Federal Employees Health Benefits Program: Competition and Other 
Factors Linked to Wide Variation in Health Care Prices		 
(15-AUG-05, GAO-05-856).					 
                                                                 
Congress is concerned about the health care spending burden	 
facing the Federal Employees Health Benefits Program (FEHBP), the
largest private health insurance program in the country. Health  
care spending per person varies geographically, and the 	 
underlying causes for the spending variation have not been fully 
explored. Understanding market forces and other factors that may 
influence health care spending may contribute to efforts to	 
moderate health care spending. Health care spending varies across
the country due to differences in its components, the utilization
and price of health care services. A wide body of research	 
describes extensive geographic variation in utilization. However,
less is known about private sector geographic variation in	 
prices. This report examined prices and spending in FEHBP	 
Preferred Provider Organizations (PPOs) to determine (1) the	 
extent to which hospital and physician prices varied		 
geographically, (2) which factors were associated with geographic
variation in hospital and physician prices, and (3) the extent to
which hospital and physician price variation contributed to	 
geographic variation in spending. We analyzed claims data from	 
several large national PPOs participating in FEHBP. We used 2001 
data, the most current data available at the time of the study.  
-------------------------Indexing Terms------------------------- 
REPORTNUM:   GAO-05-856 					        
    ACCNO:   A32980						        
  TITLE:     Federal Employees Health Benefits Program: Competition   
and Other Factors Linked to Wide Variation in Health Care Prices 
     DATE:   08/15/2005 
  SUBJECT:   Comparative analysis				 
	     Cost analysis					 
	     Health insurance					 
	     Health insurance cost control			 
	     Health maintenance organizations			 
	     Hospitals						 
	     Medicaid						 
	     Medical fees					 
	     Medical services rates				 
	     Medicare						 
	     Physicians 					 
	     Price regulation					 
	     Prices and pricing 				 
	     Competition					 
	     Economic analysis					 
	     Federal Employees Health Benefits			 
	     Program						 
                                                                 

******************************************************************
** 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-05-856

     

     * Results in Brief
     * Background
          * FEHBP and Participating PPOs
          * Geographic Variation in Spending, Utilization, and Prices
          * Health Care Market Characteristics and Price
     * Large Differences in Hospital and Physician Prices across Me
          * Hospital Prices Varied More than Physician Prices
          * Hospital and Physician Prices Were Generally Higher in the M
     * Less Competition and Less HMO Capitation Linked to Higher He
          * Prices Were Higher in Metropolitan Areas with Less Competiti
          * Prices Were Higher in Metropolitan Areas with Less HMO Capit
          * No Evidence of Cost Shifting Due to Medicaid, Medicare, or t
     * Total Spending Varied 112 Percent; Price Variation Contribut
          * Spending per Enrollee Varied by 112 Percent across Metropoli
          * Price Contributed to One-third of the Variation in Spending,
     * Concluding Observations
     * Agency and Other Comments
     * Appendix I: Scope and Methodology
          * FEHBP Data and Study Eligibility Criteria
          * Hospital and Physician Price Estimates
          * Factors Affecting Health Care Prices
          * Analytical Approach
               * Price Regression Analysis-Methods and Results
          * Spending Analysis
          * Decomposing Spending Variation into Price and Utilization Ef
          * Data Reliability
     * Appendix II: FEHBP PPO Adjusted Hospital Prices in U.S. Metr
     * Appendix III: FEHBP PPO Adjusted Physician Prices in U.S. Me
     * Appendix IV: FEHBP PPO Adjusted Health Care Spending Per Enr
     * Appendix V: Comments from the Office of Personnel Management
     * Appendix VI: GAO Contacts and Staff Acknowledgments
          * GAO Contacts
          * Acknowledgments
               * Order by Mail or Phone

                 United States Government Accountability Office

GAO

                            House of Representatives

August 2005

FEDERAL EMPLOYEES HEALTH BENEFITS PROGRAM

  Competition and Other Factors Linked to Wide Variation in Health Care Prices

GAO-05-856

FEDERAL EMPLOYEES HEALTH BENEFITS PROGRAM

Competition and Other Factors Linked to Wide Variation in Health Care
Prices

  What GAO Found

FEHBP PPOs paid substantially different prices for hospital inpatient and
physician services across metropolitan areas in the United States.
Hospital prices varied by 259 percent and physician prices varied by about
100 percent across metropolitan areas. While there were some areas with
very high or low prices, most had prices that were closer to the average.

The variation in prices appeared to be affected by market characteristics.
Metropolitan areas with the least competition, areas with a higher
percentage of hospital beds in the two largest hospitals or hospital
networks, had hospital prices that were 18 percent higher and physician
prices that were 11 percent higher than areas with the most competition.
The percent of primary care physicians' reimbursement that was paid on a
capitation basis in health maintenance organizations (HMO), a proxy for
HMO price bargaining leverage, was also associated with geographic
variation in prices. Metropolitan areas with the least HMO capitation
tended to have hospital and physician prices that were about 10 percent
higher than areas with the most HMO capitation. When GAO controlled for
other factors that might be associated with geographic variation in
prices, more hospital competition and HMO capitation were still associated
with lower prices, but the effect was reduced. GAO did not find any
evidence that price variation was due to cost shifting, where providers
raise private sector prices to compensate for lower prices from other
payers.

Total health care spending per enrollee varied by over 100 percent across
metropolitan areas. For hospital and physician services, price contributed
to about one-third and utilization to about two-thirds of the variation in
spending between metropolitan areas in the highest and lowest spending
quartiles. Higher physician prices were also associated with lower
physician utilization, but higher prices were still typical in higher
spending areas.

The Office of Personnel Management provided comments on a draft of this
report and agreed with our findings.

           Distribution of Hospital and Physician Price Indices, 2001

Source: GAO analysis of FEHBP data.

Note: GAO converted prices to an index by dividing the average price in a
metropolitan area by the average price in all study metropolitan areas.

                 United States Government Accountability Office

Contents

Letter 1
Results in Brief 4
Background 5
Large Differences in Hospital and Physician Prices across Metropolitan
Areas 10
Less Competition and Less HMO Capitation Linked to Higher Health Care
Prices 18
Total Spending Varied 112 Percent; Price Variation Contributed to
One-third of the Variation in Hospital and Physician Spending 24
Concluding Observations 30
Agency and Other Comments 30
Appendix I Scope and Methodology 32
FEHBP Data and Study Eligibility Criteria 32
Hospital and Physician Price Estimates 34
Factors Affecting Health Care Prices 35
Analytical Approach 39
Spending Analysis 45
Decomposing Spending Variation into Price and Utilization Effects 46
Data Reliability 47
Appendix II FEHBP PPO Adjusted Hospital Prices in U.S. Metropolitan Areas,
2001
Appendix III FEHBP PPO Adjusted Physician Prices in U.S. Metropolitan
Areas, 2001
Appendix IV FEHBP PPO Adjusted Health Care Spending Per Enrollee in U.S.
Metropolitan Areas, 2001
Appendix V Comments from the Office of Personnel Management 72

Appendix VI           GAO Contacts and Staff Acknowledgments            74 
Tables                                                                  
                 Table 1: FEHBP PPO Hospital Price Indices in Metropolitan 
                                                                     Areas 
                             Grouped by Census Region, 2001                14 
                   Table 2: Metropolitan Areas with the Highest and Lowest 
                                                                  Hospital 
                            Price Indices in FEHBP PPOs, 2001              14 
                Table 3: FEHBP PPO Physician Price Indices in Metropolitan 
                                                                     Areas 
                             Grouped by Census Region, 2001                17 
                   Table 4: Metropolitan Areas with the Highest and Lowest 
                                                                 Physician 
                            Price Indices in FEHBP PPOs, 2001              17 
                 Table 5: FEHBP PPO Price Indices in the Least and Most    
                          Competitive Metropolitan Areas, 2001             19 
               Table 6: FEHBP PPO Price Indices in Metropolitan Areas with 
                                                                       the 
                           Least and Most HMO Capitation, 2001             21 
               Table 7: FEHBP PPO Price Indices in Metropolitan Areas in   
               the                                                         
                       Lowest and Highest Medicaid Payment Quartiles, 2001 23 
                   Table 8: FEHBP PPO Spending Per Enrollee Indices in     
                        Metropolitan Areas by Census Region, 2001          27 
                    Table 9: Price and Utilization Indices in Metropolitan 
                                                              Areas in the 
                    Highest and Lowest Quartiles of Hospital and Physician 
               Spending, 2001                                              28 
                   Table 10: Example of the Offsetting Effect of Physician 
                                                                 Price and 
                     Utilization on Physician Spending in Two Metropolitan 
                                Areas in the FEHBP, 2001                   29 
               Table 11: Factors Included in Analysis of Hospital and      
               Physician                                                   
               Price, 2001                                                 36 
                 Table 12: Results for Hospital Price Regression-Estimated 
                                                                   Effects 
                    of Selected Factors on Hospital Prices in Metropolitan 
               Areas, 2001                                                 41 
               Table 13: Results for Physician Price Regression-Estimated  
                   Effects of Selected Factors on Physician Prices in      
                                Metropolitan Areas, 2001                   42 
               Table 14: Effects of Changes in Explanatory Variables on    44 
               Prices                                                      
               Table 15: Ranking of Metropolitan Areas by Adjusted         
               Hospital                                                    
               Prices, 2001                                                49 
               Table 16: Ranking of Metropolitan Areas by Adjusted         
               Physician                                                   
               Prices, 2001                                                56 
                Table 17: Ranking of Metropolitan Areas by Adjusted Health 
                                                                      Care 
                               Spending Per Enrollee, 2001                 65 

Figures 
              Figure 1: Distribution of Hospital Price Indices across 232  
                              Metropolitan Areas, 2001                     11 
            Figure 2: Distribution of Physician Price Indices across 319   
                              Metropolitan Areas, 2001                     12 
            Figure 3: FEHBP PPO Adjusted Hospital Price Index Quartiles in 
                            232 Metropolitan Areas, 2001                   13 
           Figure 4: FEHBP PPO Adjusted Physician Price Index Quartiles in 
                            319 Metropolitan Areas, 2001                   16 
              Figure 5: Distribution of FEHBP PPO Spending Per Enrollee    
                     Indices across 232 Metropolitan Areas, 2001           25 
           Figure 6: FEHBP Adjusted Spending Per Enrollee Quartiles in 232 
                              Metropolitan Areas, 2001                     26 

Abbreviations

APR-DRG            All Patient Refined/Diagnosis Related Group 
FEHBP                Federal Employees Health Benefits Program 
GPCI            Geographic Practice Cost Index                 
HMO             Health Maintenance Organization                
OPM             Office of Personnel Management                 
PPO             Preferred Provider Organization                

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

August 15, 2005

The Honorable Paul Ryan House of Representatives

Dear Mr. Ryan:

Congress is concerned about the health care spending burden facing the
Federal Employees Health Benefits Program (FEHBP), the largest private
health insurance program in the country. Previous research has shown that
health care spending varies geographically, but has not fully explored the
underlying causes. A better understanding of market and other forces that
may influence health care spending could assist efforts to moderate health
care spending.

Geographic differences in health care spending are due to differences in
utilization-the amount and type of health services used-and price-the
amount paid to physicians, hospitals, and other providers. Most of the
geographic variations research has focused on the utilization of services.
However, less is known about the variation in prices, factors that affect
price variation, or how price variation contributes to spending variation.

You asked us to analyze geographic variation in prices and spending in
FEHBP. In August 2004, we provided you with an interim report about how
hospital and physician prices and spending in FEHBP Preferred Provider
Organizations (PPO) 1 in Milwaukee compared to other metropolitan areas. 2
In this report, we have expanded that analysis to include geographic
variation in prices and spending in metropolitan areas 3 throughout the
United States. This final report examines prices and spending in FEHBP
PPOs to determine: (1) the extent to which hospital

1

PPOs in our study refer to fee-for-service plans with preferred provider
networks. PPOs generally allow enrollees to obtain care from any provider,
but charge enrollees less if they obtain care from the plans' networks of
preferred providers.

2

GAO, Milwaukee Health Care Spending Compared to OtherMetropolitan Areas:
Geographic Variation in Spending for Enrollees in the Federal
EmployeesHealth Benefits P rogram,GAO-04-1000R (Washington, D.C.: Aug. 18,
2004).

3

A metropolitan area refers to a metropolitan statistical area, which the
Office of Management and Budget defines as a core population of at least
50,000 people with adjacent communities linked socially and economically
with that core.

and physician prices varied geographically, (2) which factors were
associated with geographic variation in hospital and physician prices, and

(3) the extent to which hospital and physician price variation contributed
to geographic variation in spending.

To estimate the extent to which hospital and physician prices varied
geographically, we analyzed health claims data from several large national
insurers participating in FEHBP in 2001, all of which were PPOs. 4 These
2001 data were the most recent that were available at the time we began
our study. We grouped all claims by the metropolitan area where care was
delivered. For hospital and physician prices, we removed the effect of
geographic differences in the costs of doing business (such as wages and
rents) and the mix of services provided, using the same methodology
Medicare uses to geographically adjust payments for hospital stays and
physician services with some modifications. 5 We then computed an average
adjusted price for hospital stays and an average adjusted price for
physician services for each metropolitan area in our study. 6 Finally, we
created hospital and physician price indices that showed how prices in
each metropolitan area compared to the average of all the metropolitan
areas in our study. The average value for each index was set at 1.00.

To determine which factors might be associated with geographic differences
in price, we examined the relationship between price and indicators of
market competition, health maintenance organization (HMO) price bargaining
leverage, and cost-shifting pressures for each metropolitan area. 7 To
measure competition among hospitals for each metropolitan area, we
estimated the percentage of beds in the two largest hospitals or hospital
networks as a percent of all acute care hospital beds in the metropolitan
area. 8 The larger the share of the hospital service

4

Price throughout this report includes both the amount the PPO pays
directly and the amount the enrollee is obligated to pay through
deductibles and coinsurance.

5

See app. I for a description of how we adjusted prices.

6

We had a sufficient volume of hospital stays to analyze hospital prices in
232 metropolitan areas, and we had a sufficient volume of physician
services to analyze physician prices in 319 metropolitan areas.

7

See app. I for a description of all of our data measures and sources.

8

Hospital networks were defined by the vendor supplying the data, Verispan,
L.L.C., as an affiliation between three or more health care organizations,
at least one of which is a hospital, with a unified marketing strategy.
Where one or both of the two largest hospitals was not affiliated with a
network, the percentage of beds in the hospital was used instead of the
percentage of beds in a network.

market controlled by a few providers, the greater the likelihood that
insurers will have to contract with those providers to ensure enrollee
access to care. We used hospital competition as a proxy for physician
competition because many physicians are affiliated with hospitals and
hospital networks. We also measured the percent of primary care physician
compensation from HMOs that was capitated. 9 Because physicians generally
prefer fee-for-service to capitation payments, the use of capitation by
HMOs demonstrates that they have the leverage to negotiate capitation
contracts with physicians. Therefore, we used HMO capitation as a proxy
measure for the strength of HMO presence in a community, and HMOs' ability
to negotiate prices with physicians, hospitals, and other providers. We
also developed indicators of cost shifting-hospitals and physicians
charging higher prices to privately insured patients to compensate for
lower payments from other patients. For each metropolitan area, we
estimated the proportions of the population who were without insurance or
who were enrolled in Medicare or Medicaid. 10 We also estimated average
physician Medicaid payment rates in each metropolitan area based on
Medicaid rates for 29 common procedures. 11 We examined the relationships
of these variables to our hospital and physician price variables.

To examine how prices affected spending, we computed the average spending
for all covered health care services per enrollee for each metropolitan
area, excluding pharmaceuticals, mental health services, and chemical
dependency services. 12 We adjusted total spending per enrollee, hospital
spending, and physician spending, for differences in the costs of doing
business and for differences in the age and sex of the enrollees in each
metropolitan area. We calculated the relative contribution of prices

9

Capitation is a payment method used by managed care organizations where
physicians are paid a fixed, predetermined payment for caring for an
enrollee for a specified period of time, regardless of the number or type
of services ultimately provided.

10

The number of individuals without health insurance in each metropolitan
area was obtained from InterStudy Publications, Inc., and was based on
statewide data; it does not include differences in the uninsured among
metropolitan areas in the same state.

11

J. Menges, et al., for The Lewin Group, Comparing Physician and Dentist
Fees Among Medicaid Programs(Oakland, Calif.: Medi-Cal Policy Institute,
2001).

12

Total spending per enrollee includes both enrollee deductible and
coinsurance obligations and PPO expenditures on behalf of the enrollee.

                                Results in Brief

and utilization to spending for hospital stays and physician services. 13
See appendix I for a more detailed description of our methodology.

We tested the data we obtained from FEHBP and other sources for
consistency and reliability, and determined that they were adequate for
our purposes. Our analysis is limited to geographic variation in 2001
spending and prices in the FEHBP PPOs in our study and to the factors
listed in appendix I. We performed our work from September 2002 through
July 2005 in accordance with generally accepted government auditing
standards.

We found that FEHBP PPO hospital prices differed by 259 percent and
physician prices differed by about 100 percent across metropolitan areas
in the United States, after we removed the geographic variation associated
with the costs of doing business such as rents and salaries, and
differences in the types of services provided. While some metropolitan
areas had hospital or physician prices that were very low or very high,
most had prices that were much closer to the average. Hospital and
physician prices tended to vary together, such that areas with higher
hospital prices tended to also have higher physician prices. Prices for
hospital stays and physician services tended to be higher in metropolitan
areas in the Midwest and lower in the Northeast.

In general, less competition and less HMO capitation were associated with
higher prices. Metropolitan areas where there was less competition-areas
with a higher percentage of beds in the two largest hospitals or hospital
networks-had higher prices, on average. Metropolitan areas with the least
competition had, on average, 18 percent higher hospital prices and 11
percent higher physician prices than areas with the most competition. 14
Metropolitan areas with the least HMO capitation had hospital and
physician prices that were both close to 10 percent higher, on average,
than areas with the most HMO capitation. 15 When we controlled for other

13

Our analysis of hospital spending and utilization may have been limited by
the small number of enrollees and admissions in some areas. Ten of the 232
metropolitan areas in this analysis had between 500 and 1,000 enrollees.

