemicals for several reasons. First, 
the Gobas model includes both benthic and pelagic food chains, thereby 
incorporating exposure of organisms to chemicals from both the 
sediments and the water column. Second, the input data needed to run 
the model can be readily defined. Third, the predicted BAFs using the 
model are in agreement

[[Page 43815]]

with field-measured BAFs for chemicals, even those with very high log 
Kows. Finally, the model predicts chemical residues in 
benthic organisms using equilibrium partitioning theory, which is 
consistent with EPA's sediment quality criteria effort.
    The Gobas model requires input of specific data on the structure of 
the food chain and the water quality characteristics of the water body 
of interest. For example, in the GLI and in these proposed revisions to 
the AWQC methodology, it is assumed that the food chain consists of 
four trophic levels. Trophic level 1 is phytoplankton, trophic level 2 
is zooplankton, trophic level 3 is forage fish (e.g., sculpin and 
smelt), and trophic level 4 are predator fish (e.g., salmonids). 
Additional assumptions must be made regarding the composition of the 
aquatic species diet (e.g., salmonids consume 10 percent sculpin, 50 
percent alewives, and 40 percent smelt), the physical parameters of the 
aquatic species (e.g., lipid values), and the water quality 
characteristics (e.g., water temperature, sediment organic carbon).
    EPA has estimated FCMs using three different potential food web 
structures. The first food web structure includes both a benthic and 
pelagic food chains. The FCMs range from 1.00 to about 27 for log 
Kow values ranging from 2.0 to 9.0. The second food web 
structure includes only the pelagic food chain. The FCMs for this food 
web structure range from 1.0 to about 4 for log Kow values 
ranging from 2.0 to 9.0. Finally, the third food web structure includes 
only the benthic food chain. The FCMs for this scenario range from 1.0 
to about 57 for log Kow values ranging from 2.0 to 9.0. The 
resulting FCMs for trophic levels 2, 3, and 4 are included in Tables 
IIID-1, IIID-2, and IIID-3. A more detailed discussion on the model and 
the input parameters for the model are included in the TSD for BAFs.
    In addition to determining FCMs for organic substances using the 
Gobas (1993) model, EPA also recommends the use of FCMs derived from 
field data where data are sufficient to enable scientifically valid and 
reliable determinations to be made. Currently, field-measured FCMs are 
the only method recommended for estimating FCMs for inorganic chemicals 
because appropriate model-derived estimates are not yet available (see 
Section D.8). Similarly, field-measured FCMs can also be determined for 
organic chemicals. Compared to the model-based FCMs described 
previously, properly derived field-based FCMs may offer some advantages 
in some situations. For example, field-measured FCMs rely on measured 
contaminant concentrations in tissues of biota and therefore inherently 
account for any contaminant metabolism which may occur. Field-measured 
FCMs may also be useful for estimating BAFs for some highly hydrophobic 
contaminants whose water column concentrations are very difficult to 
determine with accuracy and precision. Furthermore, field-measured FCMs 
may better reflect local conditions that can influence bioaccumulation, 
such as differences in food web structure, exposure pathways, water 
body type, and target species. Finally, use of field-measured FCMs in 
estimating BAFs may enable existing data on contaminant concentrations 
in aquatic organisms to be used in situations where companion water 
column data are unavailable or are judged to be unreliable for 
derivation of a BAF.

                        Table IIID-1. Food-Chain Multipliers for Trophic Levels 2, 3 & 4                        
                                         [Pelagic and Benthic Structure]                                        
----------------------------------------------------------------------------------------------------------------
                                                                  Trophic  Level     Trophic a    Trophic  Level
                            Log K ow                                     2            Level 3            4      
----------------------------------------------------------------------------------------------------------------
<2.0............................................................           1.000           1.000           1.000
2.0.............................................................           1.000           1.005           1.000
2.5.............................................................           1.000           1.010           1.002
3.0.............................................................           1.000           1.028           1.007
3.1.............................................................           1.000           1.034           1.007
3.2.............................................................           1.000           1.042           1.009
3.3.............................................................           1.000           1.053           1.012
3.4.............................................................           1.000           1.067           1.014
3.5.............................................................           1.000           1.083           1.019
3.6.............................................................           1.000           1.103           1.023
3.7.............................................................           1.000           1.128           1.033
3.8.............................................................           1.000           1.161           1.042
3.9.............................................................           1.000           1.202           1.054
4.0.............................................................           1.000           1.253           1.072
4.1.............................................................           1.000           1.315           1.096
4.2.............................................................           1.000           1.380           1.130
4.3.............................................................           1.000           1.491           1.178
4.4.............................................................           1.000           1.614           1.242
4.5.............................................................           1.000           1.766           1.334
4.6.............................................................           1.000           1.950           1.459
4.7.............................................................           1.000           2.175           1.633
4.8.............................................................           1.000           2.452           1.871
4.9.............................................................           1.000           2.780           2.193
5.0.............................................................           1.000           3.181           2.612
5.1.............................................................           1.000           3.643           3.162
5.2.............................................................           1.000           4.188           3.873
5.3.............................................................           1.000           4.803           4.742
5.4.............................................................           1.000           5.502           5.821
5.5.............................................................           1.000           6.266           7.079
5.6.............................................................           1.000           7.096           8.551
5.7.............................................................           1.000           7.962          10.209
5.8.............................................................           1.000           8.841          12.050
5.9.............................................................           1.000           9.716          13.964
6.0.............................................................           1.000          10.556          15.996
6.1.............................................................           1.000          11.337          17.783

[[Page 43816]]

                                                                                                                
6.2.............................................................           1.000          12.064          19.907
6.3.............................................................           1.000          12.691          21.677
6.4.............................................................           1.000          13.228          23.281
6.5.............................................................           1.000          13.662          24.604
6.6.............................................................           1.000          13.980          25.645
6.7.............................................................           1.000          14.223          26.363
6.8.............................................................           1.000          14.355          26.669
6.9.............................................................           1.000          14.388          26.669
7.0.............................................................           1.000          14.305          26.242
7.1.............................................................           1.000          14.142          25.468
7.2.............................................................           1.000          13.852          24.322
7.3.............................................................           1.000          13.474          22.856
7.4.............................................................           1.000          12.987          21.038
7.5.............................................................           1.000          12.517          18.967
7.6.............................................................           1.000          11.708          16.749
7.7.............................................................           1.000          10.914          14.388
7.8.............................................................           1.000          10.069          12.050
7.9.............................................................           1.000           9.162           9.840
8.0.............................................................           1.000           8.222           7.798
8.1.............................................................           1.000           7.278           6.012
8.2.............................................................           1.000           6.361           4.519
8.3.............................................................           1.000           5.489           3.311
8.4.............................................................           1.000           4.683           2.371
8.5.............................................................           1.000           3.949           1.663
8.6.............................................................           1.000           3.296           1.146
8.7.............................................................           1.000           2.732           0.778
8.8.............................................................           1.000           2.246           0.521
8.9.............................................................           1.000           1.837           0.345
9.0.............................................................           1.000           1.493           0.226
----------------------------------------------------------------------------------------------------------------
a The FCMs for trophic level 3 are the geometric mean of the FCMs for sculpin and alewife.                      


                        Table IIID-2. Food-Chain Multipliers for Trophic Levels 2, 3 & 4                        
                                             [All Benthic Structure]                                            
----------------------------------------------------------------------------------------------------------------
                                                                  Trophic  Level     Trophic a    Trophic  Level
                            Log K ow                                     2            Level 3            4      
----------------------------------------------------------------------------------------------------------------
<2.0............................................................           1.000           1.000           1.000
2.0.............................................................           1.000           1.009           1.001
2.1.............................................................           1.000           1.010           1.001
2.2.............................................................           1.000           1.011           1.001
2.3.............................................................           1.000           1.013           1.002
2.4.............................................................           1.000           1.015           1.002
2.5.............................................................           1.000           1.018           1.002
2.6.............................................................           1.000           1.022           1.003
2.7.............................................................           1.000           1.026           1.003
2.8.............................................................           1.000           1.032           1.004
2.9.............................................................           1.000           1.039           1.005
3.0.............................................................           1.000           1.048           1.006
3.1.............................................................           1.000           1.060           1.008
3.2.............................................................           1.000           1.074           1.010
3.3.............................................................           1.000           1.092           1.013
3.4.............................................................           1.000           1.114           1.017
3.5.............................................................           1.000           1.142           1.022
3.6.............................................................           1.000           1.177           1.029
3.7.............................................................           1.000           1.222           1.039
3.8.............................................................           1.000           1.277           1.053
3.9.............................................................           1.000           1.347           1.072
4.0.............................................................           1.000           1.433           1.099
4.1.............................................................           1.000           1.541           1.138
4.2.............................................................           1.000           1.676           1.195
4.3.............................................................           1.000           1.843           1.276
4.4.............................................................           1.000           2.050           1.392
4.5.............................................................           1.000           2.306           1.559
4.6.............................................................           1.000           2.620           1.796
4.7.............................................................           1.000           3.004           2.131
4.8.............................................................           1.000           3.470           2.595
4.9.............................................................           1.000           4.032           3.232
5.0.............................................................           1.000           4.702           4.087

[[Page 43817]]

