The Inpatient AIDS Unit: A Preliminary Empirical Investigation of Access, Economic, and Outcome Issues Mananne C. Fahs, PhD, MPH, George Fulop, MD, James Strain, MD, Henry S. Sacks, PhD, MD, Charlotte Muller, PhD, Paul D. Cleary, PhD, James Schmeidler, PhD, and Barbara Turner, MD, MSEd



The rapid spread of acquired immune deficiency syndrome (AIDS) poses a tremendous challenge to US hospitals. Yet there are no studies of appropriate models of hospital care.1 A lack of consensus remains over whether AIDS patients would be better served "clustered" together in designated AIDS units or "scattered" among other general medical inpatients.2-6 This is the first study to address several aspects ofthis debate. Specifically, we ask whether models of AIDS care characterized by clustered or scattered beds differ with respect to (1) equity and access, (2) economic costs, and (3) outcomes of care.

Using the computerized Hospital Information Management System, we selected all adult discharges from January 1, 1988, through December 31, 1988, having human immunodeficiency virus (HIV)related disease.8 The final study population consisted of 325 cases of HIV-related disease, 192 discharged from the scatter beds and 133 discharged from the AIDS unit. We collected information for each patient on age, sex, race, insurance status, origin (direct admission to unit or intrahospital transfer), emergency room admission, diagnoses (up to seven), discharge placement (long-term care facility or other), use of specific resources such as intensive care units, attending physician, length of stay, charges (room and board, and ancillary), outcome of hospitalization (survival vs death), and disease severity. We coded disease severity using the criteria of the Severity Classification for AIDS Hospitalizations developed by Turner and colleagues.9-12

Patins and Methods Setting The study was conducted at the Mount Sinai Medical Center in New York City. The hospital provides care to an average of 40 AIDS patients on a daily basis, both in general medical/surgical beds throughout the hospital (scatter beds) and in an AIDS cluster unit. The AIDS unit was established in October 1987 and has been described in detail elsewhere.7 Briefly, it contained 10 beds with a dedicated staff of four registered nurses (two on the day shift, two on the night shift) and one social worker. House staff are assigned to patients in the unit as part of their medical rotation. Admission to the AIDS unit is limited and requires documentation of HIV infection, AIDS-related complex, or AIDS, as well as patient awareness of the diagnosis. Patients are prioritized as follows: (1) patients admitted through the emergency room or clinic and awaiting beds, (2) patients who have been in the unit before, and (3) in-house patients wanting to transfer to the unit. Patients can refuse to be admitted to the AIDS unit.

Marianne C. Fahs and Charlotte Muller are with the Department of Community Medicine, George Fulop and James Strain are with the Department of Psychiatry, Henry S. Sacks is with the Department of Medicine, and James Schmeidler is with the Departments of Biomathematical Sciences and Psychiatry at Mount Sinai School of Medicine in New York. Paul D. Cleary iswith the Department of Health Care Policy at Harvard Medical School in Cambridge, MA, and Barbara Turner is with the Department of Medicine at Jefferson Medical College in Philadelphia, PA. Requests for reprints should be sent to Marianne C. Fahs, PhD, Associate Professor and Director, Division of Health Economics, Department of Community Medicine, Mount Sinai School of Medicine, 1 Gustave Levy Place, Box 1043, New York, NY 10029. This paper was submitted to the Journal January 22, 1991, and accepted with revisions November 27, 1991.

April 1992, Vol. 82, No. 4

Public Health Briiefs

Data Analysis The individual patient discharge was the unit of analysis. Controlling for clinical characteristics with logistic regression, we assessed access to the AIDS unit by estimating whether the probability of placement in it was associated with demographic characteristics. The predictors were age, race, sex, stage of disease, number of comorbid conditions, requirement of nursing home placement, and death. We entered only the first discharge for each study patient (n = 192) in the logistic regression to avoid the statistical bias that can occur when separate discharges are analyzed for the same patient. We assessed economic differences between the two settings by comparing the length of stay, total charges, total ancillary charges, and average daily ancillary charges for patients in the two settings. A semilog regression model was estimated using analysis of covariance to test whether the AIDS unit setting had an independent effect on length of stay and charges. The natural logs of length of stay and charges were used to normalize their distributions, increasing the precision of the coefficient estimates. Patients who died during hospitalization were excluded from the length-of-stay analysis. Finally, we assessed outcome by comparing inpatient mortality across the two settings. We compared death rates by severity level, and we used a logistic regression model to estimate differences in the probability of death, controlling for clinical and social characteristics.