14

We defined areas in the lowest 25 percent of competition as having the
least competition, and areas in the highest 25 percent of competition as
having the most competition.

15

We defined areas in the lowest 25 percent of HMO capitation as the having
the least HMO capitation, and areas in the highest 25 percent of HMO
capitation as having the most HMO capitation.

factors that might be associated with geographic variation in prices, we
found that less hospital competition and HMO capitation were still
associated with higher prices, but the effect was reduced. We found no
evidence of cost shifting-hospital and physician prices were no higher, on
average, in areas with lower Medicaid payments, a higher proportion of the
uninsured, or a higher percent of the population enrolled in Medicaid or
Medicare. Rather, we found that physician prices were, on average, lower
in areas with lower Medicaid payments and a higher percentage of
uninsured. We did not find a relationship between hospital prices and
Medicaid payments or between hospital prices and the percentage uninsured.

Total adjusted health care spending per enrollee was more than twice as
high in the highest-spending metropolitan area as it was in the
lowest-spending metropolitan area. 16 Spending in metropolitan areas in
the South was about 23 percent higher, on average, than in metropolitan
areas in the Northeast. For hospital and physician services, prices
contributed to about one-third of the variation in spending between the
areas with the highest spending and the areas with the lowest spending,
such that higher prices tended to be associated with higher hospital and
physician spending. 17 The contribution of physician prices to variation
in physician spending was partially offset by utilization of physician
services; we found higher prices in areas with lower utilization and lower
prices in areas with higher utilization. We did not find a similar
offsetting relationship between price and utilization for hospital
spending.

                                   Background

FEHBP and Participating In 2004, the federal government spent more than
$21 billion on FEHBP, which provides health insurance to federal civilian
employees, their

    PPOs

families, and retirees. Administered by the Office of Personnel Management
(OPM), FEHBP contracts with private insurers to provide

16

Total spending per enrollee includes spending for all health care services
except mental health, chemical dependency, and pharmaceuticals. We
adjusted total spending per enrollee for differences in costs of providing
service and in the age and sex of enrollees across metropolitan areas.

17

We defined areas in the highest 25 percent of spending as areas with the
highest spending and areas in the lowest 25 percent of spending as areas
with the lowest spending.

Page 5 GAO-05-856 FEHBP Health Care Prices

    Geographic Variation in Spending, Utilization, and Prices

health benefits. As such, it is the largest private health insurance
program in the country, covering nearly 8 million enrollees. Federal
employees enrolled in FEHBP can select from a number of private insurance
plans. In 2004, 183 private health insurance plans, including both local
HMOs and national PPOs, contracted with FEHBP to provide health insurance.
Nearly 75 percent of FEHBP beneficiaries were enrolled in national PPOs in
2004; the remainder were enrolled in local HMOs. The national PPOs offered
the same benefits and charged the same premiums regardless of where
enrollees lived or obtained their health care. However, the prices the
national PPOs paid to the hospitals and physicians in their networks
varied across the country depending on the prices negotiated between the
PPOs and their hospital and physician providers. Enrollee coinsurance
payments, which are based on a percentage of the negotiated prices, also
varied.

Geographic variation in prices and spending in private sector plans, such
as those participating in FEHBP, have not been extensively researched.
However, a well-established body of research has shown wide variation in
fee-for-service Medicare spending and utilization per beneficiary, even
after accounting for differences in population demographics and illness.
18 In 1996, Medicare spending per beneficiary was higher in the Midwest
and the South, especially in parts of Texas and Louisiana, than in the
North and West. Across the country, Medicare spending per beneficiary
varied by a factor of 2.9. A more recent examination of Medicare spending
showed continued geographic differences in spending per beneficiary across
the nation. 19

Geographic differences in utilization have also been found, though the
amount of utilization variation depends upon the type of service. For
instance, Medicare beneficiaries had more than twice as many nonsurgical
hospital discharges in 1995-1996 20 and more than five times as many hip
and knee replacement surgeries in some markets as in others in 2000

18

The Center for the Evaluative Clinical Sciences, Dartmouth Medical School,
The Dartmouth Atlas of Health Care 1999: The Quality of Medical Care in
the United States: A Report on theMedicare Program(Chicago, Ill.: AHA
Press, 1999).

19

GAO analysis of unadjusted 2003 Medicare spending per beneficiary data.

20The Dartmouth Atlas of Health Care 1999, p. 74.

Page 6 GAO-05-856 FEHBP Health Care Prices

2001. 21 Geographic differences in the use of inpatient services do not
appear to be caused by the substitution of other, less costly services;
markets with higher Medicare spending per enrollee for acute care hospital
services in 1996 also tended to have higher outpatient and physician
spending per enrollee. 22 Studies of other populations, such as veterans
and enrollees in Blue Cross Blue Shield of Michigan, also showed that
regional variation in hospital use occurred in those populations. 23

Unlike in the private sector, where prices may be subject to negotiation,
the prices paid to hospitals and physician providers by Medicare are not
subject to negotiation. Medicare establishes national prices and adjusts
them by using formulas that incorporate estimates of differences in input
costs, such as wages and rents across geographic areas. In the private
sector, prices are negotiated between providers 24 and health insurers.
Insurers may negotiate discounted rates with providers in exchange for an
anticipated share of patient volume from the insurers' enrollees. The
negotiated price may take into account the costs of doing business faced
by providers as well as other market characteristics affecting the
geographic area. Thus, the geographic differences in price in the Medicare
program may not be the same as in the private sector.

Characteristics of the health care markets across the country may affect
the prices that private sector insurers pay for health care services.
Market characteristics such as the extent of competition among providers,
the prevalence of managed care, and whether private sector providers shift
costs to compensate for lower reimbursements from some payers all may
contribute to variations in prices across the country.

    Health Care Market Characteristics and Price

21

J.N. Weinstein et al., "Trends and Geographic Variations in Major Surgery
for Degenerative Diseases of the Hip, Knee and Spine," Health Affairs, Web
Exclusive, (Oct. 7, 2004).
http://content.healthaffairs.org/cgi/content/full/hlthaff.var.81
(downloaded June 21, 2005).

22Th

e Dartmouth Atlas of Health Care 1999, pp. 11 and 27.

23

C.M. Ashton, et al., "Geographic Variations in Utilization Rates in
Veterans Affairs Hospitals and Clinics," The New England Journal of
Medicine,vol. 340, no. 1 (1999). The Center for the Evaluative Clinical
Sciences, Dartmouth Medical School and The Center for Outcomes Research
and Evaluation, Maine Medical Center, The Dartmouth Atlas of Health Care
in Michigan,2000, pp. 46 and 47.

24

We use the term providers to refer to hospitals, physicians, and other
providers of health care services unless otherwise specified.

Some but not all studies have shown that recent decreases in competition
among providers have been associated with increased prices. 25 , 26
Research shows that since 1995, the hospital industry has become
increasingly consolidated, and physicians have become increasingly aligned
with health systems and hospital networks. For example, in 1995, 51
percent of all private acute care hospitals were part of a hospital
system. By 2000, the percent of hospitals in systems had risen to 57
percent. 27 Consolidation reduces the number of competitors in a market,
giving the consolidated competitors a larger market share. Competition
also may be limited in markets with small populations because less
populated markets naturally have fewer hospitals or providers and hence
few competitors. Some studies have shown that consolidation is associated
with cost savings achieved by generating efficiencies and reducing excess
capacity. 28 For example, consolidated hospitals can streamline operations
by centralizing services, such as emergency care or intensive care units.
29 However, other studies of hospital mergers and acquisitions have not
found evidence that they result in any reductions in costs. 30

Other research has shown that the presence of HMOs in a metropolitan area
may also influence the price of health care services. 31 HMOs have

25

See for example, A.E. Cuellar and P.J. Gertler, "How the Expansion of
Hospital Systems Has Affected Consumers," Health Affairs,vol. 24, no. 1
(2005); C. Capps and D. Dranove, "Hospital Consolidation and Negotiated
PPO Prices," Health Affairs,vol. 23, no. 2 (2004);

L.M. Nichols, et al., "Are Market Forces Strong Enough to Deliver
Efficient Health Care Systems? Confidence is Waning," Health Affairs, vol.
23, no. 2 (2004); and H.R. Spang, G.J. Bazzoli, and R.J. Arnould,
"Hospital Mergers and Savings for Consumers: Exploring New Evidence,"
Health Affairs,vol. 20, no. 4 (2001).

26

Hospitals may compete on dimensions other than price, such as services,
amenities, and quality. See for example, M.A. Morrisey, "Competition in
Hospital and Health Insurance Markets: A Review and Research Agenda,"
Health Services Research, vol. 36, no. 1 (2001).

27

Cuellar and Gertler, "How the Expansion of Hospital Systems Has Affected
Consumers,"

p. 213.

28

See for example, Spang, Bazzoli, and Arnould, "Hospital Mergers and
Savings for Consumers," p. 150; and G.J. Bazzoli et al., "Hospital
Reorganization and Restructuring Achieved Through Merger," Health Care
Management Review, vol. 27, no. 1 (2002).

29

Bazzoli et al., "Hospital Reorganization and Restructuring Achieved
Through Merger," pp. 2 and 6.

30

D. Dranove, A. Durkac, and M. Shanley, "Are Multihospital Systems More
Efficient?" Health Affairs,vol. 15, no. 1 (1996).

31

L. Baker, "Measuring Competition in Health Care Markets," Health Services
Research, April (2001); and M.A. Morrisey, "Competition in Hospital and
Health Insurance Markets,"

p. 191.

typically attempted to moderate spending by introducing controls on both
utilization and price. One of the controls HMOs have used is to compensate
their primary care physicians with a capitated payment-a fixed,
predetermined payment for caring for an enrollee for a specified period of
time, regardless of the number or type of services ultimately provided. In
addition, research indicates HMOs have been able to secure deeper
discounts from hospitals and physicians than other insurers. HMOs have
tended to have smaller, exclusive provider networks and have been able to
channel their enrollees to a limited number of providers in exchange for
the lower rates. Toward the end of the 1990s, in response to resistance
against managed care from providers and patients alike, HMOs relaxed the
policies they had imposed to control utilization, price, and spending. For
example, one study reported a sharp decline from 1999 to 2001 in the
controls typically used by HMOs. Of more than 50 HMOs in the study,
virtually all reported a trend toward broader provider networks and some
reported decreased use of financial incentives, such as capitation. 32

Cost shifting-the theory that providers charge higher prices to one set of
payers to compensate for lower revenues from other payers-has been debated
for decades. Some researchers, for example, have found that when Medicare
and Medicaid reimbursements fall, private sector reimbursements rise. 33
Yet, other researchers have found no evidence of cost shifting. 34 More
recent articles on this subject note that cost shifting is possible, but
only when providers have had sufficient and untapped market power to raise
prices. 35 Without sufficient market power, providers

32

D.A. Draper, et al., "The Changing Face of Managed Care," Health
Affairs,vol. 21, no. 1 (2002).

33

J.S. Lee, et al., "Medicare Payment Policy: Does Cost Shifting Matter?"
Health Affairs, Web Exclusive, (Oct. 8, 2003).
http://content.healthaffairs.org/cgi/content/full/hlthaff.w3.480v1
(downloaded June 21, 2005); and Congressional Budget Office, "Responses to
Uncompensated Care and Public Program Controls on Spending: Do Hospitals
`Cost-Shift'?" (Washington, D.C.: 1993).

34

See for example, T. Rice, et al., "Do Physicians Cost Shift," Health
Affairs,vol. 15, no. 3 (1996); and J. Hadley, S. Zuckerman, L.I. Iezzoni,
"Financial Pressure and Competition: Changes in Hospital Efficiency and
Cost-Shifting Behavior," Medical Care,vol. 34, no. 3 (1996).

35

See for example, M.A. Morrisey, "Cost Shifting: New Myths, Old Confusion,
and Enduring Reality," Health Affairs, Web Exclusive (Oct. 8, 2003).
http://content.healthaffairs.org/cgi/content/full/hlthaff.w3.489v1
(downloaded June 21, 2005); and P.B. Ginsburg, "Can Hospitals and
Physicians Shift the Effects of Cuts in Medicare Reimbursement to Private
Payers?" Health Affairs,Web Exclusive (Oct. 8, 2003).
http://content.healthaffairs.org/cgi/content/full/hlthaff.w3.472v1
(downloaded June 21, 2005).

that cost shift and raise private sector prices might lose privately
insured patients. Alternatively, providers might also react to a decrease
in prices from payers by lowering private sector prices, as was reported
to be the case for Medicaid dependent hospitals in California. 36

Prices paid by FEHBP PPOs varied by 259 percent for hospital stays and by
about 100 percent for physician services across the metropolitan areas in
our study. Prices for both hospital stays and physician services tended to
be higher in metropolitan areas in the Midwest and lower in metropolitan
areas in the Northeast.

  Large Differences in Hospital and Physician Prices across Metropolitan Areas

    Hospital Prices Varied More than Physician Prices

Adjusted hospital prices paid by FEHBP PPOs varied considerably across
metropolitan areas. In the lowest-priced metropolitan area, hospital
prices were 51 percent of the national average (index value of 0.51) and
in the highest-priced metropolitan area, they were 83 percent above the
national average (index value of 1.83)-a difference of 259 percent. In
five of the 232 metropolitan areas, FEHBP PPOs paid hospital prices that
were more than 50 percent above the national average. While there were
other metropolitan areas with very high and very low prices, most had
prices much closer to the average. Half of the metropolitan areas in our
study, those in the second and third quartiles, had hospital prices that
were no more than 14 percent above or below the national average, 37 and
80 percent had hospital prices ranging from 22 percent below average to 27
percent above average. The distribution of hospital price indices among
232 metropolitan areas is presented in fig. 1.

36

D. Dranove and W.D. White, "Medicaid-dependent Hospitals and Their
Patients: How Have They Fared?" Health Services Research, (June 1998).

37

Quartiles divide the distribution of prices from lowest to highest into
four equal groups. The lowest quartile represents metropolitan areas
ranked in the lowest 25 percent of price, and the highest quartile
represents metropolitan areas ranked in the highest 25 percent of price.

Page 10 GAO-05-856 FEHBP Health Care Prices

Figure 1: Distribution of Hospital Price Indices across 232 Metropolitan Areas,
                                      2001

Source: GAO analysis of FEHBP data.

Note: We adjusted hospital prices to remove the effect of geographic
differences in the costs of doing business (wages, rents, etc.) and
differences in the severity of illnesses and mix of diagnoses among
metropolitan areas. We converted hospital prices to an index by dividing
the average price for a hospital stay in a metropolitan area by the
average price for all hospital stays in 232 metropolitan areas. The
average hospital price index value is 1.00.

Prices paid by FEHBP PPOs for physician services also varied substantially
but less than hospital prices, after adjusting them for geographic
differences in the costs of doing business and the mix of services. In the
lowest-priced metropolitan area, Baltimore, Maryland, physician prices
were 73 percent of the national average (index value of 0.73), and in the
highest-priced metropolitan area, La Crosse, Wisconsin, 38 they were
nearly 50 percent above the national average (index value of 1.48).
Overall, the percentage difference in prices between the lowest- and the
highest-priced metropolitan areas was about 100 percent. Half of the
metropolitan areas in our study, those in the second and third quartiles,
had physician prices that were no more than 9 percent above or below the
national average, and 80 percent had physician prices that were no more
than 16 percent above or below the national average. The distribution of
physician prices among 319 metropolitan areas is presented in fig. 2. 39
In addition, metropolitan areas with higher physician prices tended to
have higher hospital prices, and metropolitan areas with lower physician
prices tended to have lower hospital prices.

38

The La Crosse, Wisconsin metropolitan area includes areas in Minnesota.

39

We had sufficient data to analyze more metropolitan areas for physician
prices than for hospital prices.

Page 11 GAO-05-856 FEHBP Health Care Prices

Figure 2: Distribution of Physician Price Indices across 319 Metropolitan
Areas, 2001

Source: GAO analysis of FEHBP data.

Notes: We adjusted physician prices to remove the effect of geographic
variation in the costs of doing business (wages, rents, etc.) and
differences in the mix of services among metropolitan areas. We converted
physician prices to an index by dividing the average physician price per
service in a metropolitan area by the average physician price in 319
metropolitan areas. The average physician price index value is 1.00.

We had sufficient data to analyze more metropolitan areas for physician
prices than for hospital prices.

    Hospital and Physician Prices Were Generally Higher in the Midwest and Lower
    in the Northeast

On average, FEHBP PPOs paid higher prices for hospital stays in
metropolitan areas in the Midwest and lower prices in the Northeast. (See
fig. 3.) Hospitals in the Midwest were paid about 14 percent more, on
average, than hospitals in the Northeast (table 1), but there was a
considerable range of hospital prices within regions. In fact, several
metropolitan areas with hospital prices in the highest quartile were
located in the same state as metropolitan areas with hospital prices in
the lowest quartile. For example, hospital prices in Buffalo-Niagara
Falls, New York were 45 percent higher than average, but prices in
Syracuse, New York were 20 percent below average. Similarly, prices in
Salinas, California were 50 percent higher than average, but prices in
Orange County, California were 48 percent below average. The 10
metropolitan areas with the highest and lowest hospital prices are listed
in table 2. Appendix II presents the complete rankings of metropolitan
areas by hospital price.

Figure 3: FEHBP PPO Adjusted Hospital Price Index Quartiles in 232 Metropolitan
                                  Areas, 2001

Table 1: FEHBP PPO Hospital Price Indices in Metropolitan Areas Grouped by
Census Region, 2001

Average hospital price indexa for Region region

                                  Midwest 1.07

                                   West 1.00

                                   South 1.00

                                 Northeast 0.94

Percent by which prices in the Midwest exceed prices in the Northeast
13.83

Source: GAO analysis of FEHBP data.

a

We adjusted hospital prices to remove the effect of geographic differences
in the costs of doing business (wages, rents, etc.) and differences in the
severity of illnesses and mix of diagnoses among metropolitan areas. We
converted hospital prices to an index by dividing the average hospital
price in a metropolitan area by the average hospital price for all 232
metropolitan areas. The average hospital price index is 1.00.

