Measuring cost differences associated with AIDS patient heterogeneity: an examination of inequity in funding.

The federal-state Medicaid program finances health care for a broad cross-section of individuals diagnosed with the HIV virus and meeting the diagnostic criteria for Acquired Immunodeficiency Syndrome (AIDS). The relatively high incidence of public assistance (55 percent of adults living with AIDS) (HCFA, 2002a) results from a combination of two factors: the socio-economic status of the at-risk population (Makadon, Seage, Thorpe & Fineberg, 1990) and the existence of Medicaid regulations that define patients with AIDS as being presumptively disabled (Johns Hopkins, 2002).

Although only 218,000 of the 33.7 million individuals funded by Medicaid (0.6 percent) live with AIDS (HCFA, 2002a, 2002b), the cost burden of treating patients with HIV is significant. In fiscal year 2001, combined federal and state Medicaid expenditures for direct services provided to persons living with AIDS were estimated to be $7.7 billion (HCFA, 2002a). Hellinger (1998) notes that the cost of treating HIV/AIDS patients may run as high as $20,000 per year. Jasper and Kertesz (1996) estimate that the average lifetime cost of treating such a patient ranges from $150,000 to $200,000. Much of this expense is attributable to the use of improved (albeit costly) medical therapies that have fostered greater patient longevity. For example, Harvey (1997) suggests that the cost of antiviral combination therapy alone can range between $8,000 and $15,000 annually for a single patient, depending on the regimen prescribed. Covered services also include payments for medical specialists, social services, and confinement in acute care hospitals or special care nursing facilities. This study examines one of these care providers--the special care nursing facility.

Specific nursing facility reimbursement policies differ from state to state. Nevertheless, the current Medicaid prospective payment system typically reimburses special care nursing facilities treating AIDS patients a fixed payment per day, regardless of the many complications and comorbidities encountered by an individual patient. Per-diem systems were initially developed to provide an incentive for more efficient provision of care by shifting the loss associated with inefficient care to the provider. Per-diem systems may, however, foster undesirable behavior. Specifically, when a facility is operating at or near full capacity, the provider has an incentive to avoid admitting patients requiring more complex (and thus more costly) care.

This study uses both cost structure and healthcare delivery data collected from an AIDS special care nursing facility to assess the appropriateness of the current reimbursement system in such an environment. The experiences of this institution are used to demonstrate that patient heterogeneity is systematically associated with cost differences and that measures are readily available to capture and report these differences. We argue, in conclusion, that use of these measures for reimbursement purposes might reduce the implicit rationing of scarce health care resources.

BACKGROUND

Since the initial enactment of Medicare and Medicaid legislation in 1965, the federal government has assumed an increasingly proactive role in the provision of health care in the United States. Medicare was established to provide health insurance to people sixty-five and over as well as to younger individuals with end-stage renal disease or certain permanent disabilities (CMS, 2002). In contrast, Medicaid is a federal-state matching entitlement program that provides medical care to certain individuals and families with low levels of income and financial resources. Within broad guidelines established by the federal government, each state: (1) establishes its own Medicaid eligibility standards; (2) determines the type, amount, duration, and scope of covered services; (3) sets the rate of payment for services; and (4) administers its own program. Medicaid reimbursement policies tend to be complex and can differ greatly from state to state, as well as over time within a specific state (HCFA, 2002c).

Initially, when the Medicaid and Medicare programs were established, public concern focused on redistributing health care so that all segments of the population received quality treatment at minimal personal cost and providers were reimbursed on a fee-for-service basis (Preston, 1992; Preston, Chua & Neu, 1997). Precipitous increases in health care spending (from 7.4 percent of the Gross National Product in 1970 to 10.7 percent in 1983) diverted public focus from the provision of services to cost containment (Fetter, 1991). In 1980, the Boren Amendment (Public Law 96-499, Section 962) mandated that the Medicaid reimbursement of nursing care facilities be changed from a retrospective cost-based reimbursement system to a prospective payment system. Shortly thereafter, the passage of the Social Security Amendment Act instituted a prospective cost reimbursement system for Medicare recipients based on predetermined rates for 474 distinct diagnosis-related groups (DRGs) (Preston et al., 1997).

The development of the extant DRG system began in 1969. A group of Yale University Hospital physicians asked local researchers, led by Robert B. Fetter, to devise a means by which the output of their hospital facility could be better measured and evaluated. This research culminated in the development of 474 DRGs that define hospital output. Fetter (1991) argues that all patients with a given diagnosis should receive similar bundles of goods and services and incur similar costs during the course of their hospital treatment. Thus, by grouping service outputs by DRG definitions, it is theoretically possible for hospital administrators to use the resulting data as a management tool to improve both the efficiency and effectiveness of hospital operations.

