This is the accessible text file for GAO report number GAO-07-383 
entitled 'Ambulance Providers: Costs and Expected Medicare Margins Vary 
Greatly' which was released on May 23, 2007. 

This text file was formatted by the U.S. Government Accountability 
Office (GAO) to be accessible to users with visual impairments, as part 
of a longer term project to improve GAO products' accessibility. Every 
attempt has been made to maintain the structural and data integrity of 
the original printed product. Accessibility features, such as text 
descriptions of tables, consecutively numbered footnotes placed at the 
end of the file, and the text of agency comment letters, are provided 
but may not exactly duplicate the presentation or format of the printed 
version. The portable document format (PDF) file is an exact electronic 
replica of the printed version. We welcome your feedback. Please E-mail 
your comments regarding the contents or accessibility features of this 
document to Webmaster@gao.gov. 

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. 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. 

Report to Congressional Committees: 

United States Government Accountability Office: 

GAO: 

May 2007: 

Ambulance Providers: 

Costs and Expected Medicare Margins Vary Greatly: 

GAO-07-383: 

GAO Highlights: 

Highlights of GAO-07-383, a report to congressional committees 

Why GAO Did This Study: 

In 2002, Medicare implemented a national fee schedule designed to 
standardize payments for ambulance services. The Medicare Prescription 
Drug, Improvement and Modernization Act of 2003 (MMA) required GAO to 
study ambulance service costs. GAO examined providers’ costs of ground 
ambulance transports in 2004 and factors that contributed to cost 
differences; average Medicare ambulance payments expected under the 
national fee schedule in 2010 and how those payments will relate to 
providers’ costs per transport; and changes that occurred in Medicare 
beneficiaries’ use of ambulance transports from 2001 to 2004. GAO 
estimated costs of ambulance transports based on a nationally 
representative survey of 215 ambulance providers that did not share 
costs with nonambulance services. Providers that shared costs with 
other institutions or services and could not report their costs for 
ambulance services separately, such as fire departments, were excluded 
because their reported costs appeared unreliable. GAO used its survey, 
Medicare claims, and other data for its analyses. 

What GAO Found: 

Costs of ground ambulance services were highly variable across 
providers that did not share costs with nonambulance services in 2004, 
reflecting differences in certain provider and community 
characteristics. Costs per transport among these providers varied from 
$99 per transport to $1,218. Providers without shared costs that had 
higher costs per transport typically had fewer transports per year, a 
greater percentage of transports in which more than a basic medical 
intervention occurred, more transports in super-rural areas (rural 
counties with lowest population density), lower productivity—measured 
as number of transports furnished per staffed hour, and a greater 
percentage of revenues from local tax support. 

Average payments under the national fee schedule in 2010 are expected 
to be higher than historical payments, but providers’ Medicare margins 
will vary greatly. GAO could not assess whether, on average, providers 
without shared costs would break even, lose, or profit under the 
national fee schedule, because the average Medicare margin for 
providers without shared costs was estimated to fall from negative 14 
percent to positive 2 percent. However, GAO estimated that 
approximately 39 to 56 percent of providers without shared costs would 
have average Medicare payments above their average cost per transport 
under the national fee schedule in 2010. 

From 2001 to 2004, utilization of ambulance transports per beneficiary 
increased 16 percent overall. However, use declined by 8 percent in 
super-rural areas. 

Declining utilization coupled with potentially negative Medicare 
margins in super-rural areas, which could be exacerbated when the MMA 
temporary payment provisions expire, raise questions as to whether 
Medicare payments will be adequate to support beneficiary access in 
super-rural areas. 

Figure: Distribution of Cost per Transport for Providers without Shared 
Costs, 2004: 

[See PDF for Image] 

Source: 2005 GAO Survey of Ambulance Services. 

Note: Based on a sample of 215 providers, weighted to represent more 
than 5,200 providers in the United States that did not share costs with 
nonambulance services. 

[End of section] 

What GAO Recommends: 

GAO recommends that the Administrator of CMS monitor utilization of 
ambulance transports to ensure that Medicare payments 
are adequate to provide for beneficiary access to ambulance services, 
particularly in super-rural areas. CMS agreed with GAO’s 
recommendation. 

[Hyperlink, http://www.gao.gov/cgi-bin/getrpt?GAO-07-383]. 

To view the full product, including the scope and methodology, click on 
the link above. For more information, contact Kathleen M. King at (202) 
512-7119 or kingk@gao.gov. 

[End of section] 

Contents: 

Letter: 

Results in Brief: 

Background: 

Costs per Transport Were Highly Variable, Reflecting Differences in 
Certain Provider Characteristics: 

Average Payments under the National Fee Schedule Will Be Greater Than 
Average Historical Payments, but Providers' Expected Medicare Margins 
Will Vary Greatly: 

MMA Provisions Resulted in Greater Average Payments for Higher-Cost 
Super-Rural Transports and Adjusted Payments Regionally Where No 
Significant Cost Differences Were Observed: 

Medicare Beneficiaries' Use of Ambulance Transports Increased from 2001 
to 2004, Except in Super-Rural Areas: 

Conclusions: 

Recommendation for Executive Action: 

Agency and External Comments and Our Evaluation: 

Appendix I: Data and Methods: 

Appendix II: Comments from the Centers for Medicare & Medicaid 
Services: 

Appendix III: GAO Contact and Staff Acknowledgments: 

Related GAO Products: 

Tables: 

Table 1: Summary of MMA Temporary Payment Provisions, Implemented July 
1, 2004: 

Table 2: Estimated Average Cost per Transport for Provider 
Characteristics That Affect Costs: 

Table 3: Payments Prior to and under the National Fee Schedule after 
MMA Provisions Expire: 

Table 4: Expected Average Medicare Margins under the National Fee 
Schedule for Providers without Shared Costs in 2004 Dollars: 

Table 5: Average Payments prior to MMA Implementation and after 
Implementation: 

Table 6: Percentage Changes in Average Payments prior to MMA 
Implementation and after Implementation, by Region: 

Table 7: Ambulance Transports per 1,000 Beneficiaries in Urban, Rural, 
and Super-Rural Areas: 

Table 8: Provider and Local Area Characteristics Included in Analysis 
of Average Cost per Transport, 2004: 

Table 9: Results for Average Cost of an Ambulance Transport Regression-
-Estimated Effects of Selected Provider and Local Area Characteristics 
on the Average Cost of Ambulance Transports for Providers, Not 
Including Impact of Productivity and Community Tax Support: 

Table 10: Results for Average Cost of an Ambulance Transport 
Regression--Estimated Effects of Selected Provider and Local Area 
Characteristics on the Average Cost of Ambulance Transports for 
Providers, Including Impact of Productivity and Community Tax Support: 

Table 11: Census Divisions: 

Table 12: Average Mile per Transport, First Half of 2004: 

Table 13: Mileage Rates: 

Figures: 

Figure 1: Medicare Ambulance Payment Formula under the National Fee 
Schedule: 

Figure 2: National Fee Schedule and Regional Fee Schedule: 

Figure 3: Distribution of Cost per Transport for Providers without 
Shared Costs in 2004: 

Figure 4: Expected Medicare Margins for Urban, Rural, and Super-Rural 
Providers without Shared Costs: 

Abbreviations: 

ALS: advanced life support: 
BLS: basic life support: 
CF: conversion factor: 
CMS: Centers for Medicare & Medicaid Services: 
CPI-U: Consumer Price Index for All Urban Consumers: 
EMS: emergency medical services: 
EMT: emergency medical technician: 
GPCI: geographic practice cost index: 
MMA: Medicare Prescription Drug, Improvement and Modernization Act of 
2003: 
MSA: metropolitan statistical area: 
NECMA: New England county metropolitan area: 
NFS: national fee schedule: 
RVU: relative value unit: 

United States Government Accountability Office: 
Washington, DC 20548: 

May 23, 2007: 

The Honorable Max Baucus: 
Chairman: 
The Honorable Charles E. Grassley: 
Ranking Member: 
Committee on Finance: 
United States Senate: 

The Honorable John D. Dingell: 
Chairman: 
The Honorable Joe Barton: 
Ranking Member: 
Committee on Energy and Commerce: 
House of Representatives: 

The Honorable Charles B. Rangel: 
Chairman: 
The Honorable Jim McCrery: 
Ranking Member: 
Committee on Ways and Means: 
House of Representatives: 

In 2005, ambulance service providers completed more than 12.6 million 
ground transports for Medicare beneficiaries.[Footnote 1] Medicare paid 
more than $4 billion for ground ambulance transports and is likely the 
largest single payer of ambulance services in the United States. 

As part of a series of Medicare payment reforms in 1997, Congress 
required the Health Care Financing Administration, now the Centers for 
Medicare & Medicaid Services (CMS), to develop a national fee schedule 
for Medicare ambulance services, which was implemented in 
2002.[Footnote 2] Historically, CMS had used two methods to pay for 
ambulance services, which resulted in wide variations in payment for 
the same service among different types of ambulance service providers. 
In particular, CMS had used one method--reasonable costs[Footnote 3]-- 
to pay hospital-based providers. It used another method--reasonable 
charges[Footnote 4]--to pay other, nonhospital-based types of ambulance 
service providers. This meant that hospital-based and nonhospital-based 
providers were paid different amounts for the same ambulance services. 

In 2002, CMS began phasing in a national fee schedule that established 
a single payment method for all ambulance services regardless of the 
type of provider.[Footnote 5] This fee schedule standardized Medicare 
payments for ambulance services. In general, providers strive to keep 
their costs of delivering a service at or below the standard fee 
schedule rate for that service. Under the Medicare ambulance national 
fee schedule, providers that have costs of delivering ambulance 
services above the fee schedule payment lose the difference between the 
payment amount and their costs, while providers with costs below the 
fee schedule payment are able to keep the difference between the 
payment amount and their costs. In aggregate, these differences are 
known as Medicare margins and express whether the provider makes a 
profit or loss on its Medicare transports. Some providers rely heavily 
on Medicare revenues and adequate Medicare margins help ensure the 
continuing availability of beneficiaries' access to ambulance services. 

CMS phased in the ambulance national fee schedule from April 2002 
through December 2005. During this transition, the new fee schedule 
payments were blended with the previous reasonable-cost payments for 
hospital-based providers and reasonable-charge payments for nonhospital-
based providers. In 2003, Congress passed the Medicare Prescription 
Drug, Improvement and Modernization Act (MMA), which introduced several 
temporary payment provisions, including a regional fee schedule that 
overlapped with the transition to the national fee schedule.[Footnote 
6] Beginning in July 2004, these temporary payment provisions were 
expected to add about $840 million to Medicare payments for ambulance 
services through December 2009, when the last of these provisions are 
set to expire. 

The Medicare, Medicaid, and SCHIP Benefits Improvement and Protection 
Act of 2000[Footnote 7] and the MMA required GAO to study ambulance 
service costs. As discussed with the congressional committees of 
jurisdiction, in this report we (1) examined the differences in 
providers' costs of ambulance transports in 2004 and the factors that 
contributed to these cost differences, (2) assessed how the ambulance 
national fee schedule in 2010 is expected to affect average ambulance 
payments and how those payments will relate to providers' costs per 
transport, (3) determined the effect of MMA temporary payment 
provisions on ambulance payments, and (4) described the change that 
occurred in Medicare beneficiaries' use of ambulance transports from 
2001 to 2004. 

To examine differences in costs of providing ambulance transports, we 
conducted a national survey of ambulance providers in 2005. In our 
survey, we requested information about providers' costs, revenues, 
transports, and organizational characteristics for their most recently 
completed fiscal year.[Footnote 8] We selected a stratified,[Footnote 
9] random sample of 500 eligible providers that billed Medicare, and we 
received 321 completed questionnaires for a response rate of 64 
percent. We used this nationally representative sample of 321 providers 
to describe the ambulance industry.[Footnote 10] However, after 
excluding two cost outliers, our analysis was further limited to a 
subgroup of providers that (1) did not share costs with other 
institutions or services or (2) shared costs but reported costs of 
ambulance services separately from the costs of their other services. 
Our analysis and findings are nationally representative of this 
subgroup of ambulance providers, which we refer to as providers without 
shared costs.[Footnote 11] We excluded 104 providers that shared costs 
of ambulance services with other institutions or nonambulance services 
and could not distinguish their costs for providing ambulance services 
from other costs, including but not limited to all fire departments. We 
excluded these providers because their reported costs appeared to be 
unreliable. The resulting sample size for our analysis was 215 
providers without shared costs. Although our sample is nationally 
representative of an estimated 5,200 providers without shared costs, 
the small sample size along with the variability of responses reduces 
the precision of our estimates, increasing the range of the 95 percent 
confidence intervals we report. A 95 percent confidence interval is the 
range within which we expect the true population estimate to fall 95 
percent of the time, and it is the range of the confidence interval 
that expresses the precision of our estimates. 

To examine factors that contributed to differences in costs, we used 
our survey data and Medicare data supplemented by data from two other 
sources. The Oil Price Information Service was our source for the 
average annual retail price of fuel by zip code, and the United States 
Postal Service supplied building rents because it tracks its facility 
costs in each zip code. We used regression analysis to analyze the 
relationships between various provider and local area characteristics 
and cost per transport among providers without shared costs. We also 
compared Medicare claims data for all nonrespondents with those of 
respondent providers without shared costs and determined that our cost 
estimates were not biased by nonresponse. See appendix I for details 
regarding our survey, other data sources, data limitations, and the 
analytic methods we employed. 

To assess the effect of the ambulance national fee schedule on 
payments, we used Medicare claims data to compute average payments for 
ambulance transports in 2001, before the implementation of the 
ambulance national fee schedule, and in 2004, 2 years after the phase- 
in of the fee schedule had begun. For Medicare payment analyses, 
payments were expressed in 2004 dollars to exclude the effects of 
inflation. We also compared average payments for urban, rural, and 
super-rural transports.[Footnote 12] We used Medicare claims data and 
payment formulas as specified in federal regulations to simulate 
average payments under the national fee schedule in 2010, after all of 
the MMA provisions expire, but computed these payments in 2004 dollars, 
the year that best reflects the cost data collected in our survey. To 
compare the simulated Medicare payments under the national fee schedule 
for providers without shared costs with the costs per transport of 
those providers, we computed providers' Medicare margins--the 
percentage difference between average Medicare payments and providers' 
costs per transport.[Footnote 13] All costs per transport and provider 
margins are based solely on our sample of providers without shared 
costs, and for this reason, these estimates are reported with their 
confidence intervals. 

