oversight

Review of Veterans Administration's Methodology for Determining Hospital Bed Size

Published by the Government Accountability Office on 1977-05-20.

Below is a raw (and likely hideous) rendition of the original report. (PDF)

                          DOCUMENT RISC!E                         9
02227 - [11532512] "9sLtrti   tj-
(Review of Veterans Administrationts Methodology for Determining
Hospital Bed Size]. HRD-77-104; B-133044. May 20, 1977. 2 pp. +
enclosure (25 pp.).

Report to Sen. William Proxkire, Chairman, Senate Committee on
Appropriations: HUD-Independent Agencies Subcommittee; by Elmer
B. Staats, Comptroller General.

Issue Area: Health Programs ( 100); Health Programs: Health
    Facilities (1203).
Contac*: Human Resources and Development Div.
Budget Function: Veterans Benefits and Services: Hospital and
    Medical Care for Veterans (703).
Organization Concerned: Veterans Administration.
Congressional Relevance: House Committee on Veterans' Affairs;
    Senate Committee on Veterans' Affairs; Senate Committee on
    Ippropriations: HUD-Independent Agencies Subcommittee.

          The process used by the Vet'Zrans Administration to
determine the bed size of new and replacement health care
facilities was evaluated. Three of eight hospitals currently
au+'.orized for construction were analyzed.
FinaLngs/Conclusions: GAO's results from its computer-based
mcle] nearly agrees with the VA's proposed number of beds, but
shoeau thet the mix of beds was wrong. The VA was planning too
many acute ,;are beds and too few nursing home care beds. Given
the significant cost differentials, building and operating costs
can be reduced if GAO analyses were used to deterriae the mix of
hospital beds.   Recommendations: The VA should revise the nix of
beds for the proposed Bay Pines (Florida), L.ttle Rock
(Arkansas), and Richmond (Virginia) hospitals in light of the
computer analyses, and discard its  present hospital sizing
criteria and use a method similar to the one described for all
future hospital construction. The Congress should explore to
what extent future VA hospitals should treat veterans with
nonservice-connecte$ illnesses. (DJM)
                                                                   3 f. &'!
                  REMSRICTEDP ,-Mot   to     e ct'-'e          ce away__,
                                                               + Og
                                                                              i   nri
                                                                                   oval
      Ad)                C TE tee ,
                 tGou-rtjrag               -e-    .*

                           ,%PO   tOLLER         ?KfRAL O11;+. UNnID srAT=
                                       WASHfLN.         MrC.    o or



so,             a1~MAY                                                                  2 0 1977
rJ          B-133044


            The Honorable. William Proxmire
            Chairman, SuDcommit.ee on
              HUD-Independent Agencies
            Committee on Appropriations
            United States Senate
            Dear Mr. Chairman:
                 Your letter dated January 4, 1977, requested that we
            make a comprehensive evaluation of the process the Veterans
            Administration uses to determil"e the bed size of new and
            replacement health care facilities.

                 The enclosure to this letter describes in detail the
            results of our analyses of three of the eight hospitals
            currently authorized for construction--the Bay Pines, Little
            Rock, and Richmond hospitals. Subsequent reports will dis-
            cuss the results of our analyses of the other five hospitals.

                 Using our computer-based model which we developed to
            determine the acute care bed needs in hospitals, we estimate
            that the total number- of beds required for the three hospitals
            and the Veterans Admi; istration's proposed number of beds are
            nearly equal. Howeve., our analysis showed that the mix of
            beds Dropused by the Veterans Administration is iuproper--
            the Veterans P.lministration is planninr too many acute care
            beds and too few nursing home care beds. Based upon the
            recognized significant differences in the cost of con.struc-
            ting acute care beds instead of nursing home care beds, we
            believe that the construction cost for the three hospitals
            could be reduced if the mix of hospital beds were determined
            on the basis of our analyses. Further, operating costs
            could be reduced significantly over the life of the facilities.
                  Accordingly, we are recommending that the Administrator
             of Veterans Affairs
                  -- revise the mix of beds for the proposed Bay Pines,
                     Little Rock, and Richmcnd hospitals providing for
                     the appropriate mix computed by using our computer-
                     based model as described in this report, and



                                                                                          HED-77-104
      -- witbdraw the Veterans Administra.ion's
         sizing criteria now used and implement hospital
         methodology similar to the one describeda planning
                                                   in this
         report for all future hospital construction.
     In view of the estimated surplus of beds
hospitals, the fact that VA does not consider   in community,
of other Federal hospital beds in determining the availability
Hospitals, and the large proportion of          the size of its
                                        VA hospital beds devoted
to the treatment of veterans with nonservice-connected
nesses, we recommend that the Congress                   ill-
VA hospitals authorized in the future   explore  to what extent
available to provide for the treatment should have the capacity
nonservice-connected Illnesses.         of veterars with

      The timely resolution of this policy question
a significant impact on the eight hospitals         could have
ized.                                        currently author-

     As requested by your office,
from the Veterans Administration. comments
                                    Also
                                             were not obtained
sending copies of this report today to    as  requested, we are
                                         the Chairman
House and Senate Committees on Appropriations,          of the
on Government Operations, Senate Committee        House  Committee
Affairs, House and Senate Committees         on Governmental
other Members of the Congress who haveon Veterans' Affairs;
                                        expressed an interest
in the report, to the Administrator of
                                        Veterans Affairs,
and to the Director, Office of Managetment
                                            and Budget.
                                  Sincerely yours,



                                  Comptroller General
                                  of the United States
Enclosure




                         -   2-
ENCLOSURE I                                               ENCLOSURE I



                           REVIEW OF THE SIZING
                   OF VETERANS ADMINISTRATION HOSPITALS


INTRODt'"TION
      -       -,



     In a lett¢er'"d. ed January 4, 1977, the Chairman, Subcom-
mittee on HUD-Indepenaont Agencies, Senate Committee on
Apprcpriations, requested the General Accounting Office (GAO)
to make a comprehensive evaluation of the process the Veterans
Administration (VA) uses to determine the bed size of new and
replacement health care facilities.
     According to the Chairman's letter, he was concerned
 ibout construction costs associated with.VA health care
£.1:ilities.  He referred to a May 1976 announcement to build
eight new VA hospitals at a cost in excess of $800 million
and was concerned that VA build hospitals of the appropriate
size and with the proper mix of beds (acute care and nursing
home care).