Table 2: Metropolitan Areas with the Highest and Lowest Hospital Price
Indices in FEHBP PPOs, 2001

Highest-priced Lowest-priced Rank metropolitan areas Rank metropolitan
areas

a

232 Orange County, Calif.

Dover, Del. 231 Pueblo, Colo.

Biloxi-Gulfport-Pascagoula, Miss. 230 Ventura, Calif.

St. Joseph, Mo. 229 Albany-Schenectady-Troy, N.Y.

Milwaukee-Waukesha, Wisc. 228 Newburgh, New York-Penn.

Salinas, Calif. 227 New York, N.Y.

Buffalo-Niagara Falls, N.Y. 226 Altoona, Penn.

Grand Junction, Colo. 225 Decatur, Ala.

a

224 Anniston, Ala.

La Crosse, Wisconsin-Minn. 223 Saginaw-Bay City-Midland, Mich.

Source: GAO analysis of FEHBP data.

Note: We adjusted hospital prices to remove the effect of geographic
differences in the costs of doing business (wages, rents, etc.) and
differences in the severity of illnesses and mix of diagnoses among
metropolitan areas.

a

Name withheld to protect proprietary data where the metropolitan area had
only one hospital in 2001.

As with hospital prices, FEHBP PPOs paid higher average physician prices
in metropolitan areas in the Midwest and lower average physician prices in
metropolitan areas in the Northeast (see fig. 4). Prices for physician
services were 15 percent higher, on average, in metropolitan areas in the
Midwest than in metropolitan areas in the Northeast (table 3).
Metropolitan areas in Wisconsin had physician prices ranked among the
highest in our study: of the 10 metropolitan areas with the highest
physician prices, eight were located in Wisconsin (table 4). About 80
percent of the metropolitan areas in the Northeast had below-average
prices for physician services. Also, physician prices tended to be less
variable within states than hospital prices. For example, among
metropolitan areas in New Jersey, physician prices ranged from 12 percent
below average to 19 percent below average, but hospital prices ranged from
about 4 percent below average to about 27 percent below average. Appendix
III contains a complete ranking of physician prices in 319 metropolitan
areas.

Figure 4: FEHBP PPO Adjusted Physician Price Index Quartiles in 319 Metropolitan
                                  Areas, 2001

Table 3: FEHBP PPO Physician Price Indices in Metropolitan Areas Grouped
by Census Region, 2001

Average physician price Region indexa for region

                                  Midwest 1.05

                                   South 1.02

                                   West 0.99

                                 Northeast 0.91

Percent by which prices in the Midwest exceed prices in the Northeast

Source: GAO analysis of FEHBP data.

a

We adjusted physician prices to remove the effect of geographic
differences in the costs of doing business (wages, rents, etc.) and
differences in the mix of services among metropolitan areas. We converted
physician prices to an index by dividing the average physician price per
service in a metropolitan area by the average physician price in 319
metropolitan areas. The average physician price index value is 1.00.

Table 4: Metropolitan Areas with the Highest and Lowest Physician Price
Indices in FEHBP PPOs, 2001

Highest-priced Lowest-priced Rank metropolitan areas Rank metropolitan
areas

La Crosse, Wisconsin-Minn. 319 Baltimore, Md.

Wausau, Wisc. 318 Lowell, Massachusetts-N.H.

Eau Claire, Wisc. 317 Nassau-Suffolk, N.Y.

Madison, Wisc. 316 Washington, D.C.

Jonesboro, Ark. 315 Fort Lauderdale, Fla.

Janesville-Beloit, Wisc. 314 West Palm Beach-Boca Raton, Fla.

Great Falls, Mont. 313 Miami, Fla.

Green Bay, Wisc. 312 Providence-Fall River-Warwick, Rhode Island-Mass.

Appleton-Oshkosh-Neenah, Wisc. 311 Dutchess County, N.Y.

Racine, Wisc. 310 San Francisco, Calif.

Source: GAO analysis of FEHBP data.

Note: We adjusted physician prices to remove the effect of geographic
differences in the costs of doing business (wages, rents, etc.) and
differences in the mix of services among metropolitan areas.

a

The Washington, District of Columbia metropolitan area includes parts of
Maryland, Virginia, and West Virginia.

  Less Competition and Less HMO Capitation Linked to Higher Health Care Prices

FEHBP PPOs paid higher average hospital and physician prices in
metropolitan areas with less competition among hospitals. 40 Many
metropolitan areas we studied had low levels of competition; about one in
four metropolitan areas had only one or two hospitals or hospital networks
serving the entire market. Also, FEHBP PPOs paid higher average hospital
and physician prices in metropolitan areas with less HMO capitation. HMOs
did not have capitated arrangements in more than one-third of the
metropolitan areas we studied. We found no evidence of cost
shifting-higher hospital or physician prices where there were lower
Medicaid payments or larger uninsured, Medicare, or Medicaid populations.

    Prices Were Higher in Metropolitan Areas with Less Competition

FEHBP PPO hospital and physician prices were higher, on average, in
metropolitan areas with less competition among hospitals. In the least
competitive metropolitan areas-those in the quartile with the least
competition-hospital prices tended to be about 18 percent higher and
physician prices tended to be nearly 11 percent higher than in the most
competitive metropolitan areas-those in the quartile with the most
competition. See table 5. For example, Rapid City, South Dakota, was in
the quartile with the least competition; its hospital prices were 25
percent above average, and its physician prices were 10 percent above
average. In contrast, Pittsburgh, Pennsylvania, a metropolitan area in the
quartile with the most competition, had hospital prices 14 percent below
average and physician prices 16 percent below average. When we conducted a
separate analysis that simulated the effect of increasing the level of
competition while controlling for the effects of other factors, we found
that less competition was still associated with higher prices, although
the difference was reduced by 58 percent for hospital prices and 38
percent

40

We measured competition as the percentage of hospital beds in a
metropolitan area (market share) held by the two largest hospitals or
hospital networks, where higher percentages indicated less competition and
lower percentages indicated more. Physicians are often aligned with health
systems and hospital networks. Therefore, we approximated physician
competition by measuring competition among hospitals and hospital networks
in a metropolitan area.

Page 18 GAO-05-856 FEHBP Health Care Prices

for physician prices. 41 See appendix I for a complete description of the
other factors we analyzed.

Table 5: FEHBP PPO Price Indices in the Least and Most Competitive
Metropolitan Areas, 2001

Average Average hospital price physician price Competition quartile indexa
                                                                       indexb

                          Least competitivec 1.10 1.04

                          Most competitivec 0.93 0.94

Percent by which prices in the least competitive areas exceed prices in
the most competitive areasd 18.28 10.64

Source: GAO analysis of FEHBP data.

a

We adjusted hospital prices to remove the effect of geographic differences
in the costs of doing business (wages, rents, etc.) and differences in the
severity of illnesses and mix of diagnoses among metropolitan areas. We
converted hospital prices to an index by dividing the average price for a
hospital stay in a metropolitan area by the average price for all hospital
stays in 232 metropolitan areas. The average hospital price index value is
1.00.

b

We adjusted physician prices to remove the effect of geographic
differences in the costs of doing business (wages, rents, etc.) and
differences in the mix of services among metropolitan areas. We converted
physician prices to an index by dividing the average physician price per
service in a metropolitan area by the average physician price in 319
metropolitan areas. The average physician price index value is 1.00

The competition quartiles were based on 232 metropolitan areas for the
hospital price analysis and 319 metropolitan areas for the physician price
analysis.

d

We simulated the effect of increasing competition in these metropolitan
areas from the average level of competition in the lowest quartile to the
average level of competition in the highest quartile, while controlling
for other factors such as our measures of competition, HMO capitation,
cost shifting, per capita income, percent of for-profit beds, provider
supply, and census divisions. We found that, on average, the effect of
increasing competition was to reduce the hospital price index in a
metropolitan area by 7.62 percent and the physician price index in a
metropolitan area by 6.64 percent. See app. I for a complete list of
control factors.

Other factors included in our analysis were our measures of competition,
HMO capitation, cost shifting, per capita income, percent of for-profit
beds, provider supply, and census division. See app. I for a detailed
description of each factor. When we simulated the effect of increasing
competition from the average level of competition in the lowest quartile
to the average level of competition in the highest quartile, while
controlling for other factors, our estimate of the percent difference in
the average hospital price index between the highest and lowest
competition quartiles was 7.62 percent, and our estimate of the percent
difference in the average physician price index between the highest and
lowest quartiles was 6.64 percent.

    Prices Were Higher in Metropolitan Areas with Less HMO Capitation

Overall, many metropolitan areas in our study had low levels of
competition. Several of the metropolitan areas in our study had few
competing hospitals or hospital networks. In approximately one quarter of
the 319 metropolitan areas in our study, 100 percent of the market share
was held by one or two hospitals or hospital networks. In the most
competitive metropolitan areas, about 44 percent of the market share, on
average, was held by the two largest hospitals or hospital networks.
Across all metropolitan areas, about 75 percent of the market share, on
average, was held by the two largest hospitals or hospital networks. The
least competitive metropolitan areas also tended to have smaller
populations. In the quartile with the least competition, the average
population was about 160,000. The average population of the metropolitan
areas in the quartile with the most competition was more than 1.8 million.

FEHBP PPO hospital and physician prices were higher, on average, in
metropolitan areas with less HMO capitation. 42 On average, both hospital
prices and physician prices were more than 10 percent higher in
metropolitan areas in the quartile with the least HMO capitation than in
the quartile with the most HMO capitation (table 6). For example,
Hattiesburg, Mississippi, which had no HMO capitation, had both hospital
and physician prices in the highest quartile. In contrast, Philadelphia,
Pennsylvania, was in the highest quartile of HMO capitation and in the
lowest quartiles of both hospital and physician prices. When we conducted
a separate analysis that simulated the effect of increasing the level of
HMO capitation while controlling for the effects of other factors, less
HMO capitation was still associated with higher prices, but the difference
was

42

Capitation is a payment method where physicians are paid a fixed,
predetermined payment for caring for an enrollee for a specified period of
time, regardless of the number or type of services provided. Physicians
often try to resist capitation payments. The use of capitation by HMOs
demonstrates that they have the leverage to negotiate capitation contracts
with physicians. We used HMO capitation as a proxy measure for the
strength of the HMO presence in a community, and its ability to negotiate
prices with physicians, hospitals, and other providers.

reduced by about one-third for hospital prices and two-thirds for
physician prices. 43 See appendix I.

Table 6: FEHBP PPO Price Indices in Metropolitan Areas with the Least and
Most HMO Capitation, 2001

       Average Average hospital price physician price HMO capitation quartile
                                                                indexa indexb

                        Least HMO capitationc 1.05 1.06

                         Most HMO capitationc 0.95 0.96

Percent by which prices in areas with the least capitation exceed prices
in areas with the most capitationd 10.53 10.42

Source: GAO analysis of FEHBP data.

a

We adjusted hospital prices to remove the effect of geographic differences
in the costs of doing business (wages, rents, etc.) and differences in the
severity of illnesses and mix of diagnoses among metropolitan areas. We
converted hospital prices to an index by dividing the average price for a
hospital stay in a metropolitan area by the average price for all hospital
stays in 232 metropolitan areas. The average hospital price index value is
1.00.

b

We adjusted physician prices to remove the effect of geographic
differences in the costs of doing business (wages, rents, etc.) and
differences in the mix of services among metropolitan areas. We converted
physician prices to an index by dividing the average physician price per
service in a metropolitan area by the average physician price in 319
metropolitan areas. The average physician price index value is 1.00.

HMO capitation quartiles were based on 232 metropolitan areas for the
hospital price analysis. HMO capitation data were not available in 4 of
the 319 metropolitan areas in physician price analysis, and the HMO
capitation quartiles were based on 315 metropolitan areas for the
physician price analysis.

d

We simulated the effect of increasing HMO capitation in these metropolitan
areas from the average level of HMO capitation in the lowest quartile to
the average level of HMO capitation in the highest quartile, while
controlling for other factors such as the level of competition, cost
shifting, income, percent of for-profit beds, provider supply, and census
divisions. We found that, on average, the effect of increasing HMO
capitation was to reduce the hospital price index in a metropolitan area
by

7.17 percent and the physician price index in a metropolitan area by 3.31
percent. See app. I for a complete list of control factors.

Other factors included in our analysis were our measures of competition,
HMO capitation, cost shifting, per capita income, percent of for-profit
beds, provider supply, and census division. See app. I for a detailed
description of each factor. When we simulated the effect of increasing the
level of HMO capitation from the average level of HMO capitation in the
lowest quartile to the average level of HMO capitation in the highest
quartile, while controlling for other factors, our estimate of the percent
difference in the average hospital price index between the highest and
lowest HMO capitation quartiles was 7.17 percent and our estimate of the
percent difference in the average physician price index between the
highest and lowest quartiles was 3.31 percent.

    No Evidence of Cost Shifting Due to Medicaid, Medicare, or the Uninsured

Many of the metropolitan areas in our study had low levels of HMO
capitation. 44 More than a third of the metropolitan areas had almost no
HMO capitation; on average, less than 1 percent of the payments to primary
care physicians in these areas were paid on a capitated basis. In the
metropolitan areas in the highest quartile of HMO capitation, 23 percent
of primary care physicians' compensation was capitated, on average. Among
all metropolitan areas, about 8 percent of primary care physicians'
compensation was capitated, on average. As we found with competition,
metropolitan areas with the least HMO capitation tended to be the less
populated areas. Of the metropolitan areas that had almost no HMO
capitation, the average population was about 250,000, while those in the
highest quartile of HMO capitation had an average population of nearly

1.1 million.

We found no evidence of cost shifting. FEHBP PPOs did not pay higher
prices in metropolitan areas with a higher percentage of Medicaid or
Medicare beneficiaries, a larger uninsured population, or lower Medicaid
payments. 45 When we controlled for other factors that might have been
associated with price, none of our cost-shifting factors were
significantly related to higher prices. See appendix I.

While none of these cost-shifting factors were significantly associated
with higher hospital or physician prices, physician prices were actually
lower, on average, in metropolitan areas with lower adjusted Medicaid
payment rates and proportionately larger uninsured populations. Physician
prices were nearly 10 percent lower in the metropolitan areas in the
quartile with the lowest Medicaid payment index (average of 0.65) than in
the quartile with the highest Medicaid payment index (average of 1.29).
See table 7. When we conducted a separate analysis that simulated the
effect of increasing the level of Medicaid payments, while controlling for
the effects of other factors, we found that other factors did not
significantly affect the

44

HMO capitation data were not available in 4 of the 319 metropolitan areas
in our study. Accordingly, our analysis of HMO capitation was based on 315
metropolitan areas.

45

We estimated Medicaid payment rates for each metropolitan area by taking
the average physician payment for a set of common services. Medicaid
payment rate estimates for metropolitan areas were based on statewide
payment rates. We adjusted Medicaid payment rates to remove the effect of
geographic differences in input costs and in the mix of services across
metropolitan areas. See app. I.

observed relationship between physician prices and Medicaid payments. 46
There was no significant association between Medicaid payments and
hospital prices. See appendix I.

Table 7: FEHBP PPO Price Indices in Metropolitan Areas in the Lowest and
Highest Medicaid Payment Quartiles, 2001

Average physician Medicaid payment quartile price indexa

                                  Lowest 0.92

                                  Highest 1.02

Percent by which prices in the lowest Medicaid payment areas were lower
than prices in the highest Medicaid payment areasb

Source: GAO analysis of FEHBP data.

a

We adjusted physician prices to remove the effect of geographic
differences in the costs of doing business (wages, rents, etc.) and
differences in the mix of services among metropolitan areas. We converted
physician prices to an index by dividing the average physician price per
service in a metropolitan area by the average physician price in 319
metropolitan areas. The average physician price index value is 1.00.

b

We simulated the effect of increasing Medicaid payments in these
metropolitan areas from the average Medicaid payment in the lowest
quartile to the average Medicaid payment in the highest quartile, while
controlling for other factors such as our measures of competition, HMO
capitation, other cost-shifting variables, income, percent of for-profit
beds, provider supply, and census divisions. We found that, on average,
the effect of increasing Medicaid payments was to increase the physician
price index in a metropolitan area by 9.69 percent. However, there was no
significant association between the Medicaid payments and hospital prices.
See app. I for a complete list of control factors.

The relationship between the percentage of the population uninsured and
physician price was only evident when we controlled for other factors. We
simulated the effect of increasing the percentage of the population
uninsured from the average percent uninsured in the lowest quartile to the
average percent uninsured in the highest quartile, while controlling for
other factors. 47 In this simulation, we found that the physician prices
were 6 percent lower, on average, in the quartile with the highest percent

46

Other factors included in our analysis were measures of competition, HMO
capitation, cost shifting, per capita income, provider supply, and census
division. See app. I for a detailed description of each factor. When we
simulated the effect of increasing Medicaid payments from the average
Medicaid payment in the lowest quartile to the average Medicaid payment in
the highest quartile, while controlling for other factors, we found that
the physician price index was 9.69 percent higher, on average.

47

These factors included our measures of competition, HMO capitation, other
cost-shifting variables, per capita income, percent of for-profit beds,
provider supply, and census divisions.

  Total Spending Varied 112 Percent; Price Variation Contributed to One-third of
  the Variation in Hospital and Physician Spending

uninsured (average uninsured percent of 19.5) than in the quartile with
the lowest percent uninsured (average percent uninsured of 8.5). There was
no significant association between the percent uninsured and hospital
prices. See appendix I for a complete list of control factors. 48

FEHBP PPO total spending per enrollee was more than twice as high in some
areas as in others. 49 Metropolitan areas in the South tended to have
higher spending per enrollee, while metropolitan areas in the Northeast
tended to have lower spending per enrollee. For both hospital and
physician services, variation in price contributed about one-third of the
difference in spending per enrollee between metropolitan areas in the
highest and lowest quartiles of spending. Metropolitan areas with higher
physician prices tended to have lower physician utilization, which offset
the impact of physician price on physician spending to some extent. We
found no such offsetting relationship between hospital prices and hospital
utilization.