Others (Horn, 1983; Gonnella, Hornbrook & Louis, 1984; Stem & Epstein, 1985; Cooper, 1996) counter, however, that the DRG classification system fails to consider factors that may affect the cost of providing care. Treatment protocols only represent standardized methods of treating particular medical conditions. Actual treatment (and resulting cost) depends on the severity of the disease, the physical condition of the patient, and other relevant factors. Horn (1983) specifically demonstrated that variations in the severity level of a given DRG illness create concomitant variations in resource consumption and costs for six disease conditions.

Weiner and Stevenson (1998) note that the disadvantages inherent in the Medicare DRG system are also present in the facility-specific per-diem rate typically utilized by Medicaid to prospectively reimburse nursing facilities. Although a case-mix adjustment procedure, which provides higher reimbursement for those patients who require a higher level of care, would make care providers indifferent to the relative care needs of the individuals they admit, only half of the states employ a case-mix adjusted rate. Indeed, Holahan and Liska (1997) note that the use of disproportionate share hospital payments, not case-mix adjustments, produces the greatest state-to-state variation in Medicaid spending.

Medicaid reimbursement inadequacies were further exacerbated in 1997 when the Balanced Budget Act of 1997 (Public Law 105-33) repealed the Boren Amendment (Schneider, 1997). From 1980 to 1997, the provisions of the Boren Amendment required that Medicaid payments to nursing facilities be based on minimum quality of care standards. Rates were to be "reasonable and adequate to meet costs which must be incurred by efficiently and economically operated providers in order to provide care and services in conformity with applicable state and federal laws, regulations, and quality and safety standards" (Wiener & Stevenson, 1998). However, effective October 1997, states could legally set nursing facility reimbursement rates that did not purport to provide for a minimum standard of care. The Balanced Budget Act requires only that nursing facility rates be set through a public notice process that gives providers and beneficiaries a reasonable opportunity to comment on proposed changes to the rate-setting methodology (Schneider, 1997). Such a system creates incentives to constrain patient-related costs and, thus, potentially reduce patient care services.

Rapid growth in Medicaid expenditures provided motivation for another form of cost control--the creation of managed care programs for Medicaid-eligible individuals. Under a managed care program, entities such as health maintenance organizations (HMOs) or prepaid health plans (PHPs) contract to provide specific services to enrollees in exchange for predetermined per capita payments (HCFA, 2002d). Individual states are given discretion in determining the type of managed care arrangement as well as the population categories that may or must be covered under such arrangements. The proliferation in managed care has been dramatic; from 1993 to 2000, the percentage of the total Medicaid population covered by managed care programs climbed from 14 percent to 56 percent (HCFA, 2002b).

Both Preston et al. (1997) and Eldenburg and Kallapur (1997) note that providers may forgo cost management activities in favor of strategies intended to increase donor confidence, identify new sources of funding, and/or alter the services provided to attract additional fee-paying clients. McMurtry, Netting, and Kettner (1990) argue that if alternative funding resources cannot be obtained, the provider may reduce, or even eliminate, service to that segment of the population deemed most costly. Thus, a reimbursement system that ignores patient heterogeneity may force the care provider to make strategic choices unrelated to improved efficiency in response to restrictions created by its funding environment.

HETEROGENEITY

Twenty-two states reimburse Medicaid nursing home providers using an unadjusted per-diem reimbursement rate that provides a uniform daily amount for each patient in residence (Wiener & Stevenson, 1998). This system requires that all facility operating costs pertaining to patient care be allocated to patients based on a single measure--number of patient days (PD). Assignment of costs in this manner is appropriate when costs are consumed strictly in proportion to volume. However, when activities occur for which resources are not consumed proportionally to volume, cost assignments upon which Medicaid reimbursement rates are based will be distorted. For example, Miller and Vollman (1985) argue that, in a manufacturing environment, expansion in an entity's product mix will trigger disproportionate increases in support activities that are not directly related to volume. The allocation of costs based on patient days alone may create similar problems in the Medicaid reimbursement of long-term care provider expenditures for AIDS patients.

AIDS is a disease that displays a wide range of severity. With an incidence rate of 7.3 per 100 person-years (Ostrow, 1997)and a median survival experience of less than one year after diagnosis (Turner, Markson, McKee, Houchens & Fanning, 1991), AIDS Dementia Complex (ADC) is one of the common late complications of the AIDS disease process. ADC is characterized by cognitive, motor, and behavioral dysfunction. The initial signs of dysfunction are usually decreased concentration and attention, decreased ability to perform complex tasks, and decreased motor performance. Patients may develop ataxia and generalized weakness. Later, patients often become incontinent and bed-bound, and ultimately may lapse into a vegetative state. The six-month mortality rate with advanced ADC is 67 percent (Ostrow, 1997).