     This report discusses proposed facilities at Bay Pines,
Florida; Little Rock, Arkansas; and Richmond, Virginia.
Subsequent reports will be issued on the other proposed
facilities.
Background

     VA is responsible for providing medical cart for the
Nation's 29.4 million veterans. The Department of Medicine
and Surgery (DM&S) administers VA's health care delivery
system, providing this care primarily through a system of 171
hospitals, 213 outpatient clinics, 86 nursing home care
facilities, and b1 domiciliaries. In fiscal year 1378, DM&S
estimates that it will employ over 190,000 people and its
budget is estimated to be about $4.9 billion.

     On May 11, 1976, the President announced his decision to
construct eight hospitals--seven replacements and one new.
The proposed hospitals, bed sizes, and estimated construction
costs are listed below in VA's order of construction priority:
ENCLOSURE I                                         ENCLOSURE I




                                   Bed size       Estimatea
Hospital                           (note'a)   construction costs

                                                   (millions)
Richmond, Vir-inia                    700          $111.4

Bay Pines, Florida                    710            97.2

Martinsburg, West Virginia            370            62.1

Portland, Oregon                      770           14e.7

Seattle, Washington           b/      515            90.0

Lt]tle Rock, Arkansas         c/ 1,585              125.8

Baltimore, Maryland                   400            80.8

Camden, New Jersey    (new)           360            70.2




     a/Unless otherwise noted, does not include non-hospital
       beds such as those in nursing home care facilities or
       domiciliaries.

     b/Includes 455 hospital and 60 nursing home care beds.
       Cost breakdown not available.

     c/Includes all beds proposed for both the Central Little
       Rock and Nocth Little Rock divisions.
ENCLOSURE I                                     ENCLOSURE I



     Subsequent to the President's announcement, VA's fiscal
year 1977 budget request to the Congress was amended to pro-
vide design funds for all eight hospitals and construction
funds for the proposed facilities at Bay Pines and Richmond.
Congress appropriated $221,290,000 in fiscal year 1977 for
final design and construction of these two facilities.  An,
additional $47,026,000 was appropriated for the other six
hositals for design and site preparation purposes.

Scope of'review

     Our review included discussions with officials of the VA
Central Office in Washington, D.C., and with officials at the
Bay Pines, Little Rock, and Richmond VA hospitals.

     We reviewed pertinent records, reports, and other docu-
ments available at the VA Central Office and the three hospi-
tals we visited.

     Our source of statistical data on the use of VA hospitals
was magnetic tapes maintained at VA's Data Processing Center,
Austin, Texas. The tapes--the patient treatment file--con-
tained information on all patients discharged from the three
hospitals reviewed for fiscal year 1976. The tapes were
validated by selecting a random sample of patient data and
checking it against riedical records on file at the Hospitals.

     Regarding community hospitals, the basic data for use
in this study were supplied by the Commission on Professional
and Hospital Activities (CPHA), Ann Arbor, Michigan. In
this data, the identities of individual hospitals were not
revealed in any way. Any analyses, interpretations, or
conclusions based on this data are ours, and CPHA disclaims
responsibility for any such analyses, interpretations, or
conclusions.

PRESENT VA PLANNING METHOD-RESULTS IN-OVERESTIMATIFG
ACUTE CARE HOSPITAL BED REOUIPEMENTS
     VA's current method of planning new VA hospitals results
in the wrong mix of acute and nursing home care bed require-
ments. VA generally estimates the size of a new hospital by
using historical length of stay data and projecting hospital
admissions and patient average length of stay in the hospital.
However, lengths of stay in VA acute care facilities are often
longer than necessary because many existing VA hospitals lack
an appropriate mix of acute and nursing home care bed
ENCLOSURE I                                      ENCLOSURE I



facilities. Therefore, historical utilization data may not
be appropt.ate for determining the number of acute care buds
n ede.
     When the appropriate mix of acute and nonacute care
beds is provided, the acute care hospital average rength
of stay can often be considerably reduced, thus lowering
the acute care bed requirements for the facility.
VA methodolosg   for
determining -ospital'size

     VA's methodology in sizing the proposed hospitals at
Bay Pines, Little Rock. and Richmond consisted of the
following steps.
     -- Starting with 5 year historical data on hospital
        admissions per thousand veterans in e&:h
        hospital's catchment area {geographic area
        expected to be served by a hospital), an6s
        estimates of veteran population in 1985, VA
        projected admissions ·.
                              o 1985, broken down into
        various hospital bed sections, such as medical,
        surgical, and psychiatric.
     -- Separate 1985 projections of average length of
        stay 'Richmond and Bay Pines) or average monthly
        turnover rate (Little Rock) were then made using
        5 -ear historical data for ,ach bed section.
     -- By combining projected admissions with projected
        length of stay or turnover, VA estimated 1985
        average bed occupancy for each bed section.
     -- Assuming occupancy rates of 85 percent for medical,
        85 percent for surgical, and 90 percent for
        psychiatric, .the average occupancy projections
        were converted to bed requirements for each bed
        section.
     Projecting length of stay is a critical element in
determining hospital bed requirements. Even small changes
in the assumed average length of stay can have a large
impact on the estimated number of beds required.




                             4
ENCLOSURE I                                      ENCLOSURE I

Lengths of-sta.
often excessive
in VA-hospitals
     VA's projections of acute care length of stay will be
inflated if they are based on historical data taken from
hospitals which do-not provide appr priate alternatives to
acute care needs. A previous GAO study l/ showed that VA
hospital acute care length of stay could be reduced
through

     -- discharging patfents to nursing home care
        facilities or oitpatient treatment when they no
        longer required acute care,

     -- greater use of outpatient treatment for
        diagnostic tests prior to hospital admission, and

     -- better coordination of hospital admissions with
        the availabilit, of surgical facilities.