    Spending per Enrollee Varied by 112 Percent across Metropolitan Areas

We found that total spending per enrollee varied by 112 percent across the
232 metropolitan areas in this analysis. Total spending per enrollee was
the amount spent by FEHBP PPOs per person for all health care services
except pharmaceuticals, mental health services, and substance abuse
services, after adjusting for enrollee age and sex differences as well as
geographic differences in the costs of doing business. Spending per
enrollee in the metropolitan area with the lowest spending per enrollee,
Grand Rapids-Muskegon-Holland, Michigan, was 67 percent of the national
average (index value of 0.67). Spending per enrollee in the metropolitan
area with the highest spending per enrollee, Biloxi-Gulfport-Pascagoula,
Mississippi, was 42 percent above the average (index value of 1.42). Half
of the metropolitan areas in our study, those in the second and third
quartiles, had spending per enrollee that was no more than 10 percent
above or below the national average, and 80 percent had spending per
enrollee ranging from about 16 percent below average to about 19 percent

48

The percent of the population that was uninsured was based on statewide
data and does not include differences in uninsured rates among
metropolitan areas in the same state. See app. I for a description of our
regression methodology and results.

49

Total spending per enrollee includes spending for all health care services
except mental health, chemical dependency, and pharmaceuticals. We
adjusted total spending per enrollee for differences in costs of providing
service and in the age and sex of enrollees across metropolitan areas.

above average. The distribution of spending per enrollee indices among 232
metropolitan areas is presented in figure 5. Appendix IV contains the
spending per FEHBP enrollee ranking for 232 metropolitan areas.

Figure 5: Distribution of FEHBP PPO Spending Per Enrollee Indices across
232 Metropolitan Areas, 2001

Source: GAO analysis of FEHBP data.

Note: Total spending per enrollee includes spending for all services
except mental health, chemical dependency, and pharmaceuticals. We
adjusted total spending per enrollee to remove the effect of geographic
differences in enrollee age and sex, as well as geographic differences in
the costs of doing business (such as wages and rents). The spending per
enrollee index compares spending per enrollee in a metropolitan area to
the average spending per enrollee in all study metropolitan areas,
adjusted for patients' age and sex composition, and costs. The average
spending index was 1.00.

Total spending per enrollee in FEHBP PPOs was, on average, highest among
metropolitan areas in the South and lowest in metropolitan areas in the
Northeast. About 86 percent of the metropolitan areas in the highest
spending quartile were located in the South (see fig. 6). Nearly 38
percent of the metropolitan areas in the lowest spending quartile were
located in the Northeast, and none of the metropolitan areas in the
highest spending quartile were in the Northeast. Spending per enrollee was
about 23 percent higher in metropolitan areas in the South than in the
Northeast, on average (see table 8).

  Figure 6: FEHBP Adjusted Spending Per Enrollee Quartiles in 232 Metropolitan
                                  Areas, 2001

    Price Contributed to One-third of the Variation in Spending, but the
    Contribution of Price to Spending Was Partially Offset by Utilization of
    Physician Services

Table8: FEHBP PPO Spending Per Enrollee Indices in Metropolitan Areas by
Census Region, 2001

Average spending per Region enrollee indexa or region

                                   South 1.08

                                  Midwest 0.95

                                   West 0.94

                                 Northeast 0.88

Percent by which spending in the South exceeds spending in the Northeast

Source: GAO analysis of FEHBP data.

a

Total spending per enrollee includes spending for all services except
mental health, chemical dependency, and pharmaceuticals. We adjusted total
spending per enrollee to remove the effect of geographic differences in
enrollee age and sex, as well as geographic differences in the costs of
doing business (wages, rents, etc.). The spending per enrollee index
compares spending per enrollee in a metropolitan area to the average
spending per enrollee in all study metropolitan areas, adjusted for
patients' age and sex composition, and costs. The average spending index
value was 1.00.

In FEHBP PPOs, hospital price variation contributed to about one-third of
the difference in average hospital spending per enrollee between the
highest and lowest hospital spending quartiles. 50 Similarly, physician
price variation contributed to about one-third of the difference in
average physician spending per enrollee between the highest and lowest
physician spending quartiles. Variation in utilization contributed about
two-thirds of the difference between metropolitan areas in the highest and
lowest quartiles of spending per enrollee for both hospital and physician
services. 51 Hospital prices and hospital utilization (hospital stays per
enrollee) were, on average, 26 percent higher and 55 percent higher,
respectively, in metropolitan areas in the highest hospital spending
quartile compared to metropolitan areas in the lowest hospital spending
quartile. 52 Physician prices were 12 percent higher, on average, in the

50

In order to analyze the contribution of price to geographic variation in
spending, we focused on hospital and physician spending (not total
spending), price, and utilization.

51

We did not analyze factors associated with this variation in utilization
as it was outside the scope of our research objectives.

52

The 26 percent difference between hospital prices in the highest and
lowest quartiles contributed to about one-third of the difference in
hospital spending. The 55 percent difference between hospital utilization
in the highest and lowest hospital spending quartiles contributed to about
two-thirds of the difference in hospital spending.

metropolitan areas in the highest than in the lowest physician spending
quartile. Physician utilization was 26 percent higher in the highest
physician spending quartile than it was in the lowest. 53 See table 9.

Table 9: Price and Utilization Indices in Metropolitan Areas in the
Highest and Lowest Quartiles of Hospital and Physician Spending, 2001

  Type of Average Average spending Spending quartile price indexa utilization
                                     indexb

                        Hospital stays Highest 1.12 1.24

                                Lowest 0.89 0.80

Percent by which highest hospital spending areas exceed lowest hospital
spending areas 25.84 55.00

Physician Highest 1.05 1.12services

                                Lowest 0.94 0.89

Percent by which highest physician spending areas exceed lowest physician
spending areas 11.70 25.84

Source: GAO analysis of FEHBP data.

a

We adjusted physician and hospital prices to remove the effect of
geographic differences in the costs of doing business (wages, rents, etc.)
and differences in the mix of services among metropolitan areas. For this
analysis, we converted both hospital and physician prices to an index by
dividing the average price in a metropolitan area by the average price in
232 metropolitan areas. The average price index is 1.00.

b

We removed the effect of geographic variation in enrollee age and sex in
metropolitan areas from utilization. The utilization of hospital and
physician services indices compare utilization of hospital and physician
services in a metropolitan area to the average utilization of hospital and
physician services in all study metropolitan areas, adjusted for age and
sex. The average utilization index for both hospital and physician
utilization was 1.00.

Although metropolitan areas with higher hospital and physician FEHBP PPO
spending per enrollee also tended to have higher hospital and physician
prices, respectively we found a modestly sized but statistically
significant inverse relationship between physician prices and physician
utilization. In general, there was lower utilization of physician services
where the price of physician services was higher, and higher utilization
of

The 12 percent difference between physician prices in the highest and
lowest quartiles contributed to about one-third of the difference in
physician spending. The 26 percent difference between physician
utilization in the highest and lowest physician spending quartiles
contributed to about two-thirds of the difference in physician spending.

Page 28 GAO-05-856 FEHBP Health Care Prices

physician services where the price of physician services was lower. For
example, Anchorage, Alaska and Bakersfield, California had similar
physician spending per enrollee,with both ranked in the highest spending
per enrollee quartile. Yet, Anchorage had below average utilization of
physician services and above average physician prices, while Bakersfield
had above average utilization of physician services and below average
physician prices. See table 10. The similar spending per enrollee in
Anchorage and Bakersfield occurred despite these areas having different
prices and utilization levels because of the offsetting relationship
between physician prices and physician utilization. While the off setting
relationship between physician price and physician utilization dampened
slightly the overall effect of physician price on spending, there was
still a statistically significant relationship between higher prices and
higher spending for both physician and hospital inpatient sectors. For
hospital services, we did not find an offsetting relationship between
price and utilization.

Table 10: Example of the Offsetting Effect of Physician Price and
Utilization on Physician Spending in Two Metropolitan Areas in the FEHBP,
2001

                   Anchorage, Alaska Bakersfield, California

                        Physician price indexa 1.22 0.94

                     Physician utilization indexb 0.78 1.20

                      Physician spending indexc 1.23 1.34

Source: GAO analysis of FEHBP data.

a

We adjusted physician prices to remove the effect of geographic
differences in the costs of doing business (wages, rents, etc.) and
differences in the mix of services among metropolitan areas. We converted
physician prices to an index by dividing the average physician price per
service in a metropolitan area by the average physician price in 232
metropolitan areas. The average physician price index is 1.00.

b

We removed the effect of geographic variation in enrollee age and sex in
metropolitan areas from utilization. The utilization of physician services
index compares utilization of physician services in a metropolitan area to
the average utilization of physician services in all study metropolitan
areas, adjusted for age and sex. The average utilization index is 1.00.

We removed the effect of geographic differences in enrollee age and sex,
as well as geographic differences in the costs of doing business (wages,
rents etc.) from physician spending. The physician spending per enrollee
index compares physician spending per enrollee in a metropolitan area to
the average physician spending per enrollee in all study metropolitan
areas, adjusted for patients' age and sex, and costs. The average
physician spending index is 1.00.

  Concluding Observations

Our analysis shows that an understanding of price variation is essential
to understanding geographic variation in health care spending in the
private sector. We found that market forces, not just the underlying costs
of doing business providers face, help to determine the prices FEHBP PPOs
ultimately pay hospitals and physicians. In metropolitan areas where there
was less competition among hospitals, FEHBP PPOs paid a higher price to
hospitals and physicians than in metropolitan areas where hospitals and
physicians had more competition. In metropolitan areas with less HMO
capitation, FEHBP PPOs paid higher prices,which also suggests that
hospitals and physicians in those metropolitan areas had less competition
for patient share. We found no evidence that hospitals or physicians
shifted costs, which suggests that FEHBP PPOs may have been influenced by
market forces when establishing prices, regardless of the amount of
uncompensated or undercompensated care in a metropolitan area. Further
investigation may help to explain why there were regional patterns that
appeared to be associated with private sector price variation.

  Agency and Other Comments

In written comments on a draft of this report, OPM officials agreed with
our findings that competition and other factors were linked to variation
in prices, stating that the findings confirm a long-held view of the
agency. In addition, they suggested that several issues warranted further
study and discussion. They pointed out that it would have been interesting
to examine the relationships between physician prices, Medicaid payments,
percentage of the population uninsured, and physician-prescribing
patterns. They also noted that it would be instructive to investigate
unexplained regional variations and intraregional variations. They thought
some findings could have been addressed in greater detail within the text
and in the concluding observations.

Representatives of the FEHBP PPOs were also given an opportunity to
comment on a draft of the report. Representatives of one PPO noted that
market dynamics and prices could have changed since 2001.

We agree this report addressed important issues but investigating them in
further detail was beyond the scope of our work. We agree that market
dynamics and prices could have changed since 2001, but we used the most
recent data available at the start of the study and maintain that the
relationship among the variables, specifically the linkage between
competition, HMO capitation, and prices is less likely to have changed.
Other comments provided by OPM and representatives of the FEHBP PPOs were
incorporated into the draft, as appropriate.

As arranged with your office, unless you publicly announce the contents of
this report earlier, we plan no further distribution of it until 30 days
after its issue date. At that time, we will send copies of this report to
the Director of the Office of Personnel Management and other interested
parties. We will also provide copies to others upon request. In addition,
the report is available at no charge on the GAO Web site at
http://www.gao.gov.

If you or your staff have any questions about this report, please contact
me at (202) 512- 7101 or steinwalda@gao.gov. Contact points for our
Offices 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 VI.

Sincerely yours,

A. Bruce Steinwald Director, Health Care

                       Appendix I: Scope and Methodology

  FEHBP Data and Study Eligibility Criteria

In this appendix we describe the data and methods we used to compare
geographic variations in prices and spending in metropolitan areas 1
across the United States, and to analyze patterns in the factors that
affect hospital and physician prices in these areas. We compared
differences in hospital and physician prices and in per-enrollee spending
across metropolitan areas using medical claims data from enrollees in
selected national preferred provider organizations (PPO) participating in
the Federal Employees Health Benefits Program (FEHBP). We identified
potential factors that contributed to hospital price and physician price
variation. We then examined the relationship between these factors and our
measures of hospital and physician prices. Finally, we compared total
spending per enrollee across metropolitan areas, and we examined the
contribution of hospital and physician prices to hospital and physician
spending.

We compared hospital prices, physician prices, and health care spending
per enrollee in metropolitan areas using 2001 health claims data from
FEHBP. These 2001 data were the most recent that were available at the
time we began our study. FEHBP, the health insurance program administered
by the Office of Personnel Management for federal civilian employees and
retirees, covered about 8.5 million people in 2001. FEHBP negotiates with
private insurers to provide health benefits. It is the largest
employer-sponsored insurance program in the United States.

Our study included claims data from federal civilian employees under the
age of 65 and their dependents who enrolled in selected national PPOs as
their primary insurers. 2, 3 We selected these PPOs because they had a
similar benefit structure with respect to coverage and out-of-pocket
requirements. We prorated the data for enrollees with partial year
enrollment based on their days of eligibility during 2001. We checked the
dates of service on claims to ensure that they were included only if the
service was delivered during a period when the member had insurance
coverage. We excluded pharmaceutical claims from the study, as well as

1

Metropolitan areas refer to metropolitan statistical areas, which the
Office of Management and Budget defines as a core population of at least
50,000 people and the adjacent communities linked socially and
economically with that core.

2

Our study may also have included some federal retirees under the age of
65, whose primary insurer was an FEHBP PPO.

3

We excluded PPO enrollees age 65 and over because Medicare, not FEHBP, was
their primary insurer, and consequently the PPOs did not have records of
all claim payments.

mental health and chemical dependency claims, because these services were
subcontracted to other organizations by at least one of the PPOs in our
study, and the associated claims for all service types were not available.

We aggregated payments from our claims data to metropolitan areas.
Metropolitan areas are designed to approximate market areas in general.
Actual health care markets may include larger or smaller geographic areas
and may not coincide exactly with metropolitan areas. However, we chose
metropolitan areas for our analysis because they correspond fairly closely
with heath care markets and we were able to obtain claims and other data
(see table 11) at the metropolitan area level. We did not examine prices
or spending outside of metropolitan areas because nonmetropolitan areas
are expansive and could include multiple markets that we would not be able
to distinguish between.

In 2001, there were 331 metropolitan areas in the 50 states and the
District of Columbia. We excluded some metropolitan areas from our study
because we could not obtain complete claims information due to payment
adjustments that occurred outside of the claims system or because there
was an insufficient number of hospital stays to support our price
analyses. 4 In addition, we excluded one metropolitan area because it had
a high proportion of claims from enrollees that lived outside of the area.
In our physician price analyses, we had adequate data to make comparisons
among 319 metropolitan areas. The population of these 319 metropolitan
areas accounted for 98 percent of the population living in all
metropolitan areas. In all other analyses, including physician spending
and utilization, we had adequate data to make comparisons among 232
metropolitan areas. 5 The population of these 232 metropolitan areas
accounted for 88 percent of the population living in all metropolitan
areas.

4

We excluded metropolitan areas that had fewer than 38 hospital stays.

5

Our analysis of hospital spending and utilization may have been limited by
the small number of enrollees and admissions in some areas. Ten of the 232
metropolitan areas in this analysis had between 500 and 1,000 enrollees.

Page 33 GAO-05-856 FEHBP Health Care Prices

  Hospital and Physician Price Estimates

We calculated price indices for hospital and physician services. We
selected these services because together they represented nearly
two-thirds of total health care spending and we could identify standard
units of service-hospital stays and physician procedures-to which we could
link prices. We derived our price estimates for each metropolitan area by
aggregating payments from individual claims to the metropolitan area where
the service was provided. 6

To estimate the price of a hospital stay, we first aggregated payments
from separate hospital claims to determine the total payments for that
stay. This involved combining hospital claims for the same enrollee that
had contiguous dates of service from the same provider. We excluded stays
that involved multiple hospital providers, and mental health or chemical
dependency services.

To account for differences in the types of hospital stay cases-known as
"case mix"- across metropolitan areas, we first classified each stay into
an All Patient Refined/Diagnosis Related Group (APR-DRG), using
information on length of stay, diagnoses, procedures, and the patients'
demographic characteristics. 7 Each APR-DRG is associated with a weight
that reflects the expected resources required to treat a typical privately
insured patient under age 65 in the same APR-DRG, relative to the average
resources required for that representative group. We used the APR-DRG
weight to adjust the hospital price for case mix. We excluded stays from
the analysis for which there was insufficient information on the claim to
assign a valid APR-DRG.

We adjusted hospital prices for differences in local costs of doing
business by applying Medicare's methodology of cost-adjusting hospital
payments. We applied the Medicare hospital wage index to 65 percent of the
price, which is Medicare's estimate of the wage-related component of the
costs, and applied the geographic adjustment factor to 9 percent of the
price, which is Medicare's estimate of the capital cost component. We
excluded hospital stays that had either extremely high or low prices,
because these high or low prices could distort average prices in an area.
We trimmed the cost- and service-mix-adjusted data for outliers using a
standard statistical

6

Price throughout this report includes both the amount the PPO pays
directly and the amount the enrollee is obligated to pay through
deductibles and coinsurance.

7

The APR-DRG software was provided to GAO by 3M Health Information Systems
in Murray, Utah.

Page 34 GAO-05-856 FEHBP Health Care Prices

                      Factors Affecting Health Care Prices

distribution (the lognormal) to remove observations more than three
standard deviations above or below the mean.

For our physician price analysis, we excluded laboratory, radiology,
anesthesiology, mental health and chemical dependency, unspecified
services, and services billed with certain modifiers and codes, because
these services were not uniformly classified or billed across the PPOs in
our study. This minimized the potential for aberrant billing practices in
some areas to inappropriately affect our results. We aggregated the prices
for the remaining services to the metropolitan area based on the
provider's place of service. To account for differences in the mix of
physician services across metropolitan areas, we applied the Medicare
methodology used to adjust physician payments. For each service, we
applied the appropriate relative value unit to reflect the resources
required to perform a specific service relative to an intermediate office
visit.

