Seasonal variation in stroke mortality rates.

OBJECTIVE: To identify possible contributors to the seasonal variation in stroke mortality. BACKGROUND: Stroke and respiratory disease mortality rates were calculated from vital statistics and census data for the United States from 1938 to 1988. State-specific average temperatures by month were derived from data obtained from the National Climatic Data Center for 1938 to 1987. METHODS: Each time series was decomposed into a trend, a seasonal effect, and a residual effect. Multiple regression was used to fit both a trend and a seasonal harmonic series. Cross-correlation was used to assess the relationship between the residual time series. RESULTS: There is a strong and consistent seasonal pattern of high stroke and respiratory disease mortality in the colder winter months. Stroke mortality was significantly and independently both positively associated with respiratory disease mortality and inversely associated with temperature. The sharp initial increases in both respiratory disease and stroke mortality in the late fall and early winter are synchronous, and the amplitudes are strongly associated, except for a saturation effect with extreme respiratory disease amplitudes. CONCLUSIONS: Seasonal change in stroke mortality is associated with seasonal variation in both respiratory disease and temperature. Respiratory disease and temperature may influence stroke mortality nonspuriously by affecting stroke case fatality, incidence, or both.