     The length of stagy of VA patients is generally much longer
than that of similar pat: ants in private community hospitals.
According to a Congressional Budget Office (CBO) study 2/,
when length of stay in VA hospitals is adjusted for age,
diagnosis, and surgical procedure it is usually twice that
of a community hospital. (BO's analysis of the factors
causing long length of stay showed that:

          "While some of these factors may be
      uncontrollable because of special character-
      istics of the VA patient population or be-
      cause certain facilities are not convenient
      to patients homes, it appears that u,.ther
      improvement in management and resource allo-
      cation in the VA system could significantly
      reduce the average length of stay, even if
      not by as much as experienced in the last
      decade. The VA itself, in a study showing
      the effect of selected factors on length of
      stay, concluded that certain stays in VA
      hospitals are excessive. The VA study indi-
      cated that the length of stay could have been


L/"Better Use of Outpatient Services and Nursing Care Bed
   Facilities Could Improve Healtn Care Delivery To Veterans,"
   (B-167656, April 11, 1973).

2/"Projected Acute-Care Bed Needs of Veterans Administration
   Hospitals," Congressional Budget Office, April 1977.

                             5
ENCLOSURE I                                     ENCLOSURE I


     reduced 24 percent by increasing the use of
     outpatient treatment before hospitalization
     and after surgery. Scheduling admissions
     according to the availability of the operating
     room would have reduced length of stay by
     another 7 percent."

CBO also noted that an earlier study made by McKinsey and
Company for VA reached similar conclusions.

     We believe that with proper provisions for outpatient
and nursing home care facilities in new VA hospital construc-
tion programs, VA can reduce its average acute care length of
stay to that which currently prevails in private coz-runity
hospitals among similar patient groups. By substituting
outpatient and nursing home care for acute care where
appropriate, considerable economies can be obtained in both
construction and operating costs.

GAO'S METHOD OF ESTIMATING HOSPITAL-SIZE

     During an earlier review of the Department of Defense
planning for the San Diego Naval 3ospital 1/, we developed
a conputer-based model for determining the acute care bed
needs in military hospitals. In July 1976, the Congress
adopted a conference report on the military construction
appropriations bill for fiscal year 1977 stating that acute
care hospital bed requirements for active duty members and
their dependents throughout the Department of Defense sys-
tem should be calculated using our model. The Department
of Defense is currently using the model to plan the size
Df its hospital facilities. We believe that this model,
which we modified to consider the unique characteristics
of VA, should be adopted by VA in determining its acute care
bed needs.

     In addition to the use of our model, a number of other
matters should be considered in determining VA's acute care
bed needs. These include the number of beds that should be
 provided to permit continued treatment of veterans for non-
 service-connected illness, the use of community or other
 Federal hospitals, and the issue of national health care
 insurance. These matters need to be addressed since they
 individually and/or collectively could have a significant
 impact on future VA hospital bed needs.

 1/"Policy Changes and More Realistic Planning Can Reduce Size
   of New San Diego Naval Hospital," (MWD-76-117, April 7,
   1976).


                              6
ENCLOSURE I                                        ENCLOSURE I



Descri~tion of
GAO model

     Our model provides an estimate of acute care bed needs
by accumulating the actual patient workload by diagnosis and
age group, then adjusting it to reflect data on average
length of stay in non-Federal community hospitals. The
community hospital data is available from the CPHA's
Professional Activity Study (PAS).
     PAS publishes average length of stay statistics by
diagnostic category and age for patients discharged from
PAS-member hospitals. Statistics are published for regions
of the United States and the nation as a whole. Member
hospitals use PAS data as a measure of their own efficiency
in treating patients. In analyzing the bed needs for the
three VA hospitals, we used the PAS data for the nation as .
a whole. PAS national statistics include data compiled
from 13.2 million inpatients discharged during 1974 from
1801 member hospitals having a total of 374,612 oeds--40.2
percent of all U.S. short-term non-Federal hospitals.
     The PAS system has 349 primary diagnoses categorized.
The average length of stay can be determined by knowing
(1) the primary diagnosis, (2) if the patient had a single
or multiple diagnosis, (3) if the patient underwent an
operation, and (4) the patient's age. The value of the
data is enhanced by "variance" figures which allow the
user to statistically determine their degree of reliability.
In general terms, the lower the variance, the smaller the
deviation of individual length of stay from the average.
PAS also provides length of stay figures for various
percentiles of the population. For example, the length of
stay figure at the 95 percentile is exceeded by only 5 percent
of the population. The chart on the following page is an
example of data for one diagnostic group. It illustrates,
for example, that for patients with a single diagnosis of
acute appendicitis without peritonitis (not operated on),
in the age brackets from 50-64:
     -- the total number of patients reported on was 63,
     -- the average length of stay was 4.1 days,
     -- the variance value was 9, and
     -- 5 percent of the total patients had a length of stay
        of 10 days or longer.


                              7
ENCLOSURE I                                                                                                                ENCLOSURE I




   178: Acute appendicitis without pitofti (540.0)
        TYPE OF        TOTAL AGi VAI                                                        .
          PAnTIENT              PATIENTS     STAY         AC
                                                          t                s                     S            7            m                 "
             I(1)           ~      (2)        (3)           (4)           ()           ()       I (                             (            ((11))

   1. SItILE X              j
            A. 9        j            636 1     27                 7        1                1         2           3                  7       10
           20.34                     343       2.8               6         I                1         2           4         £        7         8
           35-a49                    100 !     3.8               6         1                1         3           5         7        9       12
           50.64                      63       4.1               9         1                1         3           5         7       10       1
           65+                        31       5.3              10        <l                2         5           7        10       11       13

