Most deaths associated with pregnancy – and even more associated with induced (intentional abortion) and spontaneous (miscarriage) abortions – are not reported as such.
In 2003, the United States implemented a a revised death certificate including information on whether or not women who died had been pregnant within the last year. Many have speculated whether or not this will affect the statistics on pregnancy-related death and the myth that abortion is safer than carrying a pregnancy to term.
In the November, 2004 issue of Paediatric and Perinatal Epidemiology, there is an article which points out that, historically, the reporting of all deaths associated with pregnancy has been innacurate.
(registration required for full text)
Here is the summary from the article:
To find maternal and pregnancy-related deaths, it is important that all pregnancy-associated deaths are identified. This article examines the effect of data linkages between national health care registers and complete death certificate data on pregnancy-associated deaths. All deaths among women of reproductive age (15-49 years) in Finland during the period 1987-2000 (n = 15 823) were identified from the Cause-of-Death Register and linked to the Medical Birth Register (n = 865 988 births), the Register on Induced Abortions (n = 156 789 induced abortions), and the Hospital Discharge Register (n = 118 490 spontaneous abortions) to determine whether women had been pregnant within 1 year before death. The death certificates of the 419 women thus identified were reviewed to find whether the pregnancy or its termination was coded or mentioned. In total, 405 deaths (96.7%) were identified in registers other than the Cause-of-Death Register. Without data linkages, 73% of all pregnancy-associated deaths would have been missed; the percentage after induced and spontaneous abortions was even higher. Data linkages to national health care registers provide better information on maternal deaths and pregnancy-associated deaths than death certificates alone. If possible, pregnancies not ending in a live birth should be included in the data linkages.