We also assessed the effect of the MMA temporary payment provisions on 
payments in 2004 using Medicare claims data by examining the change in 
payments from the first half of the year, before the MMA changes went 
into effect, with the second half of the year, when MMA payment 
provisions had their maximum effect. To assess the change in Medicare 
beneficiaries' use of ambulance transports from 2001 to 2004, we used 
Medicare claims and CMS enrollment data, which contain information 
about beneficiaries, to compute transports per 1,000 beneficiaries for 
both years. 

We tested the internal consistency and reliability of our survey data 
and all non-Medicare data sources and determined that all data sources 
were adequate for our purposes. We conducted our work from July 2004 
through April 2007 in accordance with generally accepted government 
auditing standards. 

Results in Brief: 

Costs of ground ambulance transports were highly variable across 
providers without shared costs, reflecting differences in certain 
provider characteristics. Costs per transport for providers without 
shared costs averaged $415, but varied from $99 to $1,218 per 
transport--a range of more than $1,100. Contributing to the variability 
were differences in providers' volume and mix of transports; service 
area (urban, rural, or super-rural); productivity, which we defined as 
the number of transports per staffed hour; and the percentage of total 
revenue derived from local tax support. Providers without shared costs 
had higher costs per transport when they had fewer transports per year, 
a greater percentage of transports in which more than a basic medical 
assessment or intervention occurred, and more transports from super- 
rural areas than providers without shared costs that did not have these 
characteristics. In addition, providers without shared costs that had 
lower productivity and those receiving a greater percentage of revenues 
from local tax support had higher costs per transport than providers 
without shared costs that had higher productivity and less local tax 
support. Other provider and local area characteristics, such as type of 
provider, region as determined by census division, building rent, and 
price of fuel, did not significantly affect average costs per transport 
among providers without shared costs. 

Average payments under the national fee schedule in 2010, after all of 
the MMA temporary payment provisions are set to expire, are expected to 
be higher than payments in 2001, but Medicare margins for providers 
without shared costs will vary greatly. In 2010, average ambulance 
national fee schedule payments are estimated to be 3 percent higher 
overall than payments in 2001, after adjusting for inflation and 
assuming that providers bill the maximum amounts allowed. Further, a 
greater percentage than the overall increase in payments will accrue to 
rural and super-rural transports, on average, while urban transports 
will receive a decrease in payments under the national fee schedule in 
2010. We could not assess whether providers without shared costs will 
break even, lose, or profit, on average, under the ambulance national 
fee schedule after the MMA temporary payment provisions expire, because 
the 95 percent confidence interval for the average expected Medicare 
margin for providers without shared costs spanned from negative 14 
percent to positive 2 percent when we took into account the number of 
respondents in our sample and the range of their reported costs. 
However, across all providers without shared costs, we estimated that 
39 to 56 percent will have average Medicare payments above their 
average costs per transport under the ambulance national fee schedule 
in 2010. 

The MMA's temporary payment provisions, which included base-and mileage-
rate increases as well as the introduction of a regional fee schedule, 
resulted in raised average ambulance payments overall, particularly for 
super-rural transports, which we determined were typically more costly 
to provide. However, regional payment adjustments by census division 
under the MMA did not appear to be warranted on the basis of regional 
cost differences. The regional fee schedule increased payments 
substantially for some regions but not others. After controlling for 
various characteristics that affected the costs of providing ambulance 
transports, we did not discern any significant differences in average 
cost per transport across regions. 

From 2001 to 2004, Medicare beneficiaries' use of ambulance transports 
increased overall, even in those regions that had a decrease in average 
payments after the MMA was implemented. However, beneficiaries' use in 
super-rural areas decreased by 8 percent over the same period. The 
decline in the use of super-rural ambulance transports did not appear 
to be related to any significant change in the population of Medicare 
beneficiaries residing in super-rural areas. 

In light of the variability in ambulance providers' Medicare margins 
and the potential for negative margins to have an impact on beneficiary 
access, we recommend that the Administrator of CMS monitor utilization 
of ambulance transports to ensure that Medicare payments are adequate 
to provide for beneficiary access to ambulance services, particularly 
in super-rural areas. In its comments on a draft of this report, CMS 
stated that it agreed with our recommendation. External commenters 
generally agreed with our findings. 

Background: 

Ground ambulance services are provided by a wide range of organizations 
that differ in their organizational structure, types of services 
offered, staffing, and revenue sources. Local conditions--including 
whether providers are affiliated with other organizations; whether 
their service areas are predominately urban, rural, or super-rural; and 
the amount of community-dedicated revenues--contribute to this 
diversity. 

In addition, communities have few, if any, tools to help them decide 
the optimal organizational structure, staffing, or amount of funding 
for ambulance services, given local conditions. Although there have 
been efforts to establish national performance or quality 
standards,[Footnote 14] there is limited information about how to best 
evaluate the costs of providing ambulance services in a community. In 
recent years, industry associations and federal agencies have worked to 
develop a data system and mechanisms for measuring the performance of 
emergency medical services (EMS); however, these tools cannot yet be 
applied to measure performance and evaluate the efficiency of ambulance 
services.[Footnote 15] The lack of data and performance standards makes 
it difficult to assess whether any given provider is delivering quality 
care or whether services are being provided efficiently. 

Organizational structures differ in that some ambulance providers are 
affiliated with another institution, such as a hospital or fire 
department, and in that providers may or may not offer other types of 
services, such as hospital services, fire suppression, rescue, or 
wheelchair transportation. Providers affiliated with another 
institution or that offer other types of services may share resources 
and operational costs, such as building space, administrative support, 
or personnel, with these other entities and services. About two-fifths 
of the ambulance industry shared operational costs with other 
institutions or services in 2004.[Footnote 16] Nine percent of the 
ambulance industry was affiliated with a hospital, while 37 percent was 
affiliated with a fire department and 21 percent was affiliated with 
another government agency. Other communities (33 percent of the 
ambulance industry in 2004) were served by freestanding, for-profit or 
not-for-profit provider organizations. 

The types of services offered and the staff employed also vary among 
providers. Some providers perform only emergency transports, in 
response to a 911 or equivalent call, while other providers offer 
nonemergency transports, which are typically transfers from one 
facility to another and may be scheduled in advance. In 2004, 45 
percent of the ambulance industry performed only emergency transports; 
55 percent performed emergency and nonemergency transports. Some 
providers perform only basic life support (BLS)[Footnote 17] transports 
because their staff are not certified to perform more intensive medical 
assessments and interventions, such as advanced life support 
(ALS)[Footnote 18] and other more complex services. In 2004, 9 percent 
of the ambulance industry specialized in only ALS and more complex 
Medicare transports, such as those requiring one or more ALS services 
or respiratory care; 14 percent performed only BLS Medicare transports; 
and 86 percent provided a mix of BLS, ALS, and more complex Medicare 
transports.[Footnote 19] Fire departments are more likely to be 
specialized in ALS and more complex services. According to the fire 
departments that responded to our survey, 70 percent of their Medicare 
transports, on average, required ALS or more complex services; for 
other providers ALS and more complex transport services constituted 
only 49 percent of their Medicare transports. 

Ambulance providers use of a variety of staff to deliver services to 
their communities. First, providers and communities determine what 
proportion of their staff will be emergency medical technicians (EMT) 
trained to perform BLS services, and what proportion will be EMT- 
intermediates or paramedics, who have training to perform more 
intensive ALS interventions.[Footnote 20] In 2004, 77 percent of 
providers that completed our survey reported having at least one staff 
member trained as a paramedic, nurse, or physician, but the remaining 
23 percent of respondent providers relied on staff with less training 
than a paramedic. Other staffing choices include whether to employ 
career-oriented paid staff, rely on volunteers, or use a mix of paid 
and volunteer staff. Some providers choose to employ cross-trained 
staff.[Footnote 21] In 2004, two-fifths of the ambulance industry 
relied substantially on volunteer staff.[Footnote 22] In addition, 
providers and communities make different choices about whether to 
maintain backup vehicles and staff or to rely to a greater extent on 
nearby providers for backup assistance. 

Providers' service areas can be urban, rural, super-rural, or a mix of 
areas. During 2004, CMS defined urban transports as those that 
originate within metropolitan statistical areas (MSA) and New England 
county metropolitan areas (NECMA), rural transports as those that 
originate in rural counties that are outside of MSAs and 
NECMAs,[Footnote 23] and super-rural transports as those that originate 
in the bottom 25 percent of rural areas as defined by population 
density. About half of the ambulance providers served predominately 
urban areas, and the other half served predominately rural and super- 
rural areas.[Footnote 24] However, three-fourths of all Medicare 
transports originated in urban areas. Therefore, rural and super-rural 
providers performed fewer transports on average than urban providers. 
In addition, rural and super-rural transports were longer than urban 
transports, on average, requiring more time and resources per 
transport. In 2004, urban Medicare transports averaged 7 miles, while 
rural Medicare transports averaged 13 miles and Medicare transports 
from super-rural areas averaged 20 miles. 

Providers have several potential revenue sources depending on their 
communities and their choices about funding ambulance services. Revenue 
sources can include community tax support (such as revenue from local 
governments); charitable donations; state and federal grants; 
subscription programs;[Footnote 25] and reimbursements from Medicare or 
Medicaid patients, and private health insurance companies. Not all 
providers receive revenues from all sources, and the mix and amount of 
revenues available may vary.[Footnote 26] For example, 48 percent of 
the ambulance industry indicated that a state or local government 
approves the fees they may choose to charge. In communities that limit 
ambulance providers' fees, providers may bill Medicare (and other 
payers) less than the allowed amount and therefore do not receive the 
maximum Medicare payment allowed for their services. In 2004, Medicare 
payments for ambulance transports accounted for 31 percent of the 
providers' revenues, on average, while Medicare beneficiaries accounted 
for about 40 percent of their transports during that same 
year.[Footnote 27] However, the percentage of Medicare revenues among 
providers ranged from less than 1 percent to 80 percent. Furthermore, 
different communities provide different levels of tax support to 
address specific issues, such as ensuring a minimum level of service in 
remote areas or being equipped with more sophisticated transport 
vehicles or having more highly trained staff. 

Medicare Ambulance National Fee Schedule: 

The ambulance national fee schedule was part of a series of payment 
reforms to make Medicare a more equitable and prudent purchaser of 
health care services. Phased in from April 2002 through December 2005, 
the national fee schedule standardized payment rates and reduced wide 
variations in payments for the same service. 

Medicare ambulance payments under the fee schedule have two components: 
a base-rate payment and a mileage payment.[Footnote 28] The base-rate 
component of ambulance payments under the fee schedule consists of the 
relative value unit (RVU), the conversion factor (CF), and a geographic 
adjustment factor. Ambulance RVUs account for the relative resources 
needed to provide services during an ambulance transport.[Footnote 29] 
The ambulance CF converts the RVU into a payment expressed in dollars 
and is set by CMS annually. Ambulance base-rate payments are also 
adjusted by a geographic practice cost index (GPCI), which is intended 
to account for regional differences in the cost of providing ambulance 
services.[Footnote 30] The mileage component consists of the number of 
miles traveled during an ambulance transport multiplied by the 
applicable mileage rate. (See fig. 1 for an example of the payment 
formula.[Footnote 31]): 

Figure 1: Medicare Ambulance Payment Formula under the National Fee 
Schedule: 

[See PDF for image] 

Source: GAO analysis of CMS information. 

Note: Payments are expressed in 2004 dollars. 

[A] The GPCI is applied to only 70 percent of the unadjusted base rate 
payment, which is the product of the RVU and the CF. For example, the 
base rate payment equals 1.90 x $176.04 x [0.30 + (0.70 x 1.166)]. The 
RVU, CF, GPCI, and mileage rate are based on 2004 values. 

[End of figure] 

The ambulance national fee schedule was phased in from April 2002 
through December 2005 by blending new fee schedule payments with 
historical payments. During this transition, the national fee schedule 
portion constituted a greater share of the total blended ambulance 
payment each year until January 2006, when the historical payment 
portion of the blend was discontinued. For example, in the latter part 
of 2002, total ambulance payments were a blend of 20 percent under the 
national fee schedule and 80 percent under the historical payment 
system. In 2004, the blend was 60 percent national fee schedule and 40 
percent historical payment system. (See fig. 2 for further details on 
the blending of historic payments and national fee schedule payments.) 

MMA Temporary Payment Provisions: 

The MMA introduced several temporary ambulance payment provisions that 
were implemented in 2004, the last of which expires at the end of 2009. 
CMS estimates that these payment adjustments will add $840 million to 
Medicare ambulance services over the 5 years they are in effect. The 
MMA provisions increased payment rates for urban and rural transports 
and for transports 51 miles or greater. The MMA also provided a 
significant base rate increase for transports originating in super- 
rural areas and provided for a new regional fee schedule based on the 
nine census divisions.[Footnote 32] 

The regional fee schedule was designed to ease the transition from the 
historical payment system to the national fee schedule. The 
introduction of the regional fee schedule overlapped with the phase-in 
of the national fee schedule. The regional fee schedule gave 
temporarily higher ambulance payments than what would generally be paid 
under the national fee schedule to ambulance providers in census 
divisions that had historically higher payments. If the regional base- 
rate payment was determined to be greater than the national base-rate 
payment for a particular region, then the region received the more 
advantageous blend of the regional fee schedule base-rate payment and 
the national fee schedule base-rate payment. For example, in the second 
half of 2004 under the regional fee schedule, affected regions received 
a blend of 80 percent of their regional fee schedule base-rate payment 
and 20 percent of the national fee schedule base payment rate. This 
base-rate payment was then further blended with historical payments as 
a part of the gradual phase-in of the ambulance national fee 
schedule.[Footnote 33] The regional fee schedule component of the base 
rate blend was reduced each year from 2005 through 2007, and expires at 
the end of 2009. (See fig. 2 for further details on the blending of 
historical, national fee schedule, and regional fee schedule payments.) 

Figure 2: National Fee Schedule and Regional Fee Schedule: 

[See PDF for image] 

Source: GAO analysis of CMS information. 

Note: All years refer to calendar years unless otherwise specified. 

