Protective effect of pregnancy in rural South Africa : questioning the concept of "indirect cause'' of maternal death

Background: Measurement of the level and composition of maternal mortality depends on the definition used, with inconsistencies leading to inflated rates and invalid comparisons across settings. This study investigates the differences in risk of death for women in their reproductive years during and outside the maternal risk period (pregnancy, delivery, puerperium), focusing on specific causes of infectious, non-communicable and external causes of death after separating out direct obstetrical causes. Methods: Data on all deaths of women aged 15-49 years that occurred in the Agincourt sub-district between 1992 and 2010 were obtained from the Agincourt health and socio-demographic surveillance system (HDSS) located in rural South Africa. Causes of death were assessed using a validated verbal autopsy instrument. Analysis included 2170 deaths, of which 137 occurred during the maternal risk period. Findings: Overall, women had significantly lower mortality during the maternal risk period than outside it (age-standardized RR = 0.75; 95% CI = 0.63-0.89). This was true in most age groups with the exception of adolescents aged 15-19 years where the risk of death was higher. Mortality from most causes, other than obstetric causes, was lower during the maternal risk period except for malaria, cardiovascular diseases and violence where there were no differences. Lower mortality was significant for HIV/AIDS (RR = 0.29, P<0.0001), cancers (RR = 0.10, P<0.023), and accidents (RR = 0, P