To adjust physician prices for geographic differences in the cost of doing
business, we applied the Medicare methodology used to adjust physician
payments. We applied the appropriate Geographic Practice Cost Index (GPCI)
to each physician payment. However, instead of applying the GPCIs used for
Medicare payments, which are often based on geographic areas larger than a
metropolitan area, we aggregated county-level cost indices to metropolitan
areas and then applied them. We trimmed the cost and service-mix-adjusted
data using the same method we used to trim our hospital price data,
namely, using the lognormal distribution to identify and remove
observations more than three standard deviations above or below the mean.

We identified factors that might explain geographic differences in
hospital and physician prices to use in our analysis, including measures
that approximated provider competition and health maintenance organization
(HMO) capitation. We also included measures sometimes associated with cost
shifting, measures of provider supply, per capita income, and hospital
ownership status. See table 11 for a list of factors and data sources.

  Table 11: Factors Included in Analysis of Hospital and Physician Price, 2001

Source of data to calculate Factor Measurement measurement

Competition Percent hospital beds of the two Verispan, L.L.C. largest
hospitals or hospital networksa

HMO capitation Percent of primary care physicians' InterStudy Publications
and compensation from capitationb U.S. Census Bureau

Cost shifting Percent of population enrolled in InterStudy Publications
and Medicare U.S. Census Bureau

Percent of population enrolled in InterStudy Publications and Medicaid
U.S. Census Bureau

Percent of population uninsuredc InterStudy Publications and

U.S. Census Bureau

Average Medicaid payment The Lewin Group, Centers for Medicare and
Medicaid Services, and U.S. Census Bureau

Supply of providers Hospital beds per capita Verispan, L.L.C. and U.S.
Census Bureau

Per capita income Population's real per capita Bureau of Economic incomed
Analysis and Centers for Medicare and Medicaid Services

Hospital ownership Percent beds in for-profit hospitals Verispan, L.L.C.
status

Census division Indicator of the presence or U.S. Census Bureau absence of
the metropolitan area in the census divisions

Source: GAO analysis of FEHBP data.

a

If a hospital was a member of more than one hospital network in a
metropolitan area, we averaged the percent of hospital beds in the two
largest hospitals or hospital networks across each combination of network
affiliation.

b

We estimated the percent of primary care physicians' compensation from
capitation in each metropolitan area by multiplying the percent of HMO
compensation to primary care physicians on a capitation basis by the
percent of the population enrolled in HMOs.

InterStudy Publications based the percent uninsured in a metropolitan area
on state uninsured rates.

d

We computed real income by dividing per capita income by the Centers for
Medicare and Medicaid Services hospital wage index for each metropolitan
area.

We measured health care provider competition by the percentage of hospital
beds in a metropolitan area that were owned by the two largest hospitals
or hospital networks. 8 While this value specifically measures
concentration in the hospital services market, we used this same variable
to explain both hospital and physician prices because physicians are often
aligned with health systems and hospital networks.

We measured HMO capitation by the percentage of physician compensation
that came from capitated payments. 9 Physicians generally tend to prefer
fee-for-service arrangements to capitation, which requires them to assume
the financial risk of treating patients whose costs may exceed the
capitation amount paid by the insurer. Therefore, we assumed that areas
that had a higher percentage of physicians paid under capitation had a
strong HMO presence with leverage to negotiate prices with physicians.

We examined our data for evidence of cost shifting-hospitals and
physicians charging higher prices for privately insured patients in order
to offset lower payments from other patients. We used several variables to
determine whether there was cost shifting. To estimate Medicare's
influence on prices, we analyzed the relationship between hospital and
physician prices, and the percentage of the metropolitan area's population
who were Medicare beneficiaries. To measure Medicaid's impact, we analyzed
the relationship between prices, and both the percentage of Medicaid
beneficiaries and the average Medicaid payment. Our measure of the average
Medicaid payment in an area was constructed by first identifying commonly
provided physician services and Medicaid payment rates for those services
using data reported by The Lewin Group, and then applying the GPCI and
relative value units unique to each service. 10 We then weighted each
Medicaid service using utilization estimates from the

8

If a hospital was a member of more than one hospital network in a
metropolitan area, we averaged the percent of hospital beds in the two
largest hospitals or hospital networks across each combination of network
affiliation.

9

We estimated the percent of primary care physicians' compensation by
multiplying the percent of HMO compensation to primary care physicians on
a capitation basis by the percent of the population enrolled in HMOs.

10

Some Medicaid payments for a given service varied depending on criteria
such as patient age, sex, provider specialty, and practice setting.
Researchers at The Lewin Group, who developed the statewide payments that
we used in estimating metropolitan area Medicaid prices, reported that
they focused on the payments most commonly made to a physician in private
practice.

state of California. Our analysis assumed that the relative difference in
payments across metropolitan areas for common procedures included in our
Medicaid price variable was similar to that for other procedures not
included in our analysis. We used the statewide percentages of people
without health insurance in an area to estimate the impact of
uncompensated or charity care on hospital and physician prices. 11

We included variables to account for the effect that the supply of health
services or health service providers had on hospital and physician prices.
Metropolitan areas with larger numbers of physicians or hospital beds per
capita may have lower prices because larger numbers of providers compete
for a given amount of business. In our analysis of hospital prices, we
used hospital beds per capita to estimate this effect, and in our
physician price analysis, we used the number of physicians per capita. We
also experimented with other measures of supply, in particular, teaching
hospital beds per capita and the number of physician specialists per
capita.

We included a measure of income because variations in income can affect
beneficiaries' ability to pay and thus may affect prices. Income data were
unavailable for FEHBP enrollees, so we used per capita income in the
metropolitan area. However, to account for geographic differences in
purchasing power, specifically that the cost of living was higher in some
metropolitan areas than others, we used the Centers for Medicare and
Medicaid Services wage index as a proxy for the cost of living and divided
this into dollar per capita income to calculate our income variable. We
also included hospital ownership status in our analysis. We included the
percent of hospital beds in for-profit hospitals and determined whether
this had an impact on hospital and physician prices. Finally, we included
dummy variables for each of the U.S. census divisions to account for
regional effects. 12

11

We were unable to find uninsured data at the metropolitan area level.
Therefore we used the number of uninsured from InterStudy Publications.
The estimates from InterStudy Publications of the uninsured are based on
state numbers.

12

In order for the regression to be estimated we had to omit one of the
census division dummies from our model: we chose to omit Census Division
9.

Page 38 GAO-05-856 FEHBP Health Care Prices

                              Analytical Approach

We conducted two analyses to examine the relationship between our price
variables and the factors described above. First, we grouped the
metropolitan areas into quartiles for each of the factors. 13 This enabled
us to then compare the average prices in metropolitan areas, for example,
with the highest levels of competition to those with the lowest. In
addition, we also conducted regression analyses to examine the effect of
each of the factors on price. To simplify the presentation of our results
in the body of the report, we presented only those factors that were
statistically significant in our regression analysis. 14

    Price Regression Analysis-Methods and Results

We used separate regression models to estimate the impact of our variables
on hospital and physician prices. To simplify the calculation of
independent variables' effects and to match the statistical distribution
assumption we made in our data trimming of prices, we used a log-linear
model: that is, we regressed the logarithm of price (hospital price and
physician price) on the levels of our independent variables. We were
concerned that our measures of provider supply-hospital beds per capita
and physicians per capita in the case of hospital and physician price,
respectively-were endogenous. For example, larger numbers of physicians
could lead to lower physician prices, but lower physician prices could
also make a metropolitan area less attractive to physicians and reduce
their number. In order to address this issue we used the method of
instrumental variables: a standard method to account for an endogenous
explanatory variable. 15 We also tested whether the HMO capitation
variable was endogenous and found that it was not.

Tables 12 and 13 show the results for estimating the determinants of
hospital and physician prices, respectively. The set of explanatory
variables was the same for both hospital and physician prices except that
we used hospital beds per capita and physicians per capita to measure
provider supply in the hospital and physician price models, respectively.
Our regression results for hospital price showed significant effects of

13

Quartiles divide the distribution of prices from lowest to highest into
four equal groups. The lowest quartile represents metropolitan areas
ranked in the lowest 25 percent of price, and the highest quartile
represents metropolitan areas ranked in the highest 25 percent of price.

14

We did not perform an analysis comparing prices inside and outside of
those census divisions that were significant in our regressions.

15

P. Kennedy, A Guide to Econometrics, 5th ed. (Cambridge, Mass. MIT Press,
2003), p. 188.

provider market share and managed care presence on prices: both of these
effects were consistent with the idea that raising market competitiveness
lowers prices. Our variable measuring the market share of the two largest
networks was positively related to price: that is, when the market became
more concentrated (less competitive), price tended to be higher. Also, our
HMO presence variable, the percentage of physician compensation from
capitation payments, was negatively associated with price: that is, less
HMO presence tended to increase price.

Table 12: Results for Hospital Price Regression-Estimated Effects of
Selected Factors on Hospital Prices in Metropolitan Areas, 2001

      Dependent variable is the logarithm of adjusted hospital stay pricea

       Factor Variable used to measure factor Parameter estimate t-value

Competition Percent hospital beds of the two largest hospitals or hospital
                             networks 0.1337 2.11**

HMO capitation Percent of primary care physicians' compensation from capitation
                                -0.3213 -2.22**

          Cost-shifting Percent of population uninsured -0.3621 -0.68

                      Average Medicaid payment 0.0026 1.58

            Percent of population enrolled in Medicaid -0.0538 -0.20

            Percent of population enrolled in Medicare -0.5267 -1.14

           Supply of providers Hospital beds per capita 21.5968 0.50

       Per capita income Population's real per capita income 0.0000 -0.52

 Hospital ownership status Percent of beds in for profit hospitals 0.0767 0.86

Dummy variable indicator Census Division 1 - New England 0.0625 0.78
showing the Census

Census Division 2 - Middle Atlantic -0.1158 -1.43

Division in which the

metropolitan area was Census Division 3 - East North Central -0.0572 -0.73
located

Census Division 4 - West North Central 0.0418 0.33

                Census Division 5 - South Atlantic -0.0258 -0.35

             Census Division 6 - East South Central -0.1845 -1.80*

              Census Division 7 - West South Central -0.1077 -1.14

                   Census Division 8 - Mountain -0.0428 -0.63

                          Census Division 9 - Pacificb

                           Intercept 8.8972 45.67***

                                 R-squared 0.25

                                Observations 228

                        *** significant at the 1% level

                         ** significant at the 5% level

                         * significant at the 10% level

Source: GAO analysis.

a

We adjusted hospital prices to remove the effect of geographic differences
in the costs of doing business (wages, rents, etc.) and differences in the
severity of illnesses and mix of diagnoses among metropolitan areas.

b

The Pacific Census Division was the excluded category. In order for the
regression model's parameters to be estimated, we needed to exclude one of
the Census Divisions.

Table 13: Results for Physician Price Regression-Estimated Effects of
Selected Factors on Physician Prices in Metropolitan Areas, 2001

Dependent variable is the logarithm of adjusted physician services pricea

       Factor Variable used to measure factor Parameter estimate t-value

Competition Percent hospital beds of the two largest hospitals or hospital
                            networks 0.1234 4.36***

HMO capitation Percent of primary care physicians' compensation from capitation
                                -0.1393 -2.24**

         Cost-shifting Percent of population uninsured -0.5328 -2.22**

                    Average Medicaid payment 0.0041 5.24***

             Percent of population enrolled in Medicaid 0.1081 0.91

             Percent of population enrolled in Medicare 0.0217 0.10

Hospital ownership status Percent of beds in for profit hospitals -0.0536 -1.34

       Per capita income Population's real per capita income 0.0000 0.00

Supply of providers Physicians per capita (physicians per 1000 population)
                                 -0.0002 -0.91

Dummy variable indicator Census Division 1 - New England -0.1112 -2.79***
showing the Census Census Division 2 - Middle Atlantic -0.0346 -1.01

Division in which the metropolitan area was Census Division 3 - East North
Central 0.0041 0.14

located Census Division 4 - West North Central 0.0120 0.32

                Census Division 5 - South Atlantic -0.0470 -1.58

              Census Division 6 - East South Central -0.0558 -1.61

             Census Division 7 - West South Central 0.0947 3.24***

                   Census Division 8 - Mountain -0.0240 -0.77

                          Census Division 9 - Pacificb

                           Intercept 3.7808 35.48***

                                 R-squared 0.46

                                Observations 315

                        *** significant at the 1% level

                         ** significant at the 5% level

                         * significant at the 10% level

Source: GAO analysis.

a

We adjusted physician prices to remove the effect of geographic
differences in the costs of doing business (wages, rents, etc.) and
differences in the mix of services among metropolitan areas.

b

The Pacific Census Division was the excluded category. In order for the
regression model's parameters to be estimated, we needed to exclude one of
the Census Divisions.

Our measures of cost-shifting effects were mostly not significant and none
of the results supported the claim that more Medicaid enrollees, lower
Medicaid payments, more Medicare enrollees, or more uninsured people were
associated with higher hospital or physician prices. Ideally, we would
have included an indicator of Medicare price levels for each area, such as
the wage index or the GPCI. However, we did not include these as separate
explanatory variables in the regression models because we had used the
wage index and the GPCI to adjust the hospital and physician prices,
respectively, for differences in the cost of doing business in different
areas. Therefore, our sole measure of the impact of the Medicare program
on prices was the percent of the population who were Medicare
beneficiaries. In the physician price regression, the average Medicaid
payment was significant. However, Medicaid payments were positively
associated with prices, which was inconsistent with the negative
association we would have expected if cost shifting were occurring. In the
physician price analysis, the percent of people uninsured was
significantly related to price and the result showed that where there were
more uninsured people, prices were actually lower, rather than higher, as
would have been predicted by the cost-shifting hypothesis.

Our inclusion of the set of census division dummy variables allowed us to
measure factors affecting price that were due simply to location and that
were not accounted for by the other variables included in the model. In
both price regression models, we ran an F-test that showed that the set of
census division dummy variables was jointly significant.

In the cases where our explanatory variables in the regression were
significant, we calculated the significant variables' impact on prices by
using our regression results to calculate the percent change in price for
a given increase in the explanatory variable. To do this, we simulated the
effect of increasing the significant explanatory variable from its average
in its lowest quartile to its average in its highest quartile, while
controlling for other factors. This was accomplished using the following
steps: (1) we calculated the average value of the statistically
significant explanatory variable for its lowest quartile, and input that
value into our estimated regression equation to calculate price, (2) we
calculated the average value of the key explanatory variable in its
highest quartile, and used that value in our estimated regression model to
calculate price again, and (3) we calculated the percent difference in
price using the results from (1) and (2). See table 14.

Table 14: Effects of Changes in Explanatory Variables on Prices
                                       Percent impact on    Percent impact on 
Significant explanatory variable     physician price        hospital price 
Percent hospital beds of the two largest                 
hospitals or hospital networks                      6.64              7.62 
Percent of primary care physicians'                      
compensation from capitation                       -3.31             -7.17 
Average Medicaid payment                            9.69                 a 
Percent of population uninsured                    -6.05                 a 

Source: GAO analysis.

Note: The percent impact is the change in price that would follow an
increase in the explanatory variable from its average value in its lowest
quartile to its average value in its highest quartile.

a

The average cost-adjusted Medicaid fee and the percent uninsured
explanatory variables were not statistically significant in the hospital
price regression.

We also tested and opted not to include other variables in our regression:
specifically, we tried to explain price variations by including the
percent of the labor force in the metropolitan area covered by a labor
union contract; the mortality rate for persons aged more than one but less
than 65 years in the metropolitan area-a proxy for health status; and the
effect of certificate-of-need laws. 16 We also used the number of teaching
hospital beds per capita to see if this had an independent effect on
price, separate from the effects of supply. We included this variable
because it was possible that more teaching hospital beds in a metropolitan
area might indicate more cutting-edge and higher quality services, or
teaching hospitals might conduct more tests or services, which might in
turn affect prices. We ultimately excluded labor union, mortality rates,
certificate-of-need laws, and teaching hospital variables from our
explanatory variables because they were not the focus of our analysis,
they were not statistically significant, and their inclusion did not
affect the significance of most of the other explanatory variables in the
model.

A certificate-of-need law generally requires that a hospital or nursing
home obtain approval from the state in which it is located before hospital
construction or capital improvements occur.

Page 44 GAO-05-856 FEHBP Health Care Prices

                               Spending Analysis

                       Appendix I: Scope and Methodology

To determine average total spending per enrollee in each metropolitan
area, we summed all payments for each enrollee, assigned enrollees to
their metropolitan areas of residence, and then calculated the average for
each metropolitan area. We adjusted spending service categories for
geographic input costs, removed outliers, and accounted for differences in
the age and sex distributions across metropolitan areas. After applying
our eligibility criteria and removing outliers, we had about 2.1 million
enrollees in our study.

We accounted for geographic differences in the costs of providing hospital
inpatient, 17 hospital outpatient, home health, rehabilitation, skilled
nursing facility, other outpatient, and ambulatory surgery center services
by first summing the payments per enrollee by service categories and then
applying Medicare's hospital wage index to the labor-related portion of
the total payment for each type of service. This approach is similar to
the methodology used by Medicare to adjust such provider payments. 18

We accounted for geographic differences in the cost of providing physician
services using a different methodology, but one that generally follows the
basic methodology used by Medicare. We applied the appropriate GPCIs to
the total physician payments. 19 However, our method varied slightly from
Medicare's in that instead of applying the GPCIs at the carrier/locality
level, we calculated separate cost indices for each metropolitan area. 20

We excluded enrollees with high total health care spending because
spending for those enrollees could distort average spending in an area
with low enrollment. To identify enrollees with high spending, we used a

17

Medicare adjusts hospital inpatient payments for labor and capital-related
variations in costs. In our study, we applied labor and capital
adjustments to the hospital inpatient portion of spending and to hospital
inpatient price.

18

We excluded mental health, chemical dependency services, and
pharmaceuticals from our spending analysis.

19

There are three GPCIs reflecting the cost of three different types of
inputs to physician services: physician work, physician practice expenses,
and expenses for physician liability insurance. Each GPCI is used to
adjust for the price level for related inputs in the local market where
the service is furnished.

20

There are 89 carrier/locality regions nationwide and 331 metropolitan
areas in the 50 states and District of Columbia. Thus, a carrier/locality
area is, on average, much larger than a metropolitan area. We used
county-level data for the GPCIs and aggregated those data to the
metropolitan area level.

  Decomposing Spending Variation into Price and Utilization Effects

standard statistical distribution (the lognormal). We removed enrollees
from this analysis whose spending was at least three standard deviations
above the mean.

We adjusted spending for the age and sex distribution of each metropolitan
area's population. To do this, we calculated the average age-and
sex-specific spending rates of all 232 metropolitan areas combined, and
applied these averages to the actual age and sex distribution in each
metropolitan area. This yielded an "expected" spending rate for each
metropolitan area: the spending in that metropolitan area if it had the
study average spending rate, given the age and sex distribution of that
metropolitan area's population. We then calculated the ratio of actual
cost-adjusted spending to expected cost-adjusted spending. This yielded an
index of how much higher or lower spending in the specific metropolitan
area was from what would be expected if it had average spending rates,
given its age and sex composition. An index value greater than 1.00
implies spending was higher than expected and an index value less than
1.00 implies spending was lower than expected.

We estimated the relative contribution of price and utilization variation
to spending variation in 232 metropolitan areas. To do this, we first
computed measures of price, spending, and utilization for hospital and
physician services. We then analyzed price and utilization differences
between metropolitan areas in the highest and lowest spending quartiles to
decompose spending into its component parts.

We used the same method to adjust hospital and physician spending as we
did for total spending. That is, we used the appropriate Medicare cost
adjustments and adjustments for age and sex. To estimate hospital and
physician prices, we used prices we had computed from our price analysis
for the same 232 metropolitan areas.

We defined hospital utilization as the count of hospital stays. We
excluded mental health and chemical dependency stays, and other nonacute
hospital stays, such as nursing home and rehabilitation services, in each
of the 232 metropolitan areas. Our measure of physician utilization was
simply the count of services provided by physicians, excluding pathology,
radiology, anesthesia, and psychiatric services. We aggregated the data
for service use per enrollee up to the metropolitan area, and we then
adjusted these data in a similar way to the spending data: that is, we
adjusted for age and sex composition of the area by calculating the ratio
of actual utilization to expected utilization. We calculated the physician
and

                                Data Reliability

                       Appendix I: Scope and Methodology

hospital utilization indices using the 232 metropolitan areas as the
population basis.

For both hospital and physician services, we compared the simple average
adjusted spending per enrollee in the highest spending quartile
metropolitan areas with the lowest spending quartile metropolitan areas.
Similarly, we compared the average adjusted price and the average adjusted
utilization per enrollee in the highest versus the lowest spending
quartile. The proportional difference in spending between the highest and
lowest quartiles can be divided into (1) the proportional difference in
price between the highest and lowest spending quartiles, and (2) the
proportional difference in utilization between the highest and lowest
spending quartiles. In order to divide the variation in spending between
price and utilization differences, we compared the values of

(1) to (2) above. We estimated the relative contribution of physician
price and utilization to spending by analyzing the percentage difference
between the average prices and utilization in the highest and lowest
spending quartiles, relative to the summed total of the percentage
differences, as shown in table 9.

We used multiple data sources for this report. We obtained 2001 health
care claims data from several PPOs participating in FEHBP. In addition, we
obtained data describing characteristics of metropolitan areas from
several other sources. See table 11. We determined that the data were
sufficiently reliable to address the study objectives.

We verified that our claims data were sufficiently reliable and unbiased
in several ways. First, we interviewed staff from each of the FEHBP PPOs
participating in the study to obtain an understanding of the completeness
and accuracy of the data we had requested. Upon receipt of the data from
the PPOs, we conducted numerous tests and edit checks to ensure that our
data were complete and accurate: we reviewed the documentation that
accompanied the data; we checked that essential elements of the data were
populated with credible values; we excluded enrollees and claims records
that did not match study eligibility criteria; and we examined the
internal consistency and validity of the data, coordinating with any PPO
that submitted data that required clarification or resubmission of
corrected data. To test the validity of the hospital location variable
from our claims data, we examined the proportion of hospital stays that
occurred outside of the enrollee's state of residence or an adjacent
state. For one metropolitan area, we conducted a sensitivity analysis to
quantify the impact on our price estimate of removing the admissions from
enrollees in another state. We concluded that our location data were
sufficiently reliable for the purposes of our study.

Ultimately, we excluded 12 of the 331 metropolitan areas for one of two
reasons. First, in some metropolitan areas, some PPOs made additional
"reconciliation" payments that were not recorded in the claims system, and
price estimates would have been understated in these areas. Second, if a
disproportionate number of enrollees traveled into a metropolitan area to
receive care, we excluded the metropolitan area. We also excluded some
hospital stays and physician services from our hospital and physician
price estimates, respectively, either because there were insufficient data
to case-mix adjust these services or because hospital or physician billing
conventions were inconsistent across metropolitan areas for those
services.

We verified that the data describing market forces and other factors in a
metropolitan area were sufficiently reliable and unbiased using methods
similar to those we used to verify the claims data. We discussed data
quality issues with data suppliers, reviewed the suppliers' documentation
and internal data testing, and conducted our own tests for data
completeness and credibility. Some limitations came to light through these
processes. First, because direct estimates of uninsured rates were
unavailable for all metropolitan areas in the study, we used the
InterStudy Publications' estimates of the uninsured for metropolitan
areas, which were based on statewide uninsured estimates. Similarly,
metropolitan area specific Medicaid payment rates were not available, and
Medicaid utilization rates were not available to weight the average of
Medicaid payments in metropolitan areas. Consequently, we used statewide
payment and utilization estimates for California's Medicaid program, which
were reported by The Lewin Group. 21

We performed our work from September 2002 through July 2005 in accordance
with generally accepted government auditing standards.

Where metropolitan areas overlapped several states, we prorated state
Medicaid payment rates based on U.S. census estimates of Medicaid
enrollment in each component county of the metropolitan area. We used
utilization rates in California to weight the average Medicaid payment in
each metropolitan area because utilization rates were not readily
available for any other state.

Page 48 GAO-05-856 FEHBP Health Care Prices

Appendix II: FEHBP PPO Adjusted Hospital Prices in U.S. Metropolitan Areas, 2001

The adjusted hospital price indices based on FEHBP PPO payments for
hospital stays in 232 metropolitan areas are presented below ranked in
order from highest to lowest price.

                  Page 49 GAO-05-856 FEHBP Health Care Prices

Table 15: Ranking of Metropolitan Areas by Adjusted Hospital Prices, 2001
Rank  Metropolitan area            Predominant statea    Adjusted hospital 
                                                                  price index 
1     b                            b                                 1.829 
2     Dover                        DE                                1.680 
3     Biloxi-Gulfport-Pascagoula   MS                                1.591 
4     St. Joseph                   MO                                1.578 
5     Milwaukee-Waukesha           WI                                1.568 
6     Salinas                      CA                                1.499 
7     Buffalo-Niagara Falls        NY                                1.451 
8     Grand Junction               CO                                1.431 
9     b                            b                                 1.419 
10    La Crosse, WI-MN             WI                                1.385 
11    Wichita                      KS                                1.379 
12    Manchester                   NH                                1.365 
13    Bakersfield                  CA                                1.361 
14    Sioux Falls                  SD                                1.357 
15    Bangor                       ME                                1.340 
16    Owensboro                    KY                                1.326 
17    Fort Walton Beach            FL                                1.322 
18    Portsmouth-Rochester, NH-ME  NH                                1.318 
19    Lakeland-Winter Haven        FL                                1.310 
20    South Bend                   IN                                1.285 
21    Honolulu                     HI                                1.277 
22    Albany                       GA                                1.270 
23    Oklahoma City                OK                                1.270 
24    Nashua                       NH                                1.266 
25    Olympia                      WA                                1.262 
26    Omaha, NE-IA                 NE                                1.256 
27    Duluth-Superior, MN-WI       MN                                1.252 
28    Rapid City                   SD                                1.249 
29    Terre Haute                  IN                                1.244 
30    Charleston                   WV                                1.243 
31    Wilmington-Newark, DE-MD     DE                                1.239 
32    Lynchburg                    VA                                1.237 

                  Page 50 GAO-05-856 FEHBP Health Care Prices

Rank Metropolitan area             Predominant statea    Adjusted hospital 
                                                                  price index 
33   Billings                      MT                                1.235 
34   b                             b                                 1.233 
35   Myrtle Beach                  SC                                1.231 
36   Columbia                      MO                                1.230 
37   Topeka                        KS                                1.225 
38   Evansville-Henderson, IN-KY   IN                                1.193 
39   Lawton                        OK                                1.192 
40   Missoula                      MT                                1.187 
41   Daytona Beach                 FL                                1.186 
42   Medford-Ashland               OR                                1.177 
43   Roanoke                       VA                                1.176 
44   Bismarck                      ND                                1.173 
45   Charleston-North Charleston   SC                                1.161 
46   Portland                      ME                                1.158 
47   Sioux City, IA-NE             IA                                1.157 
48   Jackson                       MS                                1.151 
49   Hattiesburg                   MS                                1.148 
50   Provo-Orem                    UT                                1.147 
51   Fort Collins-Loveland         CO                                1.144 
52   Boise City                    ID                                1.138 
53   Salt Lake City-Ogden          UT                                1.137 
54   Enid                          OK                                1.137 
55   Gainesville                   FL                                1.136 
56   San Antonio                   TX                                1.132 
57   Parkersburg-Marietta, WV-OH   WV                                1.127 
58   Boston, MA-NH                 MA                                1.123 
59   Memphis, TN-AR-MS             TN                                1.117 
60   Cedar Rapids                  IA                                1.113 
61   Jackson                       TN                                1.111 
62   Houston                       TX                                1.103 
63   Huntington-Ashland, WV-KY-OH  WV                                1.102 
64   Fayetteville                  NC                                1.102 
65   Springfield                   MA                                1.101 
66   Melbourne-Titusville-Palm Bay FL                                1.099 
67   Portland-Vancouver, OR-WA     OR                                1.098 
68   Iowa City                     IA                                1.092 
69   Florence                      SC                                1.087 

                  Page 51 GAO-05-856 FEHBP Health Care Prices

Rank Metropolitan area                   Predominant statea       Adjusted 
                                                               hospital price 
                                                                        index 
70   Fort Pierce-Port St. Lucie          FL                          1.086 
71   Tacoma                              WA                          1.086 
72   Grand Forks, ND-MN                  ND                          1.083 
73   Lubbock                             TX                          1.078 
74   New Haven-Meriden                   CT                          1.071 
75   Great Falls                         MT                          1.068 
76   Columbus, GA-AL                     GA                          1.065 
77   Fort Myers-Cape Coral               FL                          1.061 
78   Fargo-Moorhead, ND-MN               ND                          1.061 
79   Des Moines                          IA                          1.060 
80   Minneapolis-St. Paul, MN-WI         MN                          1.057 
81   Fort Smith, AR-OK                   AR                          1.052 
82   Bremerton                           WA                          1.048 
83   Richmond-Petersburg                 VA                          1.041 
84   Lincoln                             NE                          1.040 
85   Phoenix-Mesa                        AZ                          1.039 
86   Laredo                              TX                          1.033 
87   Salem                               OR                          1.031 
88   Bloomington                         IN                          1.029 
89   Lexington                           KY                          1.029 
90   Reading                             PA                          1.028 
91   Augusta-Aiken, GA-SC                GA                          1.027 
92   Fort Worth-Arlington                TX                          1.025 
93   b                                   b                           1.024 
94   Austin-San Marcos                   TX                          1.019 
95   Asheville                           NC                          1.016 
96   Wichita Falls                       TX                          1.015 
97   Little Rock-North Little Rock       AR                          1.015 
98   Las Vegas, NV-AZ                    NV                          1.013 
99   McAllen-Edinburg-Mission            TX                          1.011 
100  Jonesboro                           AR                          1.006 
101  Miami                               FL                          1.006 
102  Charlotte-Gastonia-Rock Hill, NC-SC NC                          1.002 
103  Orlando                             FL                          1.001 
104  Seattle-Bellevue-Everett            WA                          0.993 
105  Pensacola                           FL                          0.986 
106  Odessa-Midland                      TX                          0.983 

                  Page 52 GAO-05-856 FEHBP Health Care Prices

Rank Metropolitan area                        Predominant         Adjusted 
                                                   statea            hospital 
                                                                  price index 
107  Lansing-East Lansing                  MI                        0.983 
108  Johnson City-Kingsport-Bristol, TN-VA TN                        0.981 
109  Charlottesville                       VA                        0.980 
110  Knoxville                             TN                        0.978 
111  Fayetteville-Springdale-Rogers        AR                        0.978 
112  Clarksville-Hopkinsville, TN-KY       TN                        0.975 
113  Dayton-Springfield                    OH                        0.974 
114  San Angelo                            TX                        0.971 
115  Tucson                                AZ                        0.970 
116  Tampa-St. Petersburg-Clearwater       FL                        0.967 
117  Ann Arbor                             MI                        0.965 
118  Scranton-Wilkes-Barre-Hazleton        PA                        0.964 
119  Eugene-Springfield                    OR                        0.964 
120  Atlantic-Cape May                     NJ                        0.963 
121  Anchorage                             AK                        0.962 
122  Bridgeport                            CT                        0.961 
123  San Francisco                         CA                        0.960 
124  Panama City                           FL                        0.957 
125  Baltimore                             MD                        0.953 
126  Greenville-Spartanburg-Anderson       SC                        0.950 
127  Trenton                               NJ                        0.946 
128  Redding                               CA                        0.946 
129  York                                  PA                        0.942 
130  Amarillo                              TX                        0.941 
131  Lawrence, MA-NH                       MA                        0.933 
132  Springfield                           MO                        0.932 
133  Washington, DC-MD-VA-WV               VA                        0.931 
134  Las Cruces                            NM                        0.930 
135  Indianapolis                          IN                        0.928 
136  Gary                                  IN                        0.927 
137  Detroit                               MI                        0.927 
138  Tulsa                                 OK                        0.921 
139  Greensboro-Winston-Salem-High Point   NC                        0.919 
140  Nashville                             TN                        0.914 
141  Santa Fe                              NM                        0.912 
142  Raleigh-Durham-Chapel Hill            NC                        0.911 
143  Grand Rapids-Muskegon-Holland         MI                        0.906 

                  Page 53 GAO-05-856 FEHBP Health Care Prices

Rank Metropolitan area                   Predominant statea       Adjusted 
                                                               hospital price 
                                                                        index 
144  Baton Rouge                         LA                          0.905 
145  Columbia                            SC                          0.900 
146  Middlesex-Somerset-Hunterdon        NJ                          0.899 
147  Sarasota-Bradenton                  FL                          0.896 
148  Cumberland, MD-WV                   MD                          0.895 
149  Waterbury                           CT                          0.894 
150  Atlanta                             GA                          0.891 
151  b                                   b                           0.889 
152  Macon                               GA                          0.888 
153  Birmingham                          AL                          0.886 
154  Harrisburg-Lebanon-Carlisle         PA                          0.885 
155  Sacramento                          CA                          0.884 
156  Fort Wayne                          IN                          0.883 
157  New London-Norwich, CT-RI           CT                          0.876 
158  Toledo                              OH                          0.875 
159  New Orleans                         LA                          0.873 
160  Florence                            AL                          0.870 
161  West Palm Beach-Boca Raton          FL                          0.870 
162  Mobile                              AL                          0.870 
163  Columbus                            OH                          0.868 
164  Hartford                            CT                          0.867 
165  Fort Lauderdale                     FL                          0.866 
166  Corpus Christi                      TX                          0.866 
167  Savannah                            GA                          0.865 
168  Monroe                              LA                          0.864 
169  Montgomery                          AL                          0.864 
170  Houma                               LA                          0.864 
171  Galveston-Texas City                TX                          0.862 
172  Dallas                              TX                          0.861 
173  Richland-Kennewick-Pasco            WA                          0.861 
174  Norfolk-Virginia Beach-Newport      VA                          0.861 
        News, VA-NC                                            
175  Pittsburgh                          PA                          0.861 
176  Bergen-Passaic                      NJ                          0.860 
177  Denver                              CO                          0.859 
178  Bryan-College Station               TX                          0.859 
179  Colorado Springs                    CO                          0.859 
180  Monmouth-Ocean                      NJ                          0.859 

                  Page 54 GAO-05-856 FEHBP Health Care Prices

Rank Metropolitan area                   Predominant statea       Adjusted 
                                                               hospital price 
                                                                        index 
181  Reno                                NV                          0.858 
182  Texarkana, TX-Texarkana             TX                          0.857 
183  Punta Gorda                         FL                          0.853 
184  Waco                                TX                          0.853 
185  Flint                               MI                          0.847 
186  Kansas City, MO-KS                  MO                          0.838 
187  Oakland                             CA                          0.836 
188  Killeen-Temple                      TX                          0.830 
189  Tuscaloosa                          AL                          0.826 
190  Philadelphia, PA-NJ                 PA                          0.820 
191  Chattanooga, TN-GA                  TN                          0.814 
192  Providence-Fall River-Warwick,      RI                          0.813 
        RI-MA                                                  
193  Sherman-Denison                     TX                          0.812 
194  Kalamazoo-Battle Creek              MI                          0.808 
195  Jacksonville                        FL                          0.807 
196  Boulder-Longmont                    CO                          0.804 
197  Cleveland-Lorain-Elyria             OH                          0.803 
198  Shreveport-Bossier City             LA                          0.799 
199  Syracuse                            NY                          0.797 
200  Wilmington                          NC                          0.794 
201  Erie                                PA                          0.790 
202  Jersey City                         NJ                          0.787 
203  Yakima                              WA                          0.786 
204  Los Angeles-Long Beach              CA                          0.785 
205  Chicago                             IL                          0.785 
206  Huntsville                          AL                          0.780 
207  Hagerstown                          MD                          0.779 
208  Johnstown                           PA                          0.777 
209  Cincinnati, OH-KY-IN                OH                          0.776 
210  Lafayette                           LA                          0.772 
211  Gadsden                             AL                          0.769 
212  Lake Charles                        LA                          0.764 
213  Louisville, KY-IN                   KY                          0.761 
214  Allentown-Bethlehem-Easton          PA                          0.754 
215  Spokane                             WA                          0.746 
216  Athens                              GA                          0.745 
217  Albuquerque                         NM                          0.743 

Rank Metropolitan area        Predominant statea   Adjusted hospital price 
                                                                        index 
218  Nassau-Suffolk                   NY                             0.740 
219  Dothan                           AL                             0.728 
220  San Diego                        CA                             0.727 
221  Riverside-San Bernardino         CA                             0.727 
222  Newark                           NJ                             0.725 
223  Saginaw-Bay City-Midland         MI                             0.712 
224  Anniston                         AL                             0.709 
225  Decatur                          AL                             0.709 
226  Altoona                          PA                             0.678 
227  New York                         NY                             0.676 
228  Newburgh, NY-PA                  NY                             0.675 
229  Albany-Schenectady-Troy          NY                             0.674 
230  Ventura                          CA                             0.635 
231  Pueblo                           CO                             0.609 
232  Orange County                    CA                             0.515 

Source: GAO analysis of FEHBP data.

