Relationship between longitudinal changes in blood pressure and stroke incidence.

The relationship of changes in blood pressure with time to stroke incidence was examined on members of the Adult Health Study sample who have participated in biennial clinical examinations at the Radiation Effects Research Foundation since their inception in 1958. The regression coefficient of blood pressure regressed on time (the increase in blood pressure per cycle) was used as an index of the change in blood pressure with time. Cox's regression analysis, a technique which is suitable for follow-up studies was used. The data suggest that a single blood pressure measurement is not sufficient for predicting risk; the accumulated value or average over a period of time should be considered for this purpose. In addition to the actual blood pressure, the increase in blood pressure with time is a risk factor, particularly for cerebral hemorrhage. Cerebral hemorrhage was more strongly related to diastolic than to systolic blood pressure, while cerebral infarction appeared to be more strongly related to systolic than to diastolic blood pressure.