As ADC patients become increasingly immobilized by cognitive and physical impairment, they require additional care for basic needs. The tasks of feeding, cleaning, bathing, and turning patients place increased time demands on the nursing staff. In contrast, as ADC patients lose motor performance and cognitive functioning, they participate less frequently in counseling and recreational activities (social services). However, the declines in a patient's social services costs only partially offset the concomitant increases in patient care. We argue that changes in the mix of patients exhibiting ADC symptomatic problems (ADC-Mix) exhibit greater congruity with changes in patient care and social services costs than does the volume indicator patient days (PD), the current reimbursement basis utilized by Medicaid. Specifically,

Proposition A: Compared with changes in PD, changes in ADC-Mix will more fully explain changes in patient care costs.

Proposition B: Compared with changes in PD, changes in ADC-Mix will more fully explain changes in patient social services costs.

If these propositions are true, then adding the presence of ADC as a modifier will improve reimbursement equity and reduce the burden on institutions providing long-term care to the segment of the AIDS population least able to fend for itself (those individuals diagnosed with the AIDS dementia complex). Our study addresses the Viability of this assertion.

DATA AND METHODS

We collected two years of data (1996-1997) at a 60-bed, community-based long-term care AIDS nursing facility. A consortium of local hospitals established the facility to provide a multidisciplinary approach to caring for AIDS patients exhibiting a level of impairment sufficient to warrant the rendering of acute nursing care. Our evidence is based on observations from a single facility and is therefore subject to potential shortcomings typically associated with case analyses. Nevertheless, the depth of detail that can be extracted from such an investigation allows us to isolate the effects of competing hypotheses regarding the nature of the variance in health care costs.

We interviewed the administration and staff regarding activities performed in each area of operation and compiled monthly archival data pertaining to facility expenditures. Additionally, with the assistance of the nursing staff, we obtained detailed data from medical records concerning the specific activities and therapeutics required for the treatment of each patient residing at the facility during the period of our study. Specifically, for each patient at the facility, we noted the presence or absence of the AIDS dementia complex, the presence of absence of substance abuse, and the specific periods for which they received each of 11 treatments that extend beyond the scope of basic care (i.e., incontinence, psychotherapy, speech therapy, physical therapy, occupational therapy, dialysis, diabetes, tube feeding, oxygen, IV therapy and wound care).

These data, classified in terms of number of days, were then used to develop alternative activity measures. Two indicators are contrasted here. One activity measure, expressed in patient days (PD), specifies the number of days a given patient received care in the facility. It assumes that identical daily units of patient care (and identical daily costs) are provided to each patient. As noted above, this activity measure is currently used to calculate Medicaid reimbursement. The second indicator, expressed as a ratio of ADC patient days to total patient days (ADC-Mix), is a complexity measure. This activity measure allows us to recognize that different levels of patient treatment may be provided in response to an initial ADC diagnosis.

The Site under study accumulates costs by expenditure class in 17 functional areas. We used a step-down allocation procedure to allocate the cost of administrative overhead that supports patient services to specific patient-related cost centers. Non-allocable portions of expenses, such as fixed monthly interest payments, depreciation charges, and general and administrative expenses unrelated to patient care, are treated as period expenses and are not included in the patient-related cost pools. Unallocated costs ranged from a minimum of $33,507 to a maximum of $51,789, and averaged $41,036 per month.

The step-down allocation resulted in the accumulation of costs into 11 cost centers. We allocated Site housekeeping, utilities, maintenance, and miscellaneous overhead costs to other functional areas based on square footage. Allocations of business office, social services, and administrative costs were based on time spent by cost center personnel on various activities. Allocation rankings were made in descending order based on the percentage of departmental services provided to other functional areas. Each resulting cost pool consists of salaries, related employee benefits, and associated operating costs.

Our 11 cost centers were then collapsed into three broad functional areas. The largest functional area, patient care, encompasses more than half of the total facility costs. Six cost centers are contained in this functional area: registered nurses wages/benefits (RNs); licensed professional nurses wages/benefits (LPNs); certified nursing assistants wages/benefits (CNAs); other patient care costs, such as general maintenance and housekeeping costs for patient rooms, patient food, infection and quality control expenses, medical administrative salaries, and patient records administrative costs (PATCR); drugs (other than antiretroviral therapies), laboratory tests, and medical supplies (PHARM); and costs associated with occupational, physical, and speech therapy (REHAB). Two cost centers--recreation (RECSR) and social services (SOCSR)--are classified as patient social services. The remaining three cost centers (FOOD, FACIL and ADMIN) contain food, housing, and general overhead costs unrelated to patient care, and are characterized as support services. Facility costs and activities were accumulated on a calendar month basis. Thus, 24 observations were obtained for each variable.

We tabulated descriptive statistics for all variables of interest. Spearman correlation coefficients were calculated to initially assess the association between the activity measures and facility costs. Then, we developed a series of regressions to demonstrate the comparative abilities of the two measures in explaining variances in facility costs.