Note: We adjusted hospital prices to remove the effect of geographic
differences in the costs of doing business (wages, rents, etc.) and
differences in the severity of illnesses and mix of diagnoses among
metropolitan areas. We converted hospital prices to an index by dividing
the average price for a hospital stay in a metropolitan area by the
average price for all hospital stays in 232 metropolitan areas. The
average hospital price index value is 1.00.

a

Some metropolitan areas spanned more than one state. In those cases, we
assigned the state that contained the largest proportion of the population
of the metropolitan area.

b

Metropolitan area name withheld because there was only one hospital in the
metropolitan area and the data were proprietary.

Appendix III: FEHBP PPO Adjusted Physician Prices in U.S. Metropolitan Areas,
2001

The adjusted physician price indices based on FEHBP PPO payments for
physician services in 319 metropolitan areas are presented below ranked in
order from highest to lowest price.

                  Page 56 GAO-05-856 FEHBP Health Care Prices

Table 16: Ranking of Metropolitan Areas by Adjusted Physician Prices, 2001
Rank  Metropolitan area              Predominant statea Adjusted physician 
                                                                  price index 
1     La Crosse, WI-MN                       WI                      1.484 
2     Wausau                                 WI                      1.459 
3     Eau Claire                             WI                      1.418 
4     Madison                                WI                      1.414 
5     Jonesboro                              AR                      1.348 
6     Janesville-Beloit                      WI                      1.324 
7     Great Falls                            MT                      1.287 
8     Green Bay                              WI                      1.279 
9     Appleton-Oshkosh-Neenah                WI                      1.267 
10    Racine                                 WI                      1.239 
11    Sheboygan                              WI                      1.231 
12    Billings                               MT                      1.230 
13    Wichita Falls                          TX                      1.224 
14    Anchorage                              AK                      1.221 
15    Corvallis                              OR                      1.220 
16    Milwaukee-Waukesha                     WI                      1.217 
17    Jacksonville                           NC                      1.216 
18    Kenosha                                WI                      1.213 
19    Fayetteville-Springdale-Rogers         AR                      1.206 
20    Texarkana, TX-Texarkana                TX                      1.204 
21    Fort Smith, AR-OK                      AR                      1.202 
22    Monroe                                 LA                      1.198 
23    Pine Bluff                             AR                      1.194 
24    Missoula                               MT                      1.190 
25    Salem                                  OR                      1.187 
26    St. Cloud                              MN                      1.187 
27    Eugene-Springfield                     OR                      1.184 
28    Duluth-Superior, MN-WI                 MN                      1.178 
29    Medford-Ashland                        OR                      1.165 
30    Alexandria                             LA                      1.162 
31    Houma                                  LA                      1.159 
32    Sherman-Denison                        TX                      1.159 

                  Page 57 GAO-05-856 FEHBP Health Care Prices

Rank Metropolitan area               Predominant statea Adjusted physician 
                                                                  price index 
33   Wheeling, WV-OH                         WV                      1.157 
34   Shreveport-Bossier City         LA                              1.145 
35   Grand Junction                          CO                      1.144 
36   Omaha, NE-IA                    NE                              1.143 
37   Bryan-College Station           TX                              1.143 
38   Little Rock-North Little Rock   AR                              1.142 
39   Rocky Mount                     NC                              1.136 
40   Springfield                             MO                      1.135 
41   Lafayette                       LA                              1.134 
42   Lubbock                         TX                              1.129 
43   San Angelo                      TX                              1.129 
44   Lincoln                         NE                              1.129 
45   Pueblo                                  CO                      1.128 
46   Abilene                         TX                              1.121 
47   Hattiesburg                             MS                      1.119 
48   Kankakee                        IL                              1.119 
49   Fayetteville                    NC                              1.111 
50   Parkersburg-Marietta, WV-OH             WV                      1.111 
51   Jackson                         TN                              1.106 
52   Charleston                              WV                      1.105 
53   Longview-Marshall               TX                              1.103 
54   Sioux City, IA-NE               IA                              1.101 
55   Clarksville-Hopkinsville, TN-KY TN                              1.101 
56   Albany                          GA                              1.098 
57   Bismarck                        ND                              1.097 
58   Lawrence                        KS                              1.096 
59   Panama City                     FL                              1.096 
60   Rapid City                      SD                              1.096 
61   Lewiston-Auburn                         ME                      1.096 
62   Bangor                                  ME                      1.095 
63   Muncie                          IN                              1.093 
64   Baton Rouge                     LA                              1.093 
65   Grand Forks, ND-MN              ND                              1.091 
66   Portland-Vancouver, OR-WA               OR                      1.085 
67   Huntington-Ashland, WV-KY-OH            WV                      1.085 
68   Elmira                          NY                              1.084 
69   Tyler                           TX                              1.084 

                  Page 58 GAO-05-856 FEHBP Health Care Prices

Rank Metropolitan area                  Predominant statea        Adjusted 
                                                              physician price 
                                                                        index 
70   Pocatello                          ID                           1.083 
71   Dubuque                            IA                           1.082 
72   Macon                              GA                           1.081 
73   Terre Haute                        IN                           1.079 
74   Goldsboro                          NC                           1.078 
75   Greenville                         NC                           1.077 
76   Columbus, GA-AL                    GA                           1.075 
77   McAllen-Edinburg-Mission           TX                           1.074 
78   Brownsville-Harlingen-San Benito   TX                           1.072 
79   Glens Falls                        NY                           1.072 
80   Johnson City-Kingsport-Bristol,    TN                           1.072 
        TN-VA                                                 
81   Laredo                             TX                           1.072 
82   Waco                               TX                           1.069 
83   Cedar Rapids                       IA                           1.067 
84   Boise City                         ID                           1.066 
85   Greeley                            CO                           1.065 
86   Fort Walton Beach                  FL                           1.065 
87   Lawton                             OK                           1.064 
88   Iowa City                          IA                           1.063 
89   Hickory-Morganton-Lenoir           NC                           1.062 
90   Asheville                          NC                           1.060 
91   Lake Charles                       LA                           1.059 
92   Sioux Falls                        SD                           1.057 
93   Enid                               OK                           1.057 
94   Portland                           ME                           1.055 
95   Pensacola                          FL                           1.051 
96   Yuma                               AZ                           1.051 
97   Fort Myers-Cape Coral              FL                           1.050 
98   Joplin                             MO                           1.049 
99   South Bend                         IN                           1.049 
100  Fort Wayne                         IN                           1.049 
101  Lafayette                          IN                           1.046 
102  St. Joseph                         MO                           1.046 
103  Biloxi-Gulfport-Pascagoula         MS                           1.045 
104  Auburn-Opelika                     AL                           1.044 
105  Fort Worth-Arlington               TX                           1.043 
106  Odessa-Midland                     TX                           1.043 

                  Page 59 GAO-05-856 FEHBP Health Care Prices

Rank Metropolitan area                  Predominant statea        Adjusted 
                                                              physician price 
                                                                        index 
107  Fargo-Moorhead, ND-MN              ND                           1.042 
108  Flagstaff, AZ-UT                   AZ                           1.042 
109  Savannah                           GA                           1.041 
110  Knoxville                          TN                           1.041 
111  Colorado Springs                   CO                           1.040 
112  Elkhart-Goshen                     IN                           1.038 
113  Las Cruces                         NM                           1.037 
114  Evansville-Henderson, IN-KY        IN                           1.036 
115  Beaumont-Port Arthur               TX                           1.034 
116  Columbia                           MO                           1.034 
117  Topeka                             KS                           1.034 
118  Sharon                             PA                           1.034 
119  Fort Collins-Loveland              CO                           1.033 
120  Killeen-Temple                     TX                           1.033 
121  Owensboro                          KY                           1.032 
122  Sumter                             SC                           1.032 
123  Corpus Christi                     TX                           1.030 
124  Yuba City                          CA                           1.029 
125  Victoria                           TX                           1.029 
126  Jackson                            MS                           1.028 
127  Waterloo-Cedar Falls               IA                           1.027 
128  New Orleans                        LA                           1.026 
129  Yakima                             WA                           1.024 
130  Dallas                             TX                           1.022 
131  Austin-San Marcos                  TX                           1.021 
132  Utica-Rome                         NY                           1.021 
133  Portsmouth-Rochester, NH-ME        NH                           1.018 
134  Brazoria                           TX                           1.017 
135  Memphis, TN-AR-MS                  TN                           1.016 
136  Charlotte-Gastonia-Rock Hill,      NC                           1.016 
        NC-SC                                                 
137  Wichita                            KS                           1.013 
138  Lima                               OH                           1.013 
139  Amarillo                           TX                           1.011 
140  Minneapolis-St. Paul, MN-WI        MN                           1.011 
141  Yolo                               CA                           1.010 
142  Dothan                             AL                           1.010 
143  Tallahassee                        FL                           1.009 

                  Page 60 GAO-05-856 FEHBP Health Care Prices

Rank Metropolitan area       Predominant statea   Adjusted physician price 
                                                                        index 
144  Des Moines                      IA                              1.009 
145  El Paso                         TX                              1.008 
146  Atlanta                         GA                              1.008 
147  San Antonio                     TX                              1.006 
148  Bloomington                     IN                              1.006 
149  Syracuse                        NY                              1.006 
150  Redding                         CA                              1.005 
151  Albany-Schenectady-Troy         NY                              1.005 
152  Altoona                         PA                              1.003 
153  Indianapolis                    IN                              1.002 
154  Lakeland-Winter Haven           FL                              1.001 
155  Roanoke                         VA                              1.001 
156  Modesto                         CA                              0.999 
157  Punta Gorda                     FL                              0.999 
158  Augusta-Aiken, GA-SC            GA                              0.998 
159  Mansfield                       OH                              0.998 
160  Ocala                           FL                              0.997 
161  Athens                          GA                              0.997 
162  Anniston                        AL                              0.994 
163  Chico-Paradise                  CA                              0.994 
164  Burlington                      VT                              0.994 
165  Tuscaloosa                      AL                              0.993 
166  Binghamton                      NY                              0.992 
167  Florence                        SC                              0.992 
168  Boulder-Longmont                CO                              0.991 
169  Naples                          FL                              0.991 
170  Spokane                         WA                              0.991 
171  Albuquerque                     NM                              0.991 
172  Merced                          CA                              0.991 
173  Chicago                         IL                              0.990 
174  Tulsa                           OK                              0.988 
175  Gainesville                     FL                              0.983 
176  Johnstown                       PA                              0.983 
177  Denver                          CO                              0.983 
178  Wilmington                      NC                              0.982 
179  Chattanooga, TN-GA              TN                              0.981 
180  Lexington                       KY                              0.980 

                  Page 61 GAO-05-856 FEHBP Health Care Prices

Rank Metropolitan area                       Predominant          Adjusted 
                                                  statea            physician 
                                                                  price index 
181  Tacoma                               WA                         0.979 
182  Galveston-Texas City                 TX                         0.979 
183  Norfolk-Virginia Beach-Newport News, VA                         0.975 
        VA-NC                                                  
184  Houston                              TX                         0.975 
185  Gary                                 IN                         0.974 
186  Oklahoma City                        OK                         0.974 
187  Kokomo                               IN                         0.972 
188  Raleigh-Durham-Chapel Hill           NC                         0.970 
189  Sarasota-Bradenton                   FL                         0.969 
190  Mobile                               AL                         0.966 
191  Bremerton                            WA                         0.965 
192  Montgomery                           AL                         0.964 
193  Myrtle Beach                         SC                         0.964 
194  Fresno                               CA                         0.963 
195  Nashville                            TN                         0.962 
196  Bellingham                           WA                         0.962 
197  Florence                             AL                         0.959 
198  Scranton-Wilkes-Barre-Hazleton       PA                         0.959 
199  Lynchburg                            VA                         0.959 
200  Daytona Beach                        FL                         0.959 
201  Steubenville-Weirton, OH-WV          OH                         0.958 
202  Stamford-Norwalk                     CT                         0.958 
203  Charleston-North Charleston          SC                         0.956 
204  Honolulu                             HI                         0.956 
205  Richland-Kennewick-Pasco             WA                         0.956 
206  Gadsden                              AL                         0.956 
207  Greensboro-Winston-Salem-High Point  NC                         0.955 
208  Visalia-Tulare-Porterville           CA                         0.954 
209  Decatur                              AL                         0.949 
210  Danbury                              CT                         0.949 
211  New London-Norwich, CT-RI            CT                         0.948 
212  Jacksonville                         FL                         0.947 
213  Erie                                 PA                         0.946 
214  Rochester                            NY                         0.946 
215  Reno                                 NV                         0.944 
216  Bakersfield                          CA                         0.942 
217  Olympia                              WA                         0.941 

                  Page 62 GAO-05-856 FEHBP Health Care Prices

Rank Metropolitan area               Predominant statea Adjusted physician 
                                                                  price index 
218  Pittsfield                      MA                              0.941 
219  Santa Fe                        NM                              0.939 
220  Louisville, KY-IN               KY                              0.938 
221  Benton Harbor                   MI                              0.938 
222  Williamsport                    PA                              0.936 
223  Charlottesville                 VA                              0.935 
224  Salinas                         CA                              0.935 
225  Kalamazoo-Battle Creek          MI                              0.935 
226  Manchester                      NH                              0.932 
227  Youngstown-Warren               OH                              0.930 
228  Dover                           DE                              0.926 
229  Hartford                        CT                              0.923 
230  Lancaster                       PA                              0.923 
231  Canton-Massillon                OH                              0.922 
232  Sacramento                      CA                              0.920 
233  Seattle-Bellevue-Everett        WA                              0.919 
234  Jackson                         MI                              0.913 
235  Springfield                     MA                              0.913 
236  Vallejo-Fairfield-Napa          CA                              0.911 
237  Orlando                         FL                              0.909 
238  Huntsville                      AL                              0.909 
239  Grand Rapids-Muskegon-Holland   MI                              0.909 
240  Provo-Orem                      UT                              0.906 
241  Stockton-Lodi                   CA                              0.904 
242  Fitchburg-Leominster            MA                              0.904 
243  Tucson                          AZ                              0.904 
244  Birmingham                      AL                              0.903 
245  Akron                           OH                              0.901 
246  New Haven-Meriden               CT                              0.900 
247  Waterbury                       CT                              0.899 
248  Columbus                        OH                              0.899 
249  Tampa-St. Petersburg-Clearwater FL                              0.899 
250  Jamestown                       NY                              0.898 
251  Richmond-Petersburg             VA                              0.898 
252  Cincinnati, OH-KY-IN            OH                              0.897 
253  Cumberland, MD-WV               MD                              0.895 
254  York                            PA                              0.894 

                  Page 63 GAO-05-856 FEHBP Health Care Prices

Rank Metropolitan area                       Predominant          Adjusted 
                                                   statea           physician 
                                                                  price index 
255  Greenville-Spartanburg-Anderson       SC                        0.893 
256  New Bedford                           MA                        0.892 
257  Riverside-San Bernardino              CA                        0.891 
258  Saginaw-Bay City-Midland              MI                        0.890 
259  Columbia                              SC                        0.888 
260  Nashua                                NH                        0.888 
261  Hamilton-Middletown                   OH                        0.887 
262  Harrisburg-Lebanon-Carlisle           PA                        0.886 
263  Las Vegas, NV-AZ                      NV                        0.885 
264  Toledo                                OH                        0.885 
265  Kansas City, MO-KS                    MO                        0.884 
266  Cleveland-Lorain-Elyria               OH                        0.883 
267  San Luis Obispo-Atascadero-Paso       CA                        0.883 
        Robles                                                 
268  Vineland-Millville-Bridgeton          NJ                        0.882 
269  Reading                               PA                        0.876 
270  Bridgeport                            CT                        0.874 
271  Monmouth-Ocean                        NJ                        0.873 
272  Los Angeles-Long Beach                CA                        0.870 
273  Ann Arbor                             MI                        0.870 
274  Orange County                         CA                        0.870 
275  Melbourne-Titusville-Palm Bay         FL                        0.869 
276  Santa Barbara-Santa Maria-Lompoc      CA                        0.866 
277  Jersey City                           NJ                        0.865 
278  Lawrence, MA-NH                       MA                        0.861 
279  San Diego                             CA                        0.861 
280  Trenton                               NJ                        0.861 
281  State College                         PA                        0.861 
282  Lansing-East Lansing                  MI                        0.861 
283  Barnstable-Yarmouth                   MA                        0.861 
284  Phoenix-Mesa                          AZ                        0.859 
285  Allentown-Bethlehem-Easton            PA                        0.856 
286  New York                              NY                        0.854 
287  Ventura                               CA                        0.851 
288  Santa Cruz-Watsonville                CA                        0.848 
289  Worcester, MA-CT                      MA                        0.846 
290  Flint                                 MI                        0.844 
291  Pittsburgh                            PA                        0.841 

Rank Metropolitan area                      Predominant           Adjusted 
                                                 statea       physician price 
                                                                        index 
292  San Jose                            CA                          0.837 
293  Atlantic-Cape May                   NJ                          0.835 
294  Dayton-Springfield                  OH                          0.833 
295  Salt Lake City-Ogden                UT                          0.833 
296  Fort Pierce-Port St. Lucie          FL                          0.830 
297  Philadelphia, PA-NJ                 PA                          0.828 
298  Buffalo-Niagara Falls               NY                          0.823 
299  Wilmington-Newark, DE-MD            DE                          0.823 
300  Newburgh, NY-PA                     NY                          0.822 
301  Hagerstown                          MD                          0.822 
302  Newark                              NJ                          0.818 
303  Santa Rosa                          CA                          0.817 
304  Middlesex-Somerset-Hunterdon        NJ                          0.816 
305  Oakland                             CA                          0.813 
306  Detroit                             MI                          0.809 
307  Bergen-Passaic                      NJ                          0.807 
308  Brockton                            MA                          0.802 
309  Boston, MA-NH                       MA                          0.785 
310  San Francisco                       CA                          0.772 
311  Dutchess County                     NY                          0.768 
312  Providence-Fall River-Warwick,      RI                          0.763 
        RI-MA                                                 
313  Miami                               FL                          0.755 
314  West Palm Beach-Boca Raton          FL                          0.749 
315  Fort Lauderdale                     FL                          0.747 
316  Washington, DC-MD-VA-WV             VA                          0.746 
317  Nassau-Suffolk                      NY                          0.744 
318  Lowell, MA-NH                       MA                          0.743 
319  Baltimore                           MD                          0.729 

Source: GAO analysis of FEHBP data.

