Mortality seasonality in Escazú, Costa Rica, 1851-1921.

Mortality seasonality has been frequently reported in populations living under various ecological conditions. The number of deaths apparently varies by month as a result of fluctuating climatic variables. Here, I examine whether mortality was seasonally distributed in Escazú, Costa Rica, from 1851 to 1921. I also investigate which cause of death made the greatest contribution to mortality periodicity and which climatic variable caused the seasonality of deaths. Two different methodological approaches are used: chi-square and Freedman tests to determine the presence of seasonality and Box-Jenkins time series analysis. The tests to determine the presence of seasonality show that mortality was strongly seasonal in Escazú, with the highest number of deaths occurring during the early months of the rainy season. The Box-Jenkins analysis successfully modeled the series from 1851 to 1891 with a seasonal parameter. However, the more recent years of the data were modeled without seasonal parameters. Thus the time series approach indicates that mortality became less seasonally distributed in more recent years. An analysis of gastrointestinal- and respiratory-related deaths showed that gastrointestinal causes had a strong seasonal distribution (with more deaths in the early months of the rainy season) and that respiratory causes did not. Therefore rainfall appears to have been responsible for mortality seasonality through its impact on the frequency of gastrointestinal maladies. The analysis of cause of death also shows that the seasonal fluctuation of gastrointestinal-related deaths decreased in the more recent years. Such a decrease is reflected in the nonseasonal Box-Jenkins model for the 1892-1921 years.