RESULTS

A total of 233 patients diagnosed with AIDS were treated at the facility during 1996 and 1997. With a monthly mean of 1,770 patient days in 1996 and 1,784 patient days in 1997, occupancy rates remained relatively unchanged and the facility operated near full capacity (96.8 percent in 1996 vs. 97.8 percent in 1997) across the two-year period studied. The patient population mix varied, however. In 1997, 41.8 percent of all patients carried the ADC diagnosis, down from 52.2 percent in 1996.

The facility cared primarily for minorities, with a breakdown in 1997 of 62 percent black patients, 23 percent white patients, and 15 percent Hispanic patients. This distribution reflects the national increase in the minority AIDS patient population. During the two years of this study, the facility received virtually all (97.7 percent) of its funding from public programs, a significantly greater proportion than the 60 percent estimated nationally. Consequently, the case facility provides a unique opportunity to study the Medicaid AIDS population.

The Medicaid reimbursement rates are calculated from facility historical operating costs on an annual basis. During the period of this study, Medicaid rates ranged from a daily average rate of $370 in 1996 to $336 in 1997. Private insurance rates provided the facility were significantly higher, both in 1996 ($470 per patient day) and in 1997 ($410 per patient day). In 1997, $6,982,133 was received from the Medicaid program, $251,259 from private insurance, and $29,150 from charitable donations.

Although average daily Medicaid reimbursement rates remained below those accorded private patients, public-private differences in the payments per treatment element narrowed considerably during the period of the study. Medicaid payments per treatment element increased from $75 in 1996 to $79 in 1997, while private payments per treatment element declined (1996 = $99; 1997 = $84). Treatment protocols beyond basic care provided to specific patients ranged in number from zero to nine and averaged 3.9 in 1996 and 3.3 in 1997. Further details describing the care afforded the patients at this facility are presented in Exhibit 1.

During the period of the study, the facility generated an average of 1,777 patient days per month at an average monthly operating cost of $628,425. General and administrative costs totaling $41,036 were not allocated to specific cost centers, because these costs displayed no association with patient-related activities. Approximately half (55.6 percent) of the remaining $587,389 was attributable to patient care, 9.2 percent to social services, and 35.2 percent to support services. Monthly cost differences may be attributable in part to time differences (28 days vs. 31 days) and in part to complexity of care differences (ADC vs. non-ADC patients). On average, 47 percent of the patient population was diagnosed with ADC. The monthly ADC patient mix index ranged from a monthly low of 37.9 percent to a high of 64.8 percent during the two-year period of the study. Details of the 11 cost centers, as well as the two activity measures, are contained in Exhibit 2.

Exhibit 3 presents Spearman correlation coefficients used to examine the associations between facility costs (as defined by the 11 cost centers) and the two activity measures. Hypothesized relationships suggest that patient care costs will display greater positive correlation with the activity measure denoting the mix of ADC patients than with the activity measure used by Medicaid (PD). Similarly, we anticipate that patient social services costs will display a greater negative association with the ADC patient mix index than with patient days.

The activity measure currently used by Medicaid (PD) displays significant correlation with only one (RECSR) of the eight cost centers associated with patient care or social services. No correlation exists between patient days and the costs of providing any support services. The absence of significant correlation is problematic given that the uncorrelated cost centers incurred an average cost of $546,659 per month and represented 92.8 percent of all allocable facility costs.

Consistent with prior expectations, the ADC patient mix index, which measures the proportion of ADC patients present at any given point in time, exhibits significant positive correlation with four of the six patient care cost centers and significant negative association with both of the social services cost centers. These cost centers reported average monthly expenditures of $286,802 and represented 48.8 percent of all allocable costs. Our results suggest that, congruent with proposed relationships, patients diagnosed with ADC require more intensive care and fewer social services than other AIDS patients, supporting the premise that readily observable heterogeneity within the AIDS DRG can produce systematic variations in costs.

The ADC patient mix index also displayed negative associations with the three support services cost centers, although the association was only marginally significant for FOOD. These findings, however, must be interpreted with caution, as it is likely that at least the ADMIN association was driven by very specific changes in administrative costs that were experienced by the facility in 1997 (a period of proportionately fewer ADC patients). For example, accounting fees more than doubled from 1996 to 1997 when a volunteer doing part-time accounting work in the facility's business office was replaced with a temporary professional working "on loan" from the facility's audit firm. Further, the increased attention focused on the accounting function resulted in $54,000 of write-offs of patient receivables in 1997 (compared to none in 1996). Legal costs were also unusually high in 1997 (2.6 times the 1996 level), as a result of contentious and lengthy union contract renegotiations involving the nursing staff. In all, total administrative costs increased by 78 percent from 1996 to 1997, largely as the result of administrative changes which were unrelated to the treatment and care of facility patients. Accordingly, the relationship observed in our study between ADMIN and ADC-Mix is not likely to represent the ongoing reality of cost incurrence for the facility.