Resuls Study Population The average patient age was 38.1 years. In an ethnic breakdown, 38.5% were Hispanic, 28.6% were Black, 30.8% were White, and 2.2% were Oriental and other. Seventeen percent of the patients were female. Privately insured patients made up 43.4% of the caseload.

Access to the AIDS Unit Severity of illness was the major predictor of admission to the AIDS unit. In addition, several sociodemographic characteristics were associated with placement in the unit. There were smaller proportions of female patients, older patients, and Black patients on the unit than in scatter beds (P < .05). Insurance status was not associated with cluster unit placement. April 1992, Vol. 82, No. 4

However, after adjusting for severity in the logistic regression analysis, only age remained significantly associated with admission to the AIDS unit. Younger patients (up to 40 years of age) had a higher likelihood of being admitted to the unit than did older patients (P < .05).

Economic Differences between Scatter and Cluster Patients Although the unadjusted length of staywas significantly higher for patients in the AIDS unit than for those in scatter beds (18.8 days vs 11.3 days,P < .01), the multivariate log-linear regression analysis showed that patients admitted directly to the unit did not have longer lengths of stay than patients in scatter beds. Nursing staff turnover in the AIDS unit, although high (the termination rate for 1988 was 45%), was found to be comparable to termination rates in other units, such as the geriatric unit (44%) or gynecological oncology

(45%). Inpatient Mortaity Higher rates ofinpatient mortality are likely in the AIDS unit owing to higher levels of severity. However, the noted differences in this study were not significant. Logistic regression confirmed that, controlling for severity, there were no differences in the likelihood of death between patients in the unit and those in scatter beds.

Disussion This is the first empirical study of an AIDS unit that is located in a major urban setting and serves a mixofrisk groups. We found that the proportions of AIDS patients from socially and economically advantaged groups (as indicated by White race and private health insurance) in an AIDS unit were similar to those in the general hospital. We conclude that the unit is not a "dumping ground" for the less advantaged AIDS patient. However, admission to the AIDS unit was disproportionately low for older persons with AIDS. Additionally, the data suggest the possibility of limited accessibility for women and Blacks. These findings are preliminary. Further research on access to specialized AIDS inpatient units must take into account the interrelationships between patient characteristics and hospital policies. For example, policies that require patient awareness of HIV positivity and give preference to individuals previously in the unit could adversely affect access for

individuals from population groups less likely to be aware of their HIV status, such as heterosexual women. Hospital capacity constraints also can differentially affect access if, for instance, singlesex restrictions apply to four-bed rooms. Moreover, patient preferences for unit placement may vary across population groups. These access concerns deserve further study. Economic concerns, at least, appear to be unfounded. Patients in the AIDS unit did not have significantly different lengths of stay after severity was taken into account. We found similar results when both total and ancillary charges were regressed on the model. Thus, the AIDS unit is not inefficient. In addition, nursing staff turnover was not excessively high. Finally, we found no significant differences in inpatient mortality rates between the cluster and scatter settings. However, our conclusions regarding outcome differences are limited by the study design. Further research should take into account clinical improvement from admission to discharge and postdischarge outcome measures, including survival, functional status, and readmission rates. In conclusion, the data do not support many of the arguments made against AIDS units, such as higher costs, patient stigmatization, or excessive staff burnout. The results do raise some concerns, however, regarding the accessibility ofthe unit to all persons with AIDS. The responsiveness of AIDS cluster units to the increasing diversity of persons with AIDS is an area that needs continued consideration and research as hospitals, physicians, and policymakers struggle to cope effectively and compassionately with the growing crisis of AIDS care. E

Acknowledgments Support through a grant from the United Hospital Fund in New York, NY, is grateftlly acknowledged. An earlier version of this paper was presented at the Fifth International Conference on AIDS on June 5, 1989, in Montreal, Canada. The authors wish to thank David Rose, Kathleen Wade, Marcy Rosen, Lynn Jacobson, Pat Siegal, Cynthia Bournazos, Philip Gelda, Phil Thompkins, Philip Olla, Bruce Barton, Devora Leben, Danny Pagano, and Brett Penney.

References 1. Weil PA, Stam LM. Hospital administrators' response to AIDS: results of a national survey. Med Care. 1990;28:468472. 2. Presidential Commission on the Human Immunodeficiency Virus Epidemic. Reportofthe Preidential Commrrzsion on the

American Joumal of Public Health 577

Public Health Briiefs Human Immunodeficiency Virus Epidemic. Washington, DC: US Government Printing Office; 1988. Publication 0-214701:QL3. 3. McCormick K. AIDS: moving toward chronic care. Med Health Perspect. 1989;43(suppl 11):1-4. 4. Weinberg DS, Murray HW. Coping with AIDS: the special problems of New York City. NEngIJMedJ 1987;317:1469-1472. 5. Taravelia S. Reserving a place to treat AIDS patients in the hospital. ModHealth-

1989;33-37. 6. Strain JJ, Fahs M, Fulop G, Sacks H. care.