[A] January 1, 2001 through March 31, 2002. 

[B] April 1, 2002 through December 31, 2002. 

[C] The national fee schedule payment portion includes MMA increases of 
1 percent and 2 percent for urban and rural transports, respectively, 
and an approximate 23 percent base rate increase for super-rural 
transports. 

[D] The national fee schedule payment portion includes an MMA base rate 
increase of approximately 23 percent for super-rural transports only. 

[E] July 1, 2004 through December 31, 2004. 

[End of figure] 

In addition to providing for a regional fee schedule, the MMA 
temporarily required higher payment rates for super-rural transports. 
As a result, base rate payments for transports originating in super- 
rural areas increased about 23 percent. The MMA also provided for a 25 
percent increase in the mileage rate for every ambulance mile traveled 
exceeding 50. Finally, the MMA required an increase in payment rates 
for mileage and transports originating in urban and rural areas by 1 
percent and 2 percent, respectively. (See table 1 for MMA temporary 
payment provisions and their expiration dates.) 

Table 1: Summary of MMA Temporary Payment Provisions, Implemented July 
1, 2004: 

Payment provision: The regional fee schedule; 
Expiration date: December 31, 2009. 

Payment provision: An increase in the base rate for super-rural 
transports[A]; 
Expiration date: December 31, 2009. 

Payment provision: A 25 percent increase in the urban and rural mileage 
rate for every ambulance mile exceeding 50; 
Expiration date: December 31, 2008. 

Payment provision: A 1 percent and 2 percent increase for urban and 
rural transports and mileage, respectively; 
Expiration date: December 31, 2006. 

Source: 42 U.S.C. § 1395m(l)(10)-(13). 

[A] CMS determined that this increase would be approximately 23 
percent. 69 Fed. Reg. 40288 (July 1, 2004). 

[End of table] 

Medicare Margins and Costs: 

A provider's Medicare margin under a fee schedule generally depends on 
whether the provider's costs of delivering a service are below its 
Medicare payments for the service. Under the ambulance national fee 
schedule, providers with costs per transport less than the Medicare 
payment for that transport are able to retain the difference between 
the fee schedule payment and their costs per transport. Likewise, 
providers with costs per transport above the national fee schedule 
payment will lose the difference between the Medicare payment and their 
costs per transport. Therefore, ambulance providers that can control 
their costs per transport may have an advantage over those that cannot 
control their costs per transport. 

A 2003 GAO study found that transport costs are likely to be higher in 
less densely populated rural areas because rural providers furnish 
fewer transports and because fewer transports were linked to higher 
costs per transport.[Footnote 34] As a result, we recommended that CMS 
adjust payments for transports in rural counties with particularly low 
population density to help ensure Medicare beneficiaries' access to 
ambulance services in those areas. Subsequently, the MMA increased 
payments for super-rural transports from July 1, 2004, through December 
31, 2009. The report also found that the majority of ambulance 
providers' costs were related to readiness--the availability of 
ambulance and crew for immediate emergency response--and were fixed 
costs. Fixed costs, such as staff on call, vehicles, building space, 
and administration, generally do not increase as the number of 
transports increases. Fuel costs and supplies are not fixed costs 
because they increase with the number of transports. 

Costs per Transport Were Highly Variable, Reflecting Differences in 
Certain Provider Characteristics: 

Costs of ground ambulance transports were highly variable across 
providers without shared costs; an average ambulance transport ranged 
from a low of $99 to a high of $1,218 during 2004, the year for which 
we gathered data. The variability of costs per transport reflected 
differences in certain characteristics--volume and mix of transports; 
service areas (urban, rural, and super-rural); productivity, which we 
defined as transports per staffed hour; and amount of local tax 
support.[Footnote 35] As expected, low volume, a greater percentage of 
ALS and more complex transports, and more transports from super-rural 
areas were key characteristics that helped explain why some providers 
without shared costs had higher costs per transport. Two other provider 
characteristics--productivity and amount of local tax support--were 
also associated with higher costs per transport for providers without 
shared costs. Other provider and local area characteristics--such as 
type of provider, region, building rent, and price of fuel--did not 
significantly affect average costs per transport among providers 
without shared costs. 

Providers' Reported Costs per Transport Were Highly Variable: 

Providers' average costs for a ground ambulance transport varied from 
$99 to $1,218--a range of more than $1,100--across providers without 
shared costs in 2004. Figure 3 shows the wide variation in the reported 
costs per transport among providers without shared costs. Five percent 
of providers without shared costs had average costs per transport that 
were less than $152, while 5 percent of providers had average costs per 
transport more than $913. From our sample of providers without shared 
costs, we estimated the average cost per transport at $415, with a 95 
percent confidence interval--the range within which we expect the 
population average cost per transport to fall 95 percent of the time-- 
of $381 to $450.[Footnote 36] This means that the actual average cost 
per transport across ambulance providers in the United States without 
shared costs was from $381 to $450 in 2004. 

Figure 3: Distribution of Cost per Transport for Providers without 
Shared Costs in 2004: 

[See PDF for image] 

Source: 2005 GAO Survey of Ambulance Services. 

Note: Based on a sample of 215 providers, weighted to represent more 
than 5,200 providers in the United States that did not share costs with 
nonambulance services. 

[End of figure] 

When we categorized providers without shared costs by service area and 
compared the average costs per transport across the groups, average 
cost per transport among super-rural providers was statistically 
significantly different from that of urban providers, but rural 
providers' average cost per transport was not statistically 
significantly different from that of urban providers. The average cost 
per transport for super-rural providers without shared costs was 
$538,[Footnote 37] statistically significantly different from the $370 
average cost per transport for urban providers without shared costs. 
The 95 percent confidence interval for average costs per transport 
among super-rural providers without shared costs ranged from $448 to 
$628 and among urban providers without shared costs ranged from $326 to 
$414. The average cost per transport among rural providers without 
shared costs was $409 within a confidence interval spanning $354 to 
$465, an interval that overlapped with the average costs per transport 
estimates for both urban and super-rural providers without shared 
costs. 

Certain Provider Characteristics Contributed to Differences in Costs 
per Transport: 

The variability of costs per transport among providers without shared 
costs reflected differences in certain provider characteristics. The 
provider characteristics that contributed to significant differences in 
costs per transport were volume, mix of transports, service area, 
productivity, and amount of local tax support.[Footnote 38] (See table 
2 and app. I for a full description of our methods.) 

Table 2: Estimated Average Cost per Transport for Provider 
Characteristics That Affect Costs: 

Provider characteristics: Volume of transports; 
Assigned values: 2,000 or less; 
Estimated average cost per transport[A]: $464. 

Assigned values: 2,001 - 3,000; 
Estimated average cost per transport[A]: 388. 

Assigned values: 3,001 - 4,000; 
Estimated average cost per transport[A]: 327. 

Assigned values: 4,001 - 5,000; 
Estimated average cost per transport[A]: 330. 

Assigned values: 5,001 - 6,000; 
Estimated average cost per transport[A]: 276. 

Assigned values: 6,001 or more; 
Estimated average cost per transport[A]: 330. 

Provider characteristics: Mix of transports; 
Assigned values: Only BLS; 
Estimated average cost per transport[A]: 360. 

Assigned values: ALS or more intensive services; 
Estimated average cost per transport[A]: 476. 

Provider characteristics: Service area; 
Assigned values: Urban only; 
Estimated average cost per transport[A]: 358. 

Assigned values: Rural only; 
Estimated average cost per transport[A]: 420[B]. 

Assigned values: Super-rural only; 
Estimated average cost per transport[A]: 545. 

Provider characteristics: Productivity--transports per staffed hour; 
Assigned values: 1 transport per 8 hours (0.12); 
Estimated average cost per transport[A]: 437. 

Assigned values: 5 transports per 8 hours (0.64); 
Estimated average cost per transport[A]: 386. 

Provider characteristics: Local tax support; 
Assigned values: No tax support; 
Estimated average cost per transport[A]: 392. 

Assigned values: Local tax support as 81% of revenues; 
Estimated average cost per transport[A]: 632. 

Sources: GAO analysis of 2005 GAO Survey of Ambulance Services and 2004 
Medicare claims. 

[A] The estimated average cost per transport is for providers without 
shared costs. It was created by assigning a value to the provider 
characteristic of interest for all cases and using the national average 
value for the other characteristics. See app. I. for detailed 
information about our methods. 

[B] The estimated cost of a rural transport was not significantly 
different from the estimated cost of an urban transport. 

[End of table] 

Providers without shared costs that had lower transport volumes 
generally had higher average costs than providers without shared costs 
with higher transport volumes. Our analysis affirms the finding of our 
prior work, that volume of transports was the main characteristic 
affecting providers' costs per transport.[Footnote 39] Because most 
ambulance costs are fixed, and therefore do not increase significantly 
when a provider completes more transports, it is expected that as the 
number of transports provided increases, associated costs per transport 
will be lower. In 2004, the volume of transports completed by a 
provider without shared costs ranged from 21 to more than 50,000. 
Estimated average cost per transport is reduced, from $464 for 
providers without shared costs completing 2,000 or fewer transports a 
year to $327 for those completing from 3,001 to 4,000 transports. 
Although estimated average cost per transport is slightly higher for 
4,001 to 5,000 transports, rising to $330, and again above 6,000 
transports, rising to $330, every other volume category of provider 
without shared costs had lower estimated average costs per transport 
than the lowest volume group of 2,000 transports or less. 

Also, as expected, we observed that average costs per transport were 
higher for providers without shared costs that also had a greater 
percentage of ALS and more complex transports (compared with BLS 
transports) and those with a greater percentage of super-rural 
transports (compared with urban transports)--two characteristics 
incorporated into the national fee schedule to account for the 
additional costliness associated with more intensive services and 
isolated service areas. ALS and more complex transports completed by 
providers without shared costs ranged from 0 to 100 percent of 
transports provided. We estimated that providers without shared costs 
specializing in ALS and more complex transport services had average 
costs of $476, which was 32 percent higher than providers without 
shared costs specializing in BLS transport services. We estimated that 
the average cost of a super-rural transport was $545,[Footnote 40] 
while the average cost of an urban transport was $358. Rural transports 
were not significantly higher cost than urban transports. 

Two other provider characteristics--productivity and amount of local 
tax support--were also associated with higher average costs per 
transport. Costs were higher when providers without shared costs had 
lower productivity or a lower ratio of transports per staffed hour. The 
average level of productivity for providers without shared costs was 
0.12, or about one transport per 8 staffed hours, and had an estimated 
average cost per transport of $437. The second highest level of 
productivity for these providers was 0.64, or more than five transports 
per 8 staffed hours, and had an estimated average cost per transport of 
$386. The impact of productivity on average costs may be explained by 
the fixed costs incurred in maintaining readiness--having an ambulance 
and crew available to respond to emergency calls. Although providers 
may have discretion about staffing and the ability to make backup 
arrangements to substitute for additional staff, not all providers can 
increase productivity by increasing the number of transports they 
provide or reducing the number of staffed ambulance hours. For example, 
providers that operate in small or isolated communities with one 
ambulance on call may serve only their own community's needs and may 
not be able to expand their service area or increase their volume of 
ambulance transports. 

Average costs were also higher for providers without shared costs that 
derived a larger percentage of their total revenues from local tax 
support. Among providers without shared costs, those with the largest 
percentage of revenues from local tax support (81 percent of revenues) 
had estimated average costs that were $240, or 61 percent, above those 
with no local tax support. Again, this effect was independent of volume 
and mix of transports, service area, cost of labor, use of volunteers, 
productivity, and other provider and local area characteristics. The 
relationship between greater local tax support and higher average costs 
may be explained as the income effect: if an organization has more 
money, it is able to and likely to spend more. Moreover, if costs 
increase without resulting in additional transports, the average cost 
per transport will increase. 

Characteristics that did not significantly contribute to the 
variability of average costs per transport among providers without 
shared costs included type of provider; a provider's region, as 
measured by the nine census divisions that defined the regional fee 
schedule; building rent; and the price of fuel. 

Average Payments under the National Fee Schedule Will Be Greater Than 
Average Historical Payments, but Providers' Expected Medicare Margins 
Will Vary Greatly: 

Average ambulance national fee schedule payments in 2010 are estimated 
to be 3 percent higher overall than payments in 2001, after adjusting 
for inflation and assuming that providers bill the maximum amounts 
allowed. However, the Medicare margins of providers without shared 
costs--whether they make a profit or a loss on Medicare transports-- 
will vary under the national fee schedule. We cannot assess whether 
providers without shared costs will break even, lose, or profit on 
average under the ambulance national fee schedule in 2010 after all of 
the MMA temporary payment provisions have expired, because the 95 
percent confidence interval surrounding the average Medicare margin 
spans from negative 14 percent to positive 2 percent. However, across 
all providers without shared costs, we estimate that 39 to 56 percent 
will have average Medicare payments above their average costs per 
transport under the ambulance national fee schedule even after all of 
the MMA provisions expire. 

Average Payments under the National Fee Schedule Will Be Greater Than 
Average Historical Payments and Will Be Redistributed from Urban 
Transports to Rural and Super-Rural Transports: 

Compared with average historical payments in 2001, average payments 
under the national fee schedule in 2010 will be 3 percent higher, after 
adjusting for inflation and assuming providers will bill the maximum 
amount allowed under the national fee schedule. According to our 
analysis, urban transports will experience a decrease in payments, on 
average, while rural and super-rural transports will receive an 
increase that is greater than the overall increase. Average payments 
for rural and super-rural transports will increase 20 percent and 15 
percent, respectively, while average payments for urban transports will 
decline 3 percent compared with average payments prior to the fee 
schedule. (See table 3.) 

Table 3: Payments Prior to and under the National Fee Schedule after 
MMA Provisions Expire: 

Transports: Urban; 
Payment prior to national fee schedule: $309; 
Payments under national fee schedule: $301; Percentage change: -3. 

Transports: Rural; 
Payment prior to national fee schedule: 303; 
Payments under national fee schedule: 363; Percentage change: 20. 

Transports: Super-rural; 
Payment prior to national fee schedule: 379; 
Payments under national fee schedule: 437; Percentage change: 15. 

Transports: National; 
Payment prior to national fee schedule: 309; 
Payments under national fee schedule: 319; Percentage change: 3. 

