VA Health Care: Allocation Changes Would Better Align Resources  
with Workload (28-FEB-02, GAO-02-338).				 
                                                                 
The Department of Veterans Affairs (VA) spent $21 billion in	 
fiscal year 2001 to treat 3.8 million veterans--most of whom had 
service-connected disabilities or low incomes. Since 1997, VA has
used the Veterans Equitable Resource Allocation (VERA) system to 
allocate most of its medical care appropriation. GAO found that  
VERA has had a substantial impact on network resource allocations
and workloads. First, VERA shifted $921 million from networks	 
located primarily in the northeast and midwest to networks	 
located in the south and west in fiscal year 2001. In addition,  
VERA, along with other VA initiatives, has provided an incentive 
for networks to serve more veterans. VERA's overall design is a  
reasonable approach to allocate resources commensurate with	 
workloads. It provides a predetermined dollar amount per veteran 
served to each of VA's 22 health care networks. This amount	 
varies depending upon the health care needs of the veteran served
and local cost differences. This approach is designed to allocate
resources commensurate with each network's workload in terms of  
veterans served and their health care needs. GAO identified	 
weaknesses in VERA's implementation. First, VERA excludes about  
one fifth of VA's workload in determining each network's	 
allocation. Second, VERA does not account well for cost 	 
differences among networks resulting from variation in their	 
patients' health care needs. Third, the process for providing	 
supplemental resources to networks through VA's National Reserve 
Fund has not been used to analyze how the need for such resources
is caused by potential problems in VERA's allocation, network	 
inefficiency, or other factors. 				 
-------------------------Indexing Terms------------------------- 
REPORTNUM:   GAO-02-338 					        
    ACCNO:   A02804						        
  TITLE:     VA Health Care: Allocation Changes Would Better Align    
Resources with Workload 					 
     DATE:   02/28/2002 
  SUBJECT:   Health care cost control				 
	     Health care costs					 
	     Health care programs				 
	     Program evaluation 				 
	     Veterans benefits					 
	     National Reserve Fund				 
	     VA Veterans Equitable Resource			 
	     Allocation System					 
                                                                 

******************************************************************
** This file contains an ASCII representation of the text of a  **
** GAO Product.                                                 **
**                                                              **
** No attempt has been made to display graphic images, although **
** figure captions are reproduced.  Tables are included, but    **
** may not resemble those in the printed version.               **
**                                                              **
** Please see the PDF (Portable Document Format) file, when     **
** available, for a complete electronic file of the printed     **
** document's contents.                                         **
**                                                              **
******************************************************************
GAO-02-338
     
United States General Accounting Office

GAO

Report to Congressional Requesters

February 2002

VA HEALTH CARE

Allocation Changes Would Better Align Resources with Workload

GAO-02-338

Contents

Letter

Results in Brief
Background
VERA Has Had A Substantial Effect on Network Resources and

Workload VERA Provides A Reasonable Approach to Resource Allocation VERA
Changes Required to Better Align Resources With Workload Conclusions
Recommendations for Executive Action Agency Comments and Our Evaluation 1

2 4

6

9 13 32 33 34

Appendix I Scope and Methodology

      Appendix II Comments From the Department of Veterans Affairs 42

Appendix III GAO Contact and Staff Acknowledgements

Related GAO Products

Tables

Table 1: VERA Patient Workload Compared to VERA Allocation,

Fiscal Year  2001 10 Table 2:  VERA Capitation Amounts, Fiscal  Year 2001 19
Table 3: Complex Care Workload Allocations Compared With

Complex Care Expenditures, Fiscal Year 2000 20 Table 4: VA Patient Classes,
Costs, and National Capitation Amounts, Fiscal Year 2000 22 Table 5: Network
Supplemental Funding, Fiscal Years 1999 through 2001 30

Figures

Figure 1: Proportion of Veteran Population by Region, 1990 through 2020

Figure 2: Resource Allocation Shifts Resulting from VERA from Fiscal Years
1996 through 2001

Figure 3: Percentage Increase in Veterans Treated by Network, Fiscal Years
1996 through 2001

Figure 4: Growth of Priority 7 Veterans Treated, Fiscal Years 1996 through
2001

Figure 5: Priority 7 Veterans Treated by Network, Fiscal Year 2001

Figure 6: Estimated Change in VERA Allocations from Adding Priority 7 Basic
Vested Veterans to VERA Workload at Half Their National Cost, Fiscal Year
2001

Figure 7: Estimated Change in VERA Allocations Among Networks as a Result of
Using 44 Case-Mix Categories, by Network, Fiscal Year 2001

Figure 8: Estimated Change in VERA Allocations from Incorporating 44
Case-Mix Categories, Most Current Expenditure Data for Case-Mix Weights, and
Priority 7 Basic Vested Veterans Treated, Fiscal Year 2001

                                     5

                                     7

                                     9

                                   15 16

                                     18

                                     25

                                     29

Abbreviations

DCG Diagnostic Cost Group
VA Department of Veterans Affairs
VERA Veterans Equitable Resource Allocation

United States General Accounting Office Washington, DC 20548

February 28, 2002

Congressional Requesters

The Department of Veterans Affairs (VA) spent about $21 billion for health
care in fiscal year 2001 to treat about 3.8 million veterans?most of whom
had service-connected disabilities or low incomes. Since fiscal year 1997,
VA has used the Veterans Equitable Resource Allocation (VERA) system to
allocate most of its medical care appropriation as part of its overall
strategy to reform VA health care. As a health care payer, like the Medicare
and Medicaid programs, VA faces the challenge of allocating resources to
account for differences in patient workload treated, encourage efficiency,
and ensure quality. As a direct provider, VA also operates a major health
care system consisting of 22 regional health care networks that are at risk
for budget shortfalls if the cost of providing care to veterans treated by
the network is greater than available resources.

Our reviews of VERA in its first 2 years concluded that it was an important
step forward in equitable resource allocation compared to the allocation
practices it replaced.1 Prior to VERA, VA allocated resources primarily on
the basis of facilities' historical expenditures. By contrast, VERA was
intended to equitably allocate health care resources, that is, allocate
comparable resources to networks with comparable workloads. In turn, VA's
networks have budget and management responsibilities that include allocating
VERA resources to facilities, clinics, and programs within their networks
and ensuring equity of access to appropriate health care services.

Congressional stakeholders have expressed concern as to whether VERA has
been designed and implemented to allocate resources commensurate with
workload. To address concerns you raised, this report (1) describes the
effect VERA has had on network resource allocations and workloads, (2)
assesses whether VERA's design is a reasonable approach to resource
allocation, and (3) identifies weaknesses in VERA that may limit VA's
ability to allocate comparable resources for comparable workloads. As agreed
with your offices, we focused our work on VERA's allocation of

1U.S. General Accounting Office, VAHealth Care:ResourceAllocationHas
Improved, but Better OversightIs Needed, GAO/HEHS-97-178 (Washington, D.C.:
September 17, 1997) and U.S. General Accounting Office,
VAHealthCare:MoreVeteransAreBeingServed,But Better OversightIs Needed,
GAO/HEHS-98-226 (Washington, D.C.: August 28, 1998).

resources from headquarters to the networks but we did not examine the
extent to which networks in turn allocate comparable resources for
comparable workloads to their respective facilities and programs.

To examine these issues, we reviewed VA documents and consultants' reports
on VERA's original design, proposed VERA changes, and actual VERA changes.
We also interviewed VA headquarters officials and officials in 8 networks,
conducted site visits in 5 networks, surveyed all 22 network directors, and
interviewed VA and other public and private sector health care resource
allocation experts. In addition, we analyzed changes in resources allocated
among the 22 networks from fiscal year 1996 through 2001, changes in the
number of veterans treated, and the effect of making adjustments to VERA. We
also relied on our more than 10 years of work reviewing VA's resource
allocation process in addition to other health care financing work.2 For a
complete description of our scope and methodology, see appendix I. Our work
was performed from October 2000 through December 2001 in accordance with
generally accepted government auditing standards.

VERA has had a substantial impact on network resource allocations and
workloads. First, VERA shifted resources among regions. VERA shifted
approximately $921 million from networks located primarily in the northeast
and midwest to networks located in the south and west in fiscal year 2001
compared to what allocations would have been without the implementation of
VERA. Second, VERA, in concert with other VA initiatives, has provided an
incentive for networks to serve more veterans. The number of veterans
treated nationally in VA health care programs increased from 2.6 million to
3.8 million, an increase of 47 percent, from fiscal year 1996 through fiscal
year 2001. All 22 networks contributed to this increase, including networks
from which VERA shifted resources.