                                                                                                                
5.1.............................................................           1.000           5.492           5.215
5.2.............................................................           1.000           6.411           6.668
5.3.............................................................           1.000           7.462           8.501
5.4.............................................................           1.000           8.643          10.754
5.5.............................................................           1.000           9.942          13.457
5.6.............................................................           1.000          11.337          16.617
5.7.............................................................           1.000          12.800          20.213
5.8.............................................................           1.000          14.293          24.192
5.9.............................................................           1.000          15.774          28.468
6.0.............................................................           1.000          17.202          32.920
6.1.............................................................           1.000          18.539          37.405
6.2.............................................................           1.000          19.753          41.764
6.3.............................................................           1.000          20.822          45.836
6.4.............................................................           1.000          21.730          49.472
6.5.............................................................           1.000          22.469          52.544
6.6.............................................................           1.000          23.037          54.949
6.7.............................................................           1.000          23.433          56.610
6.8.............................................................           1.000          23.659          57.472
6.9.............................................................           1.000          23.717          57.501
7.0.............................................................           1.000          23.606          56.679
7.1.............................................................           1.000          23.326          55.007
7.2.............................................................           1.000          22.873          52.507
7.3.............................................................           1.000          22.246          49.227
7.4.............................................................           1.000          21.443          45.254
7.5.............................................................           1.000          20.467          40.714
7.6.............................................................           1.000          19.327          35.780
7.7.............................................................           1.000          18.040          30.657
7.8.............................................................           1.000          16.629          25.572
7.9.............................................................           1.000          15.129          20.744
8.0.............................................................           1.000          13.580          16.359
8.1.............................................................           1.000          12.026          12.547
8.2.............................................................           1.000          10.510           9.368
8.3.............................................................           1.000           9.068           6.822
8.4.............................................................           1.000           7.732           4.856
8.5.............................................................           1.000           6.522           3.387
8.6.............................................................           1.000           5.448           2.321
8.7.............................................................           1.000           4.513           1.567
8.8.............................................................           1.000           3.711           1.045
8.9.............................................................           1.000           3.032           0.689
9.0.............................................................           1.000           2.465          0.451 
----------------------------------------------------------------------------------------------------------------
a The FCMs for trophic level 3 are the geometric mean of the FCMs for sculpin and alewife.                      


                        Table IIID-3. Food-Chain Multipliers for Trophic Levels 2, 3 & 4                        
                                             [All Pelagic Structure]                                            
----------------------------------------------------------------------------------------------------------------
                                                                   Trophic Level  Trophica Level   Trophic Level
                             Log Kow                                     2               3               4      
----------------------------------------------------------------------------------------------------------------
<2.0............................................................           1.000           1.000           1.000
2.0.............................................................           1.000           1.000           1.001
2.1.............................................................           1.000           1.000           1.001
2.2.............................................................           1.000           1.000           1.001
2.3.............................................................           1.000           1.000           1.002
2.4.............................................................           1.000           1.000           1.002
2.5.............................................................           1.000           1.001           1.002
2.6.............................................................           1.000           1.001           1.003
2.7.............................................................           1.000           1.001           1.003
2.8.............................................................           1.000           1.001           1.004
2.9.............................................................           1.000           1.001           1.005
3.0.............................................................           1.000           1.002           1.006
3.1.............................................................           1.000           1.002           1.007
3.2.............................................................           1.000           1.002           1.009
3.3.............................................................           1.000           1.003           1.011
3.4.............................................................           1.000           1.004           1.013
3.5.............................................................           1.000           1.005           1.016
3.6.............................................................           1.000           1.006           1.021
3.7.............................................................           1.000           1.007           1.026
3.8.............................................................           1.000           1.009           1.032
3.9.............................................................           1.000           1.011           1.040

[[Page 43818]]

                                                                                                                
4.0.............................................................           1.000           1.014           1.050
4.1.............................................................           1.000           1.018           1.063
4.2.............................................................           1.000           1.022           1.078
4.3.............................................................           1.000           1.028           1.097
4.4.............................................................           1.000           1.034           1.121
4.5.............................................................           1.000           1.043           1.150
4.6.............................................................           1.000           1.053           1.185
4.7.............................................................           1.000           1.066           1.228
4.8.............................................................           1.000           1.081           1.280
4.9.............................................................           1.000           1.099           1.342
5.0.............................................................           1.000           1.121           1.415
5.1.............................................................           1.000           1.147           1.502
5.2.............................................................           1.000           1.176           1.603
5.3.............................................................           1.000           1.210           1.719
5.4.............................................................           1.000           1.248           1.851
5.5.............................................................           1.000           1.289           1.999
5.6.............................................................           1.000           1.333           2.162
5.7.............................................................           1.000           1.379           2.337
5.8.............................................................           1.000           1.425           2.521
5.9.............................................................           1.000           1.471           2.711
6.0.............................................................           1.000           1.514           2.900
6.1.............................................................           1.000           1.554           3.083
6.2.............................................................           1.000           1.589           3.254
6.3.............................................................           1.000           1.619           3.407
6.4.............................................................           1.000           1.643           3.536
6.5.............................................................           1.000           1.660           3.637
6.6.............................................................           1.000           1.671           3.705
6.7.............................................................           1.000           1.674           3.738
6.8.............................................................           1.000           1.669           3.733
6.9.............................................................           1.000           1.657           3.688
7.0.............................................................           1.000           1.636           3.602
7.1.............................................................           1.000           1.606           3.474
7.2.............................................................           1.000           1.567           3.305
7.3.............................................................           1.000           1.518           3.094
7.4.............................................................           1.000           1.458           2.848
7.5.............................................................           1.000           1.389           2.570
7.6.............................................................           1.000           1.308           2.270
7.7.............................................................           1.000           1.219           1.958
7.8.............................................................           1.000           1.122           1.647
7.9.............................................................           1.000           1.020           1.349
8.0.............................................................           1.000           0.915           1.076
8.1.............................................................           1.000           0.810           0.835
8.2.............................................................           1.000           0.707           0.631
8.3.............................................................           1.000           0.610           0.466
8.4.............................................................           1.000           0.520           0.336
8.5.............................................................           1.000           0.438           0.237
8.6.............................................................           1.000           0.366           0.164
8.7.............................................................           1.000           0.303           0.112
8.8.............................................................           1.000           0.249           0.075
8.9.............................................................           1.000           0.204           0.050
9.0.............................................................           1.000           0.166           0.033
----------------------------------------------------------------------------------------------------------------
a The FCMs for trophic level 3 are the geometric mean of the FCMs for sculpin and alewife.                      

    As discussed below and in the TSD, FCMs are related to and can be 
determined from biomagnification factors (BMF). For example:
[GRAPHIC] [TIFF OMITTED] TN14AU98.045

[GRAPHIC] [TIFF OMITTED] TN14AU98.046

[GRAPHIC] [TIFF OMITTED] TN14AU98.047


[[Page 43819]]


Where:

FCM=Food chain multiplier for designated trophic level (TL2, TL3, or 
TL4)
BMF=Biomagnification factor for designated trophic level (TL2, TL3, or 
TL4)

    The basic difference between FCMs and BMFs is that FCMs relate back 
to trophic level one (or trophic level two as assumed by the Gobas 
(1993) model), whereas BMFs always relate back to the next lowest 
trophic level. For nonpolar organic chemicals, biomagnification factors 
can be calculated from tissue residue concentrations determined in 
biota at a site according to the following equation.
[GRAPHIC] [TIFF OMITTED] TN14AU98.048

[GRAPHIC] [TIFF OMITTED] TN14AU98.049

[GRAPHIC] [TIFF OMITTED] TN14AU98.050

Where:

C=Lipid-normalized concentration of chemical in tissue of appropriate 
biota that occupy the specified trophic level (TL2, TL3, or TL4).

    For inorganic chemicals, BMFs are determined as shown above, except 
that tissue concentrations expressed on a wet-weight basis and are not 
lipid normalized. In calculating field-derived BMFs for determining 
FCMs, care must be taken to ensure that the biota upon which they are 
based actually represent functional predator-prey relationships at the 
study site, and therefore, would accurately reflect any 
biomagnification that may occur at the site.
    As with field-measured BAFs, the potential advantages of using 
field data for estimating bioaccumulation can be offset by improper 
collection and use of information. In calculating field-based FCMs, 
steps similar to those recommended for determining field-measured BAFs 
need to be taken to ensure that the resulting FCMs accurately represent 
potential exposures to the target population at the site(s) of 
interest. Some of the general procedural and quality assurance 
requirements that are important for determining field-measured FCMs 
include:
    1. A food web analysis should be conducted for the site from which 
the tissue concentration data are to be determined (or have been 
already been determined) to identify the appropriate trophic levels for 
the aquatic organisms and appropriate predator-prey relationships. To 
assist in trophic level determinations, EPA is in the process of 
finalizing its draft trophic level and exposure analysis documents 
(U.S. EPA, 1995b; 1995c, 1995d) which include trophic level analyses of 
numerous species in the aquatic-based food web.
    2. The aquatic organisms sampled from each trophic level should 
reflect the most important exposure pathways leading to human exposure 
via consumption of aquatic organisms. For higher trophic levels (e.g., 
3 and 4), aquatic species should also reflect those that are commonly 
consumed by humans.
    3. Collection of tissue concentration field data for a specific 
site for which criteria are to be derived and with the specific species 
of concern are preferred.
    4. If data cannot be collected from every site for which criteria 
are to be derived, the site of the field study should not be so unique 
that the FCM values cannot be extrapolated to other locations where the 
criteria and values will apply.
    5. Samples of the appropriate resident species and the water in 
which they reside should be collected and analyzed using appropriate, 
sensitive, accurate, and precise methods to determine the 
concentrations of bioaccumulative chemicals present in the tissues.
    6. For organic chemicals, the percent lipid should be either 
measured or reliably estimated for the tissue used in the determination 
of the lipid normalized concentration in the organism's edible tissues.
    7. The tissue concentrations should reflect average exposure over 
the time period required to achieve steady-state conditions for the 
contaminant in the target species.
    (d) Calculation of a Baseline BAF from a Kow and FCM. As 
the fourth tier in the data preference hierarchy for nonpolar organic 
chemicals (e.g., when acceptable, field-measured BAFs, BSAFs, or 
laboratory-measured BCFs are unavailable), EPA recommends the use of 
the Kow for a chemical and a FCM for estimating baseline 
BAFs at various trophic levels. For each trophic level, a predicted 
baseline BAF can be calculated as:
Where:

Baseline BAFfd=BAF expressed on a freely dissolved and 
lipid-normalized basis for a given trophic level
FCM=The food-chain multiplier obtained from tables IIID-1 to IIID-3 by 
linear interpolation (or from appropriate field data) for the 
appropriate trophic level
Kow=Octanol-water partition coefficient

This equation is based on the assumption that a baseline BCF is 
approximately equal to the Kow for the chemical. This 
equation was used in the final GLI and its derivation is included in 
the TSD.
[GRAPHIC] [TIFF OMITTED] TN14AU98.051

    (e) Metabolism. Many organic chemicals that are accumulated by 
aquatic organisms are transformed to some extent by the organism's 
metabolic processes, but the rate of metabolism varies widely across 
chemicals and species. For most organic chemicals, metabolism increases 
the depuration rate and reduces the BAF. Field-measured BAFs and BSAFs 
automatically take into account any metabolism that occurs and 
therefore more accurately predict bioaccumulation than predicted BAFs 
based on laboratory measurements. Because of the uncertainties 
associated with predicting chemical metabolism, EPA prefers that the 
bioaccumulation

[[Page 43820]]

potential of a chemical be determined based on field data. Predicted 
BAFs obtained by multiplying laboratory-measured BCFs by a field-
measured FCM also take into account chemical metabolism if it occurs. 
Predicted BAFs that are obtained by multiplying a laboratory-measured 
BCF by a model-derived FCM take into account the effect of metabolism 
on the BCF, but do not take into account the effect of metabolism on 
the FCM. Predicted BAFs that are obtained by multiplying a predicted 
BCF by a FCM make no allowance for metabolism.
    EPA is aware that for some chemical classes, such as PAHs, 
metabolism can have a significant effect on the bioaccumulation for the 
chemical. Unfortunately, EPA is not aware of any generalized approach 
for predicting the effects of metabolism. For this reason, EPA suggests 
that BAFs be reviewed for consistency with all available data 
concerning bioaccumulation of a chemical. In particular, information on 
metabolism, molecular size, or other physicochemical properties which 
might enhance or inhibit bioaccumulation should be considered.
7. BAFs Used in Deriving AWQC
    After the baseline BAF has been derived for a nonpolar organic 
chemical using one of the four methods described above, the next step 
is to calculate a BAF that will be used in the derivation of AWQC. This 
requires information on: (1) the baseline BAF for the chemical of 
interest using one of the four methods described above; (2) the percent 
lipid of the aquatic organisms consumed by humans at the site of 
interest; and (3) the freely dissolved fraction of the chemical in the 
ambient water of interest. For each trophic level, the equation for 
calculating a BAF for use in deriving the AWQC is:

[GRAPHIC] [TIFF OMITTED] TN14AU98.052

Where:
Baseline BAFlfd = BAF expressed on a freely 
dissolved and lipid-normalized basis for trophic level ``n''
fl(TLn) = Fraction lipid of aquatic species 
consumed at trophic level ``n''
ffd = Fraction of the total chemical in water that is freely 
dissolved
    Baseline BAF. The baseline BAFs used in this equation are those 
derived from the equations presented in Section D.6 above.
    Lipid Content of Aquatic Species Consumed by Humans. As discussed 
above, the percent lipid of the aquatic species consumed by humans is 
needed when deriving BAFs for a chemical that will be used for deriving 
AWQC. This information is needed to provide an accurate 
characterization of the potential exposure to a chemical from ingestion 
of aquatic organisms. The percent lipid fraction used when calculating 
a BAF should, if possible, be weighted by the consumption rate of those 
aquatic species consumed by the target population (e.g., general 
population, sport anglers, subsistence fishers). A consumption-weighted 
percent lipid is recommended because it provides a more accurate 
characterization of the potential exposure to humans than simply 
averaging lipid values from a variety of species in a given geographic 
area which may or may not be eaten by humans. Since baseline BAFs are 
determined for each trophic level and must be adjusted to reflect the 
lipid content of consumed aquatic species, EPA recommends that the 
consumption-weighted lipid content of consumed aquatic organisms also 
be determined for each trophic level. For each trophic level, the 
consumption-weighted fraction lipid can be determined by the following 
equation:

[GRAPHIC] [TIFF OMITTED] TN14AU98.040

where:
f 7l = Lipid fraction representative of aquatic species at a 
given trophic level eaten by the target population
CR 5i = Consumption rate of species ``i'' of a given trophic 
level eaten by the target population
CR 5tot = Consumption rate of all species at that same 
trophic level eaten by the target population
f5l,I 5= Lipid fraction of species ``i'' eaten by 
the target population
    If sufficient information is not available to derive trophic level-
specific lipid contents, then States and Tribes may choose to calculate 
an overall consumption-weighted lipid content value that combines data 
across relevant trophic levels.
    To estimate the consumption-weighted percent lipid content of 
consumed aquatic species within various trophic levels, information is 
needed on: (1) the type and quantity of aquatic biota consumed by 
humans, (2) the trophic position of those species, and (3) the percent 
lipid of the aquatic biota consumed by humans. The types and quantity 
of aquatic species eaten by individuals differ throughout the United 
States. Thus, to determine the lipid content of the aquatic species of 
interest (e.g., freshwater and estuarine finfish and shellfish) eaten 
by local populations, EPA recommends that States use available local 
information on consumption rates specific to the types and quantity of 
aquatic species eaten by target populations. Data on consumption rates 
of species may be available from fish and shellfish consumption surveys 
conducted within the State or in States or regions that have similar 
finfish and shellfish species. EPA has published the document 
Consumption Surveys for Fish and Shellfish. A Review and Analysis of 
Survey Methods (Feb. 1992, EPA 822/R-92-001) which may assist in 
conducting and analyzing the results of such surveys. If local data on 
species-specific consumption rates are not available, States may wish 
to use regional data on consumption rates of aquatic species found in 
fresh and estuarine waters, available from USDA's CSFII (USEPA, 1998). 
These regional data from the CSFII are presented in the TSD 
accompanying this Notice. Such data may be used with local data on 
lipid contents of the consumed aquatic species.
    The second type of information required is data on the trophic 
level of consumed aquatic species corresponding to the consumption rate 
survey. In order to estimate trophic position, information on the 
dietary preferences of the organisms of interest is required. The 
dietary composition (and trophic level) of aquatic organisms can vary 
with the size and age of the organism, the type of ecosystem, season, 
and other factors, which can complicate precise determinations of 
trophic level status. Therefore, whenever possible, it is recommended 
that information on such attributes (particularly size of consumed 
organisms) be obtained from the consumption survey. EPA has developed 
draft guidance on estimating trophic status of numerous aquatic 
species, in addition to the wildlife that consume them, which is 
currently being finalized (USEPA 1995b; 1995c; 1995d). Once finalized, 
this guidance is recommended in situations where sufficient local 
information on trophic status is not available.

[[Page 43821]]