Sources: GAO analysis of 2001 and 2004 Medicare claims. 

Notes: All payments are in 2004 dollars. Payments under the national 
fee schedule assume that providers charge the maximum allowed amount. 

[End of table] 

Expected Medicare Margins Will Vary Greatly: 

After all of the MMA temporary payment provisions expire, expected 
Medicare margins under the national fee schedule will vary greatly 
among providers without shared costs. When we compared expected 
payments under the national fee schedule in 2010 with providers' costs 
per transport, the resulting Medicare margins ranged from negative 194 
percent for one provider to positive 76 percent for another provider. 
This wide difference is related to the great variability in reported 
costs among providers without shared costs. 

Among providers without shared costs, we estimated that the average 
Medicare margin, or the average percentage difference between these 
providers' Medicare payments and their costs, will be about negative 6 
percent with a 95 percent confidence interval from negative 14 percent 
to positive 2 percent. (See table 4.) This span in the confidence 
interval means we cannot assess whether providers without shared costs 
would break even, lose, or profit, on average, under the national fee 
schedule in 2010 after all of the MMA temporary payment provisions have 
expired. Similarly, we estimated that the average Medicare margin for 
urban, rural, and super-rural providers without shared costs will be 
negative under the national fee schedule, but each estimate will fall 
within a broader confidence interval range that includes positive 
Medicare margins. The estimated Medicare margin for an urban provider 
without shared costs will be negative 18 percent to positive 6 percent, 
while the estimated Medicare margin for a rural provider without shared 
costs will be negative 13 percent to positive 12 percent. We estimated 
that a super-rural provider without shared costs will have an estimated 
Medicare margin from negative 35 percent to positive 2 percent, making 
it more likely that the average Medicare margin for any given super- 
rural provider without shared costs would be negative rather than 
positive. However, given the confidence intervals surrounding the 
estimated average margin for each subset of providers without shared 
costs and the lack of statistical difference between them, we cannot 
conclude with certainty that any subset of providers would have 
significantly better or worse financial experience under Medicare's 
national fee schedule than another. Rather, we can conclude only that 
Medicare margins are likely to vary even among urban, rural, and super- 
rural providers without shared costs. 

Table 4: Expected Average Medicare Margins under the National Fee 
Schedule for Providers without Shared Costs in 2004 Dollars: 

Providers' predominate service area: Urban; 
Payment under national fee schedule: $350; 
Average cost (95 percent confidence interval): $370 ( 326 to 414); 
Providers' average Medicare margins in percentage (95 percent 
confidence interval): -6 (-18 to 6 ). 

Providers' predominate service area: Rural; 
Payment under national fee schedule: 408; 
Average cost (95 percent confidence interval): 409 (354 to 465); 
Providers' average Medicare margins in percentage (95 percent 
confidence interval): -1 (-13 to 12). 

Providers' predominate service area: Super-rural; 
Payment under national fee schedule: 471; 
Average cost (95 percent confidence interval): 538 ( 448 to 628); 
Providers' average Medicare margins in percentage (95 percent 
confidence interval): -17 (-35 to 2). 

Providers' predominate service area: All; 
Payment under national fee schedule: 394; 
Average cost (95 percent confidence interval): 415 ( 381 to 450); 
Providers' average Medicare margins in percentage (95 percent 
confidence interval): -6 (-14 to 2). 

Sources: GAO analysis of 2005 GAO Survey of Ambulance Services and 2004 
Medicare claims. 

Notes: All payments and costs are in 2004 dollars. Payments under the 
national fee schedule assume that providers charge the maximum allowed 
amount. The range of the confidence interval is affected by the 
variability of costs per transport within the sample and the size of 
the sample. Providers' average Medicare margin is the average margin 
across all providers in the sample. 

[End of table] 

When we assessed the likely experiences of all providers without shared 
costs under the national fee schedule after all of the MMA temporary 
payment provisions expire, we estimated that 39 to 56 percent of them 
will have positive Medicare margins. Among urban and rural providers, 
39 to 65 percent and 34 to 64 percent, respectively, will have positive 
Medicare margins, according to our estimations. Among super-rural 
providers, however, we estimate that 18 to 51 percent will have 
positive Medicare margins, while 49 to 82 percent would have zero or 
negative Medicare margins. (See fig. 4.) The breadth of these 
confidence intervals reflects the variability of providers' costs in 
2004 and expected financial experience under the national fee schedule 
after MMA temporary payment provisions expire. 

Figure 4: Expected Medicare Margins for Urban, Rural, and Super-Rural 
Providers without Shared Costs: 

[See PDF for image] 

Source: GAO analysis of 2005 GAO Survey of Ambulance Services and 2004 
Medicare claims. 

Notes: The range of the confidence interval is affected by the 
variability of costs per transport within the sample and the size of 
the sample. Percentages and confidence intervals are rounded. 

[End of figure] 

MMA Provisions Resulted in Greater Average Payments for Higher-Cost 
Super-Rural Transports and Adjusted Payments Regionally Where No 
Significant Cost Differences Were Observed: 

The MMA temporary payment provisions, which were implemented by CMS in 
the second half of 2004, resulted in raised ambulance average payments 
overall, particularly for super-rural transports, which we found 
typically more costly to provide. Payment adjustments under the MMA's 
regional fee schedule were not justified on the basis of regional cost 
differences, as we did not find significant differences in average cost 
per transport across regions. 

Payment Increases Were Targeted to Higher-Cost Super-Rural Transports: 

When we compared ambulance payments in the first half of 2004, prior to 
the implementation of the MMA provisions, with ambulance payments in 
the second half of 2004, after the temporary payment provisions were 
implemented and had their maximum effect, we found that payments, on 
average, increased by 5 percent overall. Super-rural transports 
received more substantial payment increases than urban or rural 
transports. (See table 5.) After MMA temporary payment provisions were 
implemented, average payments for urban and rural transports increased 
by 5 and 3 percent, respectively, while average payments for super- 
rural transports rose by 12 percent, compared with average payments 
before the MMA provisions were implemented. Increased payments for 
super-rural transports under the MMA were in keeping with our finding 
that super-rural transports were more costly than urban transports, 
independent of other characteristics that affected ambulance costs. 

Table 5: Average Payments prior to MMA Implementation and after 
Implementation: 

Transports: Urban; 
Average payment per transport prior to MMA: $306; 
Average payment per transport under MMA: $322; 
Average percentage change in payment under MMA: 5. 

Transports: Rural; 
Average payment per transport prior to MMA: 358; 
Average payment per transport under MMA: 370; 
Average percentage change in payment under MMA: 3. 

Transports: Super-rural; 
Average payment per transport prior to MMA: 442; 
Average payment per transport under MMA: 497; 
Average percentage change in payment under MMA: 12. 

Transports: National; 
Average payment per transport prior to MMA: 322; 
Average payment per transport under MMA: 338; 
Average percentage change in payment under MMA: 5. 

Source: GAO analysis of 2004 Medicare claims. 

Notes: The period prior to MMA implementation for which payments were 
computed was January 1, 2004 through June 30, 2004. The period after 
implementation of the MMA for which payments were computed was July 1, 
2004 through December 31, 2004. 

[End of table] 

Regional Payment Adjustments Required by the MMA Were Not Justified on 
the Basis of Regional Cost Differences: 

Regional payment adjustments under the MMA were not warranted on the 
basis of regional cost differences. The MMA required a regional fee 
schedule, which resulted in ambulance payments for similar services 
that differed based on the region where they were provided. When 
comparing average regional payments before the implementation of MMA 
provisions to payments after implementation of the MMA, when the 
regional fee schedule had its greatest effect, we found that average 
payments increased substantially for some regions but not others. (See 
table 6.) However, we found no significant differences in costs by 
region, after controlling for differences in volume and mix of 
transports, cost of labor, service area, and other characteristics that 
may have affected costs. The regional fee schedule is due to expire on 
December 31, 2009. 

Table 6: Percentage Changes in Average Payments prior to MMA 
Implementation and after Implementation, by Region: 

Region: Pacific; 
Percentage change in average payments: 19. 

Region: New England; 
Percentage change in average payments: 12. 

Region: Mountain; 
Percentage change in average payments: 8. 

Region: West South Central; 
Percentage change in average payments: 7. 

Region: West North Central; 
Percentage change in average payments: 3. 

Region: Middle Atlantic; 
Percentage change in average payments: 3. 

Region: South Atlantic; 
Percentage change in average payments: 1. 

Region: East North Central; 
Percentage change in average payments: 1. 

Region: East South Central; 
Percentage change in average payments: 0. 

Source: GAO analysis of 2004 Medicare claims. 

Notes: East South Central received a decrease in average payments of - 
0.04 percent, which rounds to 0 percent. The period prior to MMA 
implementation for which payments were computed was January 1, 2004 
through June 30, 2004. The period after implementation of the MMA for 
which payments were computed was July 1, 2004 through December 31, 
2004. See table 11 in app. I for a listing of the regions, or census 
divisions, and their corresponding states. 

[End of table] 

Medicare Beneficiaries' Use of Ambulance Transports Increased from 2001 
to 2004, Except in Super-Rural Areas: 

Nationally, the use of ambulance transports by Medicare beneficiaries 
increased by 16 percent, from 2001, the year before the transition to 
the national fee schedule began, to 2004, the year we studied. (See 
table 7.) Medicare beneficiaries' use of ambulance transports in urban 
areas experienced the greatest growth, 19 percent, while rural areas 
experienced a modest increase of 6 percent. However, Medicare 
transports per 1,000 beneficiaries in super-rural areas decreased by 8 
percent. The decrease in Medicare beneficiaries' use of ambulance 
transports in super-rural areas was driven mostly by a decline in the 
volume of transports rather than any significant change in the number 
of beneficiaries or the demographic characteristics of beneficiaries 
residing in super-rural areas. For example, factors such as age, race, 
and gender remained stable in the super-rural Medicare population. 
Meanwhile, Medicare beneficiaries' use of ambulance transports 
increased in all regions from 2001 to 2004, including one region that 
had a decrease in average payments under the MMA compared with before 
the implementation of MMA payment provisions. 

Table 7: Ambulance Transports per 1,000 Beneficiaries in Urban, Rural, 
and Super-Rural Areas: 

Area: Urban; 
Transports, 2001: 371; 
Transports, 2004: 443; 
Percentage change, 2001-2004: 19. 

Area: Rural; 
Transports, 2001: 372; 
Transports, 2004: 396; 
Percentage change, 2001-2004: 6. 

Area: Super-rural; 
Transports, 2001: 264; 
Transports, 2004: 244; 
Percentage change, 2001-2004: -8. 

Area: Total; 
Transports, 2001: 364; 
Transports, 2004: 420; 
Percentage change, 2001-2004: 16. 

Sources: GAO analysis of 2001 and 2004 Medicare claims and CMS 
enrollment data. 

[End of table] 

Conclusions: 

The diversity of the ambulance industry is reflected in its range of 
organizations, services offered, staffing, revenue sources, and costs 
per transport. We found that certain ambulance provider 
characteristics, such as volume, mix of transports, service area, 
productivity, and amount of local tax support, affected the cost per 
transport of providers without shared costs. For some providers and 
communities, these characteristics may be self-determined and may 
reflect those communities' preferences for readiness, quality 
standards, and ambulance services offered. For example, some 
communities may prefer to fund the greater costs of operating at a 
higher level of readiness or being equipped with more sophisticated 
transport vehicles and more highly trained staff. Other providers and 
communities have little or no control over the characteristics that 
affect providers' cost per transport. These communities, particularly 
more rural areas with low population density, may be constrained by 
local conditions, including their financial resources. Therefore, local 
conditions and community preferences may explain some of the predicted 
variability in the financial experience of providers without shared 
costs under the Medicare national fee schedule. 

We are unable to discern whether providers without shared costs would 
be compensated appropriately under the national fee schedule for two 
reasons. First, when providers experience the national fee schedule 
payments in 2010 after all of the MMA temporary payment provisions 
expire, they may make changes to control or reduce their costs. The 
cost data we collected were from 2004 and may not reflect any changes 
providers may make to control or reduce their costs in response to the 
national fee schedule. Second, we did not assess if Medicare 
beneficiaries are receiving quality care that is delivered efficiently. 
There are no national performance standards to use as benchmarks for 
determining quality and efficiency of services or for assessing whether 
providers could increase productivity by increasing the number of 
transports they provide or by reducing the number of staffed ambulance 
hours. However, current efforts to develop a national data system and 
indicators for EMS systems may, in the future, yield useful tools for 
measuring efficiency and quality of ambulance services under the 
Medicare program. 

Based on our survey of ambulance costs, we were able to estimate that 
some providers without shared costs would have positive Medicare 
margins under the national fee schedule after the MMA provisions 
expire, while others would have negative Medicare margins. Among super- 
rural providers, we estimated that 18 to 51 percent would have positive 
Medicare margins. However, 49 to 82 percent would have zero or negative 
Medicare margins. Ideally, Medicare payments should be adequate to 
ensure beneficiary access to services while using the program's 
resources judiciously. The decline in use of super-rural ground 
ambulance transports from 2001 to 2004, a time when payments for super- 
rural transports were increased, suggests that Medicare payment levels 
may not be linked to the decreased utilization of transports in super- 
rural areas. However declining utilization coupled with potentially 
negative Medicare margins in super-rural areas, which could be 
exacerbated when the MMA provisions expire, raise questions as to 
whether Medicare payments will be adequate to support beneficiary 
access in super-rural areas. 

Recommendation for Executive Action: 

In light of the variability in ambulance providers' Medicare margins 
and the potential for negative margins to have an impact on beneficiary 
access, we recommend that the Administrator of CMS monitor utilization 
of ambulance transports to ensure that Medicare payments are adequate 
to provide for beneficiary access to ambulance services, particularly 
in super-rural areas. 

Agency and External Comments and Our Evaluation: 

We provided a draft of this report to CMS and to five associations that 
represent the ambulance industry: the American Ambulance Association, 
the National Association of State EMS Officials, the National Ambulance 
Coalition, the National Volunteer Fire Council, and the International 
Association of Fire Chiefs. CMS's written comments are reprinted in 
appendix II. 

