Approaches to the problems of measuring the incidence of stroke: the Auckland Stroke Study, 1991-1992.

BACKGROUND: Stroke registers are the preferred choice for determining incidence, case-fatality and severity of acute stroke in defined populations. This paper highlights some of the problems likely to be encountered in this endeavour by describing the experience of measuring acute stroke prospectively.
METHODS: The Auckland Stroke Study is a community-based study among 945,000 residents of the Auckland region, New Zealand. Standard definitions and overlapping case-finding methods were used to identify all new acute stroke events occurring during the 12-month period ending 1 March 1992. Particular attention was directed at including non-fatal strokes managed outside hospital. The latter were identified by use of a cluster sample, a technique suitable for populations where residents have a personal primary health care physician.
RESULTS: The comprehensive sources of referral to the study involved the review of 5736 records, less than one-third of which met the criteria for inclusion. The majority of included acute stroke events (n = 1803) were found through routinely available sources such as hospital admission records (63%) and death registrations (10%). The remainder (27%) were identified through intensive efforts at case-finding of stroke events managed outside hospital. The 1803 events were registered in 1761 people, 817 men and 944 women; for 587 (72%) men and 718 (76%) women, the stroke was the first ever experienced.
CONCLUSIONS: While time-consuming, costly and demanding, there appears to be no easier alternative to a register to estimate incidence. This study demonstrates the importance of the use of comprehensive case-finding sources and suggests approaches to overcoming the difficulties in monitoring stroke incidence in large populations.