Determinants of hospital-based stroke complications and mortality: An ecological approach

The purpose of this study is to develop and empirically test an ecological model that explains the variance in stroke mortality, using the POET model. This model examines patient risk attributes, organizational and environmental characteristics. Donabedian's quality paradigm is used as a contextual framework. Unlike previous uses of this model, outcomes were used as a product of structure and process. The hospital is the unit of analysis. Hospital structural and process measures were examined. Using Contingency theory, the proposed framework notes that for hospitals to achieve the best outcomes, environmental demands must be met.;Based on a review of the literature, four constructs were developed that represented determinants of stroke outcomes: environmental risk factors, patient risk factors, hospital capacity and specialized clinical capability. Data was obtained from several sources: the 1998 AHA data base, Area Resource File 1998, HCFA 1998, and HRSA 2000. Hospital samples were taken from HCUP NIS, which included more than 1000 hospitals treating over 150,000 stroke patients. Stroke outcomes were defined as either complications or mortality, and were risk adjusted and case-mix adjusted. County-level analysis were performed.;Exploratory methods, as well as analytic methods were performed to examine the causal relationship of variables with outcomes. These included ANOVA, correlation, multiple linear regression and LISREL.;For variables measuring environmental risk factors, linear regression revealed an inverse relationship between county hard water and stroke complications. LISREL revealed that the county population living under poverty, county age 65, county population black, MSAs, and Florida mandate are also environmental variables that are statistically associated with stroke mortality.;For variables measuring organizational characteristics, linear regression reveals that hospitals with stroke units have less stroke mortalities. LISREL reveals that, teaching hospitals, hospital stroke volume, the number of ICU beds, the presence of geriatric services, and participation in a nation stroke network are all statistically related to stroke mortality.;This study was performed at the organizational and environmental level. Information about patient severity of illness was not available. The study confirms using an ecological approach to explore the causal relationship between environmental and organizational variables and disease outcomes.