             0.19 YRS   I           37131      4.4               5
                                                                 6
                                                                               2
                                                                               2
                                                                                            3
                                                                                            3
                                                                                                      A
                                                                                                      4
                                                                                                                  5         7
                                                                                                                            7
                                                                                                                                     a
                                                                                                                                     s
                                                                                                                                              12
                                                                                                                                              12
            20-34           I       18910      4.                                                     5           6         8       1         15
            3549                     5298      5.5                             3            3
            50-64                    2498      6.4              12             3            3         6           8        10       13        19
            65+                       713         8.2           19             3            4         7           10       14       16        24
   2 MULTIPLE               !
            A, A YRS-i                225         3.3 |                        1            1         2            4        6        9        17
            20-34                     181         3.9            6                          1         3            5        8        9         2
            3549                       64         5.3           33             2            2         4            6        9       11        40
            50-64                      51         6.8           19             2            2         6            9       14       15        21
            65+                        45         8.3           31             2            3                     10       16       22        24
     S. Uwt                                                                                                                         14        22
             0-19 YRS                5746      6.4              18             3            3         5            7       11
            20-34                    4132      6                19             3            3          6           8       11       14        2$
            35-49                    1619      8.4              28             3            4          7          10       15       19        26
            50-64                    1182     10.7              57             4            4          9          13       19       24        41
            65+                       712     13.2                             4            5         11          16       23       29

    SUITOTALS                                 I
    1. SINGLE DX
       A. *IWfW"d
          I                          1173 1       3.0                          1                       2           4                 7        1
       B.   ,antd ,                 64550         4.7                 6        2            3          4           5           7              13
    2. MULTIPLE DX                                                16           1            1          3           5           9    11         21
       A.AULT.p.L d                   566         4.4
       B. -p --d                j   13391          7.5            29           3            3          6           9        13      17         2
    1. SNCGLE DX                    65723          4.7             6           2            3          4              5      t7                13
     . MULTiPLEr DX                 13957 !    4                  29           3            3          6              9     13       17        28
    IA.NOT OPRATED                    1739   3.4                  10           I            1          3              4         7     9 .'
    LOPERATED                        7941    5.2                  11           2            3          4              6         a    11    18
    TOTAL 0-9Y RS                   43738w   4.6                      7            2        3             4            5        7        9     14
            20-34                   23566    5.0                   9               2        3             4         6
                                                                                                                    7
                                                                                                                             8
                                                                                                                            10
                                                                                                                                     10 1 20
                                                                                                                                     13
            35249                     70611  6.1                  14               3        3             5
            50-64                     3794   7.7                  30               3        3             6         9       14       10   28
            65I                       1501  10.5                  52               4        4             9        13       1        24   37
    GRANo TOTAL                     7968           5.1       _I                    2        3             4            6        a    t0 ls




Source:       "Length of Stay ir: PAS Hospitals," Commission on
             Professional and Hospital Activities, 1974.

                                                            8
ENCLOSURE I                                  ENCLOSURE I

     Our model determines hospital size by accumulating
adjusted length of stay for each VA hospital patient. This
process is accomplished by a computer program designed to:
     -- Accumulate the actual length cf stay of each
        patient discharged from each VA hospital during
        fiscal year 1976.
     -- Extract from the data each patient's primary
        diagnosis and age, as well as whether the patient
        had a single or multiple diagnosis, and whether the
        patient underwent surgery.
     -- Match each patient's'characteristics with those of
        corresponding patients discharged from community
        hospitals during 1974 based on PAS information.
     -- Accumulate the corresponding PAS average length of
        stay for patients discharged from each VA hospital
        during fiscal year 1976.
     --Use the accumulated patient days to calculate
       acute care bed requirements.
Since PAS length of stay statistics do not include patients
who died, we used unadjusted actual length of stay for these
patients.
     Special consideration was also given to patients who had
stayed in the hospital for 100 days or longer. PAS average
length of stay figures do not include these individuals, but
PAS percentile distribution data does. We determined the
community nhospital length of stay for each patient who had
stayed 100 days or longer by using PAS data corresponding
to the 95th rercentile.
     A flowchart illustrating the sequence of operations
which leads to the hospital size determination is included
as an appendix to this enclosure.
     Using the above data, we calculated the total number of
bed days for each patient discharged from each of the
hospitals in fiscal year 1976. We then determined the number
of acute care beds needed by calculating the average number
of beds occupied on any given day and then adding a factor
to allow for an 85 percent occupancy rate in medicine and
surgery and a 90 percent occupancy rate in the psychiatric
bed sections. These occupancy rates are consistent with
those used by VA.



                            9
ENCLOSURE I                                     ENCLOSURE I



     The computer was also instructed to accumulate bed
requirements by age category. Then, by using VA estimates
of expected changes in the veteran population size and age
mix between fiscal years 1976 and 1985, we projected the
1985 bed requirements for each age category.

     The GAO model differs from VA's sizing methodology in
one fundamental way.  Instead of assuming that current and
historical length of stay represent the true acute care-bed
requirements, the GAO model analyzes each patient admission 1/
separately and compares the length of stay to that of similar
patients in community hospitals. In most cases, the GAO model
substitutes community hospital length of stay for the patient's
actual length of stay. In our opinion, community hospital
length of stay for patients of a given age and diagnosis
better zeflects true acute care bed needs, assuming VA is pro-
vided with appropriate outpatient and nursing home care facil-
ities.

     The GAO model is designed to estimate acute care bed
requirements for.medical/surgical hospitals. Therefore, the
model was not used to estimate psychiatric or nonacute care
requirements, such as nursing home care facilities, at any of
the hospitals studied.

Other matters-which
may affect VA bed needs

     Our model determines VA acute care bed needs on the
assumption that the basis under which VA now provides care
will not change. Other matters--such as the bed capacity that
should be provided to permit continued treatment of veterans
for nonservice-conrsected illnesses, the use of community or
other Federal hospital beds, and the potential impact of
national health care insurance--need to be addressed since
they could have a significant impact on future bed needs.
These matters, which will be more fully addressed in subse-
quent reports, are briefly discussed below to show their im-
portance on future planning of VA hospital needs.




1/ Patient discharges are actually used instead of admissions
   since VA patient data is based on discharges. Over time,
   admissions equal discharges (including transfers and deaths.)



