![[ Visit Client Website ]](images/banner.jpg)
BACKGROUND: Pregnancy-associated deaths, defined as deaths from any cause occurring during pregnancy or within 1 year after pregnancy, are largely preventable and have a substantial impact on families. Surveillance of pregnancy-associated deaths is necessary for comprehensive monitoring and prevention efforts. Vital Statistics (VS) mortality data are the primary source for timely monitoring of pregnancy-associated deaths; however, their accuracy is not well-understood. We examined the accuracy of pregnancy-associated death data within Ohio’s VS mortality data.
METHODS: A case was defined as any pregnancy-associated death occurring in Ohio among residents during 2008–2011. We ascertained cases in the VS mortality database in two ways: (1) checkbox indicating pregnancy at the time of death or ≤1 year before death, assigned by physicians and coroners and recorded in the Electronic Death Registration System, and (2) International Classification of Diseases, 10th Revision, cause-of-death codes in the O series (pregnancy, childbirth, and the puerperium), assigned by the National Center for Health Statistics. As the standard for comparison, we used Ohio’s Pregnancy-Associated Mortality Review (PAMR) data. PAMR is a time-intensive expert review of medical and other records for pregnancy-associated deaths ascertained from VS mortality data plus deaths ascertained from linkage of maternal death certificates to child birth or fetal death certificates, conducted ≥2 years postmortem. We calculated sensitivity and predictive value positive (PVP) of pregnancy-associated death data in the VS mortality database overall and by certifier title.
RESULTS: A total of 282 pregnancy-associated deaths were ascertained from VS mortality data. PAMR disqualified 103 (36.5%) of these cases as false-positives and ascertained 55 additional cases, leading to a PVP for pregnancy-associated deaths identified from VS mortality data of 63.5% and a sensitivity of 76.5%. Compared with physician-certified deaths, PVP for coroner-certified deaths was significantly higher (coroners, 90.2%; physicians, 39.6%; P < .01); however, the sensitivity for coroner-certified deaths was significantly lower (coroners, 72.3%; physicians, 86.8%; P = .02).
CONCLUSIONS: Despite timely availability of the data, relatively low PVP and sensitivity limit the usefulness of VS mortality data alone for pregnancy-associated death surveillance. Improvements are needed in both coroner and physician pregnancy-associated death reporting.