Climate and stroke: a controversial association.

The study recently published by Wang et al. (2009), which is based on one of the largest population-based surveys, shows a trend in the occurrence of intracerebral hemorrhages (ICH) with temperature variations in Brisbane. His work showed a statistically significant increase (15%) of ICH for every 1°C increase in daily T° max. In the literature, the topic is considered controversial, and the study raised many observations that we would like to discuss in this paper. Our group in collaboration with the Department of Neurology at Sohag University (Fawi et al. 2009) in southern Egypt, under the patronage of the Egyptian Ministry for Higher Education, reported a similar trend of higher prevalence of hemorrhagic events in patients admitted to the local hospital during a 2-year survey. In southern Egypt, we observed a peculiar spring–summer trend: our researches demonstrated an astonishing increase in hemorrhagic events. Stroke unit admissions for hemorrhagic stroke reached the surprising prevalence of 47.7% while the local spring– summer temperature range was an average of 33°C (Tmax 40°C, Tmin 26°C). The hemorrhagic events in the autumn– winter time frame reached 31%. A meta-analytic approach to the topic may be recommended in order to better focus the relationships between disease outbreaks and meteorological data on a global scale. Statistical testing for the meta-analysis should be performed with traditional methodology for homogeneity and significance; χ2 techniques (for goodness of fit to the null model of equal distribution of strokes in each climatic area) may be recommended, along with 95% confidence intervals, to evaluate the seasonal pattern of stroke hospitalization. Two strategies may be used to estimate the relative risk of stroke occurring at specific seasons. One strategy assumed that all strokes would be evenly distributed in onset among the 12 months, and therefore compared the observed proportion relative to the proportion expected, based on the total number reported. The second strategy is based on a comparison of the observed number of strokes compared with the average for the other months of the year, normalized for the month under consideration. In this particular analysis, the expected number of events that were reported to have occurred, for instance, between January and March, should be compared with the number of events occurring in the remaining 9 months of the year (divided by 3, to normalize for the number of months in the time period under consideration), and the relative risk was the number of strokes actually reported, divided by the number expected.