                            10
ENCLOSURE I                                    ENCLOSURE I



     Construction-of new facilities
     for treatmentof -nonservice-
     connected illnesses

     Section 612 of title 38 of the U.S. Code provides
that veterans who have medical disabilities incurred or
aggravated in the line of military duty are entitled to all
reasonable medical services necessary to treat such disabil-
ities. Veterans are also entitled to medical care for
nonservice-connected conditions without regard to their
ability to pay if they (1) are released or discharged from
military service for disabilities incurred or aggravated in
the line of duty, (2) have compensable service-connected
disabilities, or (3) are 65 years of age or older. Any
veteran may be provided similar treatment for a nonservice-
connected disability if he certifies he is unable to pay.

     A large proportion of the total inpatient workload is
comprised of veterans who are being treated at VA hospitals
for nonservice-connected illnesses. Nationwide, about 89 per-
cent of the patients treated during fiscal year 1976 were
treated for nonservice-connected conditions. At the three
hospitals we reviewed, the percentage of acute care beds re-
quired to treat nonservice-connected illnesses ranged from
86 to 91 percent.

     A matter which we believe needs to be addressed is the
question of whether new and replacement facilities should be
sized to accommodate the entire current workload of nonservice-
connected illnesses, or whether some limitation-should be im-
posed. Since VA follows a policy of treating nonservice-
connected illnesses on a space available basis, it is unclear
as to whether new VA hospitals should be sized to meet all,
some, or none of this demand. The bed-requirements under
several possible assumptions relating to this matter are esti-
mated for each hospital on pages 16, 20, and 23.
     VA does not consider availability-
     ofexistns   commonity-or-other
     Federal   hospital bedss

     The U.S. today has over 931,000 non-Federal hospital
beds, of which 20 percent are estimated to be surplus.
Excess bed capecity has become a national concern in recent
years. Since 1S0 the total of non-Federal hospital beds for
short-term and other care in general hospitals has increased
from 640,000 to 931,000--more than 45 percent. When related
to the national population, the ratio of beds has increased

                                11
ENCLOSURE I                                     ENCLOSURE I



from 3.6 beds per 1,000 copulation to 4.4 beds per 1,000.
Excess bed capacity is one reason that hospital costs since
1950 have risen four times as fast as the consumer price
index.

     The National Eealth Planning and Resource Development
Act of 1974,-Public Law 93-641, provides a new appro.ach to'
resolving the problems of access, cost, and quality of health
care. The law created a network of more than 200 health sys-
tem agencies (HSA) which are to, among other things, plan for
the health resources needed in their geographic areas of re-
sponsibility.

     The HSAs, in projecting the bed supply and demand, count
the total population in their geographical area but do not
include the Federal hospital beds in their area. The Federal
agencies such as VA, in making their projections, count the
population they serve but do not count community or other
Federal agencies' hospital beds available. In effect, both
the HSAs and the Federal agencies are counting the same popu-
lation twice and are building hospital beds based on projec-
tions of these populations. To illustrate the problem of
excess bed capacity, HSA projections indicate that consider-
able excess community hospital bed capacity will exist in the
Bay Pines, Little Rock, and Richmond catchment areas in 1980.

      The Florida Gulf HSA, whose catchment area falls within
that of the Bay Pines VA hospital, provided data indicating
that in 1980 the area will have 1,100 excess community hos-
pital beds. Similarly, based on HSA projections to 1980,
there are expected to be at least 1,363 excess beds in the
Little Rock catchment area and 663 excess beds in the Rich-
mond catchment area. While the Government bears the cost
 (construction, equipment, staffing, etc.) of new VA hospital
beds, it is also sharing in the increased costs resulting
from excess community hospital beds. Many were constructed
with Federal support and operating costs are paid for, in
part, through Medicare, Medicaid, and Federal health benefit
programs.

     Potential impact of
     national-health care
     insurance legislation

     The passage of some form of national health care legis-
lation could greatly reduce the demand for VA hospital beds.
All of the national health care bills now before the Congress
would significantly reduce the expenses patients must bear
for medical treatment. For calendar years 1973-1974, only
                             12
 ENCLOSURE I
                                                   ENCLOSURE I


  12.3 percent of all discharges for veterans
  stay hospitals (30 days or less) were         from short-term
                                         from  VA  hospitals.
  About 88 percent were discharged from
                                         community hospitals.
  Many veterans now using VA hospitals
                                       would
  to private physicians and local community undoubtedly turn
  care if such care were available at low    hospitals for their
                                           cost to them.
       Many veterans are already covered by
                                             some form of health
  insurance, either public or private.
                                        Under the assumption
  that national health care insurance would
 coverage than that of the average veteran'sprovide no better
                                                insurance today,
 the Congressional Budget Office (CBO)
                                        estimated that 20 per-
 cent would be the lower limit for the
                                        expected drop in VA
 hospital demand 1/. In developing this
 that the behavior of current insured      estimate, CBO assumed
                                       veterans would not
 change under national health care insurance,
 of national health care insurance on             Only the impact
                                       those veterans who have
 no coverage at present and currently
                                       use the VA system was
 measured.

     In view of the potentially significant
national health care insurance program        impact of a
                                        on the demand for
care in VA hospitals, Congress, when
                                      considering the author-
ization of new and replacement hospitals,
the role VA will play in any national      will need to explore
                                       health  care insurance
program.

SIZE ANALYSIS FOR BAY'PINES, LITTLE ROCK,
AND RICHMOND VA HOSPITALS

     As part of our review, we evaluated the
                                               hospital _ize
proposed by VA for the Bay Pines, Little
hospitals. Our estimate of the total       Rock, and Richmond
                                       number of beds required
for the three VA hospitals and VA's
                                    proposed number of beds
are nearly equal. However, our analysis
of beds proposed by VA is improper--VA     showed that the mix
                                        is
acute ca-e beds and too few nursing home planning too many
                                           care beds. Based
upon the recognized significant differences
constructing acute care                       in the cost of
                        beds instead of nursing home dare
beds, we believe that the construction
hospitals could
                                       costs for the three
                be reduced if the mix of
determined on the basis of our analyses. hospital beds were
expenses could be reduced significantly    Further, operating
                                         over the life of the
facilities. The detailed results of
                                     our size analysis for
each hospital are presented on the following
                                              pages.