Note: We adjusted physician prices to remove the effect of geographic
variation in the costs of doing business (wages, rents, etc.) and
differences in the mix of services among metropolitan areas. We converted
physician prices to an index by dividing the average physician price per
service in a metropolitan area by the average physician price in 319
metropolitan areas. The average physician price index value is 1.00.

a

Some metropolitan areas spanned more than one state. In those cases, we
assigned the state that contained the largest proportion of the population
of the metropolitan area.

Appendix IV: FEHBP PPO Adjusted Health Care Spending Per Enrollee in U.S.
Metropolitan Areas, 2001

The adjusted spending per enrollee indices based on FEHBP PPO spending in
232 metropolitan areas are presented below ranked in order from highest to
lowest spending per enrollee.

                  Page 65 GAO-05-856 FEHBP Health Care Prices

Table 17: Ranking of Metropolitan Areas by Adjusted Health Care Spending
Per Enrollee, 2001
Rank    Metropolitan area                 Predominant             Adjusted 
                                             statea            spending index 
1       Biloxi-Gulfport-Pascagoula        MS                         1.422 
2       Myrtle Beach                      SC                         1.404 
3       Monroe                            LA                         1.393 
4       Hattiesburg                       MS                         1.393 
5       Parkersburg-Marietta, WV-OH       WV                         1.343 
6       Anniston                          AL                         1.322 
7       Florence                          SC                         1.298 
8       Terre Haute                       IN                         1.297 
9       Bakersfield                       CA                         1.268 
10      San Angelo                        TX                         1.258 
11      Gadsden                           AL                         1.250 
12      Wichita Falls                     TX                         1.240 
13      Houma                             LA                         1.240 
14      Sherman-Denison                   TX                         1.235 
15      Wilmington                        NC                         1.216 
16      Huntington-Ashland, WV-KY-OH      WV                         1.216 
17      Macon                             GA                         1.213 
18      Lubbock                           TX                         1.212 
19      Dothan                            AL                         1.211 
20      Punta Gorda                       FL                         1.211 
21      Decatur                           AL                         1.200 
22      Milwaukee-Waukesha                WI                         1.197 
23      Rapid City                        SD                         1.195 
24      Albany                            GA                         1.194 
25      Fort Walton Beach                 FL                         1.187 
26      Texarkana, TX-Texarkana           TX                         1.186 
27      Oklahoma City                     OK                         1.182 
28      Charleston-North Charleston       SC                         1.180 
29      Lake Charles                      LA                         1.169 
30      Panama City                       FL                         1.167 
31      La Crosse, WI-MN                  WI                         1.163 
32      Little Rock-North Little Rock     AR                         1.163 

Appendix IV: FEHBP PPO Adjusted Health Care Spending Per Enrollee in U.S.
Metropolitan Areas, 2001 Appendix IV: FEHBP PPO Adjusted Health Care
Spending Per Enrollee in U.S. Metropolitan Areas, 2001 Appendix IV: FEHBP
PPO Adjusted Health Care Spending Per Enrollee in U.S. Metropolitan Areas,
2001 Appendix IV: FEHBP PPO Adjusted Health Care Spending Per Enrollee in
U.S. Metropolitan Areas, 2001 Appendix IV: FEHBP PPO Adjusted Health Care
Spending Per Enrollee in U.S. Metropolitan Areas, 2001 Appendix IV: FEHBP
PPO Adjusted Health Care Spending Per Enrollee in U.S. Metropolitan Areas,
2001

                  Page 66 GAO-05-856 FEHBP Health Care Prices

Rank Metropolitan area                   Predominant statea       Adjusted 
                                                               spending index 
33   Florence                            AL                          1.161 
34   Knoxville                           TN                          1.157 
35   Jacksonville                        NC                          1.155 
36   Yuma                                AZ                          1.151 
37   Shreveport-Bossier City             LA                          1.133 
38   Pine Bluff                          AR                          1.132 
39   Lafayette                           LA                          1.126 
40   Galveston-Texas City                TX                          1.122 
41   Charlotte-Gastonia-Rock Hill, NC-SC NC                          1.120 
42   Enid                                OK                          1.119 
43   Johnson City-Kingsport-Bristol,     TN                          1.118 
        TN-VA                                                  
44   Fort Worth-Arlington                TX                          1.117 
45   Lawton                              OK                          1.116 
46   Charleston                          WV                          1.116 
47   Jonesboro                           AR                          1.115 
48   McAllen-Edinburg-Mission            TX                          1.113 
49   Melbourne-Titusville-Palm Bay       FL                          1.108 
50   Nashville                           TN                          1.103 
51   Tuscaloosa                          AL                          1.102 
52   Dallas                              TX                          1.101 
53   Bryan-College Station               TX                          1.097 
54   Waco                                TX                          1.096 
55   Omaha, NE-IA                        NE                          1.092 
56   Jackson                             MS                          1.089 
57   Savannah                            GA                          1.088 
58   Springfield                         MO                          1.088 
59   New Orleans                         LA                          1.082 
60   Las Vegas, NV-AZ                    NV                          1.081 
61   Chattanooga, TN-GA                  TN                          1.079 
62   Boulder-Longmont                    CO                          1.078 
63   Duluth-Superior, MN-WI              MN                          1.077 
64   Greenville-Spartanburg-Anderson     SC                          1.077 
65   Baton Rouge                         LA                          1.076 
66   Las Cruces                          NM                          1.074 
67   St. Joseph                          MO                          1.074 
68   Owensboro                           KY                          1.073 
69   Corpus Christi                      TX                          1.073 

                  Page 67 GAO-05-856 FEHBP Health Care Prices

Rank Metropolitan area                  Predominant statea        Adjusted 
                                                               spending index 
70   Lakeland-Winter Haven              FL                           1.072 
71   Sarasota-Bradenton                 FL                           1.072 
72   Jacksonville                       FL                           1.070 
73   San Antonio                        TX                           1.067 
74   Tulsa                              OK                           1.060 
75   Odessa-Midland                     TX                           1.059 
76   Portsmouth-Rochester, NH-ME        NH                           1.057 
77   Topeka                             KS                           1.056 
78   Orange County                      CA                           1.049 
79   Pensacola                          FL                           1.049 
80   Amarillo                           TX                           1.048 
81   Fort Myers-Cape Coral              FL                           1.048 
82   Houston                            TX                           1.045 
83   Indianapolis                       IN                           1.039 
84   Colorado Springs                   CO                           1.036 
85   Montgomery                         AL                           1.034 
86   Huntsville                         AL                           1.033 
87   Orlando                            FL                           1.033 
88   Wichita                            KS                           1.030 
89   Memphis, TN-AR-MS                  TN                           1.027 
90   Anchorage                          AK                           1.025 
91   Bloomington                        IN                           1.022 
92   Monmouth-Ocean                     NJ                           1.021 
93   Cumberland, MD-WV                  MD                           1.020 
94   Lincoln                            NE                           1.020 
95   Columbus, GA-AL                    GA                           1.014 
96   Fort Smith, AR-OK                  AR                           1.012 
97   Roanoke                            VA                           1.012 
98       Norfolk-Virginia Beach-Newport VA                           1.012 
                               News, VA-NC                    
99   Mobile                             AL                           1.011 
100  Boise City                         ID                           1.010 
101  Louisville, KY-IN                  KY                           1.008 
102  Austin-San Marcos                  TX                           1.007 
103  Clarksville-Hopkinsville, TN-KY    TN                           1.004 
104  Ventura                            CA                           1.004 
105  Birmingham                         AL                           1.000 
106  Manchester                         NH                           0.999 

                  Page 68 GAO-05-856 FEHBP Health Care Prices

Rank Metropolitan area               Predominant statea  Adjusted spending 
                                                                        index 
107  Daytona Beach                   FL                              0.996 
108  Sioux Falls                     SD                              0.994 
109  Columbia                        SC                              0.994 
110  Richland-Kennewick-Pasco        WA                              0.992 
111  Atlantic-Cape May               NJ                              0.988 
112  Grand Forks, ND-MN              ND                              0.988 
113  New London-Norwich, CT-RI       CT                              0.988 
114  Trenton                         NJ                              0.987 
115  Olympia                         WA                              0.984 
116  Columbia                        MO                              0.984 
117  Atlanta                         GA                              0.983 
118  Killeen-Temple                  TX                              0.982 
119  Grand Junction                  CO                              0.982 
120  Kansas City, MO-KS              MO                              0.980 
121  Gary                            IN                              0.979 
122  West Palm Beach-Boca Raton      FL                              0.977 
123  Athens                          GA                              0.977 
124  Fayetteville-Springdale-Rogers  AR                              0.977 
125  Billings                        MT                              0.975 
126  Fort Lauderdale                 FL                              0.971 
127  Great Falls                     MT                              0.970 
128  Dover                           DE                              0.965 
129  Jackson                         TN                              0.965 
130  Lynchburg                       VA                              0.962 
131  Des Moines                      IA                              0.962 
132  Gainesville                     FL                              0.960 
133  Laredo                          TX                              0.959 
134  Augusta-Aiken, GA-SC            GA                              0.959 
135  Denver                          CO                              0.958 
136  Bremerton                       WA                              0.957 
137  Fort Pierce-Port St. Lucie      FL                              0.955 
138  Salinas                         CA                              0.952 
139  Pueblo                          CO                              0.952 
140  Tampa-St. Petersburg-Clearwater FL                              0.951 
141  Fort Wayne                      IN                              0.950 
142  Hagerstown                      MD                              0.949 
143  Los Angeles-Long Beach          CA                              0.947 

                  Page 69 GAO-05-856 FEHBP Health Care Prices

Rank Metropolitan area                   Predominant statea       Adjusted 
                                                               spending index 
144  Lexington                           KY                          0.946 
145  Middlesex-Somerset-Hunterdon        NJ                          0.942 
146  Redding                             CA                          0.942 
147  Bangor                              ME                          0.941 
148  Tacoma                              WA                          0.941 
149  Phoenix-Mesa                        AZ                          0.935 
150  Riverside-San Bernardino            CA                          0.935 
151  Cedar Rapids                        IA                          0.934 
152  Greensboro-Winston-Salem-High Point NC                          0.932 
153  Fayetteville                        NC                          0.930 
154  Miami                               FL                          0.928 
155  Sacramento                          CA                          0.927 
156  Reading                             PA                          0.927 
157  Salt Lake City-Ogden                UT                          0.925 
158  Cincinnati, OH-KY-IN                OH                          0.923 
159  Richmond-Petersburg                 VA                          0.920 
160  Detroit                             MI                          0.920 
161  Chicago                             IL                          0.918 
162  Provo-Orem                          UT                          0.918 
163  Fort Collins-Loveland               CO                          0.913 
164  Yakima                              WA                          0.913 
165  Goldsboro                           NC                          0.913 
166  Albany-Schenectady-Troy             NY                          0.913 
167  Nashua                              NH                          0.911 
168  Asheville                           NC                          0.911 
169  Nassau-Suffolk                      NY                          0.909 
170  Santa Fe                            NM                          0.908 
171  Scranton-Wilkes-Barre-Hazleton      PA                          0.906 
172  Missoula                            MT                          0.904 
173  York                                PA                          0.904 
174  Jersey City                         NJ                          0.904 
175  Raleigh-Durham-Chapel Hill          NC                          0.901 
176  Columbus                            OH                          0.901 
177  Sioux City, IA-NE                   IA                          0.899 
178  Cleveland-Lorain-Elyria             OH                          0.899 
179  Greenville                          NC                          0.897 
180  Wilmington-Newark, DE-MD            DE                          0.897 

                  Page 70 GAO-05-856 FEHBP Health Care Prices

Rank Metropolitan area                  Predominant statea        Adjusted 
                                                               spending index 
181  Tucson                             AZ                           0.897 
182  Waterbury                          CT                           0.896 
183  Portland                           ME                           0.893 
184  Salem                              OR                           0.892 
185  Bergen-Passaic                     NJ                           0.891 
186  Eugene-Springfield                 OR                           0.883 
187  Kalamazoo-Battle Creek             MI                           0.881 
188  Washington, DC-MD-VA-WV            VA                           0.881 
189  Bismarck                           ND                           0.880 
190  Flint                              MI                           0.879 
191  Newark                             NJ                           0.878 
192  Springfield                        MA                           0.876 
193  Baltimore                          MD                           0.875 
194  New Haven-Meriden                  CT                           0.874 
195  Minneapolis-St. Paul, MN-WI        MN                           0.873 
196  Philadelphia, PA-NJ                PA                           0.870 
197  San Diego                          CA                           0.869 
198  Albuquerque                        NM                           0.868 
199  Reno                               NV                           0.866 
200  Altoona                            PA                           0.866 
201  Lawrence, MA-NH                    MA                           0.862 
202  Dayton-Springfield                 OH                           0.852 
203  Portland-Vancouver, OR-WA          OR                           0.848 
204  Newburgh, NY-PA                    NY                           0.848 
205  New York                           NY                           0.845 
206  Seattle-Bellevue-Everett           WA                           0.843 
207  Medford-Ashland                    OR                           0.841 
208  Evansville-Henderson, IN-KY        IN                           0.836 
209  Charlottesville                    VA                           0.836 
210  Providence-Fall River-Warwick,     RI                           0.834 
        RI-MA                                                 
211  Lansing-East Lansing               MI                           0.833 
212  Harrisburg-Lebanon-Carlisle        PA                           0.832 
213  South Bend                         IN                           0.830 
214  Iowa City                          IA                           0.827 
215  Toledo                             OH                           0.825 
216  Allentown-Bethlehem-Easton         PA                           0.814 
217  San Francisco                      CA                           0.809 

Rank Metropolitan area             Predominant statea    Adjusted spending 
                                                                        index 
218  Hartford                      CT                                0.809 
219  Oakland                       CA                                0.807 
220  Erie                          PA                                0.803 
221  Syracuse                      NY                                0.793 
222  Spokane                       WA                                0.789 
223  Ann Arbor                     MI                                0.778 
224  Pittsburgh                    PA                                0.776 
225  Fargo-Moorhead, ND-MN         ND                                0.766 
226  Saginaw-Bay City-Midland      MI                                0.753 
227  Johnstown                     PA                                0.746 
228  Boston, MA-NH                 MA                                0.746 
229  Bridgeport                    CT                                0.732 
230  Buffalo-Niagara Falls         NY                                0.715 
231  Honolulu                      HI                                0.684 
232  Grand Rapids-Muskegon-Holland MI                                0.672 

Source: GAO analysis of FEHBP data.

Note: Total spending per enrollee includes spending for all services
except mental health, chemical dependency, and pharmaceuticals. We
adjusted total spending per enrollee to remove the effect of geographic
differences in enrollee age and sex, as well as geographic differences in
the costs of doing business (such as wages and rents). The spending per
enrollee index compares spending per enrollee in a metropolitan area to
the average spending per enrollee in all study metropolitan areas,
adjusted for patients' age and sex composition, and costs. The average
spending index was 1.00.

a

Some metropolitan areas spanned more than one state. In those cases, we
assigned the state that contained the largest proportion of the population
of the metropolitan area.

Appendix V: Comments from the Office of Personnel Management

                  Page 73 GAO-05-856 FEHBP Health Care Prices

Appendix VI: GAO Contacts and Staff Acknowledgments

A. Bruce Steinwald, (202) 512-7101 or steinwalda@gao.gov

  GAO Contacts

In addition to the contact named above, Christine Brudevold, Assistant

Director; Jennie F. Apter; Leslie Gordon; Michael Kendix; Daniel Lee;
Jennifer M. Rellick; Holly Stockdale; Ann Tynan; and Suzanne Worth made
key contributions to this report.

(290213)

  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.

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 GAO
Reports and 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, Contact:

Web site: www.gao.gov/fraudnet/fraudnet.htm

  E-mail: fraudnet@gao.gov

Federal Programs Automated answering system: (800) 424-5454 or (202)
512-7470

Gloria Jarmon, Managing Director, JarmonG@gao.gov (202) 512-4400 U.S.
Government Accountability Office, 441 G Street NW, Room 7125 Relations
Washington, D.C. 20548

Paul Anderson, Managing Director, AndersonP1@gao.gov (202) 512-4800

  Public Affairs

U.S. Government Accountability Office, 441 G Street NW, Room 7149
Washington, D.C. 20548

    PRINTED ON

RECYCLED PAPER
*** End of document. ***