We developed a series of regressions to demonstrate the relative ability of our two activity measures (ADC-Mix and PD) to explain variances in facility costs. Exhibit 4 contains the results of these tests. The first column contains the results of the regression models that include only PD. Column B provides the results of the regression models that include both PD and the ADC patient mix index.

Consistent with the posited relationships, the ADC-Mix displays greater power in explaining variances in facility costs than does PD. Patient days (Model A) displays no significant association with patient care costs (Adjusted [R.sup.2] = -0.044, p = 0.859) and only minor association with variations in the costs of providing social services (Adjusted [R.sup.2] = 0.113, p = 0.060). In contrast, after controlling for PD, ADC-Mix explains 40.1 percent of the variation in patient care costs (p = 0.002) and 33.8 percent of the variation in social services costs (p = 0.005).

Disaggregating facility costs into the 11 cost centers offers further evidence of the superiority of the ADC patient mix index in explaining changes in patient care and social services costs. PD exhibits some power in explaining changes in recreational services costs (Adjusted [R.sup.2] = 0.209, p = 0.014). However, adding ADC-Mix as an activity measure adds significant explanatory power beyond that provided by PD (Adjusted [R.sup.2] = 0.417, p = 0.001).

In contrast, changes in the ADC patient mix index explain changes in several cost centers associated with patient care. After controlling for PD, ADC-Mix provides significant information regarding LPN nursing costs (Adjusted [R.sup.2] = 0.279, p = 0.012), CNA nursing costs (Adjusted [R.sup.2] = 0.470, p = 0.001), other patient care costs (Adjusted [R.sup.2] = 0.275, p = 0:013), and pharmacy costs (Adjusted [R.sup.2] = 0.220, p = 0.028). Results suggest that increases in the prevalence of patients with ADC are associated with increases in patient care costs.

The ADC patient mix index also explains significant amounts of variance in one social services cost center. After controlling for PD, increases in ADC-Mix were associated with cost decreases in recreational services (Adjusted [R.sup.2] = 0.417, p = 0.001). Certain support services manifested patterns similar to those captured by recreational services. The negative correlation exhibited between the costs incurred in the FACIL functional area (Adjusted [R.sup.2] = 0.456, p = 0.001) may suggest that some cost savings experienced in patient care were shifted to facility upkeep to support increases in this area. As noted above, the significant negative relationship between the ADC patient mix index and administrative services (Adjusted [R.sup.2] = 0.283, p = 0.012) may be spurious.

These regression results can be expressed monetarily. The increase in a standardized 30-day month of one patient diagnosed with ADC, which is defined clinically upon admittance to the facility, will create an increase of 1.7 in the ADC patient mix index (30 patient days per month per patient / 1,800 total patient clays per month = 1.7 percent). This patient increase, in turn, creates a cost increase of $4,435 (1.7 x $2,608.86) in patient care costs and declines of $389 (1.7 x $229.02) in social services costs.

DISCUSSION AND CONCLUSIONS

Patients with ADC seem to require more intensive patient care, as evidenced by the positive correlation with numerous patient care cost centers. This increase in cost is somewhat offset by a decrease in the use of social services, but the overall result is a net increase in the cost of the services provided. However, Medicaid reimburses strictly by number of patient days, and fails to take into account severity of illness. This method of reimbursement sets the stage for health care institutions to engage in "implicit rationing of scarce health care resources" (Zuniga, 1998, 1).

It is not necessary that all patients or costs associated with these patients be the same in order for per-diem reimbursement to be successful. Reimbursement of an average daily cost will be superior if the costs of the per-diem system (inappropriate incentives for providers to manage admissions so that the percentage of patients initially diagnosed as requiting more costly therapeutics will be reduced) are less than the costs associated with a more refined reimbursement system (the costs of identifying those patients in need of more complex care). Nevertheless, the results of this study suggest that stratifying the AIDS nursing home population into two readily identifiable groups, those with ADC and those without, has the potential to alleviate some of the provider's desire to ration health care while adding minimal or no increased costs to the measurement/reimbursement process. By modifying the Medicaid system to identify ADC patients and increasing the government payment for such individuals, a more equitable system of reimbursement could be achieved. At a minimum, knowledge that a cost differential exists should enable care providers to implement improved planning and budgeting procedures.

While the evidence provided by this study is based on observations from a single facility and is subject to potential shortcomings typically associated with case analyses, it does provide the benefit of isolating the effects of competing hypotheses regarding the nature of the variance in health care costs. Further study is warranted to evaluate the extent to which these findings hold at the macro level. However, insights have been provided into the potential usefulness of cost management tools in addressing a problem that has widespread implications for the state of health care financing.

