Assessing U.S. pregnancy mortality using vital statistics and surveillance data, 1999-2005

Deaths from pregnancy complications are a sentinel public health indicator. Maternal deaths are reported in two Centers for Disease Control and Prevention systems: the National Vital Statistics System (NVSS) which reports maternal mortality and the Pregnancy Mortality Surveillance System (PMSS) which conducts epidemiologic surveillance of pregnancy deaths. Our objectives were to ascertain revised U.S. pregnancy mortality ratios based on data from both systems, and to identify changes in maternal death reporting resulting from the implementation of ICD-10 (1999) and the 2003 revised U.S. Standard Certificate of Death. For two periods, 1999-2002 and 2003-2005, we matched deaths in PMSS and NVSS by death certificate number and used information from pregnancy checkboxes, matched live birth certificates, and other sources to establish the temporal relationship of each death to pregnancy. We reviewed causes of death, excluding deaths that did not meet the WHO maternal death definition by cause. Deaths occurring within one year of pregnancy were included in the analysis (maternal and late maternal deaths). There were 2,369 pregnancy deaths in 1999-2002 and 2,231 pregnancy deaths in 2003-2005; 60-62% of the deaths were reported by both systems during the two study periods. The proportion of deaths reported only in PMSS decreased over time from 35% to 26% while the proportion of pregnancy deaths reported only by NVSS increased from 4% to 13%. The 1999-2002 pregnancy-related mortality ratio for the U.S. was 14.7/100,000 live births and for 2003-2005 it was 18.1/100,000. Improved ascertainment of pregnancy deaths has contributed to increases in pregnancy mortality ratios.