Standard method for developing stroke registers in low-income and middle-income countries: experiences from a feasibility study of a stepwise approach to stroke surveillance (STEPS Stroke)

Background Stroke is a leading global cause of death, with an estimated 5.8 million fatal events in 2005, two-thirds of which happened in low-income and middle-income countries. In these regions, epidemiological methods to establish hospital-based stroke registers for clinical audit or studies to estimate incidence are scarce. Our aim was to ascertain whether stroke registers could be set up in geographically diverse populations in low-income and middle-income countries, using standardised data-collection manuals and methods, before recommending their wider use. Methods WHO's stepwise approach to stroke surveillance (STEPS Stroke) offers an entry point for countries to register stroke patients in health-information systems. The methods proposed in this strategy were tested in a feasibility study, which focused on hospitalised stroke patients in nine different surveillance sites located in five low-income and middle-income countries. Data collection was for a median of 12 months. Observed differences between men and women were adjusted for age and surveillance site with logistic-regression analyses. Findings A total of 5557 stroke patients were registered; 91 people whose age was missing or younger than 15 years were excluded from the analyses. Mean age was 64.2 years (SD 14.6), and 2484 (45%) participants were women. Ischaemic stroke accounted for about two-thirds of events. Half of all patients were hospitalised the same day. Stroke subtype was verified in 4913 (90%) participants by diagnostic techniques. Women had lower odds of verification of stroke subtype compared with men after adjustment for age and surveillance site (odds ratio 0.69 [95% CI 0.56-0.86]; p=0.0006). Interpretation STEPS Stroke can be used in diverse populations to provide data in a standardised manner in countries with little or no previous records of stroke. Future studies should concentrate on expansion beyond hospital case series by adding information for stroke patients treated outside the hospital, linked to census data for the source population from which the cases com