EXHIBIT 1
SITE PROFILE
FOR TWO YEARS ENDED DECEMBER 31, 1997

PANEL A
               Characteristic                      1996          1997

Number of employees (mean per month)              174.3         163.1
Patient Race/Ethnicity:
  Black                                             64%           62%
  White                                              22            23
  Hispanic                                           14            15
Patient statistics (mean per month):
  Admissions                                          8             8
  Deaths                                              6             4
  Dismissed to home                                   1             3
  Left against medical advice                         1             1
Number of beds occupied (mean per month)           58.1          58.7
Patient Days (mean per month):                    1,770         1,784
  Patient Mix-
    AIDS Dementia Complex (ADC)                   52.2%         41.8%
    Substance abuse (SA)                           66.3          59.1
Breadth of Care-
  Basic care (BC)                                100.0%        100.0%
  Incontinence (INCON)                             54.4          42.7
  Psychotherapy (PSYCH)                            76.6          78.6
  Speech therapy (SPCHT)                            8.9           9.6
  Physical therapy (PHYST)                         49.6          37.8
  Occupational therapy (OCCPT)                     31.8          25.3
  Dialysis (DIALY)                                  7.9           2.5
  Diabetic (DIAB)                                  10.8          13.2
  Tube feeding (TF)                                 5.6           5.1
  Oxygen (OXY)                                     24.2          13.6
  IV therapy (IV)                                  15.1          12.7
  Wound care (WC)                                  39.9          31.3

PANEL B
               Characteristic                      1996          1997

Treatments beyond Basic care-
  Mean per month
Number of patients receiving treatments
    0-1                                               9            18
    2-3                                              29            59
    4-5                                              59            39
    6-7                                              40            25
    8-9                                              12             9
Funding:
  Total dollars
    Private                                     $62,100      $251,259
    Medicaid                                  7,815,687     6,982,133
    Donations                                       500        29,150
  $ mean per patient day-
    Private                                        $470          $410
    Medicaid                                        370           336
  $ mean per treatment element-
    Private                                      $99.04        $83.72
    Medicaid                                      75.24         78.67

EXHIBIT 2
DESCRIPTIVE STATISTICS
(PER MONTH)

                         % of
     Category         Total Costs     Mean        St. Dev.

COST CENTERS
Patient Care
  RNs                    14.5        $84,904        $9,988
  LPNs                   11.8         69,015         9,509
  CNAs                   12.1         71,342         8,137
  PATCR                  12.5         73,332         7,325
  PHARM                   3.2         18,710         7,466
  REHAB                   1.5          8,890         1,323
Social Services
  SOCSR                   2.3         13,674         1,565
  RECSR                   6.9         40,729         2,449
Support Services
  FOOD                   12.5         73,420         5,317
  FACIL                   8.4         49,479         2,517
  ADMIN                  14.3         83,893        12,682
Recap
  Patient Care           55.6        326,193        28,807
  Social Services         9.2         54,403         3,443
  Support Services       35.2        206,793        17,885
MEASURES
  PD                                   1,777            60
  ADC-Mix                                 47.0           7.6

     Category         Median        Min.         Max.

COST CENTERS
Patient Care
  RNs                 $83,880      $70,957     $104,325
  LPNs                 69,584       48,039       84,219
  CNAs                 70,838       61,378       93,097
  PATCR                74,049       53,754       89,719
  PHARM                18,220        6,519       35,290
  REHAB                 8,878        6,666       11,641
Social Services
  SOCSR                14,418       10,577       15,689
  RECSR                40,583       34,895       45,862
Support Services
  FOOD                 73,256       66,467       87,483
  FACIL                49,772       45,193       53,753
  ADMIN                79,362       68,916      107,648
Recap
  Patient Care        320,883      269,697      376,757
  Social Services      54,244       48,006       61,551
  Support Services    202,971      180,949      246,602
MEASURES
  PD                    1,779        1,628        1,866
  ADC-Mix                  44.7         37.9         64.8

EXHIBIT 3
SPEARMAN CORRELATION COEFFICIENTS
COST CENTERS WITH ACTIVITY MEASURES

    Cost Centers            PD          ADC-Mix

Patient Care
  RNs                      0.07         0.03
  LPNs                     0.13         0.58 ***
  CNAs                     0.01         0.74 ***
  PATCR                   -0.02         0.52 ***
  PHARM                   -0.05         0.45 ***
  REHAB                   -0.08         0.08
Social Services
  SOCSR                    0.15         -0.41 **
  RECSR                    0.44 **      -0.61 ***
Support Services
  FOOD                    -0.07         -0.28 *
  FACIL                    0.22         -0.68 ***
  ADMIN                    0.19         -0.65 ***
Recap
  Patient Care            -0.01          0.64 ***
  Social Services          0.38 **      -0.66 ***
  Support Services         0.17         -0.63 ***

