Medicare Hospital Pharmaceuticals: Survey Shows Price Variation
and Highlights Data Collection Lessons and Outpatient
Rate-Setting Challenges for CMS (28-APR-06, GAO-06-372).
In 2003, the Medicare Modernization Act required the Centers for
Medicare & Medicaid Services (CMS) to establish payment rates for
a set of new pharmaceutical products--drugs and
radiopharmaceuticals--provided to beneficiaries in a hospital
outpatient setting. These products were classified for payment
purposes as specified covered outpatient drugs (SCOD). The
legislation directed CMS to set 2006 Medicare payment rates for
SCODs equal to hospitals' average acquisition costs and included
requirements for GAO. As directed, GAO surveyed hospitals and
issued two reports, providing information to use in setting 2006
SCOD rates. To address other requirements in the law, this report
analyzes SCOD price variation across hospitals, advises CMS on
future surveys it might undertake, and examines both lessons from
the GAO survey and future challenges facing CMS.
-------------------------Indexing Terms-------------------------
REPORTNUM: GAO-06-372
ACCNO: A52723
TITLE: Medicare Hospital Pharmaceuticals: Survey Shows Price
Variation and Highlights Data Collection Lessons and Outpatient
Rate-Setting Challenges for CMS
DATE: 04/28/2006
SUBJECT: Cost analysis
Data collection
Data integrity
Hospitals
Payments
Pharmaceutical industry
Prescription drugs
Prices and pricing
Surveys
Comparative analysis
******************************************************************
** This file contains an ASCII representation of the text of a **
** GAO Product. **
** **
** No attempt has been made to display graphic images, although **
** figure captions are reproduced. Tables are included, but **
** may not resemble those in the printed version. **
** **
** Please see the PDF (Portable Document Format) file, when **
** available, for a complete electronic file of the printed **
** document's contents. **
** **
******************************************************************
GAO-06-372
* Results in Brief
* Background
* MMA Established SCOD Payment Categories for Certain Pharmace
* MMA Required Us to Survey Hospitals to Determine Their Acqui
* MMA Defined ASP, Which Is Reported by Manufacturers and Used
* Radiopharmaceuticals Can Be Purchased in Different Forms
* Hospitals' Teaching Status, Location, and Size Affected Pric
* Teaching Status, Location, and Size Were Each Significant Fa
* Hospitals with Combination of Major Teaching Status, Urban L
* Our Survey of Hospitals Suggests that the Burden of Large-Sc
* Using Hospitals as Data Source for SCOD Prices Had A Major A
* Survey Indicates that Accounting for Dynamic Drug Market and
* CMS Faces Challenges in Future Data Collection Efforts to Se
* Validating ASP Would Pose Challenges for CMS Because of Lack
* Radiopharmaceuticals Pose Unique Challenges for Obtaining Ac
* Conclusions
* Recommendations for Executive Action
* Agency Comments and Our Evaluation
* SCOD Price Data
* Factors Potentially Affecting SCOD Prices
* Methods and Results of Price Analysis
* GAO Contact
* Acknowledgments
* GAO's Mission
* Obtaining Copies of GAO Reports and Testimony
* Order by Mail or Phone
* To Report Fraud, Waste, and Abuse in Federal Programs
* Congressional Relations
* Public Affairs
Report to Congressional Committees
United States Government Accountability Office
GAO
April 2006
MEDICARE HOSPITAL PHARMACEUTICALS
Survey Shows Price Variation and Highlights Data Collection Lessons and
Outpatient Rate-Setting Challenges for CMS
GAO-06-372
Contents
Letter 1
Results in Brief 4
Background 6
Hospitals' Teaching Status, Location, and Size Affected Prices for SCOD
Products by Different Magnitudes 9
Our Survey of Hospitals Suggests that the Burden of Large-Scale Annual
Surveys Could Outweigh Gains in Data Accuracy 11
CMS Faces Challenges in Future Data Collection Efforts to Set SCOD Payment
Rates Accurately 15
Conclusions 18
Recommendations for Executive Action 19
Agency Comments and Our Evaluation 19
Appendix I Methodology for Analysis of SCOD Price Differences among
Hospital Types 22
Appendix II Purchase Prices for Drug SCODs 30
Appendix III Purchase Prices for Radiopharmaceuticals SCODs 38
Appendix IV Comments from the Department of Health and Human Services 43
Appendix V GAO Contact and Staff Acknowledgments 47
Tables
Table 1: Factors Accounting for Variation in SCOD Prices among Hospitals
10
Table 2: Factors Included in Analysis of Price Variation among Hospitals
Purchasing SCODs 23
Table 3: Estimated Effects of Selected Factors on Prices Hospitals Paid
for Drug SCODs 26
Table 4: Estimated Effects of Selected Factors on Prices Hospitals Paid
for Radiopharmaceutical SCODs 28
Table 5: Purchase Prices for SCODs Accounting for 86 Percent of Medicare
Spending on SCODs 32
Table 6: Purchase Prices for Radiopharmaceutical Accounting for 9 Percent
of Medicare Spending on SCODs 40
Abbreviations
ASP average sales price CMS Centers for Medicare & Medicaid Services HHS
Department of Health and Human Services MMA Medicare Prescription Drug,
Improvement, and Modernization Act of 2003 MSA metropolitan statistical
area NDC national drug code OPPS outpatient prospective payment system
SCOD specified covered outpatient drug
This is a work of the U.S. government and is not subject to copyright
protection in the United States. It may be reproduced and distributed in
its entirety without further permission from GAO. However, because this
work may contain copyrighted images or other material, permission from the
copyright holder may be necessary if you wish to reproduce this material
separately.
United States Government Accountability Office
Washington, DC 20548
April 28, 2006
The Honorable Charles E. Grassley Chairman The Honorable Max Baucus
Ranking Minority Member Committee on Finance United States Senate
The Honorable Joe Barton Chairman The Honorable John D. Dingell Ranking
Minority Member Committee on Energy and Commerce House of Representatives
The Honorable William M. Thomas Chairman The Honorable Charles B. Rangel
Ranking Minority Member Committee on Ways and Means House of
Representatives
In 2003, federal legislation required the establishment of Medicare
payment rates for a particular set of new pharmaceutical products that
were provided to beneficiaries in hospital outpatient settings but were
generally paid for differently than other services paid under Medicare's
hospital outpatient prospective payment system (OPPS). These products were
newly introduced drugs, biologicals, and radiopharmaceuticals used to
treat and in some cases diagnose serious conditions such as cancer.1
Specifically, the Medicare Prescription Drug, Improvement, and
Modernization Act of 2003 (MMA) required the Centers for Medicare &
Medicaid Services (CMS) in the Department of Health and Human Services
(HHS) to set rates for these pharmaceuticals. MMA classified them for
payment purposes as specified covered outpatient drugs (SCOD).2 In
addition, MMA defined a SCOD as a drug or radiopharmaceutical, used in
hospital outpatient departments, covered by Medicare, and paid for as an
individual product for which CMS established a separate payment category
rather than placing it in a category that included other services.
1In this report, the term drugs refers to both drugs and biologicals.
Biologicals are products derived from living sources, including humans,
animals, and microorganisms. Radiopharmaceuticals are radioactive
substances used for diagnostic or therapeutic purposes.
The MMA directed CMS to set 2006 payment rates for SCOD products equal to
hospitals' average acquisition costs-the cost to hospitals of acquiring a
product, net the cost of rebates.3 In several related requirements, the
MMA directed us to provide information on SCOD costs and CMS's proposed
rates.4 First, we were required to conduct a survey of a large sample of
hospitals to obtain data on their acquisition costs of SCODs and provide
information based on these data to the Secretary of Health and Human
Services for his consideration in setting 2006 Medicare payment rates.5 We
provided information from this survey in two reports6-one on drugs and
another on radiopharmaceuticals. These reports presented systematic
information on hospitals' purchase prices of SCODs and limited information
on rebates.7 Second, we were required to evaluate CMS's proposed rates for
SCODs and comment on their appropriateness in light of the survey of SCOD
prices we conducted. We provided our comments in a report issued in
October 2005.8
2Pub. L. No. 108-173, sec. 621(a), S: 1833(t)(14), 117 Stat. 2066, 2307-08
(to be codified at 42 U.S.C. S: 1395l(t)(14)).
3Specifically, the MMA required that payment rates equal the average
acquisition costs as determined by the Secretary of Health and Human
Services, unless hospital acquisition cost data are not available. If such
data are not available, the law permitted payment rates to equal one of
several amounts, including average sales price, as calculated and adjusted
by the Secretary. MMA 117 Stat. 2307.
4MMA 117 Stat. 2308-09. The law also required the Medicare Payment
Advisory Commission (MedPAC) to report on overhead and related expenses
(such as pharmacy services and handling costs) and authorized the
Secretary to adjust the SCOD rates for these costs. MMA 117 Stat. 2309.
See ch. 6, "Payment for pharmacy handling costs in hospital outpatient
departments," in MedPAC's mandated report, Issues in a Modernized Medicare
Program (Washington, D.C.: June 2005).
5The Secretary of HHS considered the price data we provided but elected
not to use these data as the basis for 2006 rates.
6GAO, Medicare: Drug Purchase Prices for CMS Consideration in Hospital
Outpatient Rate Setting, GAO-05-581R (Washington, D.C.: June 30, 2005),
and GAO, Medicare: Radiopharmaceutical Purchase Prices for CMS
Consideration in Hospital Outpatient Rate Setting, GAO-05-733R
(Washington, D.C.: July 14, 2005).
7The term purchase price refers to the price that hospitals paid upon
receiving a product. The term rebates refers to price concessions given to
hospitals by manufacturers subsequent to receipt of the product.
Two other MMA requirements had a role for us-to report on any variation
found in our survey results in acquisition costs among hospitals and to
advise on future data collection efforts by CMS based on our survey
experience.9 This report addresses these requirements and examines (1) the
extent to which SCOD prices identified in our survey differed among
hospitals with different characteristics, (2) lessons the MMA-mandated
survey experience provided for the methodology and frequency of future
collection of SCOD price data, and (3) the challenges CMS faces in
collecting data to set SCOD payment rates accurately after 2006.
To examine price variation among a sample consisting of 1,157 hospitals
purchasing SCOD products, we conducted a multivariate statistical analysis
and grouped hospitals by certain key characteristics, including teaching
status, location, and size. We defined a hospital's teaching status as
major, other teaching, or nonteaching, based on the hospital's
intern/resident-to-bed ratio;10 location as urban or rural based on
metropolitan statistical areas (MSA); and size as a hospital's total
Medicare outpatient charges, classifying a hospital as large if its
Medicare charges were at or above the 80th percentile of all hospital
outpatient charges. The prices we examined were drawn from our survey of
hospitals' purchase prices for 62 SCODs for the period July 1, 2003,
through June 30, 2004.11 We determined that our survey data were reliable
for estimating SCOD prices. For details on our methodology, see appendix
I.
8GAO, Medicare: Comments on CMS Proposed 2006 Rates for Specified Covered
Outpatient Drugs and Radiopharmaceuticals Used in Hospitals, GAO-06-17R
(Washington, D.C.: Oct. 31, 2005).
9MMA 117 Stat. 2308-09.
