UNDERREPORTING OF PREGNANCY-ASSOCIATED DEATHS

Abstract

In the March issue of the Journal, Chang et al. call attention to the finding that homicide is a leading cause of death among pregnant and postpartum women.1 However, their findings substantially underestimate the magnitude of the problem, because data on pregnancy-associated deaths collected by the Centers for Disease Control and Prevention’s Division of Reproductive Health (DRH) through the Pregnancy Mortality Surveillance System are incomplete. To collect data for the Pregnancy Mortality Surveillance System, the DRH asks states to voluntarily send death certificates for all maternal deaths, that is, deaths resulting from medical causes related to the pregnancy that occur during pregnancy or within 42 days of delivery or termination of pregnancy. The DRH also asks states to send information on pregnancy-associated deaths, that is, deaths from any cause that occur during pregnancy or within a year of delivery or termination of pregnancy. This would include homicides. Because it is impossible to identify all pregnancy-associated deaths with only the information contained on death certificates, and few states use additional sources of data to identify pregnancy-associated deaths,2 the information reported to the DRH is incomplete. Our Maryland study showed that only a small proportion of pregnancy-associated deaths can be identified from death certificates alone and that comprehensive identification of pregnancy-associated deaths requires collection of data from additional sources, including medical examiners’ records and linkage of death records with birth and fetal death records.3 Medical examiners’ records are a critical source of information on homicides that occur among women who are pregnant at the time of death, but Maryland may be the only state that routinely reviews medical examiners’ records for this purpose. Linkage of death records with birth and fetal death records is important in identifying homicides among postpartum women, but few states routinely link these records at all and even fewer do so for a full year following pregnancy. Further evidence of underreporting in Chang’s study is that their reported pregnancy-associated homicide rate of 1.7 per 100000 live births is substantially lower than rates cited in other reports. The pregnancy-associated homicide rate in Maryland was found to be 10.5 per 100000 live births when death records, linkage of records, and medical examiner records were used to identify deaths.3 Using only death certificates and linked records to identify deaths, Parsons and Harper4 in North Carolina and Nannini et al.5 in Massachusetts found rates of 7.2 and 3.5, respectively. Both states and the Centers for Disease Control and Prevention must improve their efforts to collect complete and accurate data on pregnancy-associated deaths. This is a critical step in the prevention of pregnancy-associated mortality from all causes.