Abstract

We examined the relation between the daily numbers of deaths ascribed to sudden infant death syndrome (SIDS) (n = 6226) and daily temperature in England and Wales over the five year period 1979-83. When the data were filtered to remove the dominant seasonal trend, and residual autocorrelation, we found a significant negative correlation of deaths with both the level and rate of change of temperature four to six days earlier, irrespective of age at death. Place of usual residence was obtained for 909 SIDS cases occurring during the unusually severe winter of 1981-82, and, using space-time clustering techniques, we confirmed previous findings of the lack of 'epidemicity' for this condition. These results are compatible with several previous hypotheses of the relation between the weather and SIDS and directly incriminate drops in temperature in the occurrence of the condition.