Correcting the undercount in maternal mortality.
The actual number of maternal deaths occurring in the US each year is believed to be 20-30% greater than that recorded in national vital statistics, largely because death certificates often do not contain adequate information to identify maternal deaths. To assess the completeness of maternal death reporting in New York State, all deaths among women 10-49 years of age assigned to causes other than the International Classification of Diseases (ICD) codes 630-76 (the traditional codes used for classifying maternal deaths) for the years 1980-86 were identified and a computer search was undertaken to determine if the woman had terminated a pregnancy within 90 days of her death. Computerized matching resulted in the identification of an additional 193 deaths that occurred within 90 days of a pregnancy termination. A subsequent review of hospital records resulted in an additional 31 direct and 42 indirect maternal deaths in the 135 deaths for which records could be located. These additionally identified deaths served to increase the maternal mortality rate in 1980-86 from the official rate of 6.7/100,000 live births to 11.9/100,000 in upstate New York and from 29.0/100,000 to 31.9/100,000 in New York City. 41% of the deaths identified by record matching occurred within 1 week of pregnancy termination. The leading causes of death among the 73 maternal deaths revealed through record matching were subarachnoid hemorrhage and pulmonary embolism. To facilitate identification of maternal deaths, the latest version of the death certificate use in upstate New York contains a direct question on whether there was a recent pregnancy.