10Major teaching hospitals were defined as having an
intern/resident-to-bed ratio of 0.25 or more. Hospitals with other
teaching programs had an intern/resident-to-bed ratio above 0 but less
than 0.25.
11The products in these SCOD categories represented 95 percent of all
Medicare spending on SCOD products (53 drugs and 9 radiopharmaceuticals)
during the first 9 months of 2004. The nine radiopharmaceuticals accounted
for 90 percent of all Medicare hospital outpatient spending on
radiopharmaceutical SCODs.
To identify lessons learned from our hospital survey experience as well as
challenges for CMS's future data collection,12 we reviewed the findings
from our issued reports on SCOD drug prices,13 SCOD radiopharmaceutical
prices,14 and CMS's proposed SCOD rates;15 consulted on methodological
issues with an advisory panel of experts in pharmaceutical economics,
pharmacy, medicine, survey sampling, and Medicare payment;16 interviewed
officials from CMS and several dozen hospitals; and reviewed CMS's final
rule on Medicare's 2006 payment rates for SCODs.17 In particular, we
reviewed CMS's published method for collecting the average sales prices
(ASP) of drug SCODs: manufacturers report their ASPs quarterly to CMS,
which uses them as a proxy for average acquisition costs in setting drug
SCOD payment rates. We performed our work according to generally accepted
government auditing standards from September 2005 through April 2006.
Results in Brief
In an analysis of price data collected from our survey of hospitals, we
found that prices hospitals paid for the SCOD products they purchased
varied across hospitals. Certain factors-namely, whether the hospital had
a major teaching program or not, was in an urban or rural area, and had a
large or small hospital outpatient department-were associated with whether
hospitals paid higher or lower prices for the SCOD products they
purchased. Specifically,
o compared with nonteaching hospitals, major teaching hospitals
paid prices that were, on average, an estimated 3.2 percent lower
for drug SCODs;
o compared with urban hospitals, rural hospitals paid prices that
were, on average, an estimated 4.4 percent higher for
radiopharmaceutical SCOD;, and
o compared with smaller hospitals, large hospitals paid prices
that were, on average, an estimated 1.4 percent lower for drug
SCODs and 3.1 percent lower for radiopharmaceutical SCODs.
Combining the three factors, we found that large, urban, major
teaching hospitals generally paid lower prices, on average, for
all SCOD products than did hospitals grouped by other combinations
of factors.
A key lesson for CMS that we learned from conducting the 2004
MMA-mandated hospital survey is that such a survey would not be
practical for collecting the data needed to set and update SCOD
rates routinely. However, it would be useful, on occasion, for CMS
to survey hospitals so that the rate-setting data it obtained from
other sources could be validated by an independent source. Our
2004 hospital survey produced accurate hospital drug price data,
but it also created a considerable burden for hospitals as data
suppliers and considerable costs for us as the data
collector-signaling the difficulties that CMS would face in
implementing similar surveys in the future. Hospitals told us
that, to submit the required price data, they had to divert staff
from their normal duties, thereby incurring additional costs.
Similarly, we incurred substantial staff and contractor costs to
make data obtained from diverse information systems comparable and
usable for SCOD rate-setting. Nevertheless, we found that the
benefit of obtaining data on actual prices paid by hospitals could
make such surveys advantageous for validating, on an occasional
basis-possibly every 5 or 10 years-ASP data that manufacturers
report to CMS for developing SCOD payment rates.
CMS will face important challenges as it seeks to obtain accurate
data on hospitals' acquisition costs for both drug and
radiopharmaceutical SCODs.
o With regard to drug SCODs, CMS lacks the detail on
manufacturers' ASP data needed to determine if the Medicare
payment rates developed from these data are appropriate
specifically for hospitals. Manufacturers report ASP as a single
price paid by all purchasers-as defined by law-but do not identify
purchasers by type or share of purchases. Therefore, CMS could not
determine whether hospitals pay more or less than physicians, for
example, for drug SCODs. If other providers paid more or less than
hospitals, that could result in an average that was either higher
or lower than what hospitals paid. In our October 2005 report, we
recommended that CMS collect information on manufacturers' ASP
that would identify purchaser types.18 In addition, CMS instructs
manufacturers to report ASP net of rebates but does not provide
guidance on how to allocate to an individual product rebates that
are based on purchases of more than one product.
o With regard to radiopharmaceutical SCODs, their complex nature
as compared with drugs poses challenges for collecting and
interpreting cost data. Because radiopharmaceuticals consist of a
radioisotope and a medicine or pharmaceutical agent, hospitals can
purchase them in ready-to-use unit dose form, as most hospitals
do, multidose, or as separate components to be subsequently
compounded. The different purchase options available to hospitals
make pricing radiopharmaceuticals uniformly across hospitals
infeasible. In addition, the short half-life of certain
radioisotopes, which causes these products to decay over time,
makes the hospital's distance from its supplier a factor in how
much is purchased. This can lead to differences among hospitals in
the amount purchased per beneficiary served. Given the
complexities of radiopharmaceuticals, it is also important to note
that the amount spent on radiopharmaceuticals is less than 1.5
percent of total Medicare spending on hospital outpatient
services. This small percentage together with the complexities of
radiopharmaceuticals complicate CMS's ability to select a data
source that can provide reasonably accurate data efficiently.
In this report, we make recommendations to the Secretary of Health
and Human Services regarding both drugs and radiopharmaceuticals.
We recommend that CMS occasionally validate manufacturers'
reported ASPs as a measure of hospitals' acquisition costs, using
hospital purchases obtained from a survey or other method. We also
recommend the use of ready-to-use unit-dose prices as the data
source for radiopharmaceutical SCOD rate-setting. In commenting on
a draft of this report, HHS agreed to consider our
recommendations, but expressed several reservations. In
particular, it was concerned about the burden of a hospital survey
for both hospital staff and the agency. We recognize the burden of
hospital surveys and for this reason recommended only occasional
hospital surveys-or an alternative method-to validate price data
reported by manufacturers.
Background
In the period following the enactment of legislation establishing
Medicare's OPPS and leading up to the MMA in 2003, concerns were
expressed about the adequacy of payments for innovative
pharmaceutical products. The MMA addressed these concerns by
establishing a payment policy for SCODs. As mandated by the MMA,
we conducted a hospital survey and provided HHS with information
about prices hospitals paid for SCOD products. Details follow on
the background of SCODs, our survey, CMS's new rates for drug
SCODs, and the nature of radiopharmaceutical products.
MMA Established SCOD Payment Categories for Certain Pharmaceutical
Products to Ensure Beneficiary Access to New Products
CMS uses OPPS to pay hospitals for services that Medicare
beneficiaries receive as part of their treatment in hospital
outpatient departments. Under OPPS, Medicare pays hospitals
predetermined rates for most services. When OPPS was first
developed as required by the Balanced Budget Act of 1997,19 the
rates for hospital outpatient services, drugs, and
radiopharmaceuticals were based on hospitals' 1996 median costs.
However, these rates prompted concerns that payments to hospitals
would not reflect the costs of newly introduced pharmaceutical
products used to treat, for example, cancer, rare blood disorders,
and other serious conditions. In turn, congressional concerns were
raised that beneficiaries might lose access to some of these
products if hospitals avoided providing them because of a
perceived shortfall in payments. In response to these concerns,
the Medicare, Medicaid, and SCHIP Balanced Budget Refinement Act
of 1999 authorized pass-through payments, which were a way to
temporarily augment the OPPS payments for newly introduced
pharmaceutical products first used after 1996.20 The MMA modified
this payment method for some of these pharmaceutical products. As
part of the modification, the MMA defined the new SCOD payment
category, which includes many of these newly introduced
pharmaceutical products. The MMA requires that SCODs be placed in
separate payment categories-that is, not packaged with related
services.
MMA Required Us to Survey Hospitals to Determine Their Acquisition
Costs for SCOD Products
As directed by the MMA, we conducted a survey of a large sample of
hospitals to determine their acquisition costs for SCOD products.
We surveyed 1,400 hospitals and received usable data from 83
percent of the hospitals for drug SCODs and from 61 percent of the
1,322 hospitals that had submitted Medicare claims for
radiopharmaceutical SCODs in the first 6 months of 2003. We found
that we could not obtain data that would permit calculation of
hospitals' acquisition costs, because, in general, hospitals were
unable to report accurately or comprehensively on rebates.21
Consequently, we reported average purchase prices for drug and
radiopharmaceutical SCODs, which are prices net of discounts but
not rebates.22 Of the 251 SCODs that we identified, we reported
average purchase prices for the 62 SCODs that accounted for 95
percent of Medicare spending on all SCODs in the first 9 months of
2004. (These prices and related information are included as app.
II and app. III.)
MMA Defined ASP, Which Is Reported by Manufacturers and Used to Set
Rates for Drug SCODs
ASP is a price measure established in the MMA to provide a basis
for payment rates for physician-administered drugs and now used by
CMS in setting rates for drug SCODs.23 CMS instructs
pharmaceutical manufacturers to report ASP data to CMS within 30
days after the end of each quarter. The MMA defined ASP as the
average sales price for all U.S. purchasers of a drug, net of
volume, prompt pay, and cash discounts; free goods contingent on a
purchase requirement; and charge-backs and rebates.24 Under CMS's
final rule governing 2006 payment rates for hospital outpatient
services, including SCOD products, CMS uses manufacturers' ASPs in
setting drug SCOD rates.25 For radiopharmaceuticals, CMS has set
2006 rates based on an estimate of hospitals' costs derived from
charges, but the agency has not decided how to pay for
radiopharmaceutical SCODs after 2006.26
Radiopharmaceuticals Can Be Purchased in Different Forms
Hospitals can purchase radiopharmaceuticals, which consist of a
radioisotope and a medicine or pharmaceutical agent, in different
forms. They can purchase vials of the product in ready-to-use unit
doses or in multidoses, or they can purchase a product's
radioactive and nonradioactive components separately and compound
them in-house. In a survey conducted by the Society of Nuclear
Medicine and the Society of Nuclear Medicine Technologist Section,
76 percent of hospitals reported that they purchased their
radiopharmaceuticals in unit doses.27
Hospitals� Teaching Status, Location, and Size Affected Prices for
SCOD Products by Different Magnitudes
Using our hospital survey of prices hospitals paid for SCOD drugs
and radiopharmaceuticals, we examined the extent to which prices
varied among the approximately 1,200 hospitals that submitted
survey data. To do this, we looked at several hospital
characteristics, or factors-including teaching status, location,
and size of the outpatient department-while controlling for
differences in the costliness of the mix of SCODs that hospitals
purchased. We analyzed both (1) the separate effect of each
factor, controlling for other factors; and (2) the effect of the
three factors combined. We found that teaching status had the
largest separate effect on drug SCOD prices, whereas location had
the largest effect on radiopharmaceutical SCOD prices. Combining
the three factors, we found, for example, that large, urban,
hospitals with major teaching programs paid lower prices, on
average, for drug SCODs-compared with small urban hospitals with
other teaching programs.