    The third critical piece of information is the percent lipid values 
of the aquatic biota consumed by humans. The lipid content of a 
particular aquatic species may vary by geographic region, possibly a 
result of different dietary composition. Therefore, lipid values based 
on good- quality data from species consumed by the local population of 
interest are more appropriate than nationally derived values. If local 
data on both aquatic species consumption rates and lipid contents are 
not available, States may wish to use national default lipid values 
calculated by EPA. Using the general relationship in Equation IIID-30 
and information on national finfish and shellfish consumption rates at 
various trophic levels, EPA has developed a national default 
consumption- weighted mean lipid values of 2.3% at trophic level 2, 
1.5% at trophic level 3, and 3.1% at trophic level 4 (rounded to two 
significant digits for convenience).
    It should be noted that if a national default lipid value was 
determined based only on the species with the highest mean lipid 
content within each CSFII species category and trophic level (e.g., 
giving 100 percent of the weighting to lake trout which has the highest 
lipid content among the species in the trout category), the resulting 
consumption-weighted lipid values are 3.0% at trophic level 2, 2.2% at 
trophic level 3, and 6.2% at trophic level 4. The reason that there is 
not greater difference between the mean and high estimates of the 
default lipid values within each trophic level is probably due to the 
fact that the national mean consumption rates in the CSFII survey are 
weighted heavily by relatively lean aquatic organisms such as shrimp, 
crab, perch, and flounder. Because local or regional consumption 
patterns may deviate from national norms, it is further recommended 
that local and regional data on consumption patterns be used whenever 
available. When such local consumption data are used, however, 
information on lipid content of those locally-consumed species is also 
required (national default consumption-weighted lipid content values do 
not necessarily apply to local consumption data). Additional 
description of the data and methods to derive the default lipid values 
are provided in the TSD accompanying this Notice.
    Freely Dissolved Fraction. Equation IIID-15 for estimating the 
fraction freely dissolved for baseline BAFs is also used here. In this 
case, however, the POC and DOC values should be based on the site where 
the BAF and the criterion will be applied and not where the samples 
were collected. If the POC and DOC values are not available for that 
site, then data from sites expected to be similar to those to which the 
AWQC is being applied can be used. If such data are unavailable, then 
the default values for POC and DOC can be used. EPA has developed 
national default values of 0.48 mg/L (4.8 x 10-7 kg/L) for 
POC and 2.9 mg/L (2.9 x 10-6 kg/L) for DOC. Both of these 
values are 50th percentile values (medians) based on an analysis of 
over 132,000 DOC values and 81,000 POC values contained in EPA's STORET 
data base. These default values reflect the combination of values for 
streams, lakes and estuaries across the United States. Based on these 
data, EPA has also derived default values at a more disaggregated level 
(e.g., for individual States and water body types) which, in some 
situations, may provide more appropriate estimates of POC and DOC 
concentrations associated with the field BAF study than the national 
default medians listed above. Additional description of the STORET DOC/
POC data base used to derive the default values, including POC and DOC 
information presented at a more disaggregated level, is provided in the 
TSD. The Kow value for the chemical will be the same as used 
for deriving the baseline BAF for the chemical.
    As noted above, standardizing BAFs based on the freely dissolved 
concentration in water allows a common basis for averaging BAFs from 
several studies. However, for use in criteria development, these BAFs 
must be converted back to values based on the total concentration in 
the water to be consistent with monitored water column and effluent 
concentrations, which are typically based on total concentrations of 
chemicals in the water. This is done simply by multiplying the freely 
dissolved baseline BAF by the fraction of the freely dissolved chemical 
in water bodies where criteria are to be set, as shown in Equation 
IIID-29.
8. Inorganic Substances
    For inorganic chemicals, either (1) a field-measured BAF; (2) a 
laboratory-measured BCF multiplied by a field-measured FCM; or (3) a 
laboratory-measured BCF should be used. These measured values are 
recommended because no method is available for reliably predicting BCFs 
or BAFs for inorganic chemicals; BCFs and BAFs vary from one 
invertebrate to another, from one fish to another, and from one tissue 
to another. Unlike nonpolar organic chemicals, lipid normalization does 
not apply. For many inorganic chemicals, the BCF will be equal to the 
BAF. In other words, for these chemicals there is no measurable 
bioaccumulation from food or other nonwater sources. There are 
exceptions however, such as mercury and selenium, which can 
bioaccumulate substantially.
9. SAB Comments
    EPA's Science Advisory Board has reviewed the BAF methodology three 
times since 1992. In December of 1992, SAB issued the report 
``Evaluation of the Guidance for the Great Lakes Water Quality 
Initiative'' (EPA-SAB-EPEC/DWC-93-005). The SAB reviewed four technical 
guidance documents for developing water quality criteria in the Great 
Lakes Basin as a part of the Proposed Great Lakes Water Quality 
Initiative including the proposed GLI BAF methodology. The 1992 SAB 
report stated that:
    The subcommittee finds the BAF procedure is more advanced and 
scientifically credible than existing BCF procedures. The use of the 
BCF, FCM, and BAF approach appear to be fundamentally sound. However, a 
major inconsistency exists between field data for some chemicals 
(Reinert, 1970) and the conceptual model of Thomann (1989) for food 
chain derived residues. Efforts should be devoted to clarifying and 
improving the documentation and the issues discussed below with a view 
to presenting a straight-forward procedure with associated estimates of 
confidence levels. It is the Subcommittee's opinion that with some 
modification a credible BAF estimation method can be developed 
exploiting present knowledge. Based on the SAB comments, EPA revised 
the BAF methodology and finalized the GLI in March 1995.
    The second SAB review occurred as part of the overall review of the 
Revisions to the AWQC methodology. The SAB provided a report called 
``Review of the Ongoing Revisions of the Methodology for Deriving 
National Ambient Water Quality Criteria for the Protection of Human 
Health'' which stated:

    We strongly urge the Agency to base AWQC on sound experimental 
evidence that bioaccumulation does occur, rather than on 
hypothetical assumptions that bioaccumulation might occur. The 
Committee believes that the strategy of setting AWQC by measuring 
contaminant concentrations in certain biota and then applying either 
a BCF or a BAF to calculate water concentrations may not accurately 
reflect the complex ways in which the real environment operates. 
Although we support EPA's efforts to develop well-validated BAFs, 
for the time being the Committee recommends that the Agency rely 
more heavily on BCFs rather than BAFs, because

[[Page 43822]]

of the higher likelihood of collecting an adequate BCF data base.

    Finally, in September 1995, the SAB provided a report to EPA 
entitled ``Commentary on Bioaccumulation Modeling Issues'' (SAB-EPEC/
DWC-COM-95-006). The report was the result of a April 1994 consultation 
with the SAB on approaches for estimating bioaccumulation potential of 
chemicals and to discuss various mass/balance/food web models. The SAB 
provided general advice on how and when EPA should use mass balance/
food web models to estimate bioaccumulation and what research is needed 
to improve model predictions. The SAB stated:

    In summary, while the Subcommittee agrees that mass balance/food 
web models such as the Thomann model hold promise for predicting 
bioaccumulation of certain types of chemicals, we urge the Agency to 
further field test the models for additional classes of compounds 
and for additional environmental settings and assess the 
uncertainties in model prediction prior to their wide-spread 
application in a regulatory context. Ongoing peer review should be 
an integral part of this process. Finally, the use of models, no 
matter how refined, should be augmented by appropriately designed 
laboratory and field experiments and monitoring.

    After careful consideration and review of the SAB's comments, EPA 
recommends using BAFs in the derivation of AWQC because, for highly 
lipophilic chemicals, uptake from aquatic organisms is the primary 
route of exposure. Failing to account for all routes of exposure, 
including ambient water and diet, would result in criteria which are 
under protective for a substantial portion of the population. In 
addition, the data hierarchy proposed above relies upon using the most 
scientifically sound experimental evidence of bioaccumulation. 
Specifically, the first and second preference for deriving BAFs for 
organic chemicals relies on using properly collected and analyzed field 
data over predicted bioaccumulation factors based on models. However, 
in the absence of field data for a chemical, EPA believes the use of 
bioaccumulation models can be used in establishing the regulatory 
criteria when the models have been properly validated. Using data from 
the Great Lakes, EPA has evaluated the predictability of BAFs 
determined from the Gobas model (and those determined from BSAFs). EPA 
found measured and predicted BAFs to be generally in good agreement 
when field-measured BAFs are adjusted to account for the lipid and 
freely dissolved fractions. Additional information on these comparisons 
is provided in the TSD.
10. Issues for Public Comment
    Comments are requested on the following issues in the proposal:
    1. Is the suggested hierarchy for developing BAFs appropriate? Are 
there any alternatives to the four methods that could be used to derive 
AWQC?
    2. Is the procedure for estimating the consumption-weighted default 
lipid value of 2 percent for aquatic species eaten by humans and the 
data used for deriving the value appropriate? Are there other data 
available that could be used to calculate the default lipid value?
    3. Are there alternatives to the equation used to derive the freely 
dissolved fraction of a chemical appropriate? If yes, what data support 
an alternative approach? Are there scientifically defensible 
alternatives to EPA's Kow-based estimate of KDOC 
and KPOC?
    4. Are the default POC value of 0.48 mg/L and the default DOC value 
of 2.9 mg/L used in deriving BAFs appropriate as national defaults? Are 
the water body- and State-specific POC and DOC values provided in the 
TSD appropriate? Are there additional data that could be used to derive 
these values?
    5. What approaches could be used to account for metabolism in the 
determination of a BAF and what data are available to support these 
approaches?
    6. What other models are available that could be used to predict 
FCMs? What are the data that support these models? Is EPA's choice of 
food web structures used to calculate FCMs appropriate?
    7. Is EPA's guidance on selecting reproducible Kow 
values appropriate? Which of the two options for selecting reproducible 
Kow values do you consider most appropriate?
    8. Should properly derived field-measured FCMs take precedence over 
FCMs derived using the Gobas (1993) model?

References for Bioaccumulation

American Society of Testing and Materials. 1990. Standard Practice 
for Conducting Bioconcentration Tests with Fishes and Saltwater 
Bivalve Molluscs. Designation E 1022--84. In: Annual Book of ASTM 
standards. Section 11, Water and Environmental Technology. 11(04): 
606-6622.
Barron, M.G. 1990. Bioconcentration: Will Water-Borne Organic 
Chemicals Accumulate in Aquatic Animals? Environ. Sci. Technol. 24: 
1612-1618.
Brooke, D.N., A.J. Dobbs and N. Williams. 1986. Octanol: Water 
Partition Coefficients (P): Measurement, Estimation, and 
Interpretation, Particularly for Chemicals with P > 105. Ecotoxicol. 
Environ. Safety. 11: 251-260.
Brooke, D., I. Nielsen, J. de Bruijn and J. Hermens. 1990. An 
Interlaboratory Evaluation of the Stir-Flask Method for the 
Determination of Octanol-Water Partition Coefficient (log POW). 
Chemosphere. 21: 119-133.
Connell, D.W. 1988. Bioaccumulation Behavior of Persistent Organic 
Chemicals with Aquatic Organisms. In: Review of Environmental 
Contamination and Toxicology. 101: 117-159.
de Bruijn, J., F. Busser, W. Seinen and J. Hermens. 1989. 
Determination of Octanol/Water Partition Coefficients for 
Hydrophobic Organic Chemicals with the ``Slow-stirring'' Method. 
Environ. Toxicol. Chem. 8:499-512.
DeVoe, H., M.M. Miller and S.P. Wasik. 1981. Generator Columns and 
High Pressure Liquid Chromatography for Determining Aqueous 
Solubites and Octanol-Water Partition Coefficients of Hydrophobic 
Substances. J. Res. Natl. Bur. Stand. 86: 361-366.
Garst, J.E. and W.C. Wilson. 1984. Accurate, Wide-Range, Automated, 
High-Performance Liquid Chromatographic Method for the Estimation of 
Octanol/Water Partition Coefficients. I: Effect of Chromatographic 
Conditions and Procedure Variables on Accuracy and Reproducibility 
of the Method. J. Pharm. Sci. 73: 1616-1623.
Gobas, F.A.P.C. 1993. A Model for Predicting the Bioaccumulation of 
Hydrophobic Organic Chemicals in Aquatic Food-Webs: Application to 
Lake Ontario. Ecological Modelling. 69: 1-17.
Hansch, C. and A.J. Leo. 1979. Substituent Constituents for 
Correlation Analysis in Chemistry and Biology. New York: John Wiley 
and Sons.
Honeycutt, M.E., V.A. McFarland, and D.D. McCant. 1995. Comparison 
of Three Lipid Extraction Methods for Fish. Bull. Environ. Contam. 
Toxicol. 55: 469-472.
Isnard, P. and S. Lambert. 1988. Estimating Bioconcentration Factors 
from Octanol-Water Partition Coefficients and Aqueous Solubility. 
Chemosphere. 17: 21-34.
Konemann, H., R. Zelle, F. Busser and W.E. Hammers. 1979. 
Determination of log Poct Values Chloro-Substituted 
Benzenes, Toluenes and Anilines by High Performance Liquid 
Chromatography on ODS-silica. J. Chromatogr. 178: 559-565.
Mackay, D. 1982. Correlation of Bioconcentration Factors. Environ. 
Sci. Technol. 16: 274-278.
McDuffie, B. 1981. Estimation of Octanol/Water Partition 
Coefficients for Organic Pollutants Using Reversed-Phase HPLC. 
Chemosphere. 10: 73-83.
Meylan, W. M. and P. H. Howard. 1995. Atom/Fragment Contribution 
Method for Estimating Octanol-Water Partition Coefficients. J. 
Pharm. Sci. 84: 83-92. January.
Miller, M.M., S. Ghodbane, S.P. Wasik, Y.D. Terwari and D.E. 
Martire. 1984. Aqueous Solubilities, Octanol/Water Partition 
Coefficients and Entropics of Melting of Chlorinated Benzenes and 
Biphenyls. J. Chem. Eng. Data 29: 184-190.

[[Page 43823]]

Reinert, R. 1970. Pesticide Concentrations in Great Lakes Fish. 
Pesticide Monitoring J. 3(4): 97-111.
Thomann, R.V. 1989. Bioaccumulation Model of Organic Chemical 
Distribution in Aquatic Food Chains. Environ. Sci. Technol. 23: 699-
707.
USEPA. 1991. Technical Support Document for Water Quality-Based 
Toxics Control. Office of Water. Washington, DC. EPA/505/2-90-001.
USEPA. 1992. Consumption Surveys for Fish and Shellfish: A Review 
and Analysis of Survey Methods. 822/R-92-001. February.
USEPA. 1993. Interim Report on Data and Methods for Assessment of 
2,3,7,8-tetrachlorodibenzo-p-dioxin Risks to Aquatic Life and 
Associated Wildlife. Duluth, MN: U.S. Environmental Protection 
Agency. EPA/600/R-93/055.
USEPA. 1995a. Great Lakes Water Quality Initiative Technical Support 
Document for the Procedure to Determine Bioaccumulation Factors. 
EPA-820-B-95-005. March.
USEPA. 1995b. Trophic Level and Exposure Analyses for Selected 
Piscivorous Birds and Mammals. Volume I: Analyses of Species for the 
Great Lakes. Draft. Office of Water. Washington, DC. Available in 
the GLWQI docket.
USEPA. 1995c. Trophic Level and Exposure Analyses for Selected 
Piscivorous Birds and Mammals. Volume II: Analyses of Species in the 
Conterminous United States. Draft. Office of Water. Washington, DC. 
Available in the GLWQI docket.
USEPA. 1995d. Trophic Level and Exposure Analyses for Selected 
Piscivorous Birds and Mammals. Volume III: Appendices. Draft. Office 
of Water. Washington, DC. Available in the GLWQI docket.
USEPA. 1998b. Daily Average Per Capita Fish Consumption Estimates 
Based on the Combined USDA 1989, 1990, 1991 Continuing Survey of 
Food Intakes by Individuals (CSFII). Volume I: Uncooked Fish 
Consumption National Estimates; Volume II: As Consumed Fish 
Consumption National Estimates. Prepared by SAIC under Contract #68-
C4-0046. March.
USEPA, Science Advisory Board. 1992. Evaluation of the Guidance for 
the Great Lakes Water Quality Initiative. Prepared jointly by the 
Great Lakes Water Quality Subcommittee of the Ecological Processes 
and Effects Committee and the Drinking Water Committee. EPA-SAB-
EPEC/DWC-93-005. December.
USEPA, Science Advisory Board. 1993. Review of the Methodology for 
Deriving National Ambient Water Quality Criteria for the Protection 
of Human Health. Prepared by the Drinking Water Committee of the 
Science Advisory Board. EPA-SAB-DWC-93-016. August.
USEPA, Science Advisory Board. 1995. Commentary on Bioaccumulation 
Modeling Issues. Prepared by a Joint Bioaccumulation Subcommittee 
with Representatives from the Ecological Processes and Effects 
Committee and the Drinking Water Committee of the Science Advisory 
Board. EPA-SAB-EPEC/DWC-COM-95-006. September.
Veith, G.D., D.F.L. DeFoe and B.V. Bergstedt. 1979. Measuring and 
Estimating the Bioconcentration Factor in Fish. J. Fish. Res. Board 
Can. 36:1040-1045.
Veith, G.D. and P. Kosian. 1983. Estimating Bioconcentration 
Potential from Octanol/Water Partition Coefficients. In: PCBs in the 
Great Lakes. Mackay, D., R. Patterson, S. Eisenreich, and M. Simmons 
(eds.). Ann Arbor: Science.
Woodburn, K.B., W.J. Doucette and A.W. Andren. 1984. Generator 
Column Determination of Octanol/Water Partition Coefficients for 
Selected Polychlorinated Biphenyl Congeners. Environ. Sci. Technol. 
18: 457-459.

E. Microbiology

1. Existing Microbiological Criteria
    The 1980 AWQC National Methodology did not address microbiological 
criteria for the protection of human health. However, in 1986 EPA 
published a document entitled Bacteriological Ambient Water Quality 
Criteria for Marine and Fresh Recreational Water, which updated and 
revised bacteriological criteria previously published in 1976 in 
Quality Criteria for Water.
    The microbiological criteria developed in 1986 are based on 
research conducted on beaches that were officially designated for 
swimming and had well-defined sources of human fecal pollution. 
Researchers examined the relationship between swimming-associated 
gastrointestinal (GI) illness and ambient densities of indicator 
bacteria. EPA concluded from these studies that measuring the densities 
of the indicator organism group recommended in the 1976 criteria, the 
fecal coliform, is inadequate. The enumeration of the recommended 
indicators is based on analytical procedures described in USEPA (1976). 
The EPA studies demonstrated that enterococci densities correlate far 
better with swimming illness in both marine and fresh water than fecal 
coliform densities. Also, E.coli, a specific bacterial species included 
in the fecal coliform group, correlates as well as enterococci with GI 
illness in fresh water but does not correlate as well in marine water.
    The recommended densities of indicator organisms (E.coli and 
enterococci), upon which the 1986 criteria are based, were calculated 
to approximate the degree of protection already accepted using fecal 
coliforms as indicators. The current EPA criteria are as follows:
    Fresh water: E. coli not to exceed 126/100 ml or enterococci not to 
exceed 33/100 ml;
    Marine water: enterococci not to exceed 35/100 ml.
    These criteria are calculated as the geometric mean of a 
statistically sufficient number of samples, generally no fewer than 
five, equally spaced over a 30-day period.
    No single sample should exceed a one-sided confidence limit (C.L.) 
calculated using the following as guidance:
    Designated bathing beach: 75% C.L.
    Moderate use for bathing: 82% C.L.
    Light use for bathing: 90% C.L.
    Infrequent use for bathing: 95% C.L.
    These confidence limits are based on a site-specific log standard 
deviation or, if site data are not sufficient to establish a log 
standard deviation, then using 0.4 as the log standard deviation for 
both indicators in fresh water. In marine water one would use 0.7 as 
the log standard deviation.
    The quantitative relationship between the rates of swimming-
associated health effects (acute GI infection) and bacterial indicator 
densities was determined using regression analysis. Linear 
relationships were estimated from data grouped on the basis of summers 
or trials with similar indicator densities. The data for each summer 
were analyzed by pairing the geometric mean indicator density for a 
summer bathing season at each beach with the corresponding swimming- 
associated GI illness rate for the same summer. The swimming-associated 
illness rate was determined by subtracting the GI illness rate in non 
swimmers from that in swimmers. These two variables from multiple beach 
sites were used to calculate a regression coefficient, y-intercept, and 
95 percent confidence intervals for the paired data. In the marine 
studies, the total number of points for use in regression analysis was 
increased by collecting trial days with similar indicator densities 
from each study location and placing them into groups. The swimming-
associated illness rate was determined as above, by subtracting non 
swimmers' illness rate of all the individuals included in the grouped 
trial days from the swimmers' illness rate during these same grouped 
trial days.
2. Plans for Future Work
    EPA recommends no change at this time in the stringency of its 
bacterial criteria for recreational waters; existing criteria and 
methodologies from 1986

