The Stroke Data Bank is a prospective observational study which collected data on hospitalized newly diagnosed stroke patients, at four clinical centers. The four collaborating clinical centers were responsible for the collection of acute care and longitudinal follow-up information using common definitions and procedures, under contracts N01-NS-2-2302, 2398-9, N01-NS-5-384. The general objective for the project was to provide a large and comprehensive body of data for clinical research on the factors influencing survival, morbidity and quality of life following onset of a stroke. The BFSB served as the statistical coordinating center for the project, providing an on-site front-end data entry system with interactive feedback for data editing and the data base management system for transmission, storage and retrieval of data, for monitoring of data acquisition and its quality. The data collection phase was completed and now the BFSB is responsible for statistical collaboration with the clinical investigators for the analysis of the primary research questions. The project has completed its final year and has entered 1805 patients. The first major analyses began after accrual of patients was completed and the acute care data were entered and edited. A mathematical model for identifying patients at greatest risk for progression of stroke has been developed based on the Pilot Stroke Data Bank, validated and revised based on the Main Phase SDB. The characteristics of the diagnostic subtypes of lacunar and infarcts of undetermined cause have been examined in detail. Mathematical models for both infarcts and intracerebral hemorrhages have been developed to identify patients unlikely to survive for one month following their stroke. Patient and lesion characteristics detected on CT scan have been identified that are associated with silent or unreported old strokes. A mathematical model for ischemic infarct patients has been developed to estimate the probability of early recurrence.