*** Significant at p [less than or equal to] 0.01

** Significant at p [less than or equal to] 0.05

* Significant at p [less than or equal to] 0.10

EXHIBIT 4
PARAMETER ESTIMATES
COST CENTERS

          MODELS                   A                   B

Aggregated Costs
  Patient Care
    Intercept                   293,470             39,309
    PD                               18.41              92.40
    ADC-Mix                                       2,608.86 ***
    Adjusted [R.sup.2]               -0.044              0.401
    Significance of F                 0.859              0.002
  Social Services
    Intercept                    14,578             36,890 *
    PD                               22.41              15.92
    ADC-Mix                                           -229.02 ***
    Adjusted [R.sup.2]                0.113              0.338
    Significance of F                 0.060              0.005
  Support Services
    Intercept                    80,345            205,055 *
    PD                               71.16              34.85
    ADC-Mix                                         -1,280.10 ***
    Adjusted [R.sup.2]                0.014              0.276
    Significance of F                 0.262              0.013

Patient Care
  RNs
    Intercept                    61,059             63,392
    PD                               13.42              12.74
    ADC-Mix                                                -23.94
    Adjusted [R.sup.2]               -0.039             -0.088
    Significance of F                 0.709              0.931
  LPNs
    Intercept                    50,735            -21,734
    PD                               10.29              31.38
    ADC-Mix                                            743.86 ***
    Adjusted [R.sup.2]               -0.041              0.279
    Significance of F                 0.764              0.012
  CNAs
    Intercept                   114,778 **           40,522 *
    PD                              -24.44               -2.83
    ADC-Mix                                             762.21 ***
    Adjusted [R.sup.2]               -0.012                0.470
    Significance of F                 0.400                0.001
  PATCR
    Intercept                   226,572 *            82,180
    PD                               -0.74               41.29
    ADC-Mix                                           1,482.12 ***
    Adjusted [R.sup.2]               -0.045               0.275
    Significance of F                 0.991               0.013
  PHARM
    Intercept                    24,996            -27,535
    PD                               -3.54              11.75
    ADC-Mix                                            539.21 ***
    Adjusted [R.sup.2]               -0.045              0.220
    Significance of F                 0.895              0.028
  REHAB
    Intercept                     11,455             12,673
    PD                                -1.44              -1.80
    ADC-Mix                                              52.25
    Adjusted [R.sup.2]                -0.041             -0.067
    Significance of F                  0.762              0.412

Social Services
  SOCSR
    Intercept                    9,695             16,778
    PD                               2.24               0.18
    ADC-Mix                                           -72.71
    Adjusted [R.sup.2]              -0.038              0.043
    Significance of F                0.691              0.242
  RECSR
    Intercept                        4,883             20,111
    PD                              20.17 **           15.74 **
    ADC-Mix                                          -156.31 ***
    Adjusted [R.sup.2]               0.209              0.417
    Significance of F                0.014              0.001
Support Services
  FOOD
    Intercept                   78,086 **          96,512 **
    PD                              -2.63              -7.99
    ADC-Mix                                          -189.14
    Adjusted [R.sup.2]              -0.045             -0.018
    Significance of F                0.891              0.464
  FACIL
    Intercept                   29,026 *           50,652 ***
    PD                              11.51               5.21
    ADC-Mix                                          -221.99 ***
    Adjusted [R.sup.2]               0.033              0.456
    Significance of F                0.195              0.001
  ADMIN
    Intercept                  -26,266             57,891
    PD                              62.27              37.63
    ADC-Mix                                          -868.97 ***
    Adjusted [R.sup.2]               0.045              0.283
    Significance of F                0.163              0.012

*** Significant at p [less than or equal to] 0.01

** Significant at p [less than or equal to] 0.05

* Significant at p [less than or equal to] 0.10

REFERENCES

Cooper, R. (1996). Look out, management accountants, Management Accounting, (79):20-26.

Eldenburg, L., & Kallapur, S. (1997). Changes in hospital service mix and cost allocations in response to changes in Medicare reimbursement schemes, Journal of Accounting and Economics, (23),31-51.

Fetter, R.B. (1991). Diagnosis related groups: understanding hospital performance, Interfaces, (21),6-26.

Gonnella, J.S., Hornbrook, M.C., & Louis, D.Z. (1984). Staging of disease--a case-mix measurement, Journal of American Medical Association, (251),637-644.

Harvey, J.C. (1997).The Ethics of AIDS Care. http://www.thebody.com/iapac/harvey.html. Accessed March 29, 2001.

Hellinger, J. (1998). Cost and financing of care for persons with HIV disease: an overview, Health Care Financing Review, (19), 1-14.

Holahan, J., & Liska, D. (1997).Variations In Medicaid Spending among States, http://newfederalism.urban.org/html/anf_a3.htm. Accessed March 15, 2001.

Horn, S.D. (1983). Measuring severity of illness: comparisons across institutions, American Journal of Public Health, (73),25-31.

Jasper, B., & Kertesz, L. (1996). A question of care, Modern Healthcare, (June),120-129.