CMS stated that, for the most part, the report reinforces its findings. 
CMS also stated that it agreed with our recommendation that the agency 
monitor utilization of ambulance transports to ensure that Medicare 
payments are adequate to provide for beneficiary access to ambulance 
services, particularly in super-rural areas. CMS noted that it would 
continue to monitor ambulance rates and would make adjustments should 
the original assumptions made during the development of the ambulance 
fee schedule need to be changed. In addition, CMS also highlighted its 
implementation of a refinement in the definition of rural areas that 
should enable rural areas within urban areas to receive the benefit of 
higher rural payments under the ambulance fee schedule. 

CMS noted that we should have discussed in our conclusions the 
implications of omitting "shared services" providers from our analysis, 
as these providers tend to have higher costs. As we discussed in the 
report, ambulance providers that could not separately report the costs 
of the ambulance portion of their business were excluded because their 
cost data were determined to be unreliable. Consequently, we have no 
basis or information to suggest that providers with shared services 
have higher or lower costs than other providers. 

Ambulance industry associations generally agreed with our findings. 
However, the associations raised various concerns regarding our 
calculations and assumptions. Two associations questioned the inclusion 
in our analysis of ambulance providers that used unpaid staff and 
suggested that it might have been more appropriate to focus on 
providers who bear the full cost of providing ambulance services. As we 
note in the report, use of unpaid staff by ambulance providers is 
widespread with an estimated two-fifths of the industry relying 
substantially on volunteers in 2004. Thus, in order for our analysis to 
be representative of ambulance providers, we included those that used 
volunteer staff. We recognize that use of volunteer staff affects 
ambulance providers' costs and included the percentage of volunteer 
hours as a control variable in our cost model. 

Two associations were concerned that we did not allow for the effect of 
Medicare bad debt in our analysis and may have therefore overestimated 
payments. We acknowledge that bad debt will affect the percentage of 
costs recoverable for providing ambulance services. However, we 
explicitly state that our payment estimates assume providers are paid 
the full Medicare payment amounts. It was beyond the scope of our study 
to estimate the effect of Medicare bad debt on ambulance payments or to 
determine the extent to which payments should be adjusted to reflect 
bad debt. 

Two associations expressed concern that our analysis showed a 3 percent 
increase in payments to ambulance providers with the transition from 
the historical payment system to the fully implemented national fee 
schedule and thought payments should have been relatively level. We 
note that our analysis incorporated increases in mileage rates over 
time that likely accounted for some of this increase. In addition, when 
simulating payments under the national fee schedule, we assumed that 
providers would bill Medicare for all services they were entitled to 
bill. However, as we noted in the report, nearly half of the industry 
indicated that a state or local government approves the fees they may 
charge, and some providers are required to bill Medicare less than the 
allowed amount and therefore do not receive the maximum Medicare 
payment allowed under the fee schedule. This discrepancy between the 
state or local allowed amount and the actual Medicare payment could 
also account for some of the 3 percent increase. 

CMS and the associations also provided technical comments and 
clarifications, which we incorporated as appropriate. 

We are sending copies of this report to other interested congressional 
committees and the Administrator of CMS. We will also provide copies to 
others upon request. The report will also be available at no charge on 
the GAO Web site at http://www.gao.gov. 

If you or your staff have any questions, please contact me at (202) 512-
7119 or kingk@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 III. 

Signed by: 

Kathleen M. King: 
Director, Health Care: 

[End of section] 

Appendix I: Data and Methods: 

This appendix describes, in detail, the data and methods we used to 
respond to our research objectives and evaluate ambulance providers' 
costs per transport and Medicare payments under the national ambulance 
fee schedule. We conducted a survey of ambulance costs to collect cost 
data. We relied on these survey data for much of our analyses of costs 
and supplemented our survey results with information from other 
sources, including Medicare claims data, as appropriate. We also 
analyzed Medicare claims data to determine the effect of the national 
fee schedule and the Medicare Prescription Drug, Improvement and 
Modernization Act of 2003 (MMA) temporary payment provisions on 
payments, as well as to describe changes in beneficiaries' use of 
ambulance transports over time. We tested the internal consistency and 
reliability of the data from our survey and other non-Medicare data 
sources and determined that all data sources were adequate for our 
purposes. We conducted our work from July 2004 through April 2007 in 
accordance with generally accepted government auditing standards. 

National Survey of Ground Ambulance Providers' Costs: 

To collect data on ground ambulance providers' costs, revenues, 
transports, and organizational characteristics for their most recently 
completed fiscal year, we mailed a survey to a nationally 
representative sample of 500 eligible ambulance service providers that 
billed Medicare in 2003.[Footnote 41] We used a two-stage sampling 
process to select a stratified,[Footnote 42] random sample of 
providers. In the first stage, we selected a preliminary sample of 
hospital and nonhospital-based providers for screening. In the second 
stage, we conducted telephone screening interviews to confirm 
eligibility for the study and, for nonhospital-based providers, to 
identify provider type. We obtained 321 completed surveys for a 
response rate of 64 percent. We excluded two cost outliers. We also 
excluded from our analysis providers that reported sharing ambulance 
costs with other institutions or other nonambulance services, 
including, but not limited to, all fire departments, after preliminary 
analysis revealed problems with the reliability of their reported 
costs. The resulting sample size was 215 providers. The results from 
our analysis are nationally representative of all Medicare ambulance 
providers that can distinguish their costs for providing ambulance 
services from the costs of other services they provide.[Footnote 43] 
However, the small sample size and the variability of reported costs 
reduce the precision of our estimates. 

Survey Instrument Development: 

To develop our survey instrument, we reviewed other survey instruments 
and analyses of ambulance cost data, consulted with experts in survey 
methods and the ambulance industry, and tested our survey instrument. 
We reviewed cost data collected in 1999 by Project HOPE Center for 
Health Affairs, which was a nonprofit health policy research 
organization, during a survey effort sponsored by the American 
Ambulance Association. We also reviewed other surveys of emergency 
medical services, as well as industry and association guidelines about 
emergency medical and ambulance services. In addition to surveying 
ambulance providers on their costs and revenues, we included questions 
to identify organizational and local area characteristics that might 
affect ambulance costs, such as the number of emergency transports and 
number of volunteer hours. 

Industry experts and a survey specialist reviewed and commented on the 
draft survey instrument. We conducted a pilot test of the survey with 
104 ambulance service providers as well. We redesigned and refined the 
instrument based on the experience of the pilot test. Then, to further 
refine the wording of our survey questions, we asked four types of 
ambulance service organizations and a former volunteer fire chief to 
pretest the instrument and point out any issues they noted. These 
pretests were conducted mostly by telephone--one pretest was in person. 

Sample Design: 

We developed separate lists for all hospital-based and nonhospital- 
based providers from information maintained by the Centers for Medicare 
& Medicaid Services (CMS), the agency that oversees the Medicare 
program. Next, we used a two-stage sampling process to select a 
stratified, random sample of providers. In the first stage, we sorted 
each list by census division and predominant service area (urban, 
rural, or super-rural), began sampling the list at a random starting 
point, and chose providers at regular intervals from their respective 
lists. This method of sampling implicitly included representation 
across census divisions and service areas. In the second stage, 
telephone screening interviews were conducted with each sampled 
provider to confirm eligibility for the study and, for nonhospital- 
based providers, to identify their type--volunteer, fire department, 
government, or freestanding. Once a provider's type was established, 
providers were stratified by type and randomly selected to ensure 
somewhat equal representation among all provider types. We developed 
initial sample rates for each type of provider and adjusted the rates 
midway through the screening process. Each provider representative was 
told at the close of the telephone screening call whether the provider 
had been selected to participate in our survey. Finally, a survey 
instrument was mailed to the selected providers. 

To identify nonhospital-based ambulance providers that billed Medicare, 
we contacted Medicare carriers[Footnote 44] for a list of ambulance 
providers and matched this list to 2003 carrier claims for ambulance 
services by provider identification numbers. Our nonhospital-based 
sample frame included 12,082 unique provider identification 
numbers.[Footnote 45] We later learned that a number of provider 
identification numbers in the nonhospital-based sample frame were 
duplicate entries for ambulance providers that had more than one 
Medicare provider identification number for the same location.[Footnote 
46] We analyzed the extent of duplication in the sample frame and 
reduced the estimated population size to 7,968. Although our sample 
frame included a substantial number of duplicate entries, there were 
only 28 duplicates among our first-stage sample of 900 nonhospital- 
based providers. The sample weights were adjusted to account for the 
duplicate entries in the sample frame. 

To identify the total number of hospital-based ambulance providers that 
billed Medicare, we matched Medicare's Provider of Service file with 
Medicare Part A inpatient and outpatient claims for ambulance services. 
We excluded skilled nursing facilities for a total of 828 hospital- 
based ambulance service providers. We then selected a first-stage 
sample of 150 hospital-based ambulance providers. 

To make our survey sample representative of all Medicare ambulance 
providers, the population from which the sample was drawn, we computed 
a sample weight for each respondent provider. The computation of the 
sample weight took into account the type of provider, sample rate, and 
the response rate for the type of provider--hospital, volunteer, fire 
department, government, or freestanding. 

Survey Administration: 

We contracted with CODA Inc., an independent survey research firm, to 
perform the telephone screening and administer the mailed survey 
instrument.[Footnote 47] The contractor screened the 900 nonhospital- 
based and 150 hospital-based providers we selected for our first-stage 
sample, and mailed the survey instrument to 500 eligible organizations, 
randomly sampled by provider type. In order to properly select our 
sample with even representation across all types of providers 
nationally, we designed our sampling strata using data collected from 
our pilot test. 

CODA Inc., administered the mailed surveys and conducted all follow-up 
and data coding in coordination with us. Our survey period began in 
April 2005 and ran through September 2005. The survey instruments were 
mailed using Federal Express and 2-day Priority Mail. Telephone contact 
was initiated 3 business days after the instrument was mailed to 
ascertain when the respondent could return the completed instrument. 
Prompting and follow-up requests for data were conducted by phone and 
occurred whenever providers' returned instruments were incomplete, 
vague, or included conflicting responses to key items. In some cases, 
follow-up requests involved multiple contacts, faxing survey 
instruments, and spending an hour or more on the phone with the 
respondent. On average, more than 4 phone calls were made for each 
respondent with as many as 20 calls made to one respondent. 

All requests for data were conducted by CODA Inc. staff following 
strict protocols that we developed. Respondents were encouraged to 
contact CODA Inc. and GAO via toll-free numbers, so that any questions 
or problems could be resolved. All survey data were double-key entered 
into an electronic file, and computer programs were checked for keying 
discrepancies and data inconsistencies. In all, we received 321 
completed surveys; this represents a response rate of 64 percent. 

Survey Data Validity and Reliability: 

In addition to the survey administration procedures described above, we 
took several measures to ensure that the data reported on the survey 
were valid and reliable.[Footnote 48] First, the survey instrument 
included items intended to validate the reported cost data. We also 
used strict protocols during follow-up to validate the reported cost 
data. For example, if respondents could not provide cost breakdowns by 
the categories listed in our survey instrument, a separate phone 
protocol was used to verify the reported cost data. 

Second, we tested the data for internal consistency and excluded cases 
when necessary. Computer analyses were performed to identify and, where 
possible, correct any inconsistencies in responses or other errors. We 
also excluded 2 providers that appeared to be cost outliers. These 
providers had costs per transport that were at least three standard 
deviations above the mean of the standard statistical distribution (the 
lognormal), and no other variables explained their extraordinary costs. 
Through our analyses, we determined that the costs reported by 
providers that shared costs with other institutions or offered other 
services appeared to be unreliable. We found the costs reported by 
these "shared costs" providers to be highly variable, which may reflect 
inconsistent methods for separating staff time and other resources 
across different services. Therefore, we excluded these providers-- 
including but not limited to all fire departments--from our analysis. 
The resulting sample size was 215 providers, representing a population 
of more than 5,200 providers without shared costs.[Footnote 49] 

Third, we compared information reported on the survey to information on 
Medicare claims submitted by respondents, such as the number of 
Medicare transports and percentage of emergency Medicare transports. 
All computer syntax was peer reviewed and verified by separate 
programmers to ensure that the syntax was written and executed 
correctly. 

We used providers' total costs and total transports reported on the 
survey to compute providers' average costs per transport. This cost 
information and other information about revenues and provider 
characteristics were used to model ambulance costs per transport. 
Although these survey data were self-reported and had not been audited, 
based on efforts to validate the data, computer testing, and 
corrections and comparisons with Medicare data, we have concluded that 
they were sufficiently valid and reliable for our purposes. 

Interpretation of Confidence Intervals and Analysis of Nonrespondents: 

All sample surveys are subject to sampling error--that is, the extent 
to which the survey results differ from what would have been obtained 
if we had collected responses from every ambulance provider in the 
country. Because we used a sample, it is only one of a large number of 
samples that we might have drawn. As each sample could have provided 
different estimates, we express our confidence in the precision of our 
particular sample's results as a 95 percent confidence 
interval.[Footnote 50] This is the interval that would contain the 
actual value for all providers for 95 percent of the samples we could 
have drawn. As a result, we are 95 percent confident that the reported 
confidence intervals based on the mailed survey include the true values 
for all providers. For this reason, all costs per transport and 
provider margins are reported with their confidence intervals. 

We also analyzed 2004 Medicare claims data for survey nonrespondents 
and compared this information with similar claims information for 
providers without shared costs in our sample. Nonrespondents served 
predominantly urban areas rather than rural or super-rural areas. On 
average, nonrespondents completed about half the number of Medicare 
transports in 2004 compared with providers without shared costs. 
Nonrespondents also had a higher percentage of basic life support (BLS) 
transports, as opposed to advanced life support (ALS) and more complex 
transports, and about the same percentage of nonemergency transports 
compared with providers without shared costs. 

It is unclear whether nonrespondents had higher or lower costs, on 
average, than providers without shared costs. In our regression 
analysis of the cost information for providers without shared costs, we 
found that those with fewer transports per year generally had higher 
costs per transport than those with more transports. Providers without 
shared costs that served predominately urban areas had lower costs 
compared with providers serving super-rural areas. We also found that 
providers with higher percentages of BLS transports had lower costs 
compared to providers with no BLS transports. Although nonrespondents' 
characteristics differed from those of providers without shared costs, 
these differences were associated with both higher and lower costs 
among providers without shared costs. Therefore, we have no basis for 
concluding that nonresponse has biased our cost estimates in any 
particular direction. 