Teaching Status, Location, and Size Were Each Significant Factors
Affecting Price Variation among Hospitals
The importance of the three factors in accounting for variation in
SCOD prices among hospitals differed by type of product
purchased-that is, drug or radiopharmaceutical.28 A hospital's
teaching status, for example, affected prices paid for drug SCODs
but did not matter for the radiopharmaceutical SCOD prices
pertaining to unit dose purchases in our survey. In contrast, a
hospital's location was an important factor linked to price
differences for radiopharmaceuticals but did not matter with
respect to prices for drugs. In addition, hospital size was
important in affecting price differences for both drugs and
radiopharmaceuticals. (See table 1.)
Table 1: Factors Accounting for Variation in SCOD Prices among
Hospitals
Sources: GAO analysis of GAO survey data and CMS data on hospital
characteristics.
Note: We determined the importance of these factors using a
multivariate statistical analysis that examined how prices varied
for SCODs by hospitals' teaching status, location, and size of
outpatient department, while controlling for differences in the
costliness of the mix of SCODs that hospitals purchased. Factors
marked with an "x" are statistically significant at the 5 percent
level.
In assessing the magnitude of each factor's separate effect on
prices, we found the following results:
o Teaching status: Compared with nonteaching hospitals, major
teaching hospitals paid prices that were, on average, an estimated
3.2 percent lower for drug SCODs. Teaching status had no
independent effect on the prices of radiopharmaceutical SCODs
purchased in ready-to-use unit doses.29
o Location: Compared with hospitals located in urban areas, the
prices paid by hospitals located in rural areas for
radiopharmaceutical SCODs were, on average, an estimated 4.4
percent higher.
o Size: Compared with smaller hospitals, hospitals with large
outpatient departments paid prices, on average, that were an
estimated 1.4 percent lower for drugs and 3.1 percent lower for
radiopharmaceuticals.
Certain circumstances may help explain why each factor had an
effect on price. Regarding the effect of teaching status on drug
prices, for example, manufacturers may want to influence
prescribing patterns of physicians in training and may therefore
offer drugs at lower prices to hospitals with teaching programs.
As for location's effect on radiopharmaceutical SCOD prices,
industry experts suggested that the short half-life of certain
radioactive products could make transporting them to hospitals in
rural areas more costly. As for hospital size, hospitals with
large outpatient departments may have benefited from volume
discounts.
Hospitals with Combination of Major Teaching Status, Urban Location,
and Large Size Obtained Lowest SCOD Prices
To examine the combined effect of the three key factors on prices
paid by hospitals, we compared hospitals grouped by one
combination-major teaching program, urban location, and large
outpatient department-with hospitals grouped by other
combinations. Our analysis indicates that large, urban, major
teaching hospitals generally paid lower prices, on average, for
all SCOD products than did hospitals grouped by other combinations
of factors. For example, compared with small urban hospitals with
other teaching programs, large major teaching hospitals in urban
areas paid prices, on average, that were an estimated 4 percent
lower for drugs and 3 percent lower for radiopharmaceuticals. In
contrast, compared with small urban hospitals with other teaching
programs, small rural hospitals with no teaching programs paid
prices, on average, that were about the same for drugs and 4
percent higher for radiopharmaceuticals.30
Our Survey of Hospitals Suggests that the Burden of Large-Scale
Annual Surveys Could Outweigh Gains in Data Accuracy
Our MMA-mandated survey of hospitals produced accurate hospital
price data. However, for CMS to use such a survey to routinely
collect data in the future for SCOD rate-setting, the burden could
outweigh the benefit. Instead, similar surveys of hospitals could
be a useful tool to validate price data obtained from
manufacturers, if conducted on an occasional basis.
Using Hospitals as Data Source for SCOD Prices Had A Major Advantage
and Serious Drawbacks
Based on our survey experience, we noted that hospitals as a SCOD
data source had one important advantage as well as substantial
drawbacks. We found that, as a data source for estimating
hospitals' SCOD acquisition costs, hospitals offered a key
advantage: our average purchase prices obtained from hospitals, by
definition, represent actual prices paid by hospitals.31 In this
respect, our data differ from other data sources available to
CMS-such as suggested list prices, ASPs, and hospitals' Medicare
claims. As a result, none of these alternatives provide, as our
survey data do, nationwide data on the actual purchase prices paid
by hospitals for drugs and radiopharmaceutical SCODs.
However, based on our experience, we found that there would be
drawbacks in using hospitals as an annual data source on SCOD
prices, owing primarily to the considerable burden created for
hospitals as suppliers of data and the considerable costs we
incurred as data collectors, signaling the difficulties that CMS
would face in implementing similar surveys in the future.
Hospitals told us that, to submit the required price data, they
had to divert staff from their normal duties, thereby incurring
additional staff and contractor costs. The burden was more taxing
for some hospitals than for others. Most hospitals had the
advantage of relying on price data downloaded from their drug
wholesalers' information systems. A number of hospitals, however,
either collected the data manually, provided us with copies of
paper invoices, or had automated information systems that were not
designed to retrieve the detailed price data needed and required
additional data processing effort. Hospitals' data collection
difficulties were particularly pronounced regarding information on
manufacturers' rebates, which affect a drug's net acquisition
cost. Typically, hospitals did not systematically track all
manufacturers' rebates on drug purchases, although nearly 60
percent of hospitals reported receiving one or more rebates.32
As collectors of data on SCOD prices, we also experienced
difficulties obtaining the information needed from hospitals.
Hospitals' information systems were diverse and produced data in
many different formats, causing substantial resource and timing
difficulties in the data collection process.33 Specifically, we
had to reconfigure data submitted in multiple formats to produce
data comparable across hospitals and usable for SCOD rate-setting.
This reconfiguration required us to deploy substantial resources
and to allow additional time for processing before the data could
be made available to CMS. The difficulties we encountered would
likely be faced by any organization undertaking a survey using a
similar approach.
As we previously reported, using SCOD price and related data from
drug manufacturers-as CMS is doing in 2006-is a practical strategy
for setting Medicare payment rates to hospitals for SCODs.34
However, our experience obtaining information on actual purchase
prices and our observation of the pace of change in the drug
marketplace suggest that an occasional survey of
hospitals-possibly once or twice in a decade-may be advantageous
for validating the accuracy of manufacturers' price information as
a proxy for hospital acquisition cost.35 Drawing on our experience
and using data about sampling variability from our 2004 hospital
survey,36 CMS could design a similar but streamlined hospital
survey.37 Other options available to CMS for validating the
accuracy of the price data as a proxy for hospitals' acquisition
costs include audits of manufacturers' price submissions or an
examination of proprietary data the agency considers reliable for
validation purposes.
Survey Indicates that Accounting for Dynamic Drug Market and
Infrequently Purchased Drugs Has Implications for Accuracy and
Efficiency
Our hospital survey experience not only identified data collection
issues associated with hospitals but also underscored accuracy and
efficiency concerns in collecting SCOD data from any source.
Specifically, the accuracy of the rates Medicare pays for drugs
within a SCOD payment category, based on the average price of
drugs included in the SCOD, may be compromised if the price of any
drug-that is, any national drug code (NDC)-is omitted from the
average price of the SCOD category.38 In the conduct of our 2004
survey, we began with a list, which CMS provided to us, of drug
categories that included SCODs as well as other drugs that
potentially could be considered SCODs in the future. To ensure the
accuracy of our calculation of a hospital's average purchase price
for SCODs, we took additional steps using industry experts and
data sources to classify the NDCs and assign them to the
appropriate SCOD categories.39 Since the drug market is
dynamic-new drugs enter the market and other drugs drop out in the
course of a year-CMS's list of SCOD drugs and their component NDCs
could become out of date unless updated frequently to ensure that
all SCOD drugs purchased by hospitals are identified and figured
into the calculation of a SCOD's average price.
With regard to efficiency in analyzing our survey results, we
concentrated our data processing and statistical resources on the
roughly one-quarter of SCODs that account for most of Medicare's
total SCOD spending. In particular, the 62 SCODs for which we
produced price estimates accounted for 95 percent of Medicare
spending on all 251 SCODs in the first 9 months of 2004.40 We
would not have been able to produce price estimates for all SCODs
in time for CMS to take account of our data in setting the 2006
rates. Our experience-especially the amount of time and resources
necessary for each step in the data collection and analysis
process-could be used by CMS to determine in advance the number of
SCODs on which to collect data and estimate prices. There might be
some benefit in gathering data and producing price estimates for
all SCODs; on the other hand, if resources were limited, CMS might
choose to focus on fewer SCODs.
CMS Faces Challenges in Future Data Collection Efforts to Set SCOD
Payment Rates Accurately
CMS will face important challenges in its efforts to collect
accurate data for setting SCOD payment rates. In our October 2005
report on CMS's proposed SCOD rates, we expressed reservations
about the ASP data CMS used to set 2006 payment rates for drug
SCODs. We cautioned that manufacturers' reporting of ASPs in
summary form-without any further detail-does not provide the
agency the information needed to ensure that ASPs are a
sufficiently accurate measure of hospitals' acquisition costs.
Data collection and rate-setting for radiopharmaceutical SCODs
present unique challenges because of these products' distinctive
characteristics.
Validating ASP Would Pose Challenges for CMS Because of Lack of
Detail in Data
Under CMS's current policy, manufacturers are required to report
only summary ASP data, limiting CMS's ability to validate the
data's accuracy. Specifically, manufacturers report ASP as a
single price, with no breakdown of price and volume by type of
purchaser. CMS instructs manufacturers to average together prices
for each drug paid by all U.S. purchasers. However, different
purchaser types-for example, hospitals, physicians, and
wholesalers-may receive prices that, by purchaser type, are on
average higher or lower than one another's. Because CMS does not
receive price data at this level of detail, it cannot determine
whether price differences among purchaser types exist. To the
extent that nonhospital providers pay different prices than
hospitals and account for a proportion of the SCODs purchased, ASP
will differ from the prices paid on average by hospitals.41 CMS
has not presented evidence, in its final rule or in discussions
with us, that physicians and hospitals pay the same prices.
An additional weakness in CMS's instructions for computing ASPs
compounds the challenge of testing the accuracy of the ASPs that
manufacturers report. No instruction is provided to manufacturers
on the treatment of rebates that apply to several drug products in
calculating ASP.42 This is of particular concern to the extent
that manufacturers differ in their rules for calculating these
rebates. When a rebate applies to a group of a manufacturer's
products-which may include several SCODs, other pharmaceuticals,
and other products-netting out the rebate attributable to a
specific SCOD's purchase is less than straightforward. In the
absence of clear and specific instructions, each manufacturer must
identify or develop a method for allocating rebates to each of its
drug SCOD products. To the extent that manufacturers' methods
differ, they are likely to yield inconsistent results. Moreover,
CMS's final rule does not provide for a follow-up process to check
that rebate allocations have been made or have been made
appropriately.
Radiopharmaceuticals Pose Unique Challenges for Obtaining Accurate
Cost Data Efficiently
The complex nature of radiopharmaceuticals as compared with drugs
poses special challenges for collecting and interpreting cost
data. These challenges include (1) obtaining consistent data for
radiopharmaceutical SCODs produced in very different forms and (2)
the short half-life for certain products. Moreover, since Medicare
spends relatively little on radiopharmaceuticals-less than 1.5
percent of Medicare spending on hospital outpatient services-the
challenge is to find a source of data for setting rates that is
low cost and reasonably accurate.
