Problems in the Use of Death Certificates to Identify Sudden Unexpected Infant Deaths

Death certificates on all U.S. infants who died at 3 months of age in 1968 were examined to see if sufficient information was available to identify those whose death was sudden and unexpected. Of the 2,954 deaths, 371 were coded for official mortality statistics on underlying cause of death to nonspecific causes implying, or compatible with, sudden unexpected death (ICDA eighth revision, 795, 796.2, 796.3, 796.9). An additional 151 deaths were described on the death certificate as sudden and unexpected, but because of the presence of other information in the cause of death section, they were assigned to various specific causes of death. Based on incidence rates from several earlier studies, it was estimated that there remained at least several hundred additional sudden unexpected infant deaths coded to other causes. Nonetheless, other items that might have been helpful in identifying these deaths, such as approximate interval between onset and death, place of death, and type of certifier, were either infrequently recorded or not sufficiently discriminating to establish criteria for sudden unexpected death that were both sensitive and specific. Unless changes are made in the construction of the death certificate, the completeness and accuracy with which it is filled out, or the coding of underlying cause of death, it is unlikely that accurate rates of sudden unexpected infant death will be routinely produced in the United States.