Modeling Ambulance Costs per Transport: 

We analyzed the relationship between providers' average costs of 
ambulance transports and the provider and local area characteristics 
that may have affected their average costs. We used regression analysis 
to examine the effect of these characteristics on providers' costs per 
transport. We then used the results from this regression analysis to 
predict the average costs per transport across all providers without 
shared costs, based on five key provider characteristics: (1) total 
transports per year, (2) percentage of BLS Medicare transports, (3) 
percentage of Medicare transports in rural and super-rural areas, (4) 
number of ambulance transports per staffed hour, and (5) amount of 
revenue derived from local tax support. 

To perform these analyses, we identified measures and data sources for 
each of the provider and local area characteristics that we identified 
as potentially contributing to differences in costs per transport. A 
summary of these characteristics, measures we used to assess their 
potential relationship to costs per transport, and data sources used is 
presented in table 8. 

Table 8: Provider and Local Area Characteristics Included in Analysis 
of Average Cost per Transport, 2004: 

Characteristic: Transport volume; 
Measure: Indicator of volume group: <=2,000, 2001-3000, 3001-4,000, 
4001-5000, 5001-6000, 6001+; 
Source of data: GAO Survey of Ambulance Services. 

Characteristic: Type of provider; 
Measure: Staffing and organizational structure; 
Source of data: GAO Survey of Ambulance Services. 

Characteristic: Productivity; 
Measure: Transports per staffed ambulance hour[A]; 
Source of data: GAO Survey of Ambulance Services. 

Characteristic: Price of fuel; 
Measure: Annualized average retail price of fuel in the provider's zip 
code; 
Source of data: Oil Price Information Service. 

Characteristic: Building rent; 
Measure: Price per square foot of office space in the provider's zip 
code; 
Source of data: United States Postal Service. 

Characteristic: Cost of labor used by providers; 
Measure: Geographic practice cost index (GPCI)[B]; 
Source of data: Medicare Ambulance Fee Schedule Public Use File. 

Measure: Percentage of volunteer hours, including on-call time; 
Source of data: GAO Survey of Ambulance Services. 

Characteristic: Revenue derived from local tax support; 
Measure: Community tax support as a percentage of total ambulance 
provider revenue[C]; 
Source of data: GAO Survey of Ambulance Services. 

Characteristic: Regulation of fees; 
Measure: State or local government approval of fees charged; 
Source of data: GAO Survey of Ambulance Services. 

Characteristic: Mix and intensity of transports; 
Measure: Percentage of emergency transports; 
Source of data: GAO Survey of Ambulance Services. 

Measure: Percentage of Medicare transports that are BLS; 
Source of data: Medicare claims. 

Characteristic: Service area; 
Measure: Percentage of Medicare transports that are rural; 
Source of data: Medicare claims. 

Measure: Percentage of Medicare transports that are super-rural; 
Source of data: Medicare claims. 

Characteristic: Census division in which the ambulance provider was 
located; 
Measure: Census Division 1 - New England; 
Source of data: Medicare claims. 

Measure: Census Division 2 - Middle Atlantic; 
Source of data: Medicare claims. 

Measure: Census Division 3 - East North Central; 
Source of data: Medicare claims. 

Measure: Census Division 4 - West North Central; 
Source of data: Medicare claims. 

Measure: Census Division 5 - South Atlantic; 
Source of data: Medicare claims. 

Measure: Census Division 6 - East South Central; 
Source of data: Medicare claims. 

Measure: Census Division 7 - West South Central; 
Source of data: Medicare claims. 

Measure: Census Division 8 - Mountain; 
Source of data: Medicare claims. 

Measure: Census Division 9 - Pacific; 
Source of data: Medicare claims. 

Sources: 2005 GAO Survey of Ambulance Services, Oil Price Information 
Service, United States Postal Service, Medicare Ambulance Fee Schedule 
Public Use File, and Medicare claims. 

[A] This measure was computed as the total number of transports in the 
year divided by the total number of hours that an ambulance and crew 
were staffed and available to respond to an emergency call. This 
measure is also known as unit hours of utilization. 

[B] The practice expense portion of the physician fee schedule's GPCI 
is used to adjust the ambulance national fee schedule. 

[C] This measure was computed as the total amount of revenues derived 
from community tax support divided by total revenues. 

[End of table] 

Average Cost Regression Analysis--Methods and Results: 

Our regression analysis modeled the average cost of a transport at the 
provider level as a function of the provider and local area 
characteristics described above. We modeled the lognormal distribution 
of average cost per transport for a provider, which was calculated as 
the log of total costs divided by the total number of transports for 
that provider. We estimated the model using least squares[Footnote 51] 
and applied the appropriate sample weights. We used the statistical 
program SUDAAN® in SAS to estimate this model, which takes account of 
the sample stratification and weighting to obtain appropriate parameter 
estimates and standard errors. We tested for and found no specification 
problems in the model resulting from heteroscedasticity, 
misspecification, or evidence of a particular observation having undue 
influence. 

Tables 9 and 10 show the regression results for estimating the 
determinants of cost per transport. We ran two sets of average cost 
regressions. The first regression results, in table 9, do not include 
two provider characteristics, (1) transports per staffed ambulance hour 
and (2) community tax support as a percentage of total ambulance 
provider revenue, because data on these variables were not available 
for a number of providers in our sample. Including these variables as 
explanatory independent variables reduces the number of observations in 
the regression from 205 to 157. Nevertheless, we wanted to measure the 
impact of these policy variables, in addition to maintaining a larger 
sample size, in order to have a more precise estimate of the impact of 
the other explanatory variables. Therefore, we ran the same regression 
model with and without these two explanatory variables. 

Table 9: Results for Average Cost of an Ambulance Transport Regression-
-Estimated Effects of Selected Provider and Local Area Characteristics 
on the Average Cost of Ambulance Transports for Providers, Not 
Including Impact of Productivity and Community Tax Support: 

Characteristic: Transport volume[A]; 
Variable used to measure characteristic: 2,001-3,000 per year; 
Parameter estimate: -0.18; 
t- value: -1.70*. 

Variable used to measure characteristic: 3,001-4,000 per year; 
Parameter estimate: -0.35; 
t-value: -2.98***. 

Variable used to measure characteristic: 4,001-5,000 per year; 
Parameter estimate: -0.34; 
t-value: -3.33***. 

Variable used to measure characteristic: 5,001-6,000 per year; 
Parameter estimate: -0.52; 
t-value: -2.71***. 

Variable used to measure characteristic: 6,001+ per year; 
Parameter estimate:  -0.34; 
t-value: -2.86***. 

Characteristic: Cost of labor used by providers; 
Variable used to measure characteristic: GPCI; 
Parameter estimate: 1.34; 
t-value: 2.27**. 

Variable used to measure characteristic: Percentage of volunteer hours, 
including on-call time; 
Parameter estimate: -0.28; 
t-value: -2.73***. 

Characteristic: Regulation of fees; 
Variable used to measure characteristic: State or local government 
approval of fees charged; 
Parameter estimate: -0.08; 
t-value: -1.00. 

Characteristic: Mix and intensity of transports; 
Variable used to measure characteristic: Percentage of emergency 
transports; 
Parameter estimate: 0.17; 
t-value: 0.99. 

Variable used to measure characteristic: Percentage of Medicare 
transports that are BLS; 
Parameter estimate: -0.28; 
t-value: -2.39**. 

Characteristic: Service area[B]; 
Variable used to measure characteristic: Percentage of Medicare 
transports that are rural; 
Parameter estimate: 0.16; 
t-value: 1.61. 

Variable used to measure characteristic: Percentage of Medicare 
transports that are super-rural; 
Parameter estimate: 0.42; 
t-value: 3.11***. 

Characteristic: Census division in which the ambulance provider was 
located[C]; 
Variable used to measure characteristic: Census Division 1 - New 
England; 
Parameter estimate: -0.27; 
t-value: -1.66*. 

Variable used to measure characteristic: Census Division 2 - Middle 
Atlantic; 
Parameter estimate: -0.24; 
t-value: -1.66*. 

Variable used to measure characteristic: Census Division 3 - East North 
Central; 
Parameter estimate: 0.23; 
t-value: 1.61. 

Variable used to measure characteristic: Census Division 5 - South 
Atlantic; 
Parameter estimate: 0.07; 
t-value: 0.41. 

Variable used to measure characteristic: Census Division 6 - East South 
Central; 
Parameter estimate: - 0.07; 
t-value: -0.43. 

Variable used to measure characteristic: Census Division 7 - West South 
Central; 
Parameter estimate: - 0.16; 
t-value: -0.85. 

Variable used to measure characteristic: Census Division 8 - Mountain; 
Parameter estimate: 0.15; 
t- value: 0.97. 

Variable used to measure characteristic: Census Division 9 - Pacific; 
Parameter estimate: -0.27; 
t- value: -1.39. 

Variable used to measure characteristic: Intercept; 
Parameter estimate: 4.83; 
t-value: 8.73***. 

Variable used to measure characteristic: R-squared; 
Parameter estimate: 0.33; 
t-value: [Empty]. 

Variable used to measure characteristic: Observations; 
Parameter estimate: 205; 
t-value: [Empty]. 

Sources: GAO analysis of 2005 GAO Survey of Ambulance Services, 
Medicare Ambulance Fee Schedule Public Use File, and Medicare claims. 

*** significant at the 1 percent level. 

** significant at the 5 percent level. 

* significant at the 10 percent level. 

[A] Transports less than or equal to 2,000 per year is the excluded 
category. In order for the regression model's parameters to be 
estimated, we needed to exclude one of the transport volume categories. 

[B] The percentage of Medicare transports that are urban was the 
excluded category. In order for the regression model's parameters to be 
estimated, we needed to exclude one of the service area categories. 

[C] The West North Central 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. 

[End of table] 

Table 10: Results for Average Cost of an Ambulance Transport 
Regression--Estimated Effects of Selected Provider and Local Area 
Characteristics on the Average Cost of Ambulance Transports for 
Providers, Including Impact of Productivity and Community Tax Support: 

Characteristic: Transport volume[A]; 
Variable used to measure characteristic: 2,001-3,000 per year; 
Parameter estimate: -0.17; t- value: -1.84*. 

Variable used to measure characteristic: 3,001-4,000 per year; 
Parameter estimate: -0.30; 
t-value: -2.69***. 

Variable used to measure characteristic: 4,001-5,000 per year; 
Parameter estimate: -0.34; 
t-value: -3.40***. 

Variable used to measure characteristic: 5,001-6,000 per year; 
Parameter estimate: -0.31; 
t-value: -1.72*. 

Variable used to measure characteristic: 6,001+ per year; 
Parameter estimate: -0.22; 
t-value: -1.92*. 

Characteristic: Cost of labor used by providers; 
Variable used to measure characteristic: GPCI; 
Parameter estimate: 1.29; 
t-value: 2.62***. 

Variable used to measure characteristic: Percentage of volunteer hours, 
including on-call time; 
Parameter estimate: -0.10; 
t-value: -0.79. 

Characteristic: Regulation of fees; 
Variable used to measure characteristic: State or local government 
approval of fees charged; 
Parameter estimate: -0.11; 
t-value: -1.49. 

Characteristic: Mix and intensity of transports; 
Variable used to measure characteristic: Percentage of emergency 
transports; 
Parameter estimate: 0.06; 
t-value: 0.29. 

Variable used to measure characteristic: Percentage of Medicare 
transports that are BLS; 
Parameter estimate: -0.31; 
t-value: -2.25**. 

Characteristic: Service area[B]; 
Variable used to measure characteristic: Percentage of Medicare 
transports that are rural; 
Parameter estimate: 0.06; 
t-value: 0.67. 

Variable used to measure characteristic: Percentage of Medicare 
transports that are super-rural; 
Parameter estimate: 0.26; 
t-value: 1.93*. 

Characteristic: Productivity; 
Variable used to measure characteristic: Transports per staffed 
ambulance hour; 
Parameter estimate: -0.24; 
t- value: -2.76***. 

Characteristic: Community tax support; 
Variable used to measure characteristic: Community tax support as a 
percentage of total ambulance provider revenue; 
Parameter estimate: 0.59; 
t-value: 3.81***. 

Characteristic: Census division in which the ambulance provider was 
located[C]; 
Variable used to measure characteristic: Census Division 1 - New 
England; 
Parameter estimate: -0.24; 
t-value: -1.55. 

Variable used to measure characteristic: Census Division 2 - Middle 
Atlantic; 
Parameter estimate: -0.24; 
t-value: -1.42. 

Variable used to measure characteristic: Census Division 3 - East North 
Central; 
Parameter estimate: 0.17; 
t-value: 1.04. 

Variable used to measure characteristic: Census Division 5 - South 
Atlantic; 
Parameter estimate: 0.09; 
t-value: 0.48. 

Variable used to measure characteristic: Census Division 6 - East South 
Central; 
Parameter estimate: - 0.16; 
t-value: -1.00. 

Variable used to measure characteristic: Census Division 7 - West South 
Central; 
Parameter estimate: - 0.30; 
t-value: -1.48. 

Variable used to measure characteristic: Census Division 8 - Mountain; 
Parameter estimate: 0.20; 
t- value: 1.38. 

Variable used to measure characteristic: Census Division 9 - Pacific; 
Parameter estimate: 0.02; 
t- value: 0.12. 

Variable used to measure characteristic: Intercept; 
Parameter estimate: 4.95; 
t-value: 9.81***. 

Variable used to measure characteristic: R-squared; 
Parameter estimate: 0.51; 
t-value: [Empty]. 

Variable used to measure characteristic: Observations; 
Parameter estimate: 157; 
t-value: [Empty]. 

Sources: GAO analysis of 2005 GAO Survey of Ambulance Services, 
Medicare Ambulance Fee Schedule Public Use File, and Medicare claims. 

*** significant at the 1 percent level. 

** significant at the 5 percent level. 

* significant at the 10 percent level. 

[A] Transports less than or equal to 2,000 per year is the excluded 
category. In order for the regression model's parameters to be 
estimated, we needed to exclude one of the transport volume categories. 

[B] The percentage of Medicare transports that are urban was the 
excluded category. In order for the regression model's parameters to be 
estimated, we needed to exclude one of the service area categories. 

[C] The West North Central 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. 