1/ "Projected Acute Care Bed Needs of Veterans
   Hospitals,"                                 Administration
               Congressional Budget Office, April 1577.



                            13
 ENCLOSURE I
                                                     ENCLOSURE I


 Bay Pines VA-hospital

      The existingVA hospital at
of 4 buildings constructed during Bay
                                   the
                                       Pines, Florida, consists
tain a total of 693 operating beds.     early 1 9 30's, which con-
 nally built as a domiciliary and     The    hospital was origi-
                                    is located on a 3 5 0 -acre tract
 along Florida's west coast, on the
 northwest of St. Petersburg, Florida.Boca Ciega Bay, 9 miles
grounds are a 322-bed domiciliary         Also located on the
                                     built in 1933 and a 120-bed
nursing home care facility opened
                                     in 1972.
hospital served a population of 322,288          In 1976, the VA
                                           veterans which,
according to VA estimates, will
                                  increase to 364,719 by 1985.
      The VA has proposed construction
hospital (370 medical beds and 150       of a new 520-bed
bed nursing home care facility, and   surgical  beds), a new 120-
                                       a 200-bed domiciliary on
the present site. In addition, VA
ing buildings which will provide      plans to renovate 3 exiit-
                                   space for 190 psychiatric
beds, ambulatory care, and administrative
ent 120-bed nursing home care building        offices. The pres-
cost estimate for the project including   will  be retained. VA's
facility is $110 million.                  the  nursing home care

     As shown by the following schedule,
overall, 70 more acute beds are needed   we estimate that,
Bay Pines.                             for the hospital at




                              14
ENCLOSURE I                                         ENCLOSURE I

            GAO-size estimate for Bay-Pines VA-hospital

                             Current
                            operating           Bed needs
                             -beds      vA   roposal GAO estimate
Acute   'on-psychia'tric
  Medical                      229             170                 (a)
  Surgical                     151             150                 (a)
  Rehabilitation                31              30                 (a)
    Subtotal                  -411            -350                -420
  Intermediate medicine        136             170                 (b)
    Subtotal                  -547            -520                -420
  Psychiatric                 -126             190/-              190
    Total hospital             573            -710                610
Lower levels of care
  Nursing home care            120             240         c/     240
  Additional nursing
    home care                   -               -          b/     170
  Domiciliary                 -322             200         c/- 200
    Total                   1,115            1i150              1 220

a,'/ The GAO model does not estimate medical, surgical, and
     rehabilitation separately but indicates a total acute
     care non-psychiatric bed requirement of 420 beds.
b/ We have placed VA's proposed 170 "intermediate medicine"
   beds under the label "additional nursing home care" to
   indicate that these beds do not require the staffing and
   ancillary services customarily required for the operation
   of acute care beds.
c/ Since the GAO model estimates only acute care bed require-
   ments, VA's proposal for psychiatric bed, nursing home
   care and domiciliary is used.




                               15
ENCLOSURE I                                     ENCLOSURE I


     Although 170 of the proposeed hospital beds have been
labeled "intermediate medicine," VA plans to construct these
within the same hospital structure, and with the same support
facilities as the acute care medical beds. Intermediate care
patients, however, are those with long term or chronic con-
ditions who require nursing home type of care. The average
length of stay of intermediate care patients at the Bay Pines
VA hospital during fiscal year 1975 was 184.3 days. Such
patients are generally not found in acute caLe community hos-
pitals, but a.e discharged either to less expensive nursing
home care facilities--which are far less costly to construct
and operate than acute care beds--or to their homes.
     Our estimates indicate that VA's proposed mix of acute
and nonacute care beds is inappropriate. Bay Pines requires
70 more acute care beds than were proposed. However, contrary
to VA's plans, the proposed 170 intermediate care beds should
be constructed similar to nursing home care rather than acute
care. This suggests that appropriate construction would con-
sist of 610 acute medical and surgical beds and 170 nursing
home care beds in addition to the 240 nursing home care beds
proposed by VA.
     VA estimates that the construction cost of an acute care
bed in the Bay Pines area is about $136,930 per bed and about
$39,833 per nursing home care bed. We recognize that substi-
tuting a nursing home care bed for an acute care bed will not
result in a savings of $97,097 per bed because some portion
of construction costs are attributable to ancillary services.
Nevertheless, we believe that construction costs at Bay Pines
could be reduced if the mix of hospital beds were determined
on the basis of our analysis. Further, operating expenses,
such as staffing and support services, could be reduced signif-
icantly over the life of che facilities.
     Table-of policy'assumptions

     The table on the following page presents certain policy
assumptions with regard to the treatment of eligible veterans
and their impact on the required size of the Bay Pines VA
hospital. If, for example, veterans were treated only for
service-connected illnesses, only 85 beds would be required in
the Bay Pines VA hospital. If patients with service-connected
illnesses or disabilities were treated for all their illnesses,
220 beds would be needed.




                             16
ENCLOSURE I                                        ENCLOSURE I



                    Bay Pines VA-hospital
              acute care-bed-r eruirements under
                  various-policy assumptions

                                              Bed requirements

1.   Projected requirement
     with no restriction on
     beneficiary use                                a/ 610

2.   Projected requirement
     if only service-
     connected patients are
     treated (for either
     service-connected or
     nonservice-connected
     illness)                                           22C

3.   Pojected requirement
     if only service-
     connected patients are
     treated for service-
     connected illness                                   85

a/   Consists of 420 acute medical/surgical beds and 190
     psychiatric beds as computed by GAO.

Little Rock VA-hospital

     The Little Rock VA hospital is composed of two separate
divisions. TIe Central Little Rock VA hospital, a 10-story
structure opened in 1950, provides the bulk of the acute care
beds and currently houses 460 acute care medical, surgical,
and neurological beds.

