Multivariate logistic analysis of risk factors for stroke in Tilburg, The Netherlands.

By means of a case-control study conducted between October 1, 1978, and July 31, 1981, in Tilburg, The Netherlands, various characteristics and events, including personal data, health-related behavior, and medical history, were evaluated as risk factors for stroke. The study subjects included 132 stroke patients and 239 age- and sex-matched control patients interviewed at the two city hospitals. To assess joint effects and possible interactions, and to control for multiple confounding factors, a series of multivariate logistic models for matched data were studied. From this analysis, it appeared that hypertension, acute myocardial infarction, cardiac arrhythmias, transient cerebral ischemic attacks, obesity, physical activity during leisure time, education of head of household, and Rhesus factor were all significant stroke risk factors. These risk determinants demonstrated a multiplicative effect in general; however, the influence of some variables on stroke risk was not constant with age (hypertension, acute myocardial infarction, cardiac arrhythmias, obesity, and Rhesus factor) and sex (hypertension and education of head of household). The relationship of diabetes mellitus to stroke slightly decreased and became nonsignificant after adjustment for factors besides age and sex. Stroke risk was not associated with cigarette and alcohol use, family history of stroke and related disorders, marital status, and ABO blood typing.