Two models for organizing inpatient AIDS care have emerged as hospitals in high HIV incidence cities have responded to the AIDS epidemic: dedicated AIDS units and scattered-beds located on multidiagnosis medical units. Empirical evidence has been largely lacking on whether one model is better. Evidence that high AIDS volume hospitals achieve lower mortality suggests that dedicated AIDS units might also result in lower mortality, providing an alternative to regionalizing AIDS care in order to improve outcomes. This proposal is an extension of research (HSO6858) on the outcomes of dedicated AIDS inpatient units in 20 hospitals in 11 major cities. The study is a comparative, multisite observational study in which matching is employed at the hospital level to introduce the control elements of quasi-experimental design. A rich and unique dataset has been assembled on these hospitals including detailed information on organizational attributes of the units and the hospitals, extensive data on nurses, and an array of information on a consecutive sample of AIDS patients admitted to 40 units including interviews, nurses' clinical assessments, extensive medical records data, discharge summaries, and billing information. The characteristics of the 1300 patients studied closely resemble the profile of the national AIDS population. We propose to pursue two of the central aims of our outcomes study of dedicated AIDS units: (1) improving the power of our study to detect the effects of dedicated AIDS units on mortality by extending the patient follow-tip period beyond discharge; and (2) making an important methodological contribution to health services research by improving the performance and feasibility of AIDS severity of illness measures. We show that a simple four-category scale reflecting nurses' assistants of patients' needs for assistance in basic activities of daily living is a better predictor of in-hospital mortality than the two more established AIDS severity of illness measures, and CD4 counts. The superior performance of the functional status measure may be due, in part, to informative right censoring of death information inherent in the original design using in- hospital mortality outcomes. To investigate this potential bias, we shall determine the mortality status, within 30 days of admission, of patients in our original sample, re-estimate models predicting morality on the basis of various staging systems, including function, in light of the extended observational frame, and examine functional status as a single adjuster for severity of illness and as an adjunct to existing severity of illness staging systems. In addition, we shall determine the practical utility of nurse-assessed functional status in AIDS care research by conducting a four-hospital test of its reliability when used by clinical care nurses in natural inpatient settings. The proposed research will provide the best evidence to date on the outcomes of dedicated AIDS units, and is expected to improve severity adjustment in AIDS research by increasing its predictive validity and feasibility.
| Aiken, L H; Sloane, D M; Lake, E T et al. (1999) Organization and outcomes of inpatient AIDS care. Med Care 37:760-72 |
| Aiken, L H; Sochalski, J; Lake, E T (1997) Studying outcomes of organizational change in health services. Med Care 35:NS6-18 |