     The division which provides primarily psychiatric service,
nursing home care, and care for long term, chronic patients,
is referred 'to as the North Little Rock VA hospital.     This
facility consists of 12 buildings constructed mainly during
the 1920's and 1930's, housing 1,007 hospital beds and 200
nursing home care beds. Although the Central and North Little
Rock VA hospitals are located about 7 miles apart they are
under single management. The hospitals serve a catchment
area consisting of 45 Arkansas counties with a-1976 veteran
population of 161,024. VA estimates the veteran population
will decline to 146,787 by 1985.

     VA is planning to construct a new 535-bed hospital
adjacent to the University of Arkansas Medical Center to



                              17
ENCLOSURE I                                     ENCLOSURE I



replace the existing Central Little Rock VA hospital which
is approximately 6 miles away. VA also plans to modernize
the patient care buildings at North Little Rock, and con-
struct a new multi-use recreational building. After moder..-
ization, the North Little Rock facilities will contain 850
hospital beds and 200 nursing home care beds, a decrease of
157 hospital beds from its current size.. VA'S cost estimates
for the projects are as follows:
                                                 Amount

                                                 (million)

Replacement of Central Little Rock VA
  hospital                                       $ 75.7.

Modernization of North Little Rock VA
  hospital                                          50.1

    Total                                        $125.8

     We did not use the GAO model to evaluate the proposed
size of the North Little Rock VA hospital, which involves
the refurbishing of 1,050 beds, since the model estimates
only acute care non-psychiatric bed requirements, and no
beds of this type are planned for North Little Rock l/.
The proposed replacement for the Central Little Rock VA
hospital, however, has beeln evaluated using the GAO model,
and the results are shown on the following page.




1/ While 40 neurological beds are now planned for North Little
   Rocc, these represent a distribution of beds between Central
   and North Little Rock.

                             18
ENCLOSURE I                                          ENCLOSURE I


            GAO-size estimate for-North ^nd Central
                    Little Rock VA hospitals
                             Current
                            operating           Bed needs-
                             -beds      VA proposal GAOgestimate

Central Little Rock

 Medical                       191             245            (a)
 Surgical                      247             230            (a)
 Neurological                -- 22              30           -(a)

   Subtotal                   '460           -505            -445

  Psychiatric                -30                       b/      30

   Total acute care            460      '      535           -475

North Little Rock

 All beds                    1,207          1 050      b/ 1,050

   Total                     1,667          1;585           1,525


a/ The GAO model does not estimate medical, surgical, and
   neurological beds separately.

b/ Since the GAO model estimates only non-psychiatric acute
   care requirements, 'VA's proposed bed size is used.

     Our estimate indicates a need for 60 fewer acute care
medical, surgical, and neurological beds than has been pro-
posed by VA for the new facility.. Our estimate represents a
decrease of 15 beds from the current number of operating beds
in these bed sections. This decline of 3.3 percent is con-
sistent with the expected 8.8 percent decrease in veteran
population in the catchment area between 1976 and 1985, and
the increase in the proportion of elderly veterans making up
the population.

     VA estimates that acute care beds in the Little Rock
area cost about $138,043 per bed. We recognize that elimin-
ating acute care beds will not result in a savings of $138,043
per bed because some portion of construction costs are attri-
butable to ancillary services. Nevertheless, construction of
475 acute care beds, rather than the 535 acute care beds
proposed by VA, would result in reduced construction costs.


                                   19
ENCLOSURE I                                       ENCLOSURE I


Further, operating expenses, such as staffing and support
services, could be reduced significantly over the life
of the facilities.

      Table of policy assumptions

     The table below presents certain policy assumptions
(with regard to the treatment of eligible veterans), and
their impact on the required size of the Central Little
Rock VA hospital.  If, for example, veterans were treated
only for service-connected Illnesses, only 43 beds would
be required at the Central Little Rock hospital.    If
patients with service-connecvu¢f disabilities were treated
for all their illnesses, 114       would be needed.

              Central Little Rock VA'hospital
              acute caree   reuirements under
                various policy-assumntions

                                             Bed requirements
1.   Projected requirement with
     no restriction on beneficiary
     use                                          a/ 475
2.   Projected requirement if only
     service-connected patients are
     treated (for either service-
     connected or nonservice-
     connected illness)                              114
3.   Projected requirement if only
     service-connected patients are
     treated for service-connected
     illness                                          43
a/ Consists of 445 acute medical, surgical, and neurological
   beds and 30 psychiatric beds as computed by GAO.

Richmond-VA'hospital

     The existi n VA hospital at Richmondr Virginia, consists
of about 80 builcilngs connected by covered corridors. The
hospital was constructed in 1944 and contains a total of 990
operating beds. In 1976 the hospital served a population of
221,882 veterans which, according to VA estimates; will decline
to 206,105 by 1985.



                              20
ENCLOSURE I                                      ENCLOSURE I


                                                        and a
     VA has proposed construction of a 700-bed hospital
                                                  site.  VA's
120-bed nursing hom' care facility at the present
cost estimate for the project is $116 million.
                                                      differs
     As shown by the following table, our estimate
                                                 intermediate
from VA's proposal only in the designation   of
                           of care. Patients categorized      as
care beds as a lower level                  of  stay in the
"intermediate care" had an average length
Richmond VA hospital of 135 days during  fiscal year 1975. In
the private sector, such patients  are generally   transferred
                  facilities  and into less  expensive  nursing
out of acute care
home care facilities or to their  homes. VA plans to con-
struct 83 intermediate care  beds within the same hospital
                                                    the    acute
structure, and with the same support facilities as hospital
care beds. We believe, consistent with  community
                                           constructed
practice, intermediate care beds should beacute care. This
similar to nursing home care rather than
suggests that appropriate construction would consist of 620
acute care beds and 80 nursing home care  beds in addition to
the 120 nursing home care beds proposed by VA.




