Determining the incidence of different subtypes of stroke: results from the Perth Community Stroke Study, 1989-1990.

Objective: To determine the incidence and case fatality of seven distinct subtypes of stroke in Perth, Western Australia.
Design and Setting: A population-based descriptive epidemiological study.
Subjects: All residents of a geographically defined segment of the Perth metropolitan area (estimated population 138,708 persons) who had a stroke or transient ischaemic attack between 20 February 1989 and 19 August 1990, inclusive.
Main Outcome Measures: The following subtypes of stroke were classified according to standard clinical, radiological and pathological criteria: types of cerebral infarction, namely, large artery (thrombotic) occlusive infarction (LAOI), cerebral embolic infarction (EMBI), lacunar infarction (LACI) and boundary zone infarction (BZI); primary intracerebral haemorrhage (PICH); subarachnoid haemorrhage (SAH); and stroke of undetermined cause.
Results: Over the 18-month study period 536 stroke events were registered, of which 86% (95% confidence interval, 83%-89%) had a defined "pathological" diagnosis on the basis of computed tomographic scanning, magnetic resonance imaging or necropsy. Cerebral infarction accounted for 71% of cases (95% CI, 68%-75%), PICH 11% (95% CI, 9%-14%) and SAH 4% (95% CI, 2%-5%). The 382 cases of cerebral infarction included LAOI (in approximately 71%), EMBI (15%), LACI (10%) and BZI (5%). While the incidence of all subtypes of stroke increased with age, there were age and sex differences in their proportional frequency, management and prognosis: patients with PICH, SAH and EMBI were more likely to be admitted to hospital, and these conditions carried the highest early case fatality. Over all, the 28-day case fatality was 24% (95% CI, 20%-28%), but varied from 0 for LACI and BZI, to 37% (95% CI, 15%-59%) for SAH and 35% (CI, 23%-47%) for PICH.
Conclusions: In this study, we found considerable differences in incidence rates, the effect of age and sex on incidence rates, and prognosis for the different subtypes of stroke. Hospital-based studies are likely to be selectively biased by emphasising strokes that are severe and require admission to hospital. These data have important implications in the design and evaluation of clinical trials of therapy for stroke.