[[Page 43824]]

will still apply. The Agency plans to conduct national studies on 
improving indicators together with epidemiology studies for new 
criteria development.
    EPA will consider revising the criteria with the possible inclusion 
of criteria for other primary-contact waters with reduced swimming or 
full-body-contact use. The Agency will perform critical evaluation of 
studies of the health effects of recreational water microbiology. EPA 
will also form a group of experts from EPA program offices, ORD, and 
the regions to initiate development of consensus recommendations on the 
development of policy and criteria methodology, research and 
implementation strategy for a comprehensive recreational waters 
program.
    The Agency expects to make final recommendation for action as soon 
as possible. A separate Federal Register proposal with revised criteria 
and methodology is anticipated for publication after improved indicator 
methods and associated exposure risks are established. In 1997, EPA 
will approve a new 24-hour enterococcus test for recreational waters 
that may be used as an alternative to the 48-hour test.
3. SAB Comments
    (a) The SAB believes that it would be highly beneficial to 
establish and implement a multi- organizational working group made up 
of representatives from EPA, CDC, FDA, academia, the water and 
wastewater industry, and the public.
    (b) The SAB believes that despite the desirability of and need for 
a comprehensive and integrated approach to ambient water quality, it is 
unrealistic, perhaps inappropriate, and in all likelihood impossible to 
address all of the water-related exposure routes of microbial health 
effects concerns under this regulatory initiative.
    (c) The SAB recommends that the process of developing and 
evaluating water quality criteria for microbes should include microbes 
causing fecally transmitted diseases other than gastroenteritis. Such a 
process should also include microbes causing diseases of the skin, 
respiratory tract, eye, ear, nose, throat, and perhaps other sites of 
entry and infection.
    (d) The current recreational-water quality criteria are neither 
appropriate for nor transferable to other ambient waters. These 
criteria were intended to address only those pathogens causing enteric 
(GI) illness.
    (e) The SAB recommends that the likelihood of human exposure to 
different types of ambient water be the basis for identifying the types 
of ambient waters for which criteria need to be developed. The need for 
quality criteria for recreational waters has been established; however, 
the need for such criteria for some other waters has not been 
established.
    (f) The SAB believes that a risk-based approach to criteria for 
pathogenic microorganisms in ambient waters is both appropriate and 
feasible for at least some pathogens. However, the SAB believes that 
this approach has limited applicability to the quality criteria for 
microbial pathogens in ambient waters.
    (g) The SAB believes that further research has to be done on 
identifying, characterizing, and measuring the virulence determinants 
of microbial pathogens; on the factors governing or influencing the 
expression of these determinants under different environmental 
conditions; and on the role of other factors in virulence expression, 
such as host factors.
    (h) The SAB believes that the currently approved indicator 
organisms in beach waters are probably appropriate for the safety of 
bathing waters against GI disease. The SAB believes that the currently 
accepted levels of the bacterial indicators are not uniformly and 
adequately protective of health risks from non-GI pathogens in bathing 
waters.
    (i) The SAB believes that there are candidate alternative 
indicators worthy of consideration and deserving of investigation for 
improving ambient-water monitoring.
    The EPA Office of Water agrees with the SAB comments for all the 
above points. The Agency makes the following recommendations:
     Future criteria development should consider the risk of 
diseases other than gastroenteritis. The nature and significance of 
other than the classical waterborne pathogens are to some degree tied 
to the particular type of ambient water.
     EPA needs to consider and evaluate such water-related 
exposure routes as inhalation and dermal absorption when addressing 
microbial health effects.
     A new set of indicator organisms may need to be developed for 
tropical water if it is proven that the current fecal indicators can 
grow in pristine waters or on plants in the tropics. Some potential 
alternative indicators to be fully explored are coliphage, other 
bacteriophage, and Clostridium perfringens.
     Because animal sources of pathogens of concern for human 
infection such as Giardia, Cryptosporidium, and Salmonella may be 
waterborne or washed into water and thus become a potential source for 
infection, they must not be ignored in risk assessment. One possible 
approach to estimating levels of pathogens from animal sources is to 
determine the ratios of conventional indicators from human sources and 
from animal sources. Alternatively, new indicators could be developed 
that are specific to or can discriminate animal sources. The presence 
of such indicator pathogens together with a predominance of indicators 
of animal wastes would help define types of risks.
     EPA needs to develop additional data on secondary infection 
routes and infection rates from prospective epidemiology studies and 
outbreaks.
     EPA needs to improve sampling, strategies for recreational 
water monitoring including consideration of rain fall and pollution 
events to trigger sampling.

References for Microbiology

USEPA. 1976. Test Methods for Eschericia coli and Enterococci in 
Water by the Membrane Filter Procedure. EPA 600/4-85/076.

F. Other Considerations

1. Minimum Data Considerations
    For many of the preceding technical areas, considerations have been 
presented for data quality in developing toxicological and exposure 
assessments. For greater detail and discussion of minimum data 
recommendations, the reader is referred to the TSD which accompanies 
this Federal Register document.
2. Site-Specific Criterion Calculation
    The 1980 AWQC National Guidelines allowed for site-specific 
modifications to reflect local environmental conditions and human 
exposure patterns. The methodology stated that ``local'' may refer to 
any appropriate geographic area where common aquatic environmental or 
exposure patterns exist. Thus ``local'' may signify a Statewide, 
regional, river reach or entire river.
    In today's Notice, site-specific criteria may be developed as long 
as the site-specific data, either toxicological or exposure related is 
justifiable. For example, a State should use a site-specific fish 
consumption rate that represents at least the central tendency (median 
or mean) of the population surveyed (either sport or subsistence, or 
both). If a site-specific fish consumption rate for sport anglers or 
subsistence anglers is lower than an EPA default value, it may be used 
in calculating AWQC. To justify such a level (either

[[Page 43825]]

higher or lower than EPA defaults) the State should present survey data 
it used in arriving at the site-specific fish consumption rate. The 
same conditions apply to site-specific calculations of BAF, percent 
fish lipid, or the RSC. In the case of deviations from toxicological 
values (IRIS values: verified noncancer and cancer assessments), EPA 
recommends that the data upon which the deviation is based be presented 
to and approved by the Agency before a criterion is developed.
3. Organoleptic Criteria
    The 1980 AWQC National Guidelines provided for the development of 
organoleptic criteria if organoleptic data were available for a 
specific contaminant. The methodology also made a clear distinction 
that organoleptic criteria and toxicity-based criteria are derived from 
completely different endpoints and that organoleptic criteria have no 
demonstrated relationship to potential adverse human health effects. 
The 1992 National Experts Workshop participants and the Great Lakes 
Committees of the Initiative both recommended EPA to place highest 
priority on setting toxicity-based criteria, rather than using limited 
resources to set organoleptic criteria. Both efforts, the GLI and the 
National Experts Workshop concluded that organoleptic effects, while 
significant from an aesthetic standpoint, were not a significant health 
concern and did not merit significant expenditures of time and effort. 
While it can be argued that organoleptic properties indirectly affect 
human health (people may drink less water or eat less fish due to 
objectionable taste or odor), they have not been demonstrated to result 
in direct adverse effects, such as cancer or other types of toxicity.
    In today's Notice, EPA is not recommending a methodology for 
developing organoleptic criteria, but rather is asking for comment on 
the following questions: 1. How would organoleptic criteria be used if 
the Agency were to develop new criteria? (Could they be used in a 
similar fashion to the secondary standards developed by the Agency's 
National Drinking Water program?) 2. Would organoleptic criteria 
ultimately be counterproductive if they are much lower than toxicity-
based criteria?
4. Criteria for Chemical Classes
    The 1980 AWQC National Guidelines allowed for the development of 
criteria for chemical classes. A chemical class was defined as any 
group of chemical compounds which were reviewed in a single risk 
assessment document. The Guidelines also stated that in criterion 
development, isomers should be regarded as part of a chemical class 
rather than as a single compound. A class criterion, therefore, was an 
estimate of risk/safety which applied to more than one member of a 
class. It involved the use of available data on one or more chemicals 
of a class to derive criteria for other compounds of the same class in 
the event that there were insufficient data available to derive 
compound-specific criteria. The criterion applied to each member of the 
class, rather than to the sum of the compounds within the class. The 
1980 methodology also acknowledged that, since relatively minor 
structural changes within the class of compounds can have pronounced 
effects on their biological activities, reliance on class criteria 
should be minimized.
    The 1980 methodology prescribed the following analysis when 
developing a class criterion:
     A detailed review of the chemical and physical properties of 
the chemicals within the group should be made. A close relationship 
within the class with respect to chemical activity would suggest a 
similar potential to reach common biological sites within tissues. 
Likewise, similar lipid solubilities would suggest the possibility of 
comparable absorption and distribution.
     Qualitative and quantitative data for chemicals within the 
group are examined. Adequate toxicological data on a number of 
compounds with a group provides a more reasonable basis for 
extrapolation to other chemicals of the same class than minimal data on 
one chemical or a few chemicals within the group.
     Similarities in the nature of the toxicological response to 
chemicals in the class provides additional support for the prediction 
that the response to other members of the class may be similar. In 
contrast, where the biological response has been shown to differ 
markedly on a qualitative and quantitative basis for chemicals within a 
class, the extrapolation of a criterion to other members is not 
appropriate.
     Additional support for the validity of extrapolation of a 
criterion to other members of a class could be provided by evidence of 
similar metabolic and pharmacokinetic data for some members of the 
class.
    Today's Notice allows for the development of a criterion for 
classes of chemicals, as long as the 1980 methodology guidance is 
followed and a justification is provided through the analysis of 
mechanistic data, pharmacokinetic data, structure-activity relationship 
data, and limited acute and chronic toxicity data. When potency 
differences between members of a class is great (such as in the case of 
chlorinated dioxins and furans), toxicity equivalency factors (TEFs) 
may be more appropriately developed than one class criterion. The 
Agency requests comments on the practice of developing criteria for 
classes of compounds and whether the guidance provided here is 
sufficient to ensure that class criteria are derived appropriately.
5. Criteria for Essential Elements
    The 1980 AWQC National Guidelines acknowledged that developing 
criteria for essential elements, particularly metals, must be a 
balancing act between toxicity and essentiality. The 1980 guidelines 
state:

    that the criteria must consider essentiality and cannot be 
established at levels which would result in deficiency of the 
element in the human population. The difference between the RDA and 
the daily doses causing a specified risk level for carcinogens or 
the ADIs (now RfDs) for noncarcinogens defines the spread of daily 
doses which the criterion may be derived. Because errors are 
inherent in defining both essential and maximum-tolerable levels, 
the criterion is derived from the dose levels near the center of 
such dose ranges.