VERA's overall design is a reasonable approach to allocate resources
commensurate with workloads. It provides a predetermined dollar amount per
veteran served to each of VA's 22 health care networks. This amount varies
depending upon factors beyond networks' control, namely the health care
needs of the veteran served and certain local cost differences. This
approach is designed to allocate resources commensurate with each network's
workload in terms of veterans served and their health care

Results in Brief

2See the Related GAO Products page at the end of this report.

needs. It also aims to hold networks accountable for efficient service
provision by attempting to vary resource allocation only for costs beyond
their control. To protect patients from the risk that a health care network
cannot deliver needed services with the resources allocated, VERA includes a
National Reserve Fund to provide supplemental resources to networks
experiencing budget shortfalls.

Although VERA's design is a reasonable approach to resource allocation, we
identified weaknesses in its implementation. First, VERA excludes about one
fifth of VA's workload in determining each network's allocation. The
excluded veterans are those with higher incomes who do not have
service-connected disabilities. Second, VERA does not account for cost
differences among networks resulting from variation in their patients'
health care needs as well as it could. For example, VERA does not use enough
categories to adjust for patient health care needs in order to adequately
account for patient cost differences among networks. These two weaknesses
compromise VERA's ability to allocate comparable resources for comparable
workloads. Third, the process for providing supplemental resources to
networks through VA's National Reserve Fund has not been used to analyze the
extent to which the need for such resources is caused by potential problems
in VERA's allocation, network inefficiency, or other factors. VA uses the
National Reserve Fund to supplement networks' VERA allocations when they
have difficulty operating within their available resources. The lack of
information on why networks need assistance from the National Reserve Fund
limits VA's ability to provide assurance that supplemental funding of
networks is appropriate or to take corrective action.

We are recommending that VA correct weaknesses in VERA to better allocate
comparable resources for comparable workloads. In addition, we are
recommending that VA use the National Reserve Fund process to learn why
networks need supplemental resources beyond their VERA allocations and take
action to correct problems identified in this process.

In commenting on a draft of our report, VA agreed with our conclusions and
concurred with our recommendations. However, pending completion of ongoing
studies and further consideration, VA did not commit to specific actions and
timelines for implementing our recommendations to correct weaknesses in
VERA's implementation. VA's response also did not fully address our
recommendation to improve the supplemental adjustment process. Delay in
implementing our recommendations to make needed improvements to VERA means
that approximately $200 million annually will not be allocated as well as
they could be to align

Background

resources with workloads. Moreover, until improvements are made in the
supplemental funding process, VA will not be able to provide assurance that
its supplemental funding of some networks is appropriate.

VA policy is to allocate comparable resources for comparable workloads in
its 22 health care networks as an important step in ensuring equitable
access to care for the nation's veterans. To achieve this allocation in its
national health care system, VA has used VERA since fiscal year 1997 to
prospectively allocate resources to the networks. VERA allocates nearly 90
percent of VA's medical care appropriation in six categories: complex
patient care, basic patient care, equipment, nonrecurring maintenance,
education support, and research support.3 Resources for the first four
categories are allocated on the basis of patient workload and account for
approximately 96 percent of the resources VERA allocates.4 Allocations for
education support and research support are based on workload measures
specific to those activities within the VA health care system.

Developed in response to a legislative mandate,5 VERA was designed to
correct regional inequities in resource allocation created by shifts in the
veteran population from the northeast and midwest to the south and west (see
fig. 1) without a corresponding shift in resources.6 The resources did not
shift before VERA was implemented because resource allocation was based
primarily on facilities' historical expenditures. VA expects that veteran
population shifts from the northeast and midwest to the south and west will
continue at least through 2020.

3Networks and their facilities also receive resources from the medical care
appropriation not allocated through VERA for such things as prosthetics,
homeless programs, and readjustment counseling. In addition, VA facilities'
budgets include collections for insurance reimbursements, copayments, and
deductibles for the care of some veterans.

4We examined these four categories in our analysis. We did not examine the
education support and research support categories which constitute
approximately 4 percent of VERA's allocation.

5The 1997 appropriation act for the Department of Veterans Affairs required
VA to establish a plan to equitably allocate health care resources. Pub. L.
No. 104-204, Section 429.

6We classified VA's 22 health care networks into geographic regions
according to U.S. Census definitions.

Figure 1: Proportion of Veteran Population by Region, 1990 through 2020

70 Percentage

60

50

40

30

20

10

0

1990 2000 2020a

Northeast and midwest

a2020 numbers are projections.

Note: Veteran assignment to a particular region is based on the state of
residence.

Source: GAO calculations based on VA data.

Two other major changes to VA health care provision accompanied the
implementation of VERA as a result of the Veterans' Health Care Eligibility
Reform Act of 1996. The first change was a major shift in VA health care
delivery from an inpatient to an outpatient emphasis that was consistent
with changes in health care delivery outside of VA. The act eliminated
restrictions that previously prevented VA from treating some veterans in
outpatient care settings, allowing VA to shift its focus from inpatient to
outpatient care delivery. For example, VA no longer had to admit certain
veterans to an inpatient setting to make them eligible for outpatient
treatment or to receive prosthetic devices, such as crutches. As a result of
eligibility reform, VA has been successful in shifting medical care to

outpatient settings by taking advantage of advances in medical technology
and practices, such as laser, endoscopic surgery, and other less invasive
surgical techniques. VA has also identified alternatives to inpatient care,
such as home-based care, for many chronically ill patients. From fiscal year
1996 through fiscal year 2000, VA closed almost 24,000 acute inpatient beds,
a 52 percent reduction systemwide. During this time period, VA's inpatient
admissions decreased and outpatient visits increased from approximately 29
million to 40 million visits, a 36 percent increase systemwide.

The second change was the introduction of a veterans' enrollment system to
manage access in relation to available resources due to the expected
increase in demand on the VA system as a result of the new eligibility
rules. As required by the act, VA established seven priority categories for
enrollment. A higher priority for enrollment is given to veterans who have
service-connected disabilities, lower incomes, or other statuses such as
former prisoners of war. These higher priority enrollees are ranked in
priority order from 1 through 6. The lowest enrollment priority is given to
veterans not included in priorities 1 through 6, referred to as Priority 7
veterans. These veterans are primarily nonservice-connected veterans with
higher incomes. The act requires VA to restrict enrollment consistent with
these enrollment priorities if sufficient resources are not available to
provide care that is timely and acceptable in quality to all priority
categories. If needed, enrollment restrictions would begin with the lowest
priority category. For, fiscal year 2002, VA has decided to continue
enrolling veterans in all priority categories.

VERA has been a key part of VA's strategy to change its health care system.
First, VERA shifted substantial resources among regions reflecting shifts in
workload. Second, VERA, in concert with other VA initiatives, has provided
an incentive for networks to serve more veterans.

VERA has shifted substantial resources from networks located primarily in
the northeast and midwest to networks located in the south and west (see
fig. 2). VERA shifted approximately $921 million among networks in fiscal
year 2001 compared to what allocations would have been if networks received
the same proportion of funding they received in fiscal year 1996, the year
before VERA was implemented. This included additional resources Congress
appropriated from fiscal year 1996 through fiscal year 2001. VERA shifted
the most resources-approximately $198 million?to Network 8 (Bay Pines), and
VERA shifted the most resources from Network 3 (Bronx)?approximately $322
million. The shift occurred

VERA Has Had A Substantial Effect on Network Resources and Workload

VERA's implementation resulted in 10 of VA's 22 networks receiving a smaller
share of VA's medical care appropriation in fiscal year 2001 than in fiscal
year 1996. However, because VA's total medical care appropriation rose 22
percent during this period, all but two of these networks received more
resources in fiscal year 2001 than in fiscal year 1996. The two networks
with fewer resources from fiscal year 1996 to 2001 were Network 1 (Boston)
and Network 3 (Bronx), which experienced 1 percent and 10 percent declines,
respectively.

VA has also used VERA as one component of a larger strategy to improve
access to care by increasing the number of veterans treated.7 Because VERA
allocates resources based on workload, it provides incentives for networks
to increase the number of veterans treated. The number of veterans treated
nationally in VA, in all priority groups, increased from 2.6 million in
fiscal year 1996, the year before VERA was implemented, to 3.8 million in
fiscal year 2001, an increase of 47 percent. All 22 networks contributed to
this increase (see fig. 3). This includes networks from which VERA shifted
resources. VA's reduction in inpatient care, closure of acute care beds,
shift in emphasis to less expensive outpatient care delivery, and a 22
percent increase in VA's annual medical care appropriation since fiscal year
1996 have provided additional capacity allowing networks to increase
workloads.

7Eligibility Reform also added to the number of veterans enrolled to receive
care. The number of enrolled veterans increased from 4.2 million in the
first full year of enrollment in fiscal year 1999 to 6.0 million in fiscal
year 2001.

 Figure 3: Percentage Increase in Veterans Treated by Network, Fiscal Years
                             1996 through 2001

90 Percentage increase

                                  tlanta) )

go)a

                                      )

)

)e

6 (

s)

                                     )

h)

ton)

h)g

                                  nnati) )

)n

                     ) )) ) tland)ham) )ille) ity allas

lbany

                                 erlis ix o

xn

nl

                              boaltimor e ckso

si

22 (Long Beac

                                 coenco envC

ni

                              inneaporo bur hv

os

n Arn

                                     hic

y Pa

an FrancS21 (

                                    sur

inL

14 (

D17 (

(B3

ic

orP

20 (

A2 (

A7 (

                                     as

hP18 (

                                     sa

D19 (

B1 (

ittsP4 (

                                    (JaD

C12 (

                                     in

N9 (

                                     an

B5 (

B(

8

C10 (

A11 (

61

M13 (

K15 (Network

VERA Provides A Reasonable Approach to Resource Allocation

Source: GAO analysis of VA data.

VERA's design promotes the allocation of comparable resources for comparable
workloads to VA's 22 health care networks consistent with principles used by
other payers, such as the Medicare and Medicaid programs, and expert views
on the design of payment systems.8 VERA allocates resources based primarily
on networks' patient workloads. To ensure the comparability of networks'
resources with their workloads,

8For a discussion of payment system principles and implementation, see
Medicare Payment Advisory Commission, Reporttothe
Congress:MedicarePaymentPolicy(Washington, D.C.: March 2001) and Nigel Rice
and Peter C. Smith "Capitation and Risk Adjustment in Health Care Financing:
An International Progress Report," TheMilbank Quarterly, Volume 79, No. 1,
2001.

VERA adjusts these allocations for factors beyond networks' control, namely
patient health care needs and certain local costs. By adjusting allocations
only for costs beyond a network's control, VERA holds networks accountable
for providing services efficiently. Also, VERA provides protection for
patients from the risk that a health care network would not be able to
provide services because its expenditures exceed available resources.

VERA allocates resources primarily on the basis of network patient workload.
Each network receives an allocation based on a predetermined dollar amount
per veteran served.9 This is consistent with how other federal health care
payers allocate resources to managed care plans to care for their patient
workload. Because VERA uses workload to allocate resources, networks that
have more patients generally receive more resources than networks that have
fewer patients (see table 1). However, allocation adjustments result in some
situations in which networks with fewer patients receive higher total and
per patient allocations. For example, Network 3 (Bronx) received a larger
VERA allocation in fiscal year 2001 than Network 9 (Nashville) even though
Network 3 (Bronx) had a smaller workload. By receiving funding based on
workload, VA's health care networks have an incentive to focus on aligning
facilities and programs to attract patients rather than focusing on
maintaining existing operations and infrastructure regardless of the number
of patients served.

Table 1: VERA Patient Workload Compared to VERA Allocation, Fiscal Year 2001

                             Network (location)

          Fiscal Year 2001 VERA patient workload Fiscal Year 2001 VERA total
      allocation (in millions) Fiscal Year 2001 VERA per patient allocation

9VERA allocated
about $16.2
billion in fiscal
year 2001 for
basic care and
complex care and
$878 million for
equipment and
nonrecurring
maintenance based
on patient workload. In addition, VERA allocated about $688 million for
research support and education support based on other workload measures.

                             Network (location)

          Fiscal Year 2001 VERA patient workload Fiscal Year 2001 VERA total
      allocation (in millions) Fiscal Year 2001 VERA per patient allocation

Source:
Department of
Veterans Affairs,
Veterans
Equitable
Resource
Allocation:
Equity of Funding
and Access to
Care Across
Networks
(Washington,
D.C.: 2001), pp.
27 and 64 and GAO
calculations.

VERA seeks to
ensure that
comparable
resources are allocated for comparable workloads by adjusting for
differences in networks' patient health care needs and certain local costs
in calculating networks' allocations. Without these adjustments, networks
with justifiably higher costs could face pressure to compromise access to
care or lower health care quality, while networks with lower costs could
receive more resources than needed. To prevent this problem, VERA, like
other federal health care payment systems, makes adjustments to its per
patient allocations or capitation amounts.

VERA adjusts for patient health care needs?case mix?by first classifying
patients into categories by overall level of health care need and then by
setting capitation amounts for each of these patient categories. VERA
classifies patients into one of three categories according to the level of
health care needs and associated costs. The first category is complex care,
which includes patients who generally require significant high-cost
inpatient care as an integral part of their rehabilitation or functional
maintenance, and is about 4 percent of VA's workload. This category includes
most patients in VA's special disability programs such as those

with spinal cord injuries and serious mental illness.10 The second category
is basic vested care, which includes patients who have relatively routine
health care needs and are principally cared for in an outpatient care
setting. These patients?84 percent of VA's workload?rely primarily or
completely on VA for meeting their health care needs, may require short-term
inpatient admissions, and typically require significantly fewer resources
than complex care patients. The third category is basic non-vested care
which is 12 percent of VA's workload. This category includes patients who
also have relatively routine health care needs but receive only part of
their care through VA, are less costly to VA than basic vested patients, and
have not undergone a comprehensive medical evaluation by a VA practitioner.

The adjustments to capitation amounts for each category reflect whether
patients in a category are more or less costly than patients in another
category. These adjustments, or case-mix weights, determine what proportion
of VERA resources will be allocated to networks to care for patients in each
case-mix category, such as complex care. As a result, VERA's patient
case-mix adjustment provides more funding to networks with greater
proportions of complex care patients. For example, if two networks have the
same number of patients but one has more complex care patients, it will
receive a greater allocation because the VERA case-mix weight for complex
care is higher.

In addition, VERA adjusts for uncontrollable geographic price differences in
the resources it allocates. These differences result primarily from
variations in federal employee pay rates in different parts of the country.
VERA makes this adjustment by applying a price adjustment factor to each
network's allocation. The adjustment lowers the VERA allocation for networks
located in lower cost areas and raises the allocation for networks located
in higher cost areas. In fiscal year 2001, Network 8 (Bay Pines) had the
largest decrease resulting from the geographic price adjustment? 2.8
percent. Network 21 (San Francisco) had the largest increase resulting from
the geographic price adjustment? 6.3 percent. Through fiscal year 2001, this
adjustment was for services provided only by VA employees. Beginning in
fiscal year 2002, VA expanded the

10VA's special disability programs provide services to veterans disabled by
spinal cord dysfunction, traumatic brain injury, blindness, amputations,
serious mental illness, and post-traumatic stress disorder.

geographic price adjustment to all VERA allocations by including contract
labor costs and contract nonlabor purchases, such as energy.

VERA's allocation of resources based on workload with adjustments only for
costs beyond the networks' control aims to promote equity and efficiency. To
promote equity, VERA adjusts network allocations by case mix and geographic
price to standardize measures of workload and resources so that each network
receives comparable allocations for comparable workloads. To create an
efficiency incentive, VERA provides fixed capitation amounts for patient
categories that are the same for each network and are intended to reflect
VA's average costs instead of historical local costs. Using fixed capitation
amounts is consistent with how other health care payers provide managed care
plans with an incentive to operate efficiently by placing them at risk if
their expenses exceed the payment amount.

VERA also provides protection of patients from the risk that a health care
network would not be able to provide services because its expenditures
exceed available resources. VERA does this annually through the National
Reserve Fund which provides supplemental resources to networks when they
have difficulty operating within their available resources. VA's National
Reserve Fund is used to cover network requests for supplemental allocations
over and above networks' annual VERA allocations and other sources of
revenue. For fiscal years 1999 through 2001, VA has set the National Reserve
Fund amount at $100 million using a combination of annual and carry-over
funds. Since fiscal year 1999, resources distributed through the National
Reserve Fund have averaged approximately 1 percent of total VERA allocations
and supplemented VERA allocations in six networks.

Although VERA's overall design is a reasonable approach to allocate
resources, we identified weaknesses in its implementation. First, VERA's
calculation to ensure the comparability of networks' resources with their
workloads and their patient health care needs is not as accurate as it could
be. Second, the process for providing supplemental resources through the
National Reserve Fund process does not provide adequate information to
determine the extent to which networks need supplemental funding as a result
of potential problems in VERA, network inefficiency, or other factors.

VERA Changes Required to Better Align Resources With Workload

VERA's Calculation of Networks' Workloads and Patient Health Care Needs Is
Not as Accurate as It Could Be

Excluding Most Higher Income Veterans Without a Service-Connected Disability
Compromises the Comparability of Network Allocations

VERA's calculation of networks' workloads excludes most higher income
veterans without a service-connected disability-a growing proportion of VA's
users. In addition, VERA does not account for variation in patients' health
care needs and related costs among networks as accurately as it could.

When VERA was established, the number of higher income veterans treated
without a service-connected disability was small-approximately 108,000 or
about 4 percent of the total number of veterans treated in fiscal year 1996.
Because of their small numbers and the expectation that collections from
copayments, deductibles, and third-party insurance reimbursements would
cover the majority of their costs, VERA did not include most of these higher
income veterans in basic care workload. However, the number of these
veterans treated increased greatly in recent years and represent about 95
percent of VA's Priority 7 health care enrollment category. The number of
Priority 7 veterans treated increased to approximately 827,722 users (see
fig. 4).11 Priority 7 veterans comprised 22 percent of VA's total fiscal
year 2001 patient workload. This rapid growth in the number of Priority 7
veterans treated has occurred even though networks do not receive additional
VERA allocations for the majority of this workload and collections covered
only 24 percent of Priority 7 veterans' costs in fiscal year 2000.12
Networks pay for most of the costs of Priority 7 services through VERA
allocations made mostly on the basis of non-Priority 7 workload.

11VERA does include some Priority 7 veterans in its workload measure. In
fiscal year 2000, about 8 percent of Priority 7 veterans treated were
included in VERA's workload measure because they were complex care patients
or basic care patients with service-connected conditions.

12Facility collections from copayments, deductibles, and third-party
insurance reimbursements cover some of the costs of services for Priority 7
veterans.

Figure 4: Growth of Priority 7 Veterans Treated, Fiscal Years 1996 through
2001

900,000 Patients

800,000

700,000

600,000

500,000

400,000

300,000

200,000

100,000

0

1996 1997 1998 1999 2000 2001 Fiscal year

Note: Because the Priority 7 classification was not developed until fiscal
year 1999, we used VA's previous classification that most closely represents
this priority group, Category C. The Category C data we used for fiscal
years 1996 through 1998 slightly underestimate the number of Priority 7
veterans for those years.

Source: VA.

The omission of these veterans from VERA's workload calculation creates an
inequitable allocation of resources across networks because networks'
proportion of Priority 7 veterans treated varies (see fig. 5). For example,
in fiscal year 2001, Priority 7 users were 32 percent of Network 14's
(Lincoln) total veterans treated compared to the VA average of 22 percent.
Consequently, networks with a higher proportion of Priority 7 veterans, like
Network 14 (Lincoln), have fewer resources per patient to treat veterans
than networks with a lower proportion of Priority 7 veterans.

Figure 5: Priority 7 Veterans Treated by Network, Fiscal Year 2001

40 Percentage of overall workload

e)

                               ham) ille)) )

                                   nnati)i

ton) )

                              gh) tlanta) ) )

)n

                                    )) )

go)a

                                     )

h)

                                ) ))nx Pines

lbany

                                  is City

oc

lno

                                  tlandbor

hoenixP18 (

                                    allas

ltimora

                                   eckso

ong BeacL

                                inneapol env

or

                                   bur is

                                   22 ( hv

os

incL

14 (

                                  Ar hicur

an FrancS21 (

B3 (

                                     as

D17 (

orP

20 (

                                     as

A2 (

                                     inc

A7 (

B1 (

ittsP4 (

D19 (

                                   ay (Ja

D6 (

                                     nn

C12 (

                                     ans

N9 (

B5 (

B8 (

C10 (

A11 (

61

M13 (

K15 (Network

Source: VA.

VA assessed the possibility of including Priority 7 veterans in VERA's basic
vested workload. However, it had concerns that including Priority 7 veterans
in VERA workload would create a possible incentive to serve higher income
veterans at the expense of service-connected and low-income veterans. VA
considered providing a capitation amount for Priority 7 veterans that was
less than the average cost of their care. However, rather than including
Priority 7 veterans in the workload calculations with a reduced capitation
amount, VA decided instead to pursue other options.

One of the options VA is examining is the effect of changing the income
threshold used to classify enrolled veterans. Specifically, the current
uniform national income standard used in part for determining Priority 7

status would be replaced with a regional income standard to account for
regional differences in the cost of living. This would change the status of
some Priority 7 veterans in high-cost regions to low-income veterans- who
are included in VERA's workload calculation. Although adopting a regional
income threshold could improve the equity of resource allocation, the
alignment of workload with resources would still be compromised if some
networks continue to have disproportionate numbers of the remaining Priority
7 veterans.

Inclusion of Priority 7 veterans in VERA basic vested care workload would
increase the comparability of resources among networks per patient
treated.13 This would move resources from networks with a smaller proportion
of Priority 7 veteran workload to networks with a larger proportion of
Priority 7 veteran workload. If, for example, Priority 7 basic vested
veterans?those who rely primarily or completely on VA for meeting their
health care needs-were capitated at half the average national cost of their
care, as VA had considered, this would have increased the allocation to 9
networks in the northeast and midwest and decreased the allocation to 10
networks in the south and west in the fiscal year 2001 VERA allocation (see
fig. 6).

13VA's Office of Inspector General also recommended that VA include Priority
7 workload in the VERA model. See Office of Inspector General, Department of
Veterans Affairs, Audit of The Availability ofHealthcareServicesin
theFlorida/Puerto RicoVeteransIntegrated ServiceNetwork (VISN)8, Report
Number 99-00057-55 (Washington, D.C.: August 13, 2001).

Figure 6: Estimated Change  in VERA Allocations from Adding Priority 7 Basic
Vested Veterans  to VERA Workload at  Half Their National  Cost, Fiscal Year
2001

Network 
2 (Albany)

3 (Bronx)

4 (Pittsburgh)

5 (Baltimore)

6 (Durham)

7 (Atlanta)

8 (Bay Pines)

9 (Nashville)

10 (Cincinnati)

11 (Ann Arbor)

12 (Chicago)

13 (Minneapolis)

14 (Lincoln)

15 (Kansas City)

16 (Jackson) -5,495,659

17 (Dallas)

18 (Phoenix)

19 (Denver)

20 (Portland)

21 (San Francisco)

22 (Long Beach)

                                     $0

Note: For this simulation we used VERA fiscal year 2001 workload numbers for
basic  vested care,  which  are the  total unduplicated  number  of veterans
served for fiscal years 1997, 1998, and 1999.

Source: GAO analysis of VA data.

VERA Does Not Adjust for Justifiable Cost Differences As Accurately As It
Could

Although VERA adjusts network allocations for cost differences resulting
from the mix of patients networks serve, it does not do so as accurately as
it could. This is because the case-mix weights assigned to each category of
patients are based on historical cost data from fiscal year 1995 and VERA
only uses three case-mix categories to allocate resources.

Case-Mix Weights Based On Historic Data Do Not Reflect Changes In VA Health
Care

VERA uses case-mix weights based on VA health care expenditures in fiscal
year 1995 to allocate resources for basic and complex care workload. These
weights are determined by the share of resources spent on basic and complex
care in that year ? 61.6 percent of expenditures for basic care and 38.4
percent for complex care. For the VERA allocation in fiscal year 2001, for
example, $6.2 billion was available for complex care (38.4 percent) and
$10.0 billion was available for basic care (61.6 percent).

These case-mix weights, however, have not been updated to reflect the health
care that VA is providing. Because of VERA, VA Eligibility Reform, and other
VA initiatives, the number of basic care patients has increased since fiscal
year 1995 while the number of complex care patients has remained relatively
constant. The rising proportion of basic care patients has contributed to a
greater proportion of VA expenditures for basic care and a smaller
proportion of expenditures for complex care. By fiscal year 1999, 66.9
percent of expenditures were for basic care and 33.1 percent were for
complex care. Adjusting capitation amounts to reflect current expenditures
for basic and complex care would result in an approximately 9 percent
increase in the basic care capitation amounts and about a 14 percent
reduction in the complex care capitation amount (see table 2).

             Table 2: VERA Capitation Amounts, Fiscal Year 2001

      Capitation amount if most current data used for 2001 VERA allocationsb

                             Case-mix category

Actual 2001 VERA capitation amounta

                      Basic non-vested care $121 $132

                       Basic vested care 3,126 3,395

                         Complex care 42,765 36,869

aCapitation amounts based on fiscal year 1995 VA expenditures.

bCapitation amounts based on fiscal year 1999 VA expenditures.

Source: Actual fiscal year 2001 capitation amounts from Department of
Veterans Affairs, Veterans Equitable Resource Allocation: Equity of Funding
and Access to Care Across Networks (Washington, D.C.: 2001), p. 9.
Capitation amounts for most current data calculated by GAO.

VA considered updating the weights for basic and complex care based on the
most recent available costs. VA officials told us they have maintained the
fiscal year 1995 case-mix weights because using more current expenditure
data that would lower the allocation for complex care and increase the
allocation for basic care could be seen as a weakening of VA's commitment to
serve veterans with complex care needs, such as those with spinal cord
injuries or serious mental illness. However, continuing to base VERA
case-mix weights on fiscal year 1995 expenditures has not ensured that
resources were spent on complex care patients. VERA, like other allocation
systems, provides networks with resources but it does not require networks
to spend resources in a particular way. Rather, VA program guidance and
network management decisions determine how resources are spent. Eighteen of
22 networks spent less for complex care than they received based on their
complex care workload in fiscal year 2000, the most recent year for which
expenditure data are available (see table 3). As a result, the proportion of
VA's total expenditures on complex care has declined since fiscal year 1995
even though the proportion of VERA's allocation for complex care has
remained constant. However, VA has decided to defer action on using the most
recently available costs pending further study of how costs and workload
vary in complex care categories among networks.

Table  3:  Complex Care  Workload  Allocations  Compared With  Complex Care
Expenditures, Fiscal Year 2000

                            Dollars in millions

                            Dollars in millions

Note: The complex care allocation includes the VERA fiscal year 2000
geographic price adjustment.

Source: GAO calculations based on VA data.

Aligning VERA case-mix weights proportionally with current expenditures is
one way to better reflect how health care is delivered in VA. Doing so,
however, assumes that expenditures alone are an appropriate measure of
health care need. This is not always the case. For example, if health care
in a particular case-mix category is not being provided efficiently, using
expenditure data alone would result in a higher than necessary case-mix
weight. This would lead to excess resource allocation for the case-mix
category. On the other hand, using expenditure data only would result in a
lower case-mix weight than appropriate if health care providers are not
using more expensive treatments when needed to provide clinically
appropriate care. This would lead to insufficient resource allocation for
the case-mix category. As a result, setting case-mix weights may begin with
consideration of current expenditures, but ultimately must use the best
available data to reflect efficiency and clinically appropriate care.

The Small Number Of Case-Mix Categories In VERA Does Not Accurately Adjust
For Network Differences In Veterans' Health Care Needs

VERA uses only three case-mix categories-complex, basic vested, and basic
non-vested-to adjust for differences in health care needs and related
resource requirements for veterans. These three case-mix categories are
based on 44 patient classes VA uses to classify its patients. Using all 44
patient classes as case-mix categories would more accurately adjust for
differences in needs and related resource requirements because the average
costs of patients in the classes within the VERA categories vary
significantly and can be dramatically higher or lower than their capitation
amounts for the current three case-mix categories (see table

4).14 For example, the national average patient cost for domiciliary
care?one type of complex care?in fiscal year 2000 was roughly $17,000 less
than the $42,153 capitation amount for complex care, while the average
patient cost for ventilator-dependent care ? another type of complex
care?was about $121,000 more than the complex care capitation amount.

Table 4: VA Patient  Classes, Costs, and National Capitation Amounts, Fiscal
Year 2000

                                                           National National
                                                          average capitation
                                                 Patient classes cost amount

Complex care Chronic mental illness

    Extended and
 residential care

14RAND,
An
Analysis
of
theVeterans
EquitableResourceAllocation
(VERA)
System(Santa
Monica,
California,
2001),
pp.
21-22
discusses
the
need
for
additional
case-mix
adjustment
in
VERA
as
does
Price
Waterhouse
LLP and The Lewin Group, Inc., VeteransEquitable
ResourceAllocationAssessment-FinalReport, March 27, 1998.

                                                           National National
                                                          average capitation
                                                 Patient classes cost amount

Blind rehabilitation service 33,545 42,153

Stroke 26,533 42,153

Traumatic brain injury 24,146 42,153

Acquired Immune Deficiency Syndrome or human

immunodeficiency virus  positive with  antiretroviral therapy  19,870 42,153
Basic vested care

Acute care

Note: These are
the patient
classes VA uses
to group patients
into complex
care, basic
vested care, and
basic non-vested
care categories
in VERA's
case-mix
adjustment of
workload.

Source: GAO
calculations
based on VA data.

Our analysis
shows that
considerable
variation exists
among networks in
the type of
workload
represented by
VERA's three
case-mix
categories, which
limits VERA's ability to allocate comparable resources for comparable
workload. VERA provides more resources to networks, relative to their costs,
that have proportionately more workload in less expensive patient classes,
such as domiciliary care, than other networks. VERA provides fewer resources
to networks, relative to their costs, that have more workload in more
expensive patient classes, such as ventilator-dependent care. Using VA's
current 44 patient classes rather than the three

case-mix categories VERA used in fiscal year 2001 would result in a
significant movement of resources for some networks because of the variation
by network in the type of workload (see fig. 7).15 This would move resources
from networks having proportionately fewer patients in expensive patient
classes to networks having proportionately more patients in expensive
patient classes, resulting in an average movement of resources of 2 percent
per network.

15For our simulation we used VA's 44 patient classes because data were
readily available.

Figure 7: Estimated Change in VERA Allocations Among Networks as a Result of
Using 44 Case-Mix Categories, by Network, Fiscal Year 2001

In dollars Network

1 (Boston) 40,537,063

  2 (Albany) 3 (Bronx) 4 (Pittsburgh) 5 (Baltimore) 6 (Durham) 7 (Atlanta) 8
    (Bay Pines) 9 (Nashville) 10 (Cincinnati) 11 (Ann Arbor) 12 (Chicago) 13
       (Minneapolis) 14 (Lincoln) 15 (Kansas City) 16 (Jackson) 17 (Dallas)

18 (Phoenix) 19 (Denver) 20 (Portland) -36,471,841 21 (San Francisco) 22
(Long Beach) -32,831,165

$0

Note: We used fiscal year 1999 expenditure data for the calculations Source:
GAO analysis of VA data.

In 1998, VA conducted a similar analysis using 54 patient classes for
allocation and found that this would have moved a significant amount of
resources among networks, an average of 4 percent per network.16 The
analysis further concluded that using only 7 of the 54 classes achieved
nearly the same result. A 1998 Price Waterhouse analysis of VERA also
concluded that additional case-mix categories would increase equitable
resource allocation.17 VA officials told us they have not introduced more
than three case-mix categories because VA wants VERA to be easily understood
by stakeholders.

While using more case-mix categories can increase the accuracy of
allocations, the literature and experts we consulted suggest that a case-mix
classification system needs to address two concerns in order to prevent
providers from receiving inappropriately high levels of resources. First,
having a larger number of case-mix categories may provide more opportunities
for networks to inappropriately classify patients to receive the highest
capitation amount. However, increasing the VERA case-mix categories from
three to a higher number, but not necessarily 44, may strike an appropriate
balance between improved allocation and the need to control for potential
inappropriate coding of patients into higher capitation categories. Second,
basing case-mix categories in part or in whole on utilization of services
provides the incentive to overuse services. For instance, in VERA, a patient
who receives nine home-based primary care visits is categorized in basic
vested care with a capitation amount of $3,126; however, a patient who
receives 10 visits is categorized in complex care, which has a capitation
amount of $42,765. Consequently, if networks increase the number of such
visits, they can increase their funding more than 13-fold. Currently, 22 of
VA's 44 patient classes incorporate utilization factors in classifying
patients. These utilization factors are found primarily in the patient
classes for extended and residential long-term care and chronic mental
health services and for classifying basic non-vested patients. Replacing
utilization criteria with diagnosis and functional measures where possible
in VERA's case-mix categories would reduce the incentive to overuse
services, especially for complex care patients.

16The number of VA categories declined from 54 in 1998 to 44 in 2001 because
VA combined some of the 1998 categories.

17Price Waterhouse LLP and The Lewin Group, Inc., VeteransEquitableResource
AllocationAssessment-FinalReport,March 27, 1998.

Most VERA complex care patient classes are based in part on some measure of
service utilization because of the difficulty in predicting the costs of
these classes based solely on diagnostic data. Because complex care costs
are high and unusually difficult to predict, the literature and experts we
consulted suggest that it is prudent to partially insure networks from such
unpredictable costs. 18 Therefore, it may be advantageous to use a mechanism
to help providers, such as VA's health care networks, cope with their
highest cost complex care patients by providing additional resources for
their care based on a formula. If VA used such a funding mechanism, networks
with complex care patients in the 99th percentile of cost, for example,
would receive the network complex care capitation amount plus a
predetermined percentage of the cost above the capitation amount. The
additional funds above the capitation amount would partially offset the
network's expenses for high-cost complex care patients. Resources for this
funding mechanism could be set aside as part of the National Reserve Fund.

Currently, VA is exploring alternative case-mix classification systems, such
as Diagnostic Cost Groups (DCG), that could provide more case-mix
categories, classify patients based on nonutilization criteria, and better
predict costs for acute care patients. DCGs place patients into different
groups based on patient demographics and medical diagnoses. However, VA
researchers have found that the DCG diagnosis-based system may not be
sufficient to allocate resources for certain complex care patient classes.
Predicting the costs of many complex care patients is problematic because
complex care patients, including those with mental illness and those in
extended care settings, may have the same diagnosis but may need very
different levels of treatment and support. VA is studying the possibility of
supplementing a diagnoses-based system with utilization information in order
to better predict the costs of complex care patients.

18For discussion of predicting costs for mental health patients using
different risk adjustment models, see Susan Ettner, Richard Frank, Thomas
McGuire, and Richard Hermann, "Risk Adjustment Alternatives in Paying for
Behavioral Health Care Under Medicaid," HealthServicesResearch, Volume 36,
No. 4, 2001.

Combined Effect of Changes in Workload and Case-Mix Measures to Better Align
Resources and Workload Would Result in Significant Reallocation Among Some
Networks

Implementing changes to VERA could better align resources with workload for
VA's 22 health care networks by addressing case mix and workload issues.
Incorporating all 44 of the current case-mix categories, updating case-mix
weights to reflect the current distribution of expenditures, and funding
Priority 7 basic vested veterans at 50 percent of costs would better align
resources and workloads. Incorporating the 44 case-mix categories would have
the largest effect on resource allocation. The combined effect of these
changes would provide additional resources to some northeastern and
midwestern networks and reduce resources for some southern and western
networks (see fig. 8). The allocation change represents about 2 percent of
networks' budgets, but is more substantial for some networks. Network 1
(Boston) would get approximately a 5 percent increase and Network 20
(Portland) approximately a 5 percent decrease. These changes would better
align approximately $200 million with workload.

Figure  8:  Estimated  Change  in  VERA Allocations  from  Incorporating  44
Case-Mix Categories, Most Current Expenditure Data for Case-Mix Weights, and
Priority 7 Basic Vested Veterans Treated, Fiscal Year 2001

Network In dollars

1 (Boston) 
2 (Albany)

3 (Bronx) 41,573,159 4 (Pittsburgh)

5 (Baltimore) -23,395,750

  6 (Durham) 7 (Atlanta) 8 (Bay Pines) 9 (Nashville) 10 (Cincinnati) 11 (Ann
       Arbor) 12 (Chicago) 13 (Minneapolis) 14 (Lincoln) 15 (Kansas City) 16
                                                      (Jackson) 17 (Dallas)

-13,596,270

-541,935 -12,456,323

18 (Phoenix) -21,888,405 19 (Denver) -39,781,854 21 (San Francisco) 22 (Long Beach)

$0

Note: We allocated resources for Priority 7 basic vested care veterans at 50
percent of the national average cost which is $849 per veteran. We used
fiscal year 1999 expenditure data for these calculations.

Source: GAO analysis of VA data.

Inadequate Network Supplemental Funding Information Limits VA's Ability to
Take Corrective Action

VA has focused its process for administering the National Reserve Fund
almost solely on providing supplemental resources to networks to get through
a fiscal year but has not included in this process an examination of the
root causes of networks' needs for supplemental resources. To operate the
National Reserve Fund, VA, for the last 3 fiscal years (1999 - 2001), has
set aside about 1 percent of the VERA allocation in anticipation of networks
requiring supplemental resources.19 No networks requested supplemental
funding in fiscal years 1997 and 1998. However, six networks have requested
supplemental funding from fiscal year 1999 to fiscal year 2001 (see table
5). Supplemental allocations to four networks in fiscal year 2001 totaled
$220 million. Officials in 10 of 22 networks told us, in June 2001, that
they anticipated requesting supplemental funding at least once from fiscal
years 2002 through 2006.20

Table 5: Network Supplemental Funding, Fiscal Years 1999 through 2001

                            Dollars in millions

Source: VA.

VA has used three different approaches to determine whether networks
requesting additional resources would receive supplemental allocations. In
fiscal years 1999 and 2000, VA created teams consisting of staff from
networks not requesting supplemental funds. These teams reviewed networks'
funding requests and made recommendations regarding the amount of
supplemental allocations and efficiency initiatives networks needed to
implement in order to close gaps between their expected expenditures and
VERA allocations. In fiscal year 2001, responding in part

19Funds in the National Reserve Fund not needed for supplemental allocations
can be used for other purposes.

20We asked network directors if they expected to request supplemental
allocations under these conditions: VERA's design, current enrollment, and
third-party policies remain the same and budget increases are consistent
with recent years.

to criticisms of network staff review of allocation requests, VA replaced
the team review process with a review by VA headquarters officials. In this
process, VA headquarters officials reviewed requests for supplemental
resources from networks anticipating a budget shortfall for the year. In
fiscal year 2002, VA created a team to examine networks' need for
supplemental resources. This team consisted of headquarters and network
officials, including representatives from networks that requested
supplemental allocations and those that did not.

None of VA's approaches to supplemental allocations has systematically
evaluated the extent to which certain factors caused networks to require
supplemental allocations. In fiscal years 1999 and 2000, VA teams conducted
site visits and reviewed financial and clinical information that requesting
networks provided. The teams made recommendations for supplemental
allocations in order to prevent network shortfalls. However, VA could not
determine to what extent supplemental resources were needed due to
imperfections in VERA, lack of network efficiency, inability to predict
complex care patients' costs, or lack of managerial flexibility to close or
consolidate programs or facilities because the teams did not collect the
information needed to make this determination. 21 Although the evaluation
process changed in fiscal year 2001, VA was still unable to make such a
determination. For example, in fiscal year 2001, about half the supplemental
resources VA provided to networks was for "inflation and miscellaneous
program adjustments." All networks experienced inflation, however, and VA
did not distinguish between the level of inflation in networks that
requested supplemental resources and those that did not.

VA officials told us that the changes for fiscal year 2002 will still not
allow them to determine the extent to which various factors cause networks
to need supplemental resources. As a result, VA cannot provide adequate
assurance that supplemental allocations are appropriate or take needed
action to correct problems that cause networks to have budget shortfalls.

One of the corrective actions VA could take is to assist networks that
experience budget shortfalls because of an unusually large number of
high-cost complex care patients in a given year. This is important because

21  RAND,    An   Analysis    of   theVeterans   EquitableResourceAllocation
(VERA)System(Santa  Monica,  California, 2001),  pp.  24-28. RAND  discusses
barriers  that  some networks  may  face  to achieve  the efficiencies  VERA
incentives provide because of difficulties in consolidating facilities.

Conclusions

the methods used to predict health care costs are not as precise in
predicting the costs of many complex care patients as they are in predicting
the costs of many basic care patients. As a result some networks' budget
shortfalls could be explained in part by a higher than expected number of
high-cost complex care patients. To address this risk, some other payers
have established funding mechanisms to address the costs of these very
expensive patients. For example, some state Medicaid programs have used a
mechanism called stop-loss or reinsurance to reimburse managed care plans
for certain benefits that exceed a specified expense limit.22 If VA were to
use a similar funding mechanism as part of the National Reserve Fund, this
could help protect networks from budget shortfalls by providing additional
resources above the capitation amount for complex care patients that reach a
predetermined level of cost. VA is studying ways to address the risk that a
network may have unusually high-cost patients in a given year that are not
predicted in a resource allocation model.

VERA's overall design is a reasonable approach to resource allocation and
has helped promote more comparable resource allocations for comparable
workloads in VA. This approach is reasonable because VERA allocates
resources primarily on the basis of workload and attempts to adjust network
resources for factors beyond the control of network management, encourage
efficiency, and provide protection to patients against network budget
shortfalls. The implementation of this approach resulted in VA's shifting
resources to more closely mirror shifts in the veteran population from the
northeast and midwest to the south and west.

Although VERA's design is a reasonable approach to resource allocation, VA
could correct weaknesses in VERA's implementation to improve the
comparability of resource allocations with networks' workloads. One of
VERA's implementation weaknesses is that it does not include most Priority 7
veterans in its workload even though the Priority 7 workload now represents
about one-fifth of patients served. If the number of Priority 7 veterans VA
treats continues to increase, this may create even more serious inequities
in the future.

22John Holahan, Suresh Rangarajan, and Matthew Schirmer, MedicaidManagedCare
Payment Methodsand Capitation Rates:Resultsof aNationalSurvey,Urban
Institute, May 1999, p. 21.

VERA's adjustment for differences in patient health needs across networks
emphasizes simplicity at the cost of increased accuracy. Maintaining only
three case-mix categories in VERA does not adequately account for important
variations in health care needs among networks. Increasing the VERA case-mix
categories from three to a higher number would better account for the
variation in health care needs across networks and would have the largest
effect on resource allocation. In addition, changes are needed to update
VERA's case-mix weights to better reflect how VA health care is now
delivered. Updating case-mix weights may begin with using current
expenditure data, but additional consideration should be given to using the
best available data on appropriate clinical care and efficiency.

In addition to these weaknesses, VA has not used the supplemental funding
process for improving VERA allocations and management of VA's resources.
Although the amount of resources provided to networks through the
supplemental funding process has continued to increase, VA has not been able
to determine the relative contribution of factors, such as imperfections in
VERA, network inefficiency, inability to predict complex care costs, or lack
of managerial flexibility to close or consolidate programs or facilities, to
the need for supplemental resources. An important factor that other health
care payers have identified and account for that may contribute to VA
network budget shortfalls is the inability to accurately predict the cost of
complex care patients. Other payers have addressed this risk by using a
funding mechanism to partially offset the unanticipated costs of such
patients. Because VA has not identified the relative contribution of this
factor and other factors that could cause network budget shortfalls, VA is
unable to provide assurance that the supplemental funding is appropriate or
take needed action to correct problems that cause networks to have budget
shortfalls.

Making changes to address weaknesses in VERA will add some complexity to how
VA allocates resources. Doing so, however, will better align the allocation
of approximately $200 million with workload.

Recommendations for To continue to improve the allocation of comparable
resources for comparable workloads through VERA, we recommend that the
secretaryExecutive Action of veterans affairs direct the under secretary for
health to:

* better align VERA measures of workload with actual workload served
regardless of veteran priority group,

* incorporate more categories into VERA's case-mix adjustment,

* update VERA's case-mix weights using the best available data on clinical
appropriateness and efficiency,

* determine in the supplemental funding process the extent to which
different factors cause networks to need supplemental resources and take
action to address limitations in VERA or other factors that may cause budget
shortfalls, and

* establish a mechanism in the National Reserve Fund to partially offset the

Agency Comments
and Our Evaluation

cost of networks' highest cost complex care patients.

In comments on a draft of this report, VA agreed with our conclusions that
VERA's design is a reasonable approach to allocate resources commensurate
with workloads and that VERA, in concert with other VA initiatives, has
provided an incentive for VA to serve more veterans. VA also acknowledged
the opportunities for improvements in VERA's implementation that we
identified and concurred with our recommendations. VA's comments are in
appendix II.

VA concurred with our workload and case mix recommendations, recognizing the
substantial trend in Priority 7 workload expansion and the case-mix
limitations of having only three pricing groups within VERA. VA anticipates
that the distribution of an expected fiscal year 2002 supplemental
appropriation will consider the Priority 7 workload, but as of February 22,
2002, Congress has not provided VA with the supplemental appropriation it
anticipates. Further, VA is evaluating the appropriateness of expanding the
number of VERA price groups to include corresponding updates of case-mix
weights, but will not make a decision about these potential fiscal year 2003
VERA modifications until September 2002. In its comments, VA also indicated
that it plans to wait for further study of VERA's workload and case-mix
measures to determine whether all Priority 7 workload and case-mix
refinements should be incorporated in the fiscal year 2003 VERA model. Given
the extensive study of most of these issues already conducted by VA and
others, we encourage VA to implement our recommended VERA workload and
case-mix improvements in its fiscal year 2003 allocations to networks and to
further refine these improvements in the future as needed. Delaying these
needed improvements to VERA means that approximately $200 million will be
allocated annually in a manner that does not align workload and resources as
equitably as possible among networks.

VA also concurred with our recommendation to determine in the supplemental
funding process the extent to which different factors cause networks to need
supplemental resources, but the actions VA discussed to

improve the supplemental funding process do not address our recommendation.
VA used a new supplemental adjustment process in fiscal year 2002 to better
identify different factors that cause networks to require supplemental
resources. However, this process does not identify the root causes of a
network's need for additional resources as we recommended. Specifically,
VA's new supplemental process does not provide VA information on the
relative contributions of specific factors to network shortfalls such as
network inefficiency, imperfections in VERA, and the inability to predict
complex care costs. Until VA implements our recommendation, it cannot
provide assurance that supplemental resources are appropriate or take needed
actions to reduce the likelihood of network shortfalls in the future.

In addition, VA's discussion of actions for establishing a mechanism in the
National Reserve Fund to partially offset the cost of networks' highest cost
complex care patients do not fully address our recommendation. VA stated
that the resources it distributed to five networks through the fiscal year
2002 supplemental adjustment process are expected to meet these networks'
supplemental funding needs, including the cost of their highest cost
patients. To address our recommendation, however, VA would have to identify
individual complex care patients with unexpectedly high costs over the
course of the fiscal year and provide stop-loss coverage for such patients
to each network. VA's current process does not do this. However, as we have
noted, ongoing VA studies could develop ways to provide stop loss coverage
to networks for unpredictable high-cost complex care patients. Until VA
establishes such a funding mechanism, some networks may experience budget
shortfalls as a result of these unpredictable complex care costs.

We are sending copies of this report to the secretary of veterans affairs,
interested congressional committees, and other interested parties. We will
make copies of the report available to others upon request.

If you or your staffs have any questions about this report, please call me
at (202) 512-7101. Another contact and key contributors are listed in
appendix III.

Cynthia A. Bascetta Director, Health Care?Veterans' Health and Benefits
Issues

List ofRequesters

The Honorable Lane Evans
Ranking Minority Member
Committee on Veterans' Affairs
House of Representatives

The Honorable James T. Walsh
Chairman
The Honorable Alan B. Mollohan
Ranking Minority Member
Subcommittee on VA, HUD,

and Independent Agencies
Committee on Appropriations
House of Representatives

The Honorable Charles E. Grassley
The Honorable Chuck Hagel
The Honorable Tom Harkin
United States Senate

The Honorable Doug Bereuter
The Honorable Leonard Boswell
The Honorable Tom Latham
The Honorable James A. Leach
The Honorable Jim Nussle
The Honorable Lee Terry
House of Representatives

                      Appendix I: Scope and Methodology

We reviewed the Department of Veterans Affairs (VA) resource allocation for
fiscal years 1997 through 2001 to (1) describe the effect the Veterans
Equitable Resource Allocation (VERA) system has had on network resource
allocations and workloads, (2) assess whether VERA's design is a reasonable
approach to resource allocation, and (3) identify weaknesses in VERA that
may limit VA's ability to allocate comparable resources for comparable
workloads.

We worked with VA officials from the Resource Allocation and Analysis Office
to obtain documents and data on how VERA works and how VERA has changed
since fiscal year 1997. We also relied on other VA officials for our
assessment of VERA including officials from the Office of the Under
Secretary for Health, the Office of the Assistant Deputy Under Secretary for
Health, the Office of the Chief Financial Officer, the Office of Quality and
Performance, the Office of Policy and Planning, the Spinal Cord Injury
Strategic Healthcare Group, the Geriatrics and Extended Care Strategic
Healthcare Group, the Mental Health Strategic Healthcare Group, Health
Services Research and Development Service, and the Northeast Program
Evaluation Center. In addition, we interviewed officials from veteran
service organizations and payment system experts outside of government.

We obtained information on VERA's effect and how it could be improved
through interviews and documents from VA's networks. We visited five network
offices: Albany (2); Bay Pines (8); Bronx (3); Lincoln (14); and Minneapolis
(13) and conducted telephone interviews with officials in three additional
networks: Denver (19), Kansas City (15), and Phoenix (18). We chose these
networks because they were geographically diverse and had different
financial experiences under VERA. We also obtained information from network
officials about potential improvements to VERA including adding basic vested
care Priority 7 users, basing VERA case-mix weights on more current
expenditures, using more categories to adjust for case-mix differences, and
other factors. We had follow-up telephone interviews with network officials
in Bronx (3), Minneapolis (13), and Lincoln (14) regarding fiscal year 2001
supplemental resources from the National Reserve Fund.

We conducted an electronic mail survey to obtain the input of all 22 network
directors about VERA. We also obtained information on network directors'
anticipation of future supplemental funding requests.

To assess the reasonableness of VERA as an approach to resource allocation,
we performed a literature review using works published primarily within the
last 5 years. We searched the following databases:

Appendix I: Scope and Methodology

MEDLINE, ABI/Inform Global, and Econlit, and relied on publications from
other federal agencies. We focused our search on finding information on
similar health care payment systems, case-mix adjustment for acute and
extended care populations, and managing risk for mental health and special
care populations.

To assess the effect VERA has had on network resource allocations and
workload, we identified how resources have shifted among regions and
increased veterans' access to care as measured by the number of veterans
treated. We calculated the resources VERA shifted by projecting what
allocations would have been in fiscal year 2001 if networks received the
same proportion of funding in that year that they received in fiscal year
1996, the year preceding VERA's implementation. We calculated the resource
shifts by subtracting networks' actual 2001 VERA allocations from their
projected allocations. To calculate the total amount of resources VERA
shifted, we summed the absolute value of networks' net gains or losses and
divided by two. We divided the total by two to avoid double counting because
a dollar transferred to one network is the same dollar transferred from
another.

To determine weaknesses in VERA, we first examined our more than 10 years of
work reviewing VA's resource allocation processes. In addition, we relied on
external evaluations of VERA completed by Price Waterhouse, LLP and The
Lewin Group, Inc., AMA Systems, Inc., and the RAND Corporation. We
constructed simulation models to estimate the effect in fiscal year 2001 of
1) funding basic vested care Priority 7 patient users, 2) basing the
case-mix weights on current expenditures, 3) using all 44 VERA patient
classes to allocate network resources, and 4) the result of combining all
three of these simulations. VA provided workload and expenditure data from
fiscal years 1996 through 2000 and the actual VERA allocations received by
networks during fiscal years 1997 through 2001. These data were obtained
from VA's Office of the Chief Financial Officer, Allocation Resource Center,
and Office of the Assistant Deputy Under Secretary for Health.

To estimate the effect on VERA allocations of funding basic vested care
Priority 7 patients, we used the total unduplicated number of basic vested
care Priority 7 veterans served for fiscal years 1997, 1998, and 1999. In
addition, we assumed that these patients would be funded at 50 percent of
the national average cost or $849 in fiscal year 1999. In our simulation, we
chose to fund them at 50 percent of the national average cost based on
documentation from a prior recommendation by the Veterans Health

Appendix I: Scope and Methodology

Administration Policy Board. Funding these patients at less than full cost
lessens the incentive for networks to serve more Priority 7 veterans.

To estimate the effect on network allocations of basing case-mix weights on
current expenditures, we compared VERA's fiscal year 2001 allocations made
on the basis of fiscal year 1995 expenditure data to what fiscal year 2001
allocations would have been if they were based on fiscal year 1999
expenditure data. The fiscal year 1999 expenditure data were the most recent
available for case-mix weight calculations for the fiscal year 2001 VERA
allocation. Using fiscal year 1999 data, we computed new capitation amounts
for basic non-vested care, basic vested care, and complex care.

To estimate the effect on network allocations of using all 44 VERA patient
classes, we used fiscal year 1999 expenditures based on VERA workload. To
calculate new capitation amounts, we first calculated the percent of
expenditures spent on each of the 44 classes. Second, we calculated the
amount of resources available for each class by multiplying the new
percentages for each class by the total fiscal year 2001 resources that VERA
allocated. Third, we calculated the new capitation amounts by dividing the
amount of resources available by the corresponding VERA workload.

To estimate the combined effect on network allocations of making each of
these changes, we calculated capitation amounts for each of the 44 classes
and funded basic vested care Priority 7 users at 50 percent of the national
average cost based on fiscal year 1999 expenditures related to VERA
workload. In our simulation we created a separate category for basic vested
care Priority 7 patients because we did not have data on Priority 7 veterans
for each basic care patient class.

We tested VA computer-based data used in our analysis and concluded that it
was adequate for our purposes. To do this, we assessed the reliability of
workload (VERA and non-VERA) and expenditure data we obtained from VA that
were used in our analyses. When we identified inconsistencies between
databases, we tried to resolve them by interviewing officials responsible
for creating or maintaining the databases, updating the databases with
additional information VA provided, and requesting special data runs with
parameters that we specified. In addition, we confirmed that VA's Allocation
Resource Center verifies all workload and expenditure data used in the VERA
allocation process. We did not, however, verify whether these processes were
adequate. We relied on previous work to determine what limitations, if

Appendix I: Scope and Methodology

any, VA's  data may have had on the analyses  we completed. We performed our
review from October 2000  through December 2001 in accordance with generally
accepted government auditing standards.

Appendix II: Comments From the Department of Veterans Affairs

Appendix II: Comments From the Department of Veterans Affairs

Appendix II: Comments From the Department of Veterans Affairs

Appendix II: Comments From the Department of Veterans Affairs

Appendix III: GAO Contact and Staff Acknowledgments

GAO Contact James C. Musselwhite, (202) 512-7259

Staff Acknowledgments In addition to the contact named above Marcia A. Mann,
Jacquelyn T. Clinton, Thomas A. Walke, Diana Shevlin, Maria Vargas, Leslie
D. Blevins, Deborah L. Edwards, and Susan Lawes made key contributions to
this report.

Related GAO Products

MedicareManaged Care: Better Risk Adjustment Expectedto Reduce Excess
Payments Overall WhileMaking ThemFairertoIndividual Plans. GAO/T-HEHS-99-72.
Washington, D.C.: February 25, 1999.

MedicareManaged Care: Payment Rates, Local Fee-for-Service Spending, and
Other Factors Affect Plans'Benefit Packages. GAO/HEHS-99-9R. Washington,
D.C.: October 9, 1998.

VA Health Care: More VeteransAre Being Served, but Better OversightIs
Needed. GAO/HEHS-98-226. Washington, D.C.: August 28, 1998.

VA Health Care: Resource Allocation Has Improved, but Better Oversight Is
Needed. GAO/HEHS-97-178. Washington, D.C.: September 17, 1997.

Veteran'sHealthCare:Facilities'Resource Allocations Could Be More Equitable.
GAO/HEHS-96-48. Washington, D.C.: February 7, 1996.

VA Health Care: Resource AllocationMethodology Has Had Little Impact on
Medical Centers'Budgets. GAO/HRD-89-93. Washington, D.C.: August 18, 1989.

VA HealthCare: Resource AllocationMethodology ShouldImprove VA's
FinancialManagement. GAO/HRD-87-123BR. Washington, D.C.: August 31, 1987.

GAO's Mission

Obtaining Copies of GAO Reports and Testimony

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

The fastest and easiest way to obtain copies of GAO documents is through the
Internet. GAO's Web site (www.gao.gov) contains abstracts and full-text
files of current reports and testimony and an expanding archive of older
products. The Web site features a search engine to help you locate documents
using key words and phrases. You can print these documents in their
entirety, including charts and other graphics.

Each day, GAO issues a list of newly released reports, testimony, and
correspondence. GAO posts this list, known as "Today's Reports," on its Web
site daily. The list contains links to the full-text document files. To have
GAO e-mail this list to you every afternoon, go to www.gao.gov and select
"Subscribe to daily e-mail alert for newly released products" under the GAO
Reports heading.

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. General Accounting Office P.O. Box 37050 Washington, D.C. 20013

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

Visit GAO's Document GAO Building

Distribution  Center  Room 1100,  700 4th  Street, NW  (corner of 4th  and G
Streets, NW) Washington, D.C. 20013

To Report Fraud, Contact: Web site: www.gao.gov/fraudnet/fraudnet.htm,

Waste, and Abuse in E-mail: [email protected], or

Federal Programs 1-800-424-5454 or (202) 512-7470 (automated answering
system).

Jeff   Nelligan,   Managing  Director,   [email protected]  (202)   512-4800

Public Affairs U.S.  General Accounting Office, 441 G. Street NW, Room 7149,
Washington, D.C. 20548
*** End of document. ***