In our hospital survey, we faced the challenge of uniformly
pricing products purchased in very different forms. We focused on
prices for radiopharmaceuticals purchased in unit doses. Most of
the hospitals purchased radiopharmaceuticals in this ready-to-use
form, and only a small fraction of hospitals purchased
radiopharmaceuticals in separate components (the radioisotope and
the nonradioactive substance), which need to be compounded.43 We
were unable to make prices for separately purchased components
comparable to those obtained for unit doses, as the labor costs
for compounding the products are included in hospitals' reported
prices of ready-to-use products but not in their reported prices
of products they purchased as separate components.
The short half-life of certain radiopharmaceutical SCODs can also
pose challenges for collecting and interpreting price data.
Because the radioactive component decays over time, the amount of
the product purchased for a given patient may vary with the
distance between where the radiopharmaceutical is compounded and
where it is administered. The result is that for those short-lived
radiopharmaceuticals paid on a per-dose basis, the cost per dose
is more for the doses prepared far from the point of
administration than for those prepared closer by, as more of a
radioactive product must be purchased to account for its decay in
transit. This applies most commonly to F-18 radiopharmaceuticals,
the most common of which, F-18 FDG, has a half-life of 1.8
hours.44 F-18 radiopharmaceuticals, including F-18 FDG, are used
in the diagnosis of various diseases, such as cancer, heart
disease, and liver disease.
Finally, CMS faces the challenge of balancing accuracy and
efficiency in obtaining price data on radiopharmaceutical SCODs.
Our approach in estimating prices from our survey data was to use
only information on unit dose prices, the form purchased by most
hospitals.45 CMS, as stated in the 2006 final rule governing
payment rates for SCODs, has not found what it considers a
satisfactory method for obtaining data on acquisition costs of
radiopharmaceuticals and is continuing to explore both ASP and
other alternatives.46 Hospitals and manufacturers47 are the most
direct source of price data because both are parties to the
transactions in which the hospitals acquire the
radiopharmaceuticals.48 In its notice of proposed rulemaking for
radiopharmaceutical SCODs, CMS proposed collecting ASPs from
manufacturers for use in setting 2007 payment rates.49 In light of
many comments regarding the difficulty of this undertaking, CMS
decided not to collect radiopharmaceutical ASPs for 2007 rates,
but left open the possibility of using ASP in the future.
CMS has also discussed the possibility of using charges from
hospitals' Medicare claims to approximate acquisition costs for
radiopharmaceutical SCODs, rather than obtaining price data from
invoices provided by hospitals or from manufacturers. Using claims
data may be a more efficient but less accurate means of obtaining
price estimates than obtaining price data directly from
manufacturers or from hospitals' invoices. In its final rule, CMS
stated that it was basing 2006 payments on hospitals' charges
(derived from outpatient claims) for radiopharmaceuticals. CMS
plans to adjust these charges to reflect costs and noted that it
did not plan to use this methodology permanently. For rate-setting
after 2006, CMS also noted the possibility of using invoice data
submitted to Medicare by physicians who administer
radiopharmaceuticals in their offices.50 In its final rule, CMS
did not present evidence that hospitals and physicians pay similar
prices for these radiopharmaceutical drugs nor, if these prices
differ, whether using these physician data would be appropriate
for use in setting hospital outpatient rates.
Conclusions
Basing Medicare's payment rates for hospitals' SCOD purchases on
current, accurate price data is important both to ensuring that
Medicare pays appropriately-neither too much nor too little-and to
ensuring beneficiary access to these innovative pharmaceutical
products. As we previously reported, we agree with CMS that ASP is
a practical data source for setting and updating rates for drug
SCODs on a routine basis. However, we remain concerned about
whether CMS can determine that ASP accurately represents purchases
made by hospitals and believe that CMS should implement our
October 2005 recommendation to collect sufficient information on
ASP to make such a determination. We are also concerned about the
likelihood that ASPs are not calculated consistently across all
manufacturers, owing to CMS's lack of detailed instructions. As
for validating the data CMS collects to set payment rates equal to
hospitals' acquisition costs, an examination of hospitals' actual
purchase prices, by definition, is optimal for assessing accuracy.
Recognizing the operational difficulties of a hospital survey and
using the knowledge gained from our survey, CMS could conduct a
similar but streamlined hospital survey, possibly once or twice in
a decade. Other options available to CMS for validating price data
could include audits of manufacturers' price submissions or an
examination of proprietary data the agency considers reliable for
validation purposes.
In contrast, we found that the diversity of forms in which
radiopharmaceutical SCODs can be purchased-ready-to-use unit
doses, multidoses, or separately purchased radioactive and
nonradioactive components-complicates CMS's efforts to select a
data source that can provide reasonably accurate price data
efficiently. Our experience suggests that the best option
available to CMS, in terms of accuracy and efficiency, is to
collect price data on radiopharmaceuticals purchased in
ready-to-use unit doses, the form in which an estimated
three-quarters of hospitals purchase these products.
Recommendations for Executive Action
To ensure that Medicare payments for SCOD products are based on
sufficiently accurate data, we recommend that the Secretary of
Health and Human Services take the following two actions:
o validate, on an occasional basis, manufacturers' reported drug
ASPs as a measure of hospitals' acquisition costs using a survey
of hospitals or other method that CMS determines to be similarly
accurate and efficient; and
o use unit-dose prices paid by hospitals when available as the
data source for setting and updating Medicare payment rates for
radiopharmaceutical SCODs.
Agency Comments and Our Evaluation
We received written comments on a draft of this report from HHS
(see app. IV), which noted that it had considered information from
our survey of hospitals in developing 2006 hospital outpatient
payment policy and expressed appreciation for our effort and
analysis.
Regarding the first recommendation-that HHS validate ASPs as a
measure of hospital acquisition costs through occasional hospital
surveys or other methods-HHS highlighted our finding that an
annual hospital survey could place considerable burdens on both
the agency and hospital staff. However, HHS agreed to consider
this recommendation, saying that it would continue to analyze the
best approach for setting payment rates for drugs and
radiopharmaceutical SCODs in view of our recommendation. It will
also continue to analyze the adequacy of paying for drugs at ASP+6
percent in the light of claims data, which persuaded HHS that for
2006 ASP + 6 percent was the best available proxy for hospital
acquisition and handling costs.
Regarding the second recommendation-that HHS use unit-dose prices
to set and update payment rates for radiopharmaceuticals-HHS
agreed with us that the multiple forms in which
radiopharmaceuticals can be purchased makes setting their payment
rates difficult. While agreeing to consider our recommendation,
HHS expressed several reservations. First, it noted that we had
not specified whether the survey to collect acquisition cost data
should be a survey of hospitals or manufacturers and asked that we
clarify this point. Second, it noted that we had emphasized the
burden of annual surveys of hospital drug prices and expressed the
concern that an annual survey of hospital radiopharmaceutical
prices would be equally burdensome. Finally, HHS noted that we had
confined our report to 9 of the approximately 55
radiopharmaceuticals that are paid separately, and questioned
whether unit-dose data would be available for all or most
radiopharmaceuticals.
Our recommendation that HHS validate ASPs through occasional
surveys or by using other methods is based in considerable part on
our experience of the difficulty of a hospital survey. The burden
that annual surveys would place on both hospitals and the agency
is the reason that we rejected annual surveys as a source of
acquisition cost data and instead proposed only occasional surveys
to validate ASPs. Furthermore, as we noted in the recommendation,
HHS could use a method other than a survey if that method were
similarly accurate and efficient.
In our recommendation on radiopharmaceuticals, we did not comment
on whether the survey to collect acquisition cost data should be a
survey of hospitals or manufacturers, because we have not analyzed
the feasibility of obtaining these data from manufacturers. We
recognize the potential burden of hospital surveys; this burden
would need to be taken into account in weighing the merits of a
hospital survey versus other alternatives. Regarding our
recommendation to collect unit-dose prices, we have clarified it,
saying that unit-dose prices should be used when available. In our
survey, we used unit-dose data when we reported purchase prices
for the 9 radiopharmaceuticals that accounted for 90 percent of
Medicare's costs for hospital outpatient drugs. For
radiopharmaceuticals that are prepared exclusively in-house HHS
could, if necessary, establish an alternative method for
determining payment rates.
We are sending copies of this report to the Secretary of Health
and Human Services, the Administrator of the Centers for Medicare
& Medicaid Services, and other interested parties. We will also
make copies available to others upon request. In addition, the
report will be available at no charge on the GAO Web site at
http://www.gao.gov .
If you or your staff members have any questions about this report,
please contact me at (202) 512-7119 or at [email protected] .
Contact points for our Office of Congressional Relations and
Public Affairs may be found on the last page of this report. GAO
staff who made major contributions to this report are listed in
appendix V.
A. Bruce Steinwald Director, Health Care
Appendix I: Methodology for Analysis of SCOD Price Differences
among Hospital Types
This appendix describes the data and methods we used to examine
SCOD price variation among hospitals purchasing SCOD products. In
particular, we describe (1) the SCOD price data we analyzed, (2)
the factors potentially affecting SCOD prices and the measurement
of these factors, and (3) the methods underlying the statistical
analysis of prices we conducted and the statistical results we
obtained.
SCOD Price Data
Drawing on data from our survey of 1,157 hospitals,1 we examined
hospitals' purchase prices for 53 drug SCODs and 9
radiopharmaceutical SCODs for the period July 1, 2003, through
June 30, 2004.2 Combined, these 62 SCOD categories represented 95
percent of Medicare spending on SCOD products during the first 9
months of 2004. We analyzed invoice data that hospitals submitted
to us; specifically, our analysis included one SCOD price for each
SCOD purchase listed on an invoice. As a result, for a hospital
that purchased SCODs and other drugs once a month, our analysis
included 1 price for each month's purchase of a particular SCOD or
a total of up to 12 invoice prices for that SCOD during the
12-month period. We were advised in our analysis by an expert
panel consisting of Joseph P. Newhouse, John D. MacArthur
Professor of Health Policy and Management, Harvard University;
Robert A. Berenson, Senior Fellow, Urban Institute; Ernst R.
Berndt, Professor of Applied Economics, Sloan School of
Management, Massachusetts Institute of Technology; Andrea G.
Hershey, Clinical Coordinator and Pharmacy Residency Program
Director, Union Memorial Hospital (Baltimore, Md.); and Richard L.
Valliant, Senior Research Scientist, University of Michigan.
Factors Potentially Affecting SCOD Prices
To analyze SCOD price variation among hospitals purchasing SCODs,
we identified characteristics of hospitals that could plausibly
explain why prices vary: teaching status, location, and size. We
also identified a fourth factor: differences in the costliness of
the mix of SCODs that hospitals purchased. Table 2 lists these
factors and describes operational measures of these factors and
the sources of data used to calculate these measures.
12For setting SCOD payment rates after 2006, the Secretary was directed to
conduct periodic surveys to obtain cost information.
13 GAO-05-581R .
14 GAO-05-733R .
15 GAO-06-17R .
16See app. I.
