Changes in the timing of SIDS deaths in 1989 and 1999: indirect evidence of low homicide prevalence among reported cases

An unknown proportion of cases diagnosed as sudden infant death syndrome (SIDS) are misdiagnosed, and in some cases are homicides. Because recent SIDS prevention measures were unlikely to reduce homicides, changes in the reported timing of SIDS cases provide an indirect measure of covert homicides in this group. This paper uses United States vital statistics microdata to explore these questions. The sample includes all reported infant deaths to singletons with birthweight > 500 g in the 1989 and 1999 US birth cohorts. Deaths attributed to SIDS (n = 7708), homicide (n = 597), or object inhalation and mechanical suffocation (n = 860) are specifically examined. If reported SIDS cases were a mixture of 'true' cases and misdiagnosed homicides, it is hypothesised that the age-at-death distribution of SIDS deaths would have changed to reflect greater prevalence of misdiagnosed homicide. We find that the age-at-death distribution of reported SIDS cases was virtually unchanged in the two cohorts, showing no increase during periods of infancy when relative homicide risk is most pronounced. One cannot reject the hypothesis that the timing was drawn from the same distribution (chi(2)(52) = 62.2, P = 0.157). Analogous results hold for infants born in circumstances associated with high homicide risk (chi(2)(50) = 61.5, P = 0.12). The stable age-at-death distribution of reported SIDS cases between 1989 and 1999 suggests that covert homicides are a small fraction of reported SIDS cases.