Johns Hopkins (2002). Medicaid overview: basic program, Johns Hopkins AIDS Service, Baltimore, MD: Johns Hopkins University, http://www.hopkins-aids.edu/manage/medicaid.html. Accessed March 22, 2002.

Makadon, H.J., Seage, G.R., Thorpe, K.E., & Fineberg, L.V. (1990). Paying the medical cost of the HW epidemic: a review of policy options, Journal of Acquired Immune Deficiency Syndromes, (3),123-133.

McMurtry, S.L., Netting, F.E., & Kettner, P.M. (1990). Critical inputs and strategic choice in non-profit human service organizations, Administration in Social Work, (14),67-82.

Miller, J.G., & Vollmann, T.E. (1985). The hidden factory, Harvard Business Review, (Sept-Oct),142150.

Ostrow, D.G. (1997). Psychiatric considerations in human immunodeficiency virus disease, In V.T. DeVita, S. Hellman, S.A. Rosenberg, J. Curran, M. Essex, & A.S. Fauci (Eds.) AIDS Etiology, Diagnosis, Treatment and Prevention, Philadelphia, PA: Lippincott-Raven.

Preston, A.M. (1992). The birth of clinical accounting: a study of the emergence and transformations of discourses on costs and practices of accounting in U.S. hospitals, Accounting, Organizations and Society, (17),63-100.

Preston, A.M., Chua, W.F., & Neu, D. (1997). The diagnosis-related group-prospective payment system and the problem of the government rationing health care to the elderly, Accounting, Organizations and Society, (22), 147-164.

Schneider, A. (1997). Overview of Medicaid Provisions in the Balanced Budget Act of 1997, P.L. 10533. Washington, DC: Center on Budget and Policy Priorities. http://www.cbpp.org/908mcaid.htm. Accessed January 8, 2001.

Stern, R.S., & Epstein, A.M. (1985). Institutional responses to prospective payment based on diagnosis-related groups, The New England Journal of Medicine, (312),621-627.

Turner, B.J., Markson, L.E., McKee, L., Houchens, R., & Fanning, T. (1991). The AIDS-defining diagnosis and subsequent complications: a survival-based severity index, Journal of Acquired Immune Deficiency Syndromes, 4(10), 1059-1071.

US Department of Health and Human Services. Centers for Medicare and Medicaid Services (CMS) (2002). Medicare Basics, http://www.medicare.gov/Basics/Overview.asp. Accessed March 22, 2002.

US Department of Health and Human Services. Health Care Financing Administration (HCFA) (2002a). Medicaid and Acquired Immune Deficiency Syndrome (AIDS) and Human Immunodeficiency Virus (HIV) Infection, http://www.hcfa.gov/medicaid/obs11.htm. Accessed March 10, 2002.

US Department of Health and Human Services. Health Care Financing Administration (HCFA) (2002b). 2000 Medicaid Managed Care Enrollment Report, http://www.hcfa.gov/medicaid/trends00.pdf. Accessed March 22, 2002.

US Department of Health and Human Services. Health Care Financing Administration (HCFA) (2002c). Overview of the Medicaid Program, http://www.hcfa.gov/medicaid/mover.htm. Accessed March 4, 2002.

US Department of Health and Human Services. Health Care Financing Administration (HCFA) (2002d). Medicaid: A Brief Summary, http://www.hcfa.gov/pubforms/actuary/ormedmed/default4.htm. Accessed March 22, 2002.

Wiener, J.M., & Stevenson, D.G. (1998). Repeal of the Boren Amendment: Implications for Quality of Care in Nursing Homes, New Federalism: Issues and Options for States, (Number A-30) Urban Institute. http://newfederalism.urban.org/html/anf30.html. Accessed January 16, 2001.

Zuniga, J. (1998). Examining the Necessity to Ration Healthcare Resources for HIV/AIDS and Other Life Threatening Illnesses, http://www.iapac.org/policy/conferences/rationing.html. Accessed March 10, 2002.

Kelly R. Lowther
Kelly R. Lowther, M.D. (USA)

Joyce A. Strawser
Seton Hall University (USA)

Annie S. McGowan
Texas A&M University (USA)

Sarah A. Holmes
Texas A&M University (USA)

We are indebted to the Director and staff of the case study site. Without their assistance and support, this research project would not have been possible. We also wish to thank the two anonymous reviewers for their helpful comments and suggestions. In addition, we are grateful to Thomas Albright, Joseph Kerstein, Michael Kinney and Robert Strawser for their comments on prior versions of this paper. We benefited further from the comments provided by the participants of the 1998 Academy of Business Administrators Global Trends Conference, the American Accounting Association 1999 annual meeting, and the 2001 International Symposium & Workshop of the International Society for Research in Healthcare Financial Management.

Address for correspondence: Joyce A. Strawser, Stillman School of Business, Seton Hall University, South Orange, NJ 07079 USA, [e-mail].