                               21
ENCLOSURE I                                          ENCLOSURE I



              GAO'size estimate for Richmond VA-hospital

                               Current
                              operating          Bed needs
                               -beds      VA proposal GAO estimate
Acute non-psychiatric

     Medical                     330         200                (a)
     Surgical                    194         160                (a)
     Neurological                 35          40                (a)
     Rehabilitation               14          20                (a
       Subtotal                  573         420                420
Intermediate care                 66          80                 -
       Subtotal                  639         500                420
Psychiatric                      65           80           c/   -80
Spinal cord injury              161          120           S/ 120
       Total hospital           865          700                620
Lower-levels of care

     Nursing home care           -           120           c/ 120
     Additional nursing
       home care                               -b/              -80
      Total                     865          820                820
a/    The GAO model does not estimate bed requirements
      separately but indicates a total acute care non-
      psychiatric bed requirement of 420 beds.

b/    We have placed VA's proposed 80 "intermediate care" beds
      under the label "additional nursing home care" to indicate
      that these beds do not require the staffing and ancillary
      services customarily required for the operation of acute
      care beds.

c/    Since che GAO model estimates only acute care bed require-
      ments, VA's proposal for psychiatric beds, nursing home
      care, and spinal cord injury beds is used.




                                 22
ENCLOSURE I                                    ENCLOSURE I



    VA estimates that the construction cost in the Richmond
area of an acute care bed is about $159,200 per bed and about
$38,000 per nursing home care bed. We recognize that substi-
tuting a nursing home care bed for an acute care bed will not
result in savings of $121,200 per bed because some portion of
construction costs are attributable to ancillary services.,
Nevertheless, we believe that construction costs at Richmond
could be reduced if the mix of hospital beds were determined
on the basis of our analysis. Further, operating expenses,
such as staffing and support servicei, could be reduced signif-
icantly over the life of the facilities.

      Table of policy assumptions
     The table below presents certain policy assumptions with
regard to the treatment of eligible veterans and their impact
on the required size of the Richmond VA hospital. For
example, only 74 beds would be required at Richmond VA hospital
if veterans were treated only for service-connected illnesses.
If patients with service-connected disabilities were treated
for all their illnesses 114 beds would. be needed.
                      Richmond-IA'hospital
                acute care bed reguirementsunder
                   various policy-assumptions
                                             Bed requirements

1.    Projected requirement with
      no restriction on benefic-
      iary use                                     a/ 620

2.    Projected requirement if
      only service-connected
      patients are treated (for
      either service-connected
      or nonservice-connected
      illness)                                         114

 3.   Projected requirement if
      only service-connected
      patients are treated for
      service-connected illness                         74

                                                           and
 a/ Consists of 420 acute medical, surgical, neurological,
    rehabilitation beds; 80 psychiatric, and 120 spinal cord
    injury beds as computed by GAO.


                                  23
ENCLOSURE I                                      ENCLOSURE I

CONCLUSIONS

     VA's current method of planning new hospitals results
in the wrong mix of acute and nursing home care bed require-
ments. Our estimate of the total number of beds required
for the Bay Pines, Little Rock, and Richmond VA hospitals
nearly equals the total number proposed by VA. However, our
analyses showed that the mix of beds proposed by VA is im-
proper. By substituting nursing home care beds for acute
care beds where.appropriate, VA could reduce the construc-
tion costs for the three hospitals. Further, significant
savings in operating expenses could be realized over the
lice of the facilities.

      Congressional resolution of the policy question concern-
ing the appropriateness of constructing new VA facilities
with the capacity available to provide for treatment of
veterans with nonservice-connected illnesses, could greatly
 *educe the required size of new VA hospitals.

RECOMMENDATIONS TO THE-ADMINISTRATOR-OF-VETERAN-AFFAIRS

     We recommend that the Administrator

     -- revise the mix of beds for the proposed Bay Pines,
        Little Rock, and Richmond hospitals providing for
        the appropriate mix as computed by us using the
        computer-based model described in this report, and

     -- withdraw the VA hospital sizing criteria now used
        and implement a planning methodology similar to the
        one described in this report fcr all future hospital
        construction.

RECOMMENDATION TO TEE CONGRESS
     In view of the estimated surplus of beds in community
hospitals, the fact that VA does not consider the avail-
ability of other Federal hospital beds in determining the
size of its hospitals, and the large proportion of VA hos-
pital beds devoted to the treatment of veterans with non-
service-connected illnesses, we recommend that the Congress
explore to what extent VA hospitals authorized in the future
should have the capacity available to provide for the treat-
ment of veterans with nonservice-connected illnesses.

      The timely resolution of this policy question could have
a significant impact on the eight hospitals currently author-
ized.



                             24
APPENDIX I                                                                           APPENDIX I

                                 SEQUENCE OF OPERATIONS IN
                               DETERMINATION OF HOSPITAL SIZE


                        ,'~~               ·   ~~VA HOSPITAL
       PAS
                   ~~~DA~T~~A                             PATIENT
       BANK         .                                    DISCHARGE
                                                            DATA


                                                              ACtUA~C~           I TOTAL ACTUAL
                                                              PATIENT              BED DAYS FOR

CREATE PAS LENGTH                                                                                 D
   OF STAY TABLE                                          PATIENT
  BROKEN DOWN BY                                            DIE
                 TtC
DIAGNOSIS, AGE, ETC.




   USE VA HOSPITAL
 PATIENT DIAGNOSIS,
  AGE, ETC. TO FIND                             /          PATIENT
CORRESPONDING PAS                                        TAY 100 DAY
AVERAGE L.ENGTH OF                                        OR
         STAY




                                                  USE VA HOSPITAL
               LENGTH OF STAY                   PATIENT DIAGNOSIS,               USE ACTUAL LENGTH
                                                 AGE, ETC., TO FIND                 OF STAY IN VA
                                                CORRESPONDING PAS                     HOSPITAL
                                                LENGTH OF STAY AT
                                                THE 95TH PERCENTILE
                                                               ILENGTH OF STAY           I   LENITH OF
                                                                 ,_~~,  .'.~                    .STAY




                                 ACCUMULATOR




SUdMMARY   r                                                           SUMMARY

                                                           TOTAL REQUIRED
 TOTAL ADJUSTED
 BED DAYS FOR ALL                                          NUMBER OF ACUTE
     PATIEN                                                  _CARE BEDS




                                                    25