    In today's Notice, EPA endorses the guidance from the 1980 
methodology and adds that the process for developing criteria for 
essential elements should be similar to that used for any other 
chemical with minor modifications. The RfD represents concern for one 
end of the exposure spectrum (toxicity), whereas the RDA represents the 
other end (minimum essentiality). Where the RDA and RfD values might 
occasionally appear to be similar in magnitude to one another, it does 
not imply incompatibility of the two methodological approaches, nor 
does it imply inaccuracy or error in either calculation.

Appendix IV. Summary of Ambient Water Quality Criteria for the 
Protection of Human Health: Acrylonitrile 16
---------------------------------------------------------------------------

    \16\ This is a preliminary summary of a criteria document being 
prepared for the derivation of the Ambient Water Quality Criteria 
(AWQC) for the protection of human health from exposure to 
acrylonitrile. The calculated AWQC values presented in this draft 
are subject to revision pending inclusion of further information 
concerning exposure as well as possible changes in the toxicological 
information used to derive the criterion.
---------------------------------------------------------------------------

    This criteria document updates the national criteria for 
acrylonitrile using new methods and information described in this 
Federal Register document and Technical Support Document (USEPA,

[[Page 43826]]

1998a) to calculate ambient water quality criteria. These new methods 
include approaches to determine dose-response relationships for both 
carcinogenic and non-carcinogenic effects, updated information for 
determining exposure factors (e.g., values for fish consumption), 
exposure assumptions, and procedures to determine bioaccumulation 
factors. For more detailed information please refer to the U.S. EPA 
Ambient Water Quality Criteria (AWQC) document for Acrylonitrile 
(USEPA, 1998b).

Background Information

    The AWQC is being derived for acrylonitrile (CAS No. 107-13-1). The 
chemical formula is C3H3N2. 
Acrylonitrile occurrence in environmental media is not well-documented. 
Several regional and local drinking water surveys were found and one 
limited study analyzed ambient air samples. Limited information is also 
available on acrylonitrile migration into foods from packaging 
materials.
    Acrylonitrile is largely used in the manufacture of copolymers for 
the production of acrylic and modacrylic fibers. Other major uses 
include the manufacture of acrylonitrile-butadiene-styrene (ABS) and 
styrene acrylonitrile (SAN) (used in production of plastics), and 
nitrile elastomers and latexes. It is also used in the synthesis of 
antioxidants, pharmaceuticals, dyes, and surface-active agents.
    According to the U.S. Environmental Protection Agency's (EPA) Toxic 
Release Inventory, the total release of acrylonitrile into the 
environment in 1990 by manufacturers, was 8,077,470 pounds. The two 
largest pathways of release were underground injection, which accounted 
for 61% (or 4,925,276 pounds) of the total release, and emissions into 
the air, which accounted for 39% (or 3,148,049 pounds) of the total 
release. Release of acrylonitrile into water bodies was reported at 
3,877 pounds and release onto land was reported at 268 pounds.
    A baseline BAF of 1.5 was calculated for acrylonitrile. The 
baseline BAF was calculated using a value of 0.17 for the log 
Kow and 1.000 for the food-chain multiplier (FCM) at trophic 
level 4. A value of 0.17 was selected as a typical value of the log 
Kow for acrylonitrile (USEPA 1998b). A value of 1.000 was 
selected as the FCM for trophic level 4, reflective of top predator 
fish based on a log Kow of 2.0 from USEPA (1998a). Using 
these data, the baseline BAF was calculated as: Kow *  
FCM=(100.17)*1.000=1.5 (rounded to two significant digits).
    Based upon sufficient evidence from animal studies (multiple tumor 
types in several strains of rats by several routes) and limited 
evidence from human studies (lung tumors in workers), positive 
mutagenicity, acrylonitrile is considered as a likely human carcinogen 
by any route. A linear approach is used for the low dose extrapolation.

AWQC Calculation

For Ambient Waters Used as Drinking Water Sources
[GRAPHIC] [TIFF OMITTED] TN14AU98.041

    The cancer-based AWQC was calculated using the RSD and other input 
parameters listed below:

Where:

RSD=Risk specific dose (1.6 x 10-6 mg/kg-day at 
10-6 lifetime risk)
BW=Human body weight assumed to be 70 kg
DI=Drinking water intake assumed to be 2 L/day
FI=Fish intake at trophic level i, i=2,3, and 4; total intake assumed 
to be 0.01780 kg/day
BAF=Bioaccumulation factor at trophic level i (i=2,3, and 4) equal to 
1.03, 1.02, and 1.05 L/kg-tissue for trophic levels 2,3, and 4, 
respectively.

    This yields concentrations of 5.5 x 10-5 mg/L (or 0.05 
g/L), for a 102-6 (one in a million) lifetime 
cancer risk.

For Ambient Waters Not Used as Drinking Water Sources

    When the water body is to be used for recreational purposes and not 
as a source of drinking water, the drinking water value (DI above) is 
eliminated from the equation and it is substituted with an incidental 
ingestion value (II). The incidental intake is assumed to occur from 
swimming and other activities. The fish intake value is assumed to 
remain the same. The default value for incidental ingestion is 0.01 L/
day. When the above equation is used to calculate the AWQC with the 
substitution of an incidental ingestion of 0.01 L/day an AWQC of 4.0 x 
10-3 mg/L (or 4.0 g/L) is obtained for a 
10-6 lifetime cancer risk.

Site-Specific or Regional Adjustments to Criteria

    Several parameters in the AWQC equation can be adjusted on a site-
specific or regional basis to reflect regional or local conditions and/
or specific populations of concern. These include fish consumption, 
incidental water consumption as related to regional/local recreational 
activities, BAF (including factors used to derive BAFs, percent lipid 
of fish consumed by target population, and species representative of 
given trophic levels), and the relative source contribution. States are 
encouraged to make adjustments using the information and instructions 
provided in the Technical Support Document (USEPA, 1998a).

References

USEPA. 1998a. Ambient Water Quality Criteria Derivation 
Methodology--Human Health. Technical Support Document. Final Draft. 
EPA 822-B-98-005. Office of Water. Washington, DC. July.
USEPA. 1998b. Ambient Water Quality Criteria for the Protection of 
Human Health: Acrylonitrile. EPA 822-R-98-006.

Appendix V. Summary of Ambient Water Quality Criteria for the 
Protection of Human Health: 1,3-Dichloropropene 17
---------------------------------------------------------------------------

    \ 17\ This is a preliminary summary of a criteria document being 
prepared for the derivation of the Ambient Water Quality Criteria 
(AWQC) for the protection of human health from exposure to 1,3-
dichloropropene. The calculated AWQC values presented in this draft 
are subject to revision pending inclusion of further information 
concerning exposure as well as possible changes in the toxicological 
information used to derive the criterion.
---------------------------------------------------------------------------

    This criteria document updates the national criteria for 1,3-DCP 
using new methods and information described in this Federal Register 
document and Technical Support Document (USEPA, 1998a) to calculate 
ambient water quality criteria. These new methods include approaches to 
determine dose-response relationships for both carcinogenic and non-
carcinogenic effects, updated information for determining exposure 
factors (e.g., values for fish consumption), exposure assumptions, and 
procedures to determine bioaccumulation factors. For more detailed 
information please refer to the U.S. EPA Ambient Water Quality Criteria 
(AWQC) document for 1,3- Dichloropropene (1,3-DCP) (USEPA, 1998b).

Background Information

    The AWQC is being derived for 1,3-Dichloropropene (CAS No. 542-75-
6). The chemical formula is C3H4Cl2 
and molecular weight is 110.98 (pure isomers). At 25 deg.C, the 
physical state of 1,3-DCP is a pale yellow to yellow liquid. 
Dichloropropene (DCP) is used as soil fumigant in the United States to 
control soil nematodes on crops grown in sandy soils. The EPA's 
National

[[Page 43827]]

Toxics Inventory data base reported air emissions of 18,820,000 pounds/
year in the U.S. (USEPA, 1996a). Numerous studies have sampled for DCP 
(and isomers) in drinking water, groundwater and surface waters across 
the U.S. (Hall et al., 1987; Miller et al., 1990; RIDEM, 1990; 
Rutledge, 1987; STORET, 1992). All of these studies report 
concentrations of 1,3-DCP usually at or below the detection limits 
(USEPA, 1998b).
    The AWQC bioaccumulation factor (BAF) is 2.2 L/kg of tissue for 
1,3-DCP. This BAF is based on the total concentration of 1,3-DCP in 
trophic level four biota divided by the total concentration in water, 
assuming default values for the freely-dissolved fraction and lipid 
content of consumed aquatic organisms.
    The cancer risk evaluation of 1,3-DCP uses the new methods in the 
proposed cancer guidelines (USEPA, 1996), which are described in this 
Federal Register document and in the Technical Support Document (USEPA, 
1998a). Based upon sufficient evidence from animal studies (multiple 
tumor types in several species by oral, inhalation, and dermal routes), 
positive mutagenicity, and structural analogues, 1,3-DCP is considered 
``likely to be carcinogenic to humans by all routes of exposure.'' 
Based on the mutagenic mode of action, a linear low dose approach is 
recommended.

