Up to now, evaluations of care for dialysis patients have focused primarily on biomedical quality, utilization, and cost indicators. This study is aimed at identifying factors which can be used to improve health-related quality of life (HRQOL) for those patients. It will also study the impacts of HRQOL on other outcomes. Three types of questions and their policy implications will be examined. First, what are the predictors of HRQOL scales for dialysis patients? Second, do HRQOL scales vary among dialysis patients due to social categories, such as socioeconomic status, Medicaid status, uninsured status in the pre-ESRD phase of the disease, race, ethnicity, or geographic region? Third, what impacts do HRQOL scales have on other outcomes, such as utilization of hospital care, nutritional status, employment, and patient compliance? An ecological model will be tested. In this model, health status is viewed as affected by factors on five different environmental levels. These include intrapersonal factors such as comorbidities and behavior patterns, primary groups such as the family and workplace, health care system factors, community contextual factors, and larger national or regional macrosystem factors. Dependent variables will include two types. First, several HRQOL scales will be used to evaluate the predictors and distribution of HRQOL for dialysis patients. Second, several other outcomes will be used to evaluate the impact of HRQOL on these other outcomes. Independent variables representing hypotheses and statistical controls include factors representing all five ecological levels. The data source will be Wave 2 of the Dialysis Morbidity and Mortality Survey. It contains longitudinal data on a national sample of patients who began dialysis in 1996 and 1997. It merges patient questionnaires, medical record abstracts, and Medicare claims data. Statistical analysis will include both single equation models - multiple regression and logistic regression-and LISREL structural equation models. This study will explore the differences between the LISREL models and the single equation models, both in terms of their relative explanatory power and the relative complexity of constructing and interpreting each type of model.