7. 8. 9. 10.

AIDS: epidemiology and treatment issues. Mt Swi JMed 1989;56:233-237. Oberlink M. The AIDS cluster: New York's Mount Sinai establishes inpatient unit. AIDS Patient Care. 1988;2:26-29. Centers for Disease Control. Human immunodeficiencyvirus (HIV) infection classification. MMWR 1987;36(S-7):1-20. Turner BJ, Kelly JV, Ball JK. A severity classification system for AIDS hospitalizations. Med Care. 1989;27:423-437. KellyJV,BaIIJK,TurnerBJ. Durationand costs of AIDS hospitalizations in New York: variations by severity of illness. Med

Care. 1989;27:1085-1098. 11. Ball JK, Turner BJ. Distribution and characteristics of AIDS hospitalizations in the US, 1986-1987. Proceedings of the Fifth International Conference on AIDS; June 4-9, 1989; Montreal, Canada. 12. National Center for Health Services Research and Health Care Technology Assessment/Hospital Studies Program. A Severity Classification System for AIDS Hospitalization. Springfield, Va: National Technical Information Service; 1988: Publication PB88-249412:D1-1-D1O1O.

Duration of Medicaid AIDS Hospitalizations-Variation by Season, Stage, and Year Leona Enama Markson, ScD, Barbara J. Turner, MD, MSEd, and Thomas R Fanning PhD


- ~~~~~~~~~~~~-

@ :~~~~~~~~~~~~~~~R



.... E EEE __~~~~~~~~~~~~~~~~~~~~~~~~.......... -_ : R.e>;XvS.




e~%:: W . U s::: - v



Over the past decade, hospitals in several regions of the country have been heavily burdened by the dramatic rise in the number of acquired immunodeficiency syndrome (AIDS) patients.1,2 Inpatient care is one of the most expensive components of the cost of AIDS,3-5 and hospitals in New York State have one of the longest average lengths of stay for AIDS care in the country.2'6'7 Information on patterns of AIDS hospitalizations is needed to assist hospital planners in meeting the needs of this special population. In this study, we analyzed length-of-stay patterns for New York State Medicaid AIDS hospitalizations to examine temporal patterns and changes in hospital length of stay.

Medod We examined hospital claims for New York State Medicaid AIDS cases over 35 months, from October 1984 through August 1987. Using a methodology tested elsewhere,8 we identified a hospital stay as an AIDS admission if one of the following ICD-9-CM codes was recorded: Pneuocystis cainii pneumonia (136.3), selected immunodeficiency codes (279.10 and 279.19), or the AIDS codes (042.xx or 043.xx). The study population comprised AIDS patients aged 15 to 65 years who were hospitalized for nonobstetric rea-

sons. Two classes of outliers were deleted from the sample: patient admissions with stays of more than 66 days (less than 5% of the patients) and those with more than eight inpatient stays over the 35-month period (1.5% ofthe patients). The study sample consisted of 11 482patient admissions; males constituted approximately three fourths of the sample. Eighty-two percent of the sample was 26 to 45 years of age at the time of hospitalization. To adjust for differences in case mix, we determined the severity of each case, using the Staging Classification for AIDS Hospitalizations.9'10 This AIDS staging system can be summarized into three integer stages, with stage 3 indicating the most severe cases. Stage 4 is used in some analyses to distinguish patients who died in the hospital. Length of stay was analyzed as an average per month, calculated as the avLeona Enama Markson and Barbara J. Turner are with the Center for Research in Medical Education and Health Care and the Depart-

ment of Medicine, Jefferson Medical College, Philadelphia, Pa. Thomas R. Fanning is with the New York State Department of Social Services, Albany. Requests for reprints should be sent to Leona Enama Markson, ScD, Center for Research in Medical Education and Health Care, Jefferson Medical College, 1025 Walnut St, Suite 119, Philadelphia, PA 19107-5083. This paper was submitted to the Journal November 15, 1990, and accepted with revisions November 12, 1991.

April 1992, Vol. 82, No. 4

The inpatient AIDS unit: a preliminary empirical investigation of access, economic, and outcome issues.

An AIDS unit model ("cluster beds") and a general inpatient placement model ("scatter beds") in a major teaching hospital were compared to determine w...
2MB Sizes 0 Downloads 0 Views