An examination of the influence of person, disease, and provider factors on the outcomes of hospital AIDS care in an HM

This study described and investigated factors that contributed to hospital outcomes of 317 persons with AIDS (PWAs) who received care within a staff model Health Maintenance Organization (HMO) system during 1994. The goal was to determine the influence of measurable hospital care system factors on the outcomes of hospitalization using a systems framework. Multiple regression was used to determine predictors of length of stay (LOS), nursing cost, mortality, and need for additional skilled care following hospitalization. Factors contributing to the outcomes of hospitalization were organized as sociodemographic and economic, disease condition and health status, and provider related factors. LOS was predicted by number of comorbid conditions, a disease related factor and the average predicted need for nursing care score, a provider related factor. Total nursing costs were predicted by five factors, four of them being disease related factors: presence of dementia, number of comorbid conditions, disease stage, and principal diagnosis of PCP. Average predicted need for nursing care was the provider related factor predictive of nursing cost. A higher likelihood of expiration while hospitalized was associated with disease stage, principle diagnosis of PCP, total number of hospitalizations during the year, and average projected need for nursing care. Time spent on the AIDS nursing unit was a significant negative factor on in-hospital mortality. The likelihood of need for additional nursing care in a skilled nursing facility (SNF) or homecare was associated with average projected need for nursing care and long LOS during hospitalization. Notable factors that did not influence the outcomes of hospitalization were age, ethnicity, history of drug or alcohol use, physician experience with AIDS, and actual nursing care hours delivered. This study adds evidence to the role of disease burden, measured by comorbidity and projected nursing needs, as the major predictor of variability in hospital costs and utilization. It demonstrates that predicted nursing needs may underestimate actual nursing needs at the patient level, that PWAs require more nursing care than typical medical/surgical patients, and identifies the need for post hospital care services for PWAs. Finally, it provides evidence that patient placement on dedicated AIDS nursing units affects hospital mortality.