1770 Fed. Reg. 68,516 (Nov. 10, 2005).
18 GAO-06-17R .
Background
MMA Established SCOD Payment Categories for Certain Pharmaceutical Products to
Ensure Beneficiary Access to New Products
MMA Required Us to Survey Hospitals to Determine Their Acquisition Costs for
SCOD Products
19Pub. L. No. 105-33, S: 4523, 111 Stat. 251, 445-50.
20Pub. L. No. 106-113, app. F, S: 201(b), 113 Stat. 1501A-321,
1501A-337-1501A-339.
21Rebates are price concessions given to hospitals by manufacturers
subsequent to receipt of the product. For a discussion of rebates and
their relationship to hospital acquisition costs, see GAO-06-17R , p. 5.
MMA Defined ASP, Which Is Reported by Manufacturers and Used to Set Rates for
Drug SCODs
Radiopharmaceuticals Can Be Purchased in Different Forms
22Discounts are price concessions given by manufacturers and wholesalers
that are reflected in the purchase price-the price hospitals pay at the
time of delivery.
23MMA 117 Stat. 2239-45. MMA specifically required use of ASP to set rates
for drugs furnished in physicians' offices on or after January 1, 2005;
CMS began using ASP to set rates for SCOD products delivered in hospital
outpatient departments on or after January 1, 2006.
24MMA 177 Stat. 2240-41.
2570 Fed. Reg. 68,642. In total, the payment rate for drug SCODs is ASP+6
percent, which includes overhead and handling that CMS had previously
estimated at 2 percent of ASP. The implied rate for the product without
overhead is ASP+4 percent.
2670 Fed. Reg. 68,654.
Hospitals' Teaching Status, Location, and Size Affected Prices for SCOD Products
by Different Magnitudes
Teaching Status, Location, and Size Were Each Significant Factors Affecting
Price Variation among Hospitals
27See Denise A. Merlino, "Nuclear Medicine Faculty Survey: SNM 2003 Survey
Reporting on 2002 Cost and Utilization," Journal of Nuclear Medicine
Technology, vol. 32, no. 4 (2004), pp. 215-219.
28Our estimated purchase prices for radiopharmaceutical SCODs were based
on hospitals' purchases of ready-to-use unit-doses only; we did not report
prices for the generally less prevalent forms-multidoses or doses prepared
in-house using a kit.
Hospital characteristic Drugs Radiopharmaceuticals
Teaching status
Location
Size
29Compared with nonteaching hospitals, some teaching hospitals may obtain
a larger proportion of their radiopharmaceuticals by compounding
components purchased separately than by purchasing unit doses. Therefore,
the result might have been different had we been able to include the
prices hospitals paid for radiopharmaceuticals purchased as multidoses or
as separate components.
Hospitals with Combination of Major Teaching Status, Urban Location, and Large
Size Obtained Lowest SCOD Prices
Our Survey of Hospitals Suggests that the Burden of Large-Scale Annual Surveys
Could Outweigh Gains in Data Accuracy
Using Hospitals as Data Source for SCOD Prices Had A Major Advantage and Serious
Drawbacks
30The estimated percentage differences were derived from two multivariate
statistical models-one explaining variation in prices of 53 drug SCODs,
the second explaining price variation of 9 radiopharmaceutical SCODs. Each
model attributed variation in SCOD prices to three hospital
characteristics (teaching status, size, and location) and to the
particular set of SCODs purchased by each hospital.
31CMS collects ASPs from manufacturers that include prices paid by all
purchasers, not just hospitals. Average prices paid by hospitals may not
be equal to average prices paid by other purchasers, such as physicians'
offices.
32Many hospitals reported receiving rebates for a set of drugs (and
sometimes drugs and other products). In these cases, it was generally not
feasible to allocate rebates to specific drugs.
33We accepted data from hospitals in any format. We believed that we had
to make the task of submitting data as easy as possible for hospitals in
order to gain their cooperation. Reflecting on our experience, we think
that this decision was critical to achieving good response rates.
34GAO, Medicare: Comments on CMS Proposed 2006 Rates for Specified Covered
Outpatient Drugs and Radiopharmaceuticals Used in Hospitals, GAO-06-17R
(Washington, D.C.: Oct. 31, 2005). In addition to the product's ASP,
manufacturers must report the manufacturer's name, the product's National
Drug Code (NDC), and the number of units.
35Although HHS chose to use ASP data submitted by manufacturers to set
2006 payment rates, it is required to conduct hospital surveys subsequent
to ours to determine hospital acquisition costs. MMA 117 Stat. 2308.
36We refer to our survey of hospitals as the 2004 survey because data
collection began in 2004. We collected data for SCODs purchased from July
1, 2003, through June 30, 2004.
37For details on the sample design for our survey, see GAO-05-581R ,
enclosure I.
Survey Indicates that Accounting for Dynamic Drug Market and Infrequently
Purchased Drugs Has Implications for Accuracy and Efficiency
38A SCOD category may contain one or many NDCs. NDCs may differ by
manufacturer, strength, or package size.
39Each SCOD and each NDC is assigned a specific number of units (for
example, 10 mg.), and the NDC units must also be converted to SCOD units,
in order to place on the same basis all the NDCs that make up a SCOD. For
a discussion of issues in converting NDC prices to SCOD prices, see
Department of Health and Human Services, Office of Inspector General,
Calculation of Volume-Weighted Average Sales Price for Medicare Part B
Prescription Drugs, OEI-03-05-00310 (Washington, D.C.: February 2006).
40The number of SCODs can change from year to year as CMS designates
additional SCODs or combines previously separate SCODs.
CMS Faces Challenges in Future Data Collection Efforts to Set SCOD Payment Rates
Accurately
Validating ASP Would Pose Challenges for CMS Because of Lack of Detail in Data
41We recommended in a previous report that CMS collect information on ASP
by purchaser type to validate its reasonableness as a measure of hospital
acquisition cost. See GAO-06-17R .
4242 C.F.R. S:S: 414.800-414.806 (2005).
Radiopharmaceuticals Pose Unique Challenges for Obtaining Accurate Cost Data
Efficiently
43In a survey conducted by the Society of Nuclear Medicine and the Society
of Nuclear Medicine Technologist Section, 76 percent of hospitals reported
that they purchased their radiopharmaceuticals in unit doses. See Merlino,
pp. 215-219.
44Of the nine radiopharmaceuticals for which we estimated prices, F-18 FDG
is the only one that is an F-18 radiopharmaceutical. However, as more F-18
labeled products become available, the category may expand.
45See GAO-05-733R .
4670 Fed. Reg. 68,656-57.
47We consider manufacturers to include independent nuclear pharmacies and
hospitals that compound radiopharmaceuticals that they supply to other
hospitals.
48A small part of the business of some independent nuclear pharmacies, as
well as retail outlets for large radiopharmaceutical manufacturers,
involves supplying ready-to-use radiopharmaceuticals from their parent
companies and other manufacturers.
4970 Fed. Reg. 42,674, 42,727-28 (July 25, 2005).
Conclusions
5070 Fed. Reg. 68,656.
Recommendations for Executive Action
Agency Comments and Our Evaluation
Appendix I: MeSCOD PriceTypes Appendix I: Methodology for Analysis of SCOD
Price Differences among Hospital Types
SCOD Price Data
Factors Potentially Affecting SCOD Prices
1See GAO-05-581R for technical details on the survey we conducted.
2Purchase price refers to the price that hospitals paid upon receiving a
product. Purchase price incorporates a manufacturer's or other vendor's
discounts but excludes any rebates, which manufacturers may pay a hospital
purchaser at a later date. In this appendix, price refers to purchase
price, unless otherwise stated.
Table 2: Factors Included in Analysis of Price Variation among Hospitals
Purchasing SCODs
Source and date of
data used to
Factor Measure calculate measure
Teaching statusa Major teaching: Binary variable equal CMS: Medicare
to Hospital Cost
Report, 2002
o 1 if the hospital had a major
teaching program
o 0 if hospital had no major
teaching program
Other teaching: Binary variable equal
to
o 1 if the hospital had other
teaching program
o 0 if hospital had no other
teaching program
Nonteaching: Binary variable equal to
o 1 if the hospital had no
teaching program.
o 0 if hospital had a teaching
program
Location Binary variable equal to CMS: Provider of
Services File, end
o 1 if the hospital was in a of 2004
rural area-that is, outside a
metropolitan statistical area
(MSA)
o 0 if the hospital was in an
urban area-that is, in an MSA
Sizeb Binary variable equal to CMS: Health Care
Information System,
o 1 if hospital is 2003
large-indicated by outpatient
Medicare charges at or above the
80th percentile of all Medicare
hospital outpatient charges
o 0 if small-less than this
amount
Mix of SCODs Binary variable equal to GAO: Survey of
purchased Hospitals' SCOD
o 1 if the product purchased as a Prices, 2003 and
given SCOD - that is, the it h 2004
SCOD, where i = 1,...n
o 0 if the product purchased were
any other SCOD
Sources: GAO analysis of CMS and GAO information.
aMajor teaching hospitals were defined as hospitals with an
intern/resident-to-bed ratio of 0.25 or more. Hospitals with other
teaching programs were defined as hospitals with an intern/resident-to-bed
ratio above 0 but less than 0.25.
bHospitals with outpatient Medicare charges of $59.1 million or higher
were at the 80th percentile or higher of hospitals, ranked by their
outpatient Medicare charges, for our analysis of drug SCODs.
In addition to the measures listed in table 2, we considered alternative
measures for location and for size:
o We examined two geographic classification systems as
alternatives to an MSA (metropolitan statistical area)/nonMSA
classification: (1) urban influence codes, which classify counties
based on each county's largest city and its proximity to other
areas with large, urban, populations; and (2) rural-urban
continuum codes, which classify metropolitan counties (that is,
those in an MSA) by the size of the urban area and classify
nonmetropolitan counties by the size of the urban population and
proximity to a metropolitan area.3
o Before selecting our preferred measure of hospital size
(hospital outpatient charges at the 80th percentile or higher,
where hospitals were ranked by their outpatient Medicare charges),
we considered other measures of hospital size: the number of
hospital beds, the number of unique SCODs purchased by a hospital,
and the number of hospital outpatient visits.
In assessing our regression results for each of the several
measures of location and size that we considered, we took into
account statistical criteria including the statistical
significance of each measure and the overall explanatory power of
each model. We also considered qualitative factors when selecting
our preferred measures of location and size. For example, we
selected hospital outpatient charges as our measure of size,
instead of number of hospital beds, because both measures had
similar statistical properties and our analysis focuses on the
hospital outpatient setting.
In addition to conducting separate regression analyses of the
price data for drug SCODs and for radiopharmaceutical SCODs, we
analyzed price variation separately for each of four therapeutic
categories of drug SCODs. We also conducted separate regression
analyses of SCOD price variation for drugs without biologicals,
for biologicals, and for radiopharmaceuticals. We determined that
any gains in statistical properties did not outweigh the greater
complexity of these analyses.
Methods and Results of Price Analysis
In analyzing SCOD price variation, our dependent variable was the
natural logarithm of SCOD price.4 SCOD prices are not distributed
symmetrically around the average. SCOD prices are skewed to the
right and are not distributed normally, reflecting some SCODs with
particularly high prices. Taking the natural logarithm of price is
intended to take skewness into account and make the resulting
distribution consistent with the statistical assumptions of a
regression.
We weighted prices paid by hospitals for individual drugs and
radiopharmaceuticals by the purchase amount of each invoice. That
is, we weighted prices more heavily in the statistical analysis
for invoices that represented a larger proportion of total annual
purchases of a particular SCOD than for invoices that represented
a smaller proportion of purchases. In addition, our analysis took
into account the fact that multiple prices paid by a particular
hospital were not necessarily statistically independent of each
other-a phenomenon known as clustering. In estimating our
statistical models, we corrected the potential bias in our
estimates due to clustering by using the robust and cluster
options in STATA, a statistical software package.5
To gauge the effects of our explanatory factors on price variation
among hospitals, we estimated one regression model for drug SCODs
and a separate model for radiopharmaceutical SCODs. Table 3 shows
estimates of the first model, which indicate the effects of three
hospital characteristics on the natural logarithm of price of drug
SCODs.
Table 3: Estimated Effects of Selected Factors on Prices Hospitals
Paid for Drug SCODs
Source: GAO analysis.
Notes: SCOD refers to a specified covered outpatient drug. The
results in this table pertain to the top 53 drug SCOD products,
ranked by Medicare spending on SCODs during the first 9 months of
2004.
This table presents estimates from a regression model. The model's
dependent variable is the natural logarithm of the purchase price
paid by a particular hospital for a SCOD. SCOD prices are not
distributed symmetrically around the average SCOD price but are
skewed to the right, reflecting some SCODs with particularly high
prices. Taking the natural logarithm of price takes this skewness
into account. The effect of a measure, such as rural location, is
estimated relative to a reference group (urban location).
Therefore, the reference group is not explicitly included in the
model. A major teaching program refers to a hospital that has an
intern/resident-to-bed ratio of 0.25 or more. Urban refers to a
hospital inside a metropolitan statistical area. Large refers to a
hospital at or above the 80th percentile of hospitals, ranked by
Medicare outpatient charges.
aSignificant at the 5 percent level.
bNot available because the method calculates estimated
coefficients for the included groups relative to the reference
group.
To examine the separate effect of each factor, holding constant
the effects of the remaining factors, we referred to the estimated
coefficients for each factor in the model. From the estimated
coefficient, we calculated the percentage difference in price
attributable to each factor.6 For example, major teaching
hospitals paid lower prices for drugs compared to nonteaching
hospitals: major teaching hospitals paid 3.2 percent less than
nonteaching hospitals, holding constant location, size, and the
mix of SCODs purchased. In contrast, we found no statistically
significant difference in prices paid by hospitals with other
teaching programs and those paid by nonteaching hospitals, holding
the other factors constant.
Although the R-squared statistic in table 3 indicates that the
model accounts for over 99 percent of the variation in the
logarithm of the SCOD price, this feature of the estimated model
requires careful interpretation. Most of the variation in the
logarithm of the drug SCOD price was due to the particular SCODs
that were purchased-for some, hospitals paid on average about $300
per unit while for others, hospitals paid about $3 per unit.
Consequently, after accounting for differences in the mix of SCODs
purchased by different hospitals, only a small amount of variation
in price remains to be explained by other factors. As a result,
the R-squared for this model should not be interpreted as an
indicator of the three factors' success in explaining SCOD price
variation. Instead, the t-statistics associated with teaching
status, location, and size are more useful, since they signal
these factors' statistical significance-that is, whether the
difference between the estimated effect of each factor and zero is
statistically significant.
Table 4 presents the results for the second model, which estimates
the effects of the three factors on the prices of
radiopharmaceutical SCODs.
Table 4: Estimated Effects of Selected Factors on Prices Hospitals
Paid for Radiopharmaceutical SCODs
Source: GAO analysis.
Notes: SCOD refers to a specified covered outpatient drug. The
results in this table pertain to the top nine radiopharmaceutical
SCOD products, ranked by Medicare spending on SCODs during the
first 9 months of 2004. This table presents estimates from a
regression model. The model's dependent variable is the natural
logarithm of the purchase price paid by a particular hospital for
a radiopharmaceutical SCOD. SCOD prices are not distributed
symmetrically around the average SCOD price but are skewed to the
right, reflecting some SCODs with particularly high prices. Taking
the natural logarithm of price takes this skewness into account.
The effect of a measure, such as rural location, is estimated
relative to a reference group (urban location). Therefore, the
reference group is not explicitly included in the model. A major
teaching program refers to a hospital that has an
intern/resident-to-bed ratio of 0.25 or more. Urban refers to a
hospital inside a metropolitan statistical area. Large refers to a
hospital at or above the 80th percentile of hospitals, ranked by
Medicare outpatient charges.
aNot available because the method calculates estimated
coefficients for the included groups relative to the reference
group.
bSignificant at the 5 percent level.
As table 4 shows, two factors-location and size-are statistically
significant in the model examining radiopharmaceutical SCOD
prices. Other things equal, a rural hospital paid prices for
radiopharmaceutical SCODs that were an estimated 4.4 percent
higher than urban hospitals, while large hospitals paid prices an
estimated 3.1 percent lower than small hospitals.
To examine the effect of the three factors combined, while
controlling for differences in the costliness of SCODs that
hospitals purchased, we used the estimates from two models-one for
drug SCODs and one for radiopharmaceutical SCODs-to simulate the
prices that certain groups of hospitals paid. In particular, we
focused on comparing the prices paid by hospitals with one
combination of characteristics-major teaching, urban, and
large-with the prices paid by hospitals with a different
combination of characteristics-nonteaching, rural, and small.
Appendix II: Purchase Prices for Drug SCODs
Table 5 appears as table 1 in our report Medicare: Drug Purchase
Prices for CMS Consideration in Hospital Outpatient Rate-Setting,
GAO-05-581R (Washington, D.C.: June 30, 2005). The label of the
second column-HCPCS code-refers to the Healthcare Common Procedure
Coding System, which CMS uses to define SCODs.
Methods and Results of Price Analysis
3For more information on urban influence codes, see Measuring Rurality:
Urban Influence Codes, http://www.ers.usda.gov/Briefing/Rurality/urbaninf/
(downloaded Feb. 2, 2006). For more information on rural-urban continuum
codes, see Rural-Urban Commuting Area Codes,
http://www.ers.usda.gov/Briefing/Rurality/RuralurbCon/ (downloaded Feb.
14, 2006).
4Each observation of price was drawn from a particular invoice for the
purchase of a particular SCOD purchased by a particular hospital.
5StataCorp, Stata Statistical Software: Release 9 (College Station, Tex.:
StataCorp LP, 2003).
Estimated
Factor Measure of factor coefficient t-value
Teaching status Major teaching program -.0321 -5.33a
Other teaching program -.0054 -1.54
Nonteaching (reference group) n/ab n/ab
Location Rural .0009 0.17
Urban (reference group) n/ab n/ab
Size Large -.0138 -2.18a
Small (reference group) n/ab n/ab
Mix of SCODs SCOD category (one binary
purchased by a variable for each of 53 drug
particular hospital SCODs) (not reported)
Intercept 4.11 1810.16a
R-squared .9974
Number of observations 439,988
6Since each of the three "hospital characteristic" factors (teaching
status, location, and size) is measured as one or more binary variables
and the dependent variable, price, is measured as the natural logarithm,
we used a standard method to calculate the percentage difference in price
attributable to a particular measure of the factor, relative to its
comparison group. Paul Kennedy, A Guide to Econometrics, 4th Ed.
(Cambridge, Mass.: MIT Press, 1998), p. 108.
Estimated
Factor Measure of factor coefficient t-value
Teaching status Major teaching program -.0021 -.12
Other teaching program -.0001 -.01
Nonteaching (reference group) n/aa n/aa
Location Rural .0434 2.23b
Urban (reference group) n/aa n/aa
Size Large -.0311 -2.55b
Small (reference group) n/aa n/aa
Mix of SCODs SCOD category (one binary
purchased by a variable for each of 9
particular hospital radiopharmaceutical SCODs) (not reported)
Intercept 4.74 522.06
R-squared .9913
Number of observations 185,237
Appendix II: Purchase Prices for Drug SCODs Appendix II: Purchase Prices
for Drug SCODs
Table 5: Purchase Prices for SCODs Accounting for 86 Percent of Medicare
Spending on SCODs
Medicare % of
Rank in spending on Medicare
Medicare SCOD, 2004a spending Number of
spending on HCPCS ($ in on SCODs, hospitals
drug SCODs code Description millions) 2004b in sample
1 Q0136 Injection, Epoetin
Alpha (for non-ESRD
use), per 1,000 units 199.8 10.1 973
2 J9310 Rituximab, 100 mg 158.4 8.0 871
3 J2505 Injection,
Pegfilgrastim, 6 mg 144.8 7.3 759
4j Q9941 Injection, Immune
Globulin, Intravenous,
Lyophilized, 1 g k k 626
4j Q9943 Injection, Immune
Globulin, Intravenous,
Non-Lyophilized, 1 g k k 281
5 J1745 Injection, Infliximab,
10 mg 114.8 5.8 897
6 Q0137 Injection, Darbepoetin
alfa, 1 mcg (non-ESRD
use) 100.6 5.1 743
7 J9170 Docetaxel, 20 mg 73.7 3.7 829
8 J9045 Carboplatin, 50 mg 70.7 3.6 893
9 C9205 Injection, Oxaliplatin,
per 5 mg 67.0 3.4 708
10 J3487 Injection, Zoledronic
Acid, 1 mg 66.9 3.4 862