[End of table] 

We estimated the average cost per transport associated with a range of 
values for each of five provider characteristics by using our 
regression result parameter estimates and fixed values for five 
provider characteristics. For each of the five provider 
characteristics, we assumed that all providers in the sample had the 
same value for one provider characteristic and used the national 
average value for all other characteristics. The five provider 
characteristics of interest were (1) total transports per year, (2) 
percentage of BLS Medicare transports, (3) percentage of Medicare 
transports in rural and super-rural areas, (4) number of transports per 
staffed ambulance hour, and (5) amount of revenue derived from local 
tax support. We estimated the cost per transport associated with two or 
more values for each of five provider characteristics. Our estimates 
are presented in table 2 of this report. 

In order to create the estimated costs per transport for one value of a 
provider characteristic, we performed the following steps. First, we 
modified the data set we used in the regression model by assigning one 
fixed value for one provider characteristic, such as the percentage of 
super-rural Medicare transports, for all observations in the data set. 
For this characteristic, we fixed the percentage of Medicare super- 
rural transports to 100 percent and the percentage of Medicare rural 
transports to 0 percent for all observations in the data set.[Footnote 
52] Second, we used our regression result parameter estimates for all 
the other characteristics in the regression model and calculated an 
estimated average cost per transport for each provider. Third, we 
computed the mean of these estimated costs per transport. We report 
this hypothetical value as the estimated average cost per transport 
associated with the fixed value of the provider characteristic of 
interest. In the example above, we computed our estimated average cost 
per transport of a super-rural Medicare transport assuming that every 
provider in the sample provided only super-rural Medicare transports 
and their other characteristics--volume and mix of transports, 
productivity, and amount of local tax support--were the national 
average. 

Estimating Average Medicare Ambulance Payments and Use of Transports 
with Claims Data: 

To assess the effect of the ambulance national fee schedule on Medicare 
payments, we used Medicare claims data to compute average payments in 
2001, before the implementation of the fee schedule, and in 2004, 2 
years after the phase-in of the national fee schedule had begun. For 
all analyses of Medicare payments, we expressed payments in 2004 
dollars. We also used Medicare claims data and payment formulas as 
specified in federal regulations to simulate average payments under the 
national fee schedule in 2010, after all the MMA provisions are due to 
expire on December 31, 2009, but computed these payments in 2004 
dollars, the year that best reflects the cost data collected in our 
survey. We used the simulated payments to compute providers' Medicare 
margins, a comparison that assumes that providers' cost structures 
under the fee schedule would be the same as they were in 2004.[Footnote 
53] We also assessed the effect of the MMA temporary payment provisions 
on payments in 2004 using Medicare claims data by examining the change 
in payments from the first half of the year, before the MMA changes 
were implemented, with the second half of the year, when MMA payment 
provisions had their maximum effect. 

To estimate average Medicare ambulance payments in 2001 and 2004, we 
used Medicare Part A and Part B claims data from Medicare's National 
Claims History files. We constructed summary data sets with the number 
of ambulance transports, miles, and their Medicare payments, 
total[Footnote 54] and by level of service,[Footnote 55] at the zip 
code level for 2001 and 2004 to compare payments over time. These data 
sets were classified by three geographic levels: national; census 
divisions; and urban, rural, and super-rural areas. Census divisions 
were categorized according to the U.S. Census Bureau's guidelines. (See 
table 11 for a listing of census divisions.) CMS designates urban, 
rural, and super-rural areas according to population density. 

Table 11: Census Divisions: 

Census divisions: New England; 
States: Connecticut, Maine, Massachusetts, New Hampshire, Rhode Island, 
Vermont. 

Census divisions: Middle Atlantic; 
States: New Jersey, New York, Pennsylvania. 

Census divisions: East North Central; 
States: Illinois, Indiana, Michigan, Ohio, Wisconsin. 

Census divisions: West North Central; 
States: Iowa, Kansas, Minnesota, Missouri, Nebraska, North Dakota, 
South Dakota. 

Census divisions: South Atlantic; 
States: Delaware, District of Columbia, Florida, Georgia, Maryland, 
North Carolina, South Carolina, Virginia, West Virginia. 

Census divisions: East South Central; 
States: Alabama, Kentucky, Mississippi, Tennessee. 

Census divisions: West South Central; 
States: Arkansas, Louisiana, Oklahoma, Texas. 

Census divisions: Mountain; 
States: Arizona, Colorado, Idaho, Montana, Nevada, New Mexico, Utah, 
Wyoming. 

Census divisions: Pacific; 
States: Alaska, California, Hawaii, Oregon, Washington. 

Source: U.S. Census Bureau. 

[End of table] 

Medicare Payment Calculations and Simulation: 

We chose to examine payments in three distinct time periods: 2001, the 
first half of 2004, and the second half of 2004. The year before the 
national fee schedule was implemented was 2001. The first half of 2004 
represented a period after the implementation of the national fee 
schedule but before the introduction of MMA temporary payment changes. 
It also coincided with the year for which we have cost data. The second 
half of 2004 represented a period after the introduction of MMA 
provisions when the regional fee schedule had its greatest effect. 

In general, average Medicare payments were estimated as the sum of the 
average base rate payment--that is, the amount paid for the level of 
service[Footnote 56]--and the average mileage payment. Base rate 
payments and mileage payments for each period were computed separately. 
Average Medicare payments were calculated nationally; by census 
division; and by urban, rural, and super-rural areas. All payments are 
expressed in 2004 dollars. 

For the first half of 2004, the average base rate payment was 
calculated by dividing total base rate payments by the total number of 
transports provided in the first half of 2004. For 2001 and the second 
half of 2004, the average service-level base payment rate was 
calculated by dividing total base rate payments for each level of 
service by the total number of transports provided at each level of 
service. Then we computed a weighted average base rate payment for 2001 
and for the second half of 2004 by applying the percentage of 
transports for each level of service (also known as the mix of 
services) for the first half of 2004 to the service-level base payment 
rates in 2001 and the second half of 2004, respectively. The 2001 
payments were adjusted to 2004 dollars by multiplying weighted average 
base rate payment by the ratio of the 2004 Consumer Price Index for All 
Urban Consumers (CPI-U) over the 2001 CPI-U. 

For simulated 2010 base rate payments, we multiplied the 2004 
conversion factor (CF) and applicable relative value unit (RVU) for 
each level of service to compute the service level payments.[Footnote 
57] Then, we computed a weighted average base rate payment by applying 
the service mix from the first half of 2004 to the 2004 service-level 
payments. The weighted average base-rate payment was adjusted for 
regional differences in the cost of providing ambulance services by the 
GPCI.[Footnote 58] In accordance with CMS's payment methodology, only 
70 percent of the average base rate payment was adjusted by the GPCI. 

Average mileage payments were calculated by multiplying the average 
miles per transport in each geographic level from the first half of 
2004 by mileage rates in the other periods. (See tables 12 and 13.) For 
example, when estimating average 2001 mileage payments for urban 
transports, the 2001 mileage rate applicable to urban areas was 
used,[Footnote 59] but the average mile per transport in urban areas 
for the first half of 2004 was applied. For simulated payments in 2010, 
mileage rates from the first half of 2004 were applied. For average 
national and census division mileage payments, weighted mileage rates 
were applied reflecting the applicable percentage of urban and rural 
miles in the first half of 2004. Because the average mile per transport 
for super-rural areas was above 17, simulated payments for super-rural 
areas were computed using the two rural mileage rates--$8.48 for miles 
1 through 17 and $5.65 for miles 18 through 50. 

Table 12: Average Mile per Transport, First Half of 2004: 

Area: National; 
Average mile per transport: 9. 

Area: Urban; 
Average mile per transport: 7. 

Area: Rural; 
Average mile per transport: 13. 

Area: Super-rural; 
Average mile per transport: 20. 

Area: East North Central; 
Average mile per transport: 9. 

Area: East South Central; 
Average mile per transport: 12. 

Area: Middle Atlantic; 
Average mile per transport: 7. 

Area: Mountain; 
Average mile per transport: 9. 

Area: Pacific; 
Average mile per transport: 7. 

Area: New England; 
Average mile per transport: 7. 

Area: South Atlantic; 
Average mile per transport: 10. 

Area: West North Central; 
Average mile per transport: 14. 

Area: West South Central; 
Average mile per transport: 11. 

Source: 2004 Medicare claims. 

[End of table] 

Table 13: Mileage Rates: 

Period: 2001; 
Urban: $5.33; 
Rural (1-17): $5.33; 
Rural (18-50): $5.33. 

Period: 1/1/04 - 6/30/04; 
Urban: 5.65; 
Rural (1-17): 8.48; 
Rural (18- 50): 5.65. 

Period: 7/31/04 - 12/31/04; 
Urban: 5.71; 
Rural (1-17): 8.65; 
Rural (18- 50): 5.76. 

Period: Simulated fee schedule; 
Urban: 5.65; 
Rural (1-17): 8.48; 
Rural (18-50): 5.65. 

Sources: 67 Fed. Reg. 9100 and 2004 Medicare Ambulance Fee Schedule 
Public Use File. 

Notes: All mileage rates are expressed in 2004 dollars. The 2001 
mileage rate is based on the mileage rate for ground ambulance services 
used by CMS when creating the national fee schedule. Rural mileage 
rates also apply to super-rural transports. 

[End of table] 

Analysis of Medicare Margins: 

We used simulated Medicare payments and costs per transport to compute 
Medicare margins under the national fee schedule after MMA temporary 
payment provisions are due to expire, to determine how much providers 
stood to lose or gain. To do this, we merged the survey results with 
providers' Medicare payment information from 2004. For each provider, 
we computed the mix of services and the average miles per transport for 
January 2004 through June 2004. This information was used to simulate 
each provider's average Medicare payment under the national fee 
schedule in 2010 after all MMA temporary payment provisions are set to 
expire on December 31, 2009.[Footnote 60] We subtracted the provider's 
2004 cost per transport from the simulated average Medicare payment 
under the national fee schedule, and then divided by the simulated 
average payment to compute the provider's Medicare margin. 

Estimating Use of Ambulance Transports by Medicare Beneficiaries: 

To estimate transports per 1,000 Medicare beneficiaries, we 
supplemented our Medicare claims data with CMS enrollment data from 
2001 and 2004, which contained information regarding Medicare 
beneficiaries. The numbers of Medicare beneficiaries in 2001 and 2004 
were measured as the number of months beneficiaries were covered by 
Medicare in each year divided by 12. A ratio of the number of 
transports over the number of Medicare beneficiaries was estimated in 
each year and multiplied by 1,000. 

Medicare Claims Data Reliability: 

Medicare claims data, which are used by the Medicare program as a 
record of payments made to health care providers, are closely monitored 
by both CMS and Medicare carriers--contractors that process, review, 
and pay claims for Medicare Part B-covered services, including 
ambulance services. The data are subject to various internal controls, 
including checks and edits performed by the carriers before claims are 
submitted to CMS for payment approval. Although we did not review these 
internal controls, we found the data to be sufficiently reliable for 
the purposes of this report. We also assessed the reliability of CMS's 
enrollment data. We found these data to be sufficiently reliable for 
the purposes of this report. 

[End of section] 

Appendix II: Comments from the Centers for Medicare & Medicaid 
Services: 

Department Of Health & Human Services: 
Centers for Medicare and Medicaid Services: 
Administrator: 
Washington, DC 20201: 

Date: Apr 2007: 

To: Kathleen M. King: 
Director, Health Care: 
Government Accountability Office: 

From: Leslie V. Norwalk, Esq.: 
Acting Administrator: 

Subject: Government Accountability Office (GAO) Draft Report: 
"Ambulance Providers: Costs and Expected Medicare Margins Vary Greatly" 
(GAO-07-383): 

The Centers for Medicare & Medicaid Services (CMS) appreciates the 
opportunity to review and provide comments on the GAO draft report 
entitled "AMBULANCE PROVIDERS: Costs and Expected Medicare Margins Vary 
Greatly." On April 1, 2002, CMS implemented the Medicare ambulance fee 
schedule with a 5-year transition period. The fee schedule was 
established through a negotiated rulemaking process, as required by the 
statute, which allowed interested stakeholders to participate in its 
development. During the transition period, Congress enacted several 
temporary ambulance provisions in the Medicare Modernization Act of 
2003, including bonus payments for ambulance services in rural pick-up 
areas. All of these legislative provisions will sunset by December 31, 
2009. Since the ambulance fee schedule is a departure in payment 
structure from the prior methods of payment for ambulance services, GAO 
examined how ambulance companies were faring under the ambulance fee 
schedule and whether or not they have been successful and will be 
successful in the future in recovering their costs for services 
provided to Medicare beneficiaries under the ambulance fee schedule. 

We appreciate the effort that GAO has put into this report and, for the 
most part, the report reinforces CMS' own findings. Generally, 
ambulance services incur a cost of readiness, which was built into the 
ambulance fee schedule payment structure, and rural ambulance companies 
incur higher costs per trip than urban ambulance companies due to a 
lower expected volume of trips provided to the Medicare population. 

GAO Recommendation: 

CMS should monitor utilization of ambulance transports to ensure that 
Medicare payments arc adequate to provide for beneficiary access to 
ambulance services, particularly in super-rural areas. 

CMS Response: 

We agree with GAO's recommendation. In fact, we stated in the 2006 
final rule (71 FR 69717) published on December 1, 2006, that we would 
continue to monitor all ambulance rates and would make adjustments 
should the original assumptions that we made during the development of 
the ambulance fee schedule prove to need adjustment. Further, in this 
same final rule (71 FR 69714), we implemented use of the Rural Urban 
Commuting Areas (RUCAs) in conjunction with recognizing the 2000 
decennial census population data. The RUCAs allow us to recognize 
levels of rurality in every zip code across the country. Since 
ambulance payments under Medicare are based on the point of pick-up, 
the level of detail provided by the use of the RUCAs is of particular 
benefit to rural areas within urban counties, allowing these zip code 
locations to receive the benefit of the higher rural payments under the 
ambulance fee schedule. 

We note that many of the ambulance providers with "shared services" 
that were omitted from the analysis have higher than average costs. 
These include hospital-based providers and joint fire/ambulance 
services. We recommend that the GAO review the survey data and other 
data sources from prior studies for these groups. GAO should discuss 
the implications of omitting these groups in the concluding section of 
the report. 

