A population-based case-control study of 1250 stroke deaths in rural Bangladesh

Background and purpose: There are limited population-based studies to determine the risk factors for stroke in Bangladesh. Methods: A health and demographic surveillance system has been maintained in Matlab, Bangladesh (population 223886, 142 villages in 2008). All adult stroke and injury deaths (2005-2008) were monitored by verbal autopsy. Risk factors for stroke deaths were calculated using a multivariable logistic regression model with adult injury deaths as controls. Results: A total of 1250 stroke deaths (51% women; mean age 72.3years, range 20-101) occurred out of 4955 total deaths and were compared with 246 adult injury deaths (47% women, mean age 55.8years, range 20-100). The population-attributable mortality of stroke was 25.2% based on the verbal autopsy instrument and 17.8% when accounting for the reported sensitivity and specificity of a similar verbal autopsy instrument that has been validated for stroke death. Risk of stroke death was significantly increased with hypertension (OR 7.94, 95% CI 4.44-15.54, P<0.001), diabetes mellitus (OR 2.54, 1.21-6.21, P=0.02), and betel consumption (OR 2.36, 1.45-3.80, P<0.001) when adjusted for age and sex. An increased risk was not observed with heart disease (OR 1.37, 0.45-5.95, P=0.62), cigarette smoking (OR 1.41, 0.82-2.45, P=0.22), tobacco powder (OR 1.15, 0.30-7.64, P=0.86), or cigar/hookah pipe smoking 0.94 (0.45-2.18, P=0.88) when adjusted for age and sex. There were more strokes in winter (December-March) than summer (June-September) (P<0.001). Conclusions: There is a high modifiable burden of risk factors for adult stroke deaths in rural Bangladesh, most notably including hypertension. Betel consumption may be an under-recognized risk factor for stroke death.