11 J9201 Gemcitabine Hcl, 200 mg 55.0 2.8 855
12 J9206 Irinotecan, 20 mg 39.4 2.0 786
13 J2324 Injection, Nesiritide,
0.25 mg 37.6 1.9 892
14 J9265 Paclitaxel, 30 mg 32.0 1.6 792
15 J9355 Trastuzumab, 10 mg 31.4 1.6 679
16 J9217 Leuprolide Acetate (for
depot suspension), 7.5
mg 30.8 1.6 804
17 J0256 Injection, Alpha 1 -
Proteinase Inhibitor -
Human, 10 mg 20.9 1.1 38
18 J9035m Injection, Bevacizumab,
10 mg 19.8 1.0 436
19 J1441 Injection, Filgrastim
(G-CSF), 480 mcg 17.1 0.9 928
20 J1950 Injection, Leuprolide
Acetate (for depot
suspension), per 3.75
mg 16.9 0.9 541
21 J9001 Doxorubicin
Hydrochloride, all
lipid formulations, 10
mg 16.3 0.8 614
22 J2353 Injection, Octreotide,
depot form for
intramuscular
injection, 1 mg 15.7 0.8 545
23 J9055m Injection, Cetuximab,
10 mg 15.1 0.8 286
24 J9041m Injection, Bortezomib,
0.1 mg 14.1 0.7 452
25 J9350 Topotecan, 4 mg 13.9 0.7 585
26 J1440 Injection, Filgrastim
(G-CSF), 300 mcg 13.0 0.7 956
95% 95%
CMS ASP confidence confidence
payment (average Average interval of Median interval of
Total rate for sales purchase the average purchase the median
number of 2005d price)e pricef purchase priceh purchase
hospitalsc ($) ($) ($) priceg ($) ($) priceg ($)
2,758 11.09 9.25 9.74 9.55-9.94 10.12 10.11-10.13
1,418 437.83 414.92 412.31 407.43-417.20 412.30 412.13-412.52
1,177 2,448.50 2,017.55 i i i i
l 80.68 36.54 36.50 36.37-36.63 37.24 37.15-37.24
l 80.68 53.04 50.63 50.11-51.15 50.96 50.96-52.06
1,903 57.40 50.20 i i i i
1,117 3.66 3.04 3.00 2.95-3.05 3.09 3.06-3.11
1,257 312.69 278.95 295.03 294.10-295.96 294.61 294.46-294.89
1,482 129.96 71.46 132.10 131.65-132.55 132.69 132.55-132.83
1,172 82.53 77.86 75.91 74.90-76.91 77.69 77.65-77.76
1,316 197.87 187.47 185.27 183.71-186.83 190.67 190.26-191.01
1,317 105.73 108.79 105.69 105.13-106.24 106.54 106.44-106.65
1,109 127.33 119.56 116.31 113.87-118.75 122.67 122.16-123.13
1,619 66.23 69.64 i i i i
1,398 79.04 17.70 14.45 14.44-14.46 14.45 14.45-21.34
1,089 50.79 49.99 46.72 45.92-47.53 47.97 47.93-48.04
1,319 543.72 213.83 234.05 223.21-244.90 198.88 195.83-215.41
279 3.72 3.06 2.35 2.33-2.37 2.46 2.27-2.46
916 57.11 53.88 53.31 53.01-53.61 53.72 53.69-53.75
1,679 274.40 261.46 257.21 253.46-260.96 253.64 253.45-253.78
904 451.98 409.18 454.10 453.04-455.17 454.66 454.03-455.72
955 343.78 338.66 336.33 332.22-340.44 338.70 338.28-338.97
852 69.44 80.95 71.13 69.63-72.62 74.04 73.54-74.87
506 49.66 46.85 i i i i
631 28.38 26.77 i i i i
858 697.76 699.75 674.91 656.60-693.21 709.19 706.34-710.50
1,914 162.41 165.23 161.61 156.81-166.42 159.18 159.04-159.31
Medicare % of
Rank in spending on Medicare
Medicare SCOD, 2004a spending Number of
spending on HCPCS ($ in on SCODs, hospitals
drug SCODs code Description millions) 2004b in sample
27 J1785 Injection,
Imiglucerase, per unit 12.9 0.7 41
28 J3396 Injection, Verteporfin,
0.1 mg 12.3 0.6 10
29 J9202 Goserelin Acetate
Implant, per 3.6 mg 11.4 0.6 392
30 J1626 Injection, Granisetron
Hydrochloride, 100 mcg 11.1 0.6 682
31 J0585 Botulinim Toxin Type A,
per unit 10.8 0.5 480
32 J0207 Injection, Amifostine,
500 mg 10.5 0.5 477
33 J2430 Injection, Pamidronate
Disodium, per 30 mg 10.2 0.5 945
34 J9390 Vinorelbine Tartrate,
per 10 mg 9.3 0.5 568
35 J2993 Injection, Reteplase,
18.1 mg 8.9 0.4 505
36 J9293 Injection, Mitoxantrone
Hydrochloride, per 5 mg 8.4 0.4 672
37 J9185 Fludarabine Phosphate,
50 mg 7.6 0.4 669
38 C1305 Apligraf(R), per 44
square centimeters 7.0 0.4 63
39 J9395 Injection, Fulvestrant,
25 mg 6.9 0.3 468
40 J3100 Injection,
Tenecteplase, 50 mg 6.8 0.3 509
41 J9305m Injection, Pemetrexed,
10 mg 5.6 0.3 162
42 J9160 Denileukin Diftitox,
300 mcg 5.6 0.3 73
43 J0180m Injection, Agalsidase
Beta, 1 mg 5.3 0.3 29
44 Q0166 Granisetron
Hydrochloride, 1 mg,
oraln 4.8 0.2 541
45 J2469m Injection, Palonosetron
Hcl, 25 mcg 4.6 0.2 295
46 J9010 Alemtuzumab, 10 mg 4.4 0.2 236
47o Q9942 Injection, Immune
Globulin, Intravenous,
Lyophilized, 10 mg p p 626
47o Q9944 Injection, Immune
Globulin, Intravenous,
Non-Lyophilized, 10 mg p p 281
48 J7190 Factor VIII
(Antihemophilic Factor,
Human) per I.U. 4.2 0.2 55
49 J0130 Injection, Abciximab,
10 mg 4.0 0.2 570
50 J0850 Injection,
Cytomegalovirus Immune
Globulin Intravenous
(Human), per vial 3.8 0.2 156
51 J1327 Injection,
Eptifibatide, 5 mg 3.7 0.2 911
52 J9214 Interferon, Alfa-2B,
Recombinant, 1 million
units 3.6 0.2 619
53 C9201 Dermagraft(R), per 37.5
square centimeters 3.4 0.2 2
95% 95%
CMS ASP confidence confidence
payment (average Average interval of Median interval of
Total rate for sales purchase the average purchase the median
number of 2005d price)e pricef purchase priceh purchase
hospitalsc ($) ($) ($) priceg ($) ($) priceg ($)
59 3.91 3.69 3.62 3.60-3.64 3.62 3.61-3.66
45 8.49 8.48 i i i i
529 390.09 181.78 201.76 193.30-210.23 206.56 175.73-323.33
988 16.20 6.71 6.45 6.27-6.62 6.61 6.60-6.64
1,062 4.32 4.44 i i i i
705 395.75 403.84 i i i i
1,567 128.74 54.10 58.49 51.51-65.47 72.59 71.50-72.72
833 52.78 58.20 48.15 48.13-48.16 48.14 48.13-52.05
1,073 1,192.09 832.49 846.53 844.18-848.87 845.36 844.48-846.87
1,181 313.96 305.36 297.00 296.19-297.82 295.62 295.46-295.78
891 311.09 243.05 293.99 291.43-296.56 298.44 298.37-298.68
450 1,130.88 1,114.74 i i i i
778 79.65 76.78 74.63 74.45-74.80 75.03 74.95-75.18
1,181 2,350.98 1,901.29 i i i i
251 40.54 38.25 i i i i
95 1,438.80 1,144.18 i i i i
49 121.11 114.26 111.33 111.08-111.58 109.71 108.18-111.09
886 39.04 31.04 24.86 24.82-24.89 23.99 21.58-24.94
525 18.09 17.06 i i i i
356 541.46 478.73 i i i i
q 0.75 0.37 0.37 0.36-0.37 0.37 0.37-0.37
q 0.75 0.53 0.51 0.50-0.51 0.51 0.51-0.52
122 0.76 0.60 0.46 0.46-0.46 0.46 r
797 448.22 417.35 i i i i
260 622.13 632.67 i i i i
1,661 11.21 11.79 12.49 12.35-12.63 11.03 10.75-12.39
954 13.00 12.25 11.20 11.02-11.37 11.93 11.78-11.98
80 529.54 545.10 i i i i
Sources: GAO survey and CMS.
Notes: ESRD = end-stage renal disease, g = gram, I.U. = international
unit, mcg = microgram, and mg = milligram.
aMedicare spending is for the period January 1, 2004, through September
30, 2004.
bThe percentage of Medicare spending is based on Medicare spending for all
SCODs-both drugs and radiopharmaceuticals.
cThis estimate of the total number of hospitals in the population is based
on our sample.
dThis is the payment rate specified for each HCPCS for 2005. It
incorporates CMS's April 2005 update.
eCMS publishes the ASP plus 6 percent for certain drugs used in
physicians' offices. These amounts are based on data provided by
manufacturers each quarter. We are reporting ASPs for the quarter
beginning in April 2005. ASPs reported here do not include the 6 percent
added by CMS.
fThis price is based on data provided by the hospitals in our survey and
does not reflect any other costs associated with purchasing or
administering the product. We asked hospitals to report prices for drugs
purchased from July 1, 2003, through June 30, 2004. We weighted the prices
by the volume purchased as well as by the sample weights. We have excluded
prices under the 340B program, a federal program that provides drug price
discounts for certain health care entities, including those that provide
health care services for low-income individuals and individuals in
medically underserved areas. (42 U.S.C. S: 256b (2000)).
gThe confidence interval measures the precision of the estimate. The
narrower the interval, the greater the precision.
hThe median purchase price is the midpoint of all prices reported by
hospitals in our sample. Half of the prices reported by hospitals are
above the median and half are below. The median is weighted by volume
purchased and by hospital sample weights. The average purchase price
excludes prices paid under the 340B program.
iFor HCPCS codes that contain only one National Drug Code (NDC), we do not
include information on the average or median purchase price because of the
potential proprietary sensitivity of such information.
jOn April 1, 2005, CMS replaced J1563, Injection, Immune Globulin,
Intravenous, 1g, with two new codes: Q9941 and Q9943. J1563 was ranked
fourth in total Medicare spending on SCODs from January 1, 2004, to
September 30, 2004.
kJ1563, Injection, Immune Globulin, Intravenous, 1g, accounted for $127.1
million in Medicare spending from January 1, 2004, through September 30,
2004, which was 6.4 percent of total Medicare spending on SCODs for that
time period.
lOn April 1, 2005, CMS replaced J1563, Injection, Immune Globulin,
Intravenous, 1g, with two new codes: Q9941 and Q9943. Because J1563 was
replaced by two codes, we could not estimate the total number of hospitals
in the population for these new codes individually.
mOn January 1, 2005, CMS replaced C9214, C9215, C9207, C9213, C9208, and
C9210 with J9035, J9055, J9041, J9305, J0180, and J2469, respectively. The
ranks for the new codes correspond to the ranks in total Medicare spending
on SCODs from January 1, 2004, to September 30, 2004, for the former
codes.
nThe complete description for HCPCS Q0166 is "Granisetron Hydrochloride, 1
mg, Oral, Food and Drug Administration (FDA) Approved Prescription
Anti-Emetic, for Use as a Complete Therapeutic Substitute for an IV
(intravenous) Anti-Emetic at the Time of Chemotherapy Treatment, Not to
Exceed a 24 Hour Dosage Regimen."
oOn April 1, 2005, CMS replaced J1564, Injection, Immune Globulin,
Intravenous, 10 mg, with two new codes: Q9942 and Q9944. J1564 was ranked
47th in total Medicare spending on SCODs from January 1, 2004, to
September 30, 2004.
pJ1564, Injection, Immune Globulin, Intravenous, 10 mg accounted for $4.4
million in Medicare spending from January 1, 2004, through September 30,
2004, which was 0.2 percent of total Medicare spending on SCODs for that
time period.
qOn April 1, 2005, CMS replaced J1564, Injection, Immune Globulin,
Intravenous, 10 mg, with two new codes: Q9942 and Q9944. Because J1564 was
replaced by two codes, we could not estimate the total number of hospitals
in the population for these new codes individually.
rFor this SCOD, our sample data cannot be extrapolated to compute a
confidence interval for the median.
