Unintended pregnancies are associated with a variety of negative child and maternal outcomes, according to a large-scale analysis from India.1 Although some of these associations were apparent in standard regression analyses, they were especially evident and particularly robust in analyses that took into account unmeasured familial factors; in the latter models, mistimed pregnancies, unwanted pregnancies or both were positively associated with deliveries without a skilled attendant (odds ratio, 1.3), incomplete early childhood immunization (1.4–2.2), child stuntedness (1.3), and neonatal, postneonatal and early childhood mortality (1.8–5.9).
Although numerous studies have examined relationships between pregnancy intendedness and subsequent maternal and child outcomes, results have been mixed, and no studies have focused on India, where more than five million unintended births occur each year (and where a third of the world’s stunted children live). In the new study, researchers used data from the 2005–2006 Indian National Family Health Survey, which asked women about a range of social, demographic and health-related measures, including information about each pregnancy and birth the women had had in the past five years. The researchers examined four types of outcomes: whether a birth was supervised (i.e., took place in a medical institution or was attended by trained medical personnel); whether the child was stunted; whether the child had received the World Health Organization–recommended panel of vaccinations against six key infectious diseases (e.g., polio, tuberculosis); and whether the child had died. Child deaths were classified as neonatal if they had occurred within 28 days of birth, as postneonatal if they had occurred from day 29 to day 365, and as early childhood if they had occurred between the child’s first and third birthdays; pregnancies were classified according to whether they were wanted, mistimed (wanted later) or unwanted (not wanted at all) at the time of conception.
The investigators used two approaches to identify predictors of these outcomes. First, they used basic logistic regression models that controlled for 12 variables, including household wealth, the child’s age and birth order, and maternal age, education and autonomy; these analyses included the full sample of 51,555 births. Next, to account for unmeasured variables within families that might have affected the outcomes of interest, the investigators used family-fixed-effects regression models, which were restricted to births to women who had had two or more births that differed in the outcome of interest. For example, the analyses of neonatal mortality were restricted to births to women who had given birth during the past five years to both a child who had died within 28 days of birth and one who had survived. Although these analyses had greatly reduced sample sizes (465–7,619 births, depending on the outcome), their focus on sibling pairs helped control for the influence of differences in physical environment, childcare practices, and other household-specific factors that are not typically measured in large demographic surveys.
Overall, 80% of the births were the result of wanted pregnancies, 10% of mistimed pregnancies and 10% of unwanted pregnancies. Forty-one percent of the births were to mothers who had had no schooling, 71% to those aged 20–30 and 62% to those who lived in a rural area. Trained medical personnel had supervised fewer than half of the births (47%), and only 41% of children had received a full set of immunizations. About two in five children (42%) were stunted. Mortality rates were 43 per 1,000 births during the neonatal period, 22 per 1,000 in the postneonatal period and 10 per 1,000 during early childhood.
In standard regression analyses, mistimed pregnancies were less likely than wanted pregnancies to result in neonatal mortality (odds ratio, 0.8). Findings were very different, however, in the models that controlled for family fixed effects: Mistimed pregnancies were positively, rather than negatively, associated with neonatal mortality (1.8), and were also associated with elevated odds of unsupervised deliveries (1.3), incomplete immunization (1.4), and postneonatal mortality (2.6) in the fixed-effects analysis.
Unwanted pregnancies were also associated with undesirable maternal and child outcomes, and again the results differed by analysis type. In standard regression analyses, deliveries were more likely to be unsupervised if a pregnancy was unwanted rather than wanted (1.2), and children from unwanted pregnancies were more likely than those from wanted pregnancies to be stunted (1.1) and insufficiently immunized (1.3). Unwanted pregnancy was not associated with any type of child mortality. In the family-fixed-effects analyses, on the other hand, unwanted pregnancy was strongly associated with neonatal (2.2), postneonatal (3.6) and early childhood (5.9) mortality, as well as with stuntedness (1.3) and incomplete immunization (2.2); no association with unsupervised delivery was apparent, however.
The authors also performed analyses that used the same relatively small samples as the fixed-effects analyses but otherwise used standard regression methodologies. Although the associations identified in these analyses tended to mirror those found in the fixed-effects analyses, the odds ratios were uniformly (and often substantially) smaller, indicating that the differences between the results of fixed-effects regressions and those of the standard regressions were not due simply to differences in the samples, and that unmeasured familial factors influenced the outcomes observed in the study.
The researchers note that their study, like many, relied on retrospective assessments of pregnancy intentions that may not have accurately reflected women’s desires at the time of conception; for example, women may have been reluctant to acknowledge that a cherished child was the result of an unwanted pregnancy. Another potential limitation of the study is that the women whose births were included in the fixed-effects analyses tended to be poorer and less educated than women in the full sample, which may limit the generalizability of the results. Nonetheless, the investigators note that their findings are consistent with the hypothesis that unintended fertility takes "a significant toll on children and in some cases mothers," and that not only unwanted but also mistimed pregnancies may confer disadvantages. Given that "improving access to high-quality contraceptive services and fulfilling unmet need can reduce unintended pregnancies," the findings "underscore the importance of investments in family planning."—P. Doskoch
1. Singh A and Chalasani S et al., The consequences of unintended births for maternal and child health in India, Population Studies, 2012, 66(3):223–239.