AWQC Calculation

For Ambient Waters Used as Drinking Water Sources

    The cancer-based AWQC was calculated using the RSD and other input 
parameters listed below:

[GRAPHIC] [TIFF OMITTED] TN14AU98.042

Where:
RSD=Risk specific dose 1.0  x  10-5 mg/kg/day 
(10-6 risk)
BW=Human body weight assumed to be 70 kg
DI=Drinking water intake assumed to be 2 L/day
FI=Fish intake at trophic level i, i=2,3, and 4 total intake assumed to 
be 0.01780 kg/day
BAF=Bioaccumulation factor at trophic level i (i=2,3, and 4), equal to 
2.32, 1.86, and 2.78 L/kg-tissue for trophic levels 2,3, and 4, 
respectively.

    This yields a value of 3.4  x  10-4 mg/L, or 0.34 
g/L (rounded from 0.343 g/L).

For Ambient Waters Not Used as Drinking Water Sources

    When the water body is used for recreational purposes and not as a 
source of drinking water, the drinking water value is eliminated from 
the equation and it is substituted with an incidental ingestion value. 
The incidental intake is assumed to occur from swimming and other 
activities. The fish intake value is assumed to remain the same. The 
default value for incidental ingestion is 0.01 L/day. When the above 
equation is used to calculate the AWQC with the substitution of an 
incidental ingestion of 0.01 L/day an AWQC of1.4-10-2 mg/L 
(14 g/L) is obtained.

Site-Specific or Regional Adjustments to Criteria

    Several parameters in the AWQC equation can be adjusted on a site-
specific or regional basis to reflect regional or local conditions and/
or specific populations of concern. These include fish consumption; 
incidental water consumption as related to regional/local recreational 
activities; BAF (including factors used to derive BAFs, percent lipid 
of fish consumed by the target population, and species representative 
of given trophic levels); and the relative source contribution. States 
are encouraged to make adjustments using the information and 
instructions provided in the Technical Support Document (USEPA, 1998a).

References

USEPA. 1998a. Ambient Water Quality Criteria Derivation Methodology-
Human Health. Technical Support Document. Final Draft. EPA 822-B-98-
005. Office of Water. Washington, DC. July.
USEPA. 1998b. Ambient Water Quality Criteria for the Protection of 
Human Health: 1,3- Dichloropropene (1,3-DCP). EPA 822-R-98-005.

Appendix VI. Summary of Ambient Water Quality Criteria for the 
Protection of Human Health: Hexachlorobutadiene 18
---------------------------------------------------------------------------

    \18\ This is a summary of a criteria document being prepared for 
the derivation of the Ambient Water Quality Criteria (AWQC) for the 
protection of human health from exposure to HCBD. The calculated 
AWQC values presented in this draft are subject to revision pending 
inclusion of further information concerning exposure as well as 
possible changes in the toxicological information used to derive the 
criterion.
---------------------------------------------------------------------------

    This criteria document updates the national criteria for HCBD using 
new methods and information described in this Federal Register document 
and Technical Support Document (USEPA, 1998a) to calculate ambient 
water quality criteria. These new methods include approaches to 
determine dose-response relationships for both carcinogenic and non-
carcinogenic effects, updated information for determining exposure 
factors (e.g., values for fish consumption), exposure assumptions, and 
procedures to determine bioaccumulation factors. For more detailed 
information please refer to the U.S. EPA Ambient Water Quality Criteria 
(AWQC) document for hexachlorobutadiene (HCBD)(USEPA, 1998b).

Background Information

    The AWQC is being derived for hexachlorobutadiene (CAS No. 87-68-
3). The chemical formula is C4Cl6 and molecular 
weight is 260.76. At 25 deg.C, HCBD is a colorless liquid. HCBD is used 
as a solvent in chlorine gas production, as an intermediate in the 
manufacture of rubber compounds and lubricants, and as a pesticide. The 
EPA's National Toxics Release Inventory data base reported total 
emissions to the environment in 1990 of 5,591 pounds/year in the U.S., 
of which 4,906 pounds was to air. Numerous studies have sampled for 
HCBD in drinking water, ground water and surface waters across the U.S. 
(see USEPA 1998b for a summary). The vast majority of samples are at 
trace levels or below the detection limits (DL=0.1 mg/L).
    The AWQC bioaccumulation factor (BAF) is 620 L/kg of tissue for 
HCBD. This BAF is based on the total concentration of HCBD in trophic 
level four biota divided by the total concentration in water, assuming 
default values for the freely-dissolved fraction and lipid content of 
consumed aquatic organisms.
    The cancer risk evaluation of HCBD uses the new methods described 
in this Federal Register Notice and in the Technical Support Document 
(USEPA, 1998a). Based on a renal tumor finding in one chronic feeding 
study at one high dose in one species (both sexes of Sprague-Dawley 
rats), ``via oral route, HCBD is considered as likely to be 
carcinogenic to humans only at very high exposure conditions, where 
significant renal toxicity occurs.'' There is some mutagenic activity 
in the presence of metabolic activation. Thus, a mutagenic mode of 
action cannot be ruled out. As a result, both the cancer-based, linear 
low dose approach and the non-linear margin of exposure approaches are 
used for deriving the AWQC.

[[Page 43828]]

AWQC Calculation

For Ambient Waters Used as Drinking Water Sources

    The cancer-based AWQC was calculated using the RSD and other input 
parameters listed below:

 [GRAPHIC] [TIFF OMITTED] TN14AU98.043

Where:
RSD = Risk specific dose 2.5 x 10-5 mg/kg/day 
(10-6 risk)
BW = Human body weight assumed to be 70 kg
DI = Drinking water intake assumed to be 2 L/day
FI = Fish intake at trophic level i, i=2,3, and 4; total intake assumed 
to be 0.01780 kg/day
BAF = Bioaccumulation factor at trophic level i (i=2,3, and 4) equal to 
1,518, 2,389, and 1,294 L/kg-tissue for trophic levels 2,3, and 4, 
respectively.

    This yields a value of 4.6 x 10--5 mg/L, or 0.046 
g/L (rounded from 0.0462 g/L).
    The AWQC using the margin of exposure approach was calculated using 
the following equation and input parameters listed below.

 [GRAPHIC] [TIFF OMITTED] TN14AU98.044

where:
Pdp = Point of departure (0.054 mg/kg/day)
SF = Safety factor of 300
RSC = Relative source contribution from air of 1.2 x 10-4 
mg/kg-day, subtracted in this case
BW = Human body weight assumed to be 70 kg
DI = Drinking water intake assumed to be 2 L/day
FI = Fish intake at trophic level i, i=2,3, and 4; total intake assumed 
to be 0.01780 kg/day
BAF = Bioaccumulation factor at trophic level i (i=2,3, and 4) equal to 
1,518, 2,389, and 1,294 L/kg-tissue for trophic levels 2,3, and 4, 
respectively.
    This yields an AWQC of 1.1 x 10-4 mg/L (0.11 ``ug/L).

For Ambient Waters Not Used as Drinking Water Sources

    When the waterbody is used for recreational purposes and not as a 
source of drinking water, the drinking water value is eliminated from 
the equation and it substituted with an incidental ingestion value. The 
incidental intake is assumed to occur from swimming and other 
activities. The fish intake value is assumed to remain the same. The 
default value for incidental ingestion is 0.01 L/day. When the linear 
approach is used to calculate the AWQC with the substitution of an 
incidental ingestion of 0.01 L/day a cancer-based AWQC of 4.9 x 
10-5 mg/L (or 0.049 g/L, rounded from 0.0487 
g/L) is obtained. When the non-linear margin of exposure 
approach is used with the substitution of an incidental ingestion of 
0.01 L/day, the AWQC is 1.2 x 10-4 mg/L (or 0.12 g/
L, rounded from 0.117 g/L).

Site-Specific or Regional Adjustments to Criteria

    Several parameters in the AWQC equations can be adjusted on a site-
specific or regional basis to reflect regional or local conditions and/
or specific populations of concern. These include fish consumption; 
incidental water consumption as related to regional/local recreational 
activities; BAF (including factors used to derive BAFs, percent lipid 
of fish consumed by the target population, and species representative 
of given trophic levels); and the relative source contribution. States 
are encouraged to make adjustments using the information and 
instructions provided in the Technical Support Document (USEPA, 1998a).

References

USEPA. 1998a. Ambient Water Quality Criteria Derivation 
Methodology--Human Health. Technical Support Document. Final Draft. 
EPA 822-B-98-005. Office of Water. Washington, DC. July.
USEPA. 1998b. Ambient Water Quality Criteria for the Protection of 
Human Health: Hexachlorobutadiene (HCBD). EPA 822-R-98-004.
[FR Doc. 98-21517 Filed 8-13-98; 8:45 am]
BILLING CODE 6560-50-P