In rural Bangladesh, women's mortality is not directly related to the total number of children they bear or to their pace of childbearing.1 Analyses based on Matlab Demographic Surveillance System data show, however, that each birth is associated with an elevated risk of death that extends for more than two years following the immediate perinatal period. In addition, the odds of death are positively associated with age and negatively associated with height and body mass index. The authors estimate that reducing lifetime exposure to the extended risk associated with individual births could lower mortality among reproductive-age women by approximately one-quarter.
Previous research on the association between women's risk of mortality and various aspects of childbearing—for example, parity and the interval between births—has yielded mixed results. To investigate further the relationship between mortality and fertility, researchers examined data from 2,031 married women who had originally participated in a study on fertility conducted in the mid-to-late 1970s. That study had collected information on reproduction (e.g., pregnancies, terminations and births), maternal health (e.g., height, weight and body mass index) and child health each month for three years, as well as socioeconomic information (e.g., education and religion) at study entry. Women were then followed up as part of the Matlab Demographic Surveillance System, which records all births, deaths and migrations occurring each month. The researchers converted data into woman-years for each woman for each calendar year, beginning with the year of study enrollment and ending with the year of death or migration, or 1996; data from returning migrants were excluded from analysis.
Researchers examined data on the basis of age at the start of a given woman-year. By the end of the study period, the 2,031 women had contributed a total of 34,067 woman-years between ages 16 and 54—the age by which all women had completed childbearing. The average height of the women was about 148 cm and the average body mass index was 18.5. Roughly three-quarters (77%) of women had not received any schooling, and 13% were Hindu. By the end of follow-up, a total of 3,937 children had been born and 100 women had died. On average, women of reproductive age had had a total of five live births each and had given birth at a slightly faster pace than had women in Matlab in 1979 (rate ratio, 1.1).
Using a series of discrete-time models, the investigators tested the effects of fertility behavior on women's risk of dying during a given year between ages 16 and 54, with controls for background and health factors. In an analysis that included only these factors, the risk rose with increasing age and declined with increasing height and body mass index. Succeeding models that added fertility-related variables found that neither parity nor the pace of childbearing was associated with mortality risk. Furthermore, the researchers found no interaction between these two factors and age, body mass index or height. Hence, they suggest, the risk of mortality "is not directly related to cumulative effects" of parity and pace of reproduction on health—in particular, incomplete recovery between frequent, closely spaced births (a phenomenon known as "maternal depletion").
In the final model, the odds of dying were doubled among women who had given birth in the current year or in the past two years (odds ratio, 2.0), leading the analysts to suggest that each birth is associated with an elevated risk of maternal mortality that lasts for 2-3 years. In this analysis, the odds of dying remained positively associated with age (1.1), and negatively associated with height (0.96) and body mass index (0.8).
Furthermore, the researchers estimated that decreasing the lifetime exposure to the extended risk associated with each birth (i.e., by decreasing fertility) would result in a substantial reduction in mortality among women of childbearing age. For example, for a woman giving birth for the first time at age 16 and then once every three years, the probability of dying before age 50 would decrease by 19% (from one in 13 to one in 16) if she were to have a total of three instead of seven children; the reduction would be 24% if the birth interval were four years. Similarly, if a woman were to have three instead of seven children, starting at age 21, the probability of dying before age 50 would decrease by approximately one-quarter—23% with three-year birth intervals and 28% with four-year intervals.
The analysts conclude that in Matlab, "lifetime childbearing experience does affect a woman's survival," not because of maternal depletion between successive births, but because each birth poses an elevated risk of mortality, which, the authors note, lasts for several years beyond the usual postpartum observation period of six weeks. The researchers comment that the risk may be related to the pregnancy or birth itself, or to illnesses arising during or aggravated by pregnancy or birth. They add that family planning programs in Bangladesh "have surely contributed to the increase in female life expectancy...by reducing the number of times a woman was exposed to the possibility of death from extended maternal risk."
1. Menken J, Duffy L and Kuhn R, Childbearing and women's survival: new evidence from rural Bangladesh, Population and Development Review, 2003, 29(3):405-426.