In conclusion, we deeply appreciate GAO's efforts in examining these 
important issues. We look forward to working with you in the future as 
we monitor ambulance services under the ambulance fee schedule. As we 
stated in the final rule on December 1, 2006, we are committed to 
continuing to monitor these issues to ensure that the ambulance fee 
schedule accurately reflects the realities of services provided to 
Medicare beneficiaries by the ambulance industry. 

Attachment: 

[End of section] 

Appendix III: GAO Contact and Staff Acknowledgments: 

GAO Contact: 

Kathleen M. King, (202) 512-7119 or kingk@gao.gov: 

Acknowledgments: 

In addition to the contact named above, Christine Brudevold, Assistant 
Director; Jennie F. Apter; Carl S. Barden; Stella Chiang; Kevin J. 
Dietz; James M. Fields; Leslie V. Gordon; Michael Kendix; and Jessica 
Cobert Smith made key contributions to this report. 

[End of section] 

Related GAO Products: 

Ambulance Services: Medicare Payments Can Be Better Targeted to Trips 
in Less Densely Populated Rural Areas. GAO-03-986. Washington, D.C.: 
September 19, 2003. 

Ambulance Services: Changes Needed to Improve Medicare Payment Policies 
and Coverage Decisions. GAO-02-244T. Washington, D.C.: November 15, 
2001. 

Emergency Medical Services: Reported Needs Are Wide-Ranging, With A 
Growing Focus on Lack of Data. GAO-02-28. Washington, D.C.: October 12, 
2001. 

Rural Ambulances: Medicare Fee Schedule Payments Could Be Better 
Targeted. GAO/HEHS-00-115. Washington, D.C.: July 17, 2000. 

FOOTNOTES 

[1] We use the term providers to refer to all types of organizations 
that provide ambulance transports for Medicare beneficiaries. 

[2] 42 U.S.C. § 1395m(l). 

[3] Reasonable-cost payments were based on the provider's cost of 
providing ambulance services as reported on cost reports. 

[4] Reasonable-charge payments were based on the bill from the 
ambulance service provider but were subject to an upper limit. 

[5] The national fee schedule applies to ground and air ambulance 
services, but this report and our analysis are limited to ground 
ambulance services only, which include water ambulance services and 
account for nearly all ambulance services. We use the terms service, 
transport, and ambulance service to refer to ground ambulance transport 
services only. 

[6] Pub. L. No. 108-173, § 414, 117 Stat. 2066, 2278-80 (2003). 

[7] Pub. L. No. 106-554, appendix F, § 436, 114 Stat. 2763, 2763A-527 
(2000). 

[8] The GAO Survey of Ambulance Services asked providers to report on 
their organizations' costs of providing ground ambulance services. As 
such, the survey measures organizations' expenses, or spending, for 
ground ambulance services. 

[9] The sample was stratified by five types of ambulance providers: 
hospital-based, volunteer, fire-based, government, and freestanding. 

[10] To make our survey sample representative of all Medicare ambulance 
providers, we computed a sample weight for each respondent provider. 

[11] We applied our sample weights to the subgroup so the providers 
were nationally representative of all ambulance service providers 
without shared costs. 

[12] To define urban, rural, and super-rural transports, CMS uses the 
metropolitan statistical areas (MSA) definitions established by the 
Office of Management and Budget. During the period of our study, CMS 
defined urban transports as those that originate within MSAs and New 
England county metropolitan areas (NECMA), rural transports as those 
that originate in rural counties that are outside of MSAs and NECMAs as 
well as small towns and rural areas within MSAs or NECMAs that are 
isolated from central areas by distance or other features, such as 
mountains. CMS defines super-rural transports as those that originate 
in the bottom 25 percent of rural areas as defined by population 
density. 

[13] This comparison assumes that providers' cost structures under the 
fee schedule would be the same as they were in 2004. 

[14] The Commission on Accreditation of Ambulance Services and the 
Commission of Accreditation of Medical Transport Systems have 
established ambulance service accreditation standards. 

[15] The National Association of State EMS Officials has worked with 
other industry associations and federal partners to develop tools for 
measuring the performance of EMS systems: (1) EMS system indicators, 
which were released for public comment and are being finalized; (2) the 
National EMS Information Management System; and (3) a cost framework 
for cost analysis in EMS research. The American Ambulance Association 
published its Community Guide to Ensure High-Performance Emergency 
Ambulance Services in 2004. 

[16] Unless otherwise noted, all background information on the 
ambulance industry is from the GAO 2005 Survey of Ambulance Services. 
These estimates are based on the 321 total providers that responded to 
our survey. 

[17] BLS services include basic, noninvasive interventions to reduce 
morbidity and mortality associated with acute out-of-hospital medical 
and traumatic emergencies. 

[18] ALS services include advanced, invasive, and pharmacological 
interventions to reduce morbidity and mortality associated with acute 
out-of-hospital medical and traumatic emergencies. 

[19] Under the Medicare program, there are seven levels of ambulance 
transports. BLS and ALS transports each constitute two levels, 
emergency or nonemergency. The remaining levels of service--ALS Level 2 
and specialty care transport (both of which involve invasive or 
specialized care) and paramedic ALS intercept (when a paramedic 
provides ALS services but does not transport the patient)--made up less 
than 2 percent of all Medicare transports in 2004. 

[20] EMT-Basic personnel are trained in BLS services only. EMT- 
Intermediate personnel are qualified to perform essential advanced 
techniques and to administer a limited number of medications. 
Paramedics have the competencies of EMT-Intermediate personnel in 
addition to other enhanced skills and can administer additional 
interventions and medications. 

[21] Cross-trained staff typically refers to firefighters who are also 
trained as EMTs. 

[22] We defined substantial use of volunteer staff as 20 percent or 
more of staff hours spent providing ambulance services, exclusive of 
administration. 

[23] Rural areas can also be small towns and rural areas within large 
metropolitan counties that are isolated from central areas by distance 
or other features, such as mountains. 

[24] From this point forward in the report, we refer to providers that 
served predominately urban, rural, and super-rural areas as urban, 
rural, and super-rural providers, respectively. We classified providers 
as super-rural if 60 percent or more of their Medicare transports in 
2004 originated in a super-rural zip code. We classified providers as 
rural if they did not meet the super-rural definition and 60 percent or 
more of their Medicare transports in 2004 originated in rural or super- 
rural zip codes. We defined providers as urban if they did not meet the 
rural or super-rural classifications. 

[25] A subscription program is an arrangement in which an ambulance 
service provider is paid an annual fee for providing emergency 
transportation for a community. 

[26] In 2004, the annual spending of providers that responded to our 
survey ranged from less than $10,000 to more than $70 million. 

[27] This estimate is based on a smaller sample of 209 providers that 
reported Medicare revenues and total revenues on the GAO 2005 Survey of 
Ambulance Services. 

[28] Medicare ambulance transports must be deemed medically necessary 
in order for Medicare payments to be disbursed. 

[29] There is an RVU for each of the seven levels of service defined by 
Medicare. 

[30] The practice expense portion of the physician fee schedule's GPCI 
is used to adjust the ambulance national fee schedule. 

[31] See 42 C.F.R. § 414.601 et seq. 

[32] In this report we use the term regions to refer to the nine census 
divisions as defined by the U.S. Census Bureau. See table 11 in app. I 
for a description of the nine census divisions. 

[33] The national fee schedule portion blend with historical payments 
includes mileage payments. The regional fee schedule is only applicable 
to the base rate payment. 

[34] GAO, Ambulance Services: Medicare Payments Can Be Better Targeted 
to Trips in Less Densely Populated Rural Areas, GAO-03-986 (Washington, 
D.C.: Sept. 19, 2003). 

[35] Transports per staffed hour is the total number of transports 
divided by the total number of hours that an ambulance and crew were 
staffed and available to respond to an emergency call. 

[36] Because our cost information is estimated from a sample of 
providers, all costs per transport are reported with confidence 
intervals. The range of the confidence interval is affected by the 
variability of the responses within the sample and the size of the 
sample. See app. I for a full discussion of our sample, methods, and 
computations. 

[37] The average cost per transport of super-rural providers without 
shared costs was based on the average costs of survey respondents 
without shared costs that served predominantly super-rural areas, and 
reflects the different characteristics of those respondents, including 
the volume of trips they provided. It is different from the estimated 
average cost of a super-rural transport based on our regression 
analysis. 

[38] For each characteristic, we measured its effect on providers' 
average cost per transport, independent of other characteristics, by 
assuming the national average value for the other variables. 

[39] GAO-03-986. 

[40] The estimated average cost of a super-rural transport is different 
from the average cost per transport of super-rural providers without 
shared costs that was explained in footnote 37. The estimated average 
cost of a super-rural transport was calculated using regression 
analysis that assumed all providers without shared costs only performed 
super-rural transports and had the national average for the other 
variables in our analysis. See our app. I for more details on our 
methodology. 

[41] Our survey asked providers to report on their organizations' costs 
of providing ground ambulance services. 

[42] The sample was stratified by types of ambulance providers: 
hospital-based, volunteer, fire departments, government, and 
freestanding. 

[43] Our cost analysis sample includes only those providers that could 
apportion the ground ambulance component of any cost shared between 
their ambulance services and other services. Thus, our results only 
reflect the experiences of providers that did not share costs or that 
could feasibly apportion them. 

[44] Carriers are private companies that have contracts with Medicare 
to administer payment of providers' bills for covered Medicare Part B 
services. 

[45] We excluded provider identification numbers that had been selected 
to participate in the pilot study. 

[46] Duplicate entries for the same nonhospital provider were 
identified by name, address, and other information. 

[47] CODA Inc. also administered the 1999 ambulance survey for Project 
HOPE and our pilot survey. During the course of the contract, CODA Inc. 
became Survey and Epidemiology Services Division of Social & Scientific 
Systems, Inc. 

[48] For a randomly selected subsample of 30 providers--6 of each type 
of ambulance provider--we requested providers' financial records to 
match to their survey responses. We were only able to validate costs, 
revenues, and transport volumes for 2 providers in the subsample 
because providers either did not return supporting documentation or the 
categories--particularly those for costs--on providers' supporting 
documentation were different from those on the survey. As a result, we 
concluded that this method for validating survey data was not viable 
because of differences in data reporting, particularly cost categories, 
on providers' supporting documents. 

[49] We could not determine a separate response rate for our subgroup 
of providers without shared costs because we could not identify which 
nonrespondents in our original sample were providers without shared 
costs. However, we have no reason to believe that the response rate for 
providers without shared costs would be different from the overall 
survey response rate. 

[50] The range of the confidence interval is affected by the 
variability of the responses within the sample and the size of the 
sample. 

[51] Least squares is a common method of regression analysis. 

[52] The percentage of super-rural and rural Medicare transports were 
two separate variables in our model and the percentage of urban 
Medicare transports was our reference category. To assume that 100 
percent of a provider's Medicare transports were rural, we set the 
percentage of rural Medicare transports to 100 percent and the 
percentage of super-rural Medicare transports to 0 percent. To assume 
that 100 percent of a provider's Medicare transports were urban, we set 
the percentage of super-rural Medicare transports to 0 percent and the 
percentage of rural Medicare transports to 0 percent. 

[53] The Medicare margin reflects the percentage difference between the 
average Medicare payment and a provider's cost per transport. It was 
computed as the difference between the average Medicare payment and the 
cost per transport, divided by the payment. 

[54] Total ambulance payments in 2001 and 2004 included any payments 
made for supplies in addition to mileage payments and service-level 
payments. 

[55] Under the Medicare program, there are seven levels of ambulance 
transports. BLS and ALS transports each constitute two levels, 
emergency and nonemergency. The remaining levels of service are ALS 
Level 2 and specialty care transport (both of which involve invasive or 
specialized care) and paramedic ALS intercept (when a paramedic 
provides ALS services but does not transport the patient). 

[56] Estimated average base rate payments for 2001 and 2004 included 
ambulance supply payments. Under the fee schedule, the base rate 
payment was meant to incorporate the transport and any necessary 
services or supplies used during the transport. 

[57] Ambulance RVUs account for the relative resources needed to 
provide services during an ambulance transport. There is one RVU for 
each of the seven levels of service. 

[58] In many cases, more than one GPCI would apply to our average base 
rate payment. For these cases we applied an average GPCI, which was 
weighted by the total number of transports in each geographic area of 
interest--national; census division; or urban, rural, or super-rural 
area. 

[59] The 2001 mileage payments were eventually adjusted to 2004 
dollars. 

[60] We simulated the average payments under the national fee schedule 
in the same way described earlier in this appendix. 

GAO's Mission: 

The Government Accountability Office, the audit, evaluation and 
investigative arm of Congress, exists to support Congress in meeting 
its constitutional responsibilities and to help improve the performance 
and accountability of the federal government for the American people. 
GAO examines the use of public funds; evaluates federal programs and 
policies; and provides analyses, recommendations, and other assistance 
to help Congress make informed oversight, policy, and funding 
decisions. GAO's commitment to good government is reflected in its core 
values of accountability, integrity, and reliability. 

Obtaining Copies of GAO Reports and Testimony: 

The fastest and easiest way to obtain copies of GAO documents at no 
cost is through GAO's Web site (www.gao.gov). Each weekday, GAO posts 
newly released reports, testimony, and correspondence on its Web site. 
To have GAO e-mail you a list of newly posted products every afternoon, 
go to www.gao.gov and select "Subscribe to Updates." 

Order by Mail or Phone: 

The first copy of each printed report is free. Additional copies are $2 
each. A check or money order should be made out to the Superintendent 
of Documents. GAO also accepts VISA and Mastercard. Orders for 100 or 
more copies mailed to a single address are discounted 25 percent. 
Orders should be sent to: 

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

To order by Phone: Voice: (202) 512-6000 TDD: (202) 512-2537 Fax: (202) 
512-6061: 

To Report Fraud, Waste, and Abuse in Federal Programs: 

Contact: 

Web site: www.gao.gov/fraudnet/fraudnet.htm E-mail: fraudnet@gao.gov 
Automated answering system: (800) 424-5454 or (202) 512-7470: 

Congressional Relations: 

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

Public Affairs: 

Paul Anderson, Managing Director, AndersonP1@gao.gov (202) 512-4800 
U.S. Government Accountability Office, 441 G Street NW, Room 7149 
Washington, D.C. 20548: