Seasonality of mortality in Sri Lanka: biometeorological considerations.

Abstract

The seasonal variation of mortality in Sri Lanka was studied in relation to meteorological variables. Time series data of the total monthly number of deaths in men and women from 1976 to 1980 were analyzed by power spectrum analysis of the fast Fourier transformation method in the areas of Colombo (lowland area, altitude 7 m) and Nuwara-Eliya (highland area, altitude 1890 m), respectively. In the Colombo area, where the mean temperature was always high (mean, 27.5 degrees C), a seasonal variation in mortality of 6 months was validated by power spectrum analysis with peak times in June and November. These mortality peaks corresponded to those of the amounts of rainfall which also showed an identical predominant period of 6 months. In the Nuwara Eliya area, where the mean temperature was always low at about 15-16 degrees C, a seasonal variation of approximately 3 months was found in addition to the 6 months' periodicity. The comparison of the power spectrum pattern of mortality with those of meteorological variables suggested that the amount of rainfall was associated with the seasonal variation of mortality in the Nuwara-Eliya area. In conclusion, the seasonality of mortality in Sri Lanka, a typical tropical zone country, was related to seasonal variation in the amount of rainfall, or a humidity factor, but not to mean temperature.