Accuracy of cause-of-death data recorded on death certificates for pregnancy-associated deaths — Ohio, 2008–2011

Background: Pregnancy-associated deaths, defined as death of a woman from any cause during or ≤1 year after pregnancy, substantially affect families. Targeting prevention efforts requires understanding the cause of death (COD). The accuracy of COD information recorded on death certificates, the primary data source, is not well-understood; therefore, we examined COD accuracy for pregnancy-associated deaths.

Methods: Pregnancy-associated deaths were identified using the death certificate pregnancy checkbox, COD code, and/or linkage of maternal death to fetal death or child birth certificates. We obtained death certificates and medical, coroner, and other records for pregnancy-associated deaths occurring in Ohio among residents during 2008–2011. A multidisciplinary expert committee reviewed each case’s records to determine if they agreed with death certificate immediate and underlying literal CODs. We used chi square to analyze differences in percentage agreement by age, race/ethnicity, certifier title, and county type, location, and manner of death.

Results: Of 234 identified pregnancy-associated deaths, the committee has reviewed 180 to-date and agreed with death certificate CODs for 152 (84.4% of reviewed deaths). Percentage agreement for reviewed deaths differed only by location of death (hospital, 77.8% agreement of 108 deaths; other/unknown, 92.9% of 28; home, 94.9% of 39; hospice/long-term care, 100% of 5; P = .03) and manner of death (natural, 75% agreement of 96 deaths; accident, 94.3% of 53; suicide/homicide, 100% of 26; P < .01; 5 cases did not include manner).

Conclusion: Efforts to improve death certificate COD reporting for pregnancy-associated deaths should focus on natural deaths and deaths in hospitals.