Appendix III: Purchase Prices for Radiopharmaceuticals SCODs
Table 6 appears as table 1 in our report Medicare: Radiopharmaceutical
Purchase Prices for CMS Consideration in Hospital Outpatient Rate-Setting,
GAO-05-733R (Washington, D.C.: July 14, 2005). The label of the second
column-HCPCS code-refers to the Healthcare Common Procedure Coding System,
which CMS uses to define SCODs.
Table 6: Purchase Prices for Radiopharmaceutical Accounting for 9 Percent
of Medicare Spending on SCODs
Medicare
spending
on SCOD, % of
Rank in Medicare 2004a Medicare
spending on radio- spending Number of
pharmaceutical HCPCS ($ in on SCODs, hospitals
SCODs code Description millions) 2004a in sample
1 A9500 Technetium Tc 99m
Sestamibi, per dose 66.5 3.4 405
2 A9502 Technetium Tc 99m
Tetrofosmin, per
dose 38.8 2.0 174
3 C1775 Fluorodeoxyglucose
(FDG) F18, per dose
(4-40 mCi/ml) 32.1 1.6 71
4 C1083 Yttrium 90
Ibritumomab
Tiuxetan, per dose 7.1 0.4 80
5 A9505 Thallous Chloride
TL 201, per mCi 6.7 0.3 292
6 Q3005 Technetium Tc 99m
Mertiatide, per
mCig 6.2 0.3 292
7 A9507 Indium In 111
Capromab Pendetide,
per dose 4.8 0.2 56
8 Q3008 Indium In 111
Pentetreotide, per
3 mCih 4.5 0.2 193
9 A9521 Technetium Tc 99m
Exametazime, per
dose 3.8 0.2 180
95%
95% confidence confidence
interval of interval of
Total number CMS payment Average the average Median the median
of rate for purchase purchase purchase purchase
hospitalsb 2005c ($) priced ($) pricee ($) pricef ($) pricee ($)
75.58 -
2,477 106.32 75.15 73.24 - 77.06 76.47 77.85
66.23 -
964 104.58 70.70 67.92 - 73.48 67.59 70.98
263.24 - 261.83 -
687 221.11 287.90 312.55 272.80 308.52
19,498.98 - 19,459.55 -
130 20,948.25 19,614.96 19,730.95 19,516.70 19,565.02
15.06 -
1,199 18.29 17.18 16.32 - 18.05 15.49 17.06
27.56 -
1,655 31.13 27.40 26.47 - 28.34 27.58 27.60
1,760.80 - 1,703.46 -
262 1,915.23 1,801.12 1,841.43 1,841.23 1,860.22
1,198.35 - 1,395.49 -
666 1,079.00 1,279.55 1,360.76 1,423.87 1,437.61
358.29 - 379.90 -
773 778.13 455.59 552.89 456.30 523.95
Sources: GAO survey and CMS.
Notes: mCi = millicurie, ml = milliliter
aMedicare spending is for the period January 1, 2004, through September
30, 2004. The percentage of Medicare spending is based on all SCODs-both
drugs and radiopharmaceuticals.
bThis estimate of the total number of hospitals in the population is based
on our sample.
cThis is the payment rate specified for each HCPCS for 2005. It
incorporates CMS's April 2005 update.
dThis price is based on data provided by the hospitals in our survey and
does not reflect delivery fees or any other ancillary costs associated
with purchasing or administering this product. We asked hospitals to
report prices for drugs purchased from July 1, 2003, through June 30,
2004. We weighted the prices by the volume purchased as well as by the
sample weights.
eThe confidence interval measures the precision of the estimate. The
narrower the interval, the greater the precision.
fThe median purchase price is the midpoint of all prices reported by
hospitals in our sample. This price does not reflect delivery fees or any
other ancillary costs associated with purchasing or administering this
product. Half of the prices reported by hospitals are above the median and
half are below. The median is weighted by volume purchased and by hospital
sample weights.
gThe billing unit of measure for Q3005, Technetium Tc 99m Mertiatide, is
per mCi. The per mCi purchase price reported is based on purchase prices
for two commonly reported dose sizes, 5 mCi and 10 mCi. Since in our data
the 5 mCi dose is more common than the 10 mCi dose and the purchase price
of a 5 mCi dose and of a 10 mCi dose were similar, we treated a 10 mCi
dose as if it were a 5 mCi dose.
hThe billing unit of measure for Q3008, Indium In 111 Pentetreotide, is
per 3 mCi. The per mCi purchase price reported is based on purchase prices
for two commonly reported dose sizes, 3 mCi and 6 mCi. Since a 3 mCi dose
is the billing unit specified by CMS for Q3008 and since in our data the
purchase price of a 3 mCi dose and of a 6 mCi dose varied relatively
little, we treated a 6 mCi dose as if it were a 3 mCi dose.
Appendix IV: Comments from the Department of Health and Human Services
Appendix V: GAO Contact and Staff Acknowledgments
GAO Contact
A. Bruce Steinwald, (202) 512-7119 or [email protected]
Acknowledgments
Phyllis Thorburn, Assistant Director; Hannah Fein; Dae Park; Jonathan
Ratner; and Thomas Walke made key contributions to this report.
(290469)
GAO's Mission
The Government Accountability Office, the audit, evaluation and
investigative arm of Congress, exists to support Congress in meeting its
constitutional responsibilities and to help improve the performance and
accountability of the federal government for the American people. GAO
examines the use of public funds; evaluates federal programs and policies;
and provides analyses, recommendations, and other assistance to help
Congress make informed oversight, policy, and funding decisions. GAO's
commitment to good government is reflected in its core values of
accountability, integrity, and reliability.
Obtaining Copies of GAO Reports and Testimony
The fastest and easiest way to obtain copies of GAO documents at no cost
is through GAO's Web site ( www.gao.gov ). Each weekday, GAO posts newly
released reports, testimony, and correspondence on its Web site. To have
GAO e-mail you a list of newly posted products every afternoon, go to
www.gao.gov and select "Subscribe to Updates."
Order by Mail or Phone
The first copy of each printed report is free. Additional copies are $2
each. A check or money order should be made out to the Superintendent of
Documents. GAO also accepts VISA and Mastercard. Orders for 100 or more
copies mailed to a single address are discounted 25 percent. Orders should
be sent to:
U.S. Government Accountability Office 441 G Street NW, Room LM Washington,
D.C. 20548
To order by Phone: Voice: (202) 512-6000 TDD: (202) 512-2537 Fax: (202)
512-6061
To Report Fraud, Waste, and Abuse in Federal Programs
Contact:
Web site: www.gao.gov/fraudnet/fraudnet.htm E-mail: [email protected]
Automated answering system: (800) 424-5454 or (202) 512-7470
Congressional Relations
Gloria Jarmon, Managing Director, [email protected] (202) 512-4400 U.S.
Government Accountability Office, 441 G Street NW, Room 7125 Washington,
D.C. 20548
Public Affairs
Paul Anderson, Managing Director, [email protected] (202) 512-4800 U.S.
Government Accountability Office, 441 G Street NW, Room 7149 Washington,
D.C. 20548
www.gao.gov/cgi-bin/getrpt? GAO-06-372 .
To view the full product, including the scope
and methodology, click on the link above.
For more information, contact A. Bruce Steinwald at (202) 512-7119 or
[email protected].
Highlights of GAO-06-372 , a report to congressional committees
April 2006
MEDICARE HOSPITAL PHARMACEUTICALS
Survey Shows Price Variation and Highlights Data Collection Lessons and
Outpatient Rate-Setting Challenges for CMS
In 2003, the Medicare Modernization Act required the Centers for Medicare
& Medicaid Services (CMS) to establish payment rates for a set of new
pharmaceutical products-drugs and radiopharmaceuticals-provided to
beneficiaries in a hospital outpatient setting. These products were
classified for payment purposes as specified covered outpatient drugs
(SCOD). The legislation directed CMS to set 2006 Medicare payment rates
for SCODs equal to hospitals' average acquisition costs and included
requirements for GAO. As directed, GAO surveyed hospitals and issued two
reports, providing information to use in setting 2006 SCOD rates. To
address other requirements in the law, this report analyzes SCOD price
variation across hospitals, advises CMS on future surveys it might
undertake, and examines both lessons from the GAO survey and future
challenges facing CMS.
What GAO Recommends
GAO recommends that the Secretary of Health and Human Services seek to
ensure that CMS's SCOD payment rates are based on sufficiently reliable
data by (1) validating data collected on drug prices and (2) basing
payment rates for each radiopharmaceutical SCOD on the price of a
ready-to-use unit dose. Although expressing some reservations,
particularly concerning the burden of data collection, HHS agreed to
consider GAO's recommendations.
Analyzing pharmaceutical price data collected from its 2004 survey of
hospitals, GAO found that prices hospitals paid for SCOD products varied
across hospitals. Certain factors-namely, whether the hospital had a major
teaching program or not, was in an urban or rural area, and had a large or
small hospital outpatient department-were associated with whether
hospitals paid higher or lower prices for SCOD products. Major teaching
hospitals paid prices that were an estimated 3.2 percent lower than those
paid by nonteaching hospitals for drug SCODs; rural hospitals paid prices
an estimated 4.4 percent higher than those paid by urban hospitals for
radiopharmaceutical SCODs; and large hospitals paid prices an estimated
1.4 percent lower than those paid by small hospitals for drug SCODs and
3.1 percent lower for radiopharmaceutical SCODs. Combining these factors,
GAO found that large, urban, major teaching hospitals-compared with other
hospitals-generally paid lower prices, on average, for all SCOD products.
From conducting its hospital survey, GAO learned a key lesson that CMS
could use in the future: such a survey would not be practical for
collecting the data needed to set and update SCOD rates routinely but
would be useful for validating, on occasion, CMS's rate-setting data.
GAO's survey produced accurate hospital drug price data, but it also
created a considerable burden for hospitals as the data suppliers and
considerable costs for GAO as the data collector. Nonetheless, the benefit
of collecting actual prices paid by hospitals could make such surveys
advantageous for occasionally validating CMS's proxy for SCODs' average
acquisition costs--the average sales price (ASP) data that manufacturers
report.
CMS will face important challenges as it seeks to obtain accurate data on
hospitals' acquisition costs for drug and radiopharmaceutical SCODs.
o Regarding drugs, CMS lacks the detail on manufacturers' ASP
data needed to determine if rates developed from these data are
appropriate for hospitals. Manufacturers report ASP as a single
price paid by all purchasers, making it impossible to distinguish
the price paid by hospitals alone. CMS instructs manufacturers to
report ASP net of rebates but does not specify how to allocate
individual product rebates when several products are purchased.
o Regarding radiopharmaceuticals, GAO found that the diversity of
forms in which they can be purchased-ready-to-use unit doses,
multidoses, or separately purchased radioactive and
non-radioactive substances-complicates CMS's efforts to select a
data source that can provide reasonably accurate price data
efficiently. Efficiency as well as accuracy is a factor in
selecting a data source because radiopharmaceuticals account for
only 1.5 percent of Medicare hospital outpatient spending. GAO's
experience suggests that the best option available to CMS, in
terms of accuracy and efficiency, is to collect price data on
radiopharmaceuticals purchased in ready-to-use unit doses, the
form in which an estimated three-quarters of hospitals purchase
these products.
*** End of document. ***