Late Age at Childbirth: Survival Advantage or Artifact?

The article by Shadyab et al., “Maternal Age at Childbirth and Parity as Predictors of Longevity Among Women in the United States: The Women’s Health Initiative,”¹ reports that, among participants in the Women’s Health Initiative (WHI), those who first gave birth at age 25 years or older were more likely to live to age 90 years than those whose first birth occurred before age 25 years. The temporal order suggests a causal effect of birth timing on longevity, as implied in the accompanying press release, headlined: “Study finds that women giving birth for first time later in life increase chances of living to 90.” Media reports, such as a headline in The International Business Times (“Want to live past 90? Women who have children later live longer”), reinforced this interpretation.

The findings are not as definitive as such headlines imply; although intriguing, they are also puzzling. “Advanced maternal age,” which was associated with a greater likelihood of reaching 90 years of age in this sample, is a well-documented risk for childbirth-related deaths.² One puzzle is how older age at first birth acts as a risk factor for mortality at the time of birth but as a protective factor later in life. Is there a plausible mechanism for this pattern? A second puzzle is that within the group of older first mothers, those aged 25 to 29 years at delivery were significantly more likely to reach age 90 years than those who delivered before age 25 years, whereas the effect for women aged older than 30 years was not statistically significant. If older age at first birth increases chances of living to age 90 years, why doesn’t a first birth after age 30 years provide more advantage rather than less?

DESIGN LIMITATIONS

The puzzling pattern of findings may reflect biases introduced by the use of data from the WHI. The WHI recruited and followed a large sample of postmenopausal women. The sample in the current study averaged 74.6 years. Having survived earlier culling of the population, they would have had better health than those in their birth cohort who died before reaching menopause.

This survival bias does not preclude establishing some kinds of causal relationships from WHI data. The project spawned important randomized trials, such as those testing the effects of postmenopausal hormone treatments and of low-fat diets, and prospective observational studies linking events after enrollment in the study to subsequent health and survival. However, although Shadyab et al. describe their study as prospective, that is not accurate. Prospective analyses track individuals with differential exposures from the time of exposure onward. In this study, the exposure—first childbirth at a younger or older age—occurred years before enrollment in the WHI and no information is available on those who died in the interim, some number of whom died in childbirth.

Because childbirth-related deaths are not random, “immortal time bias”³ is introduced by differential selection bias involving those who delivered at different ages. Healthier women are better able to survive the rigors of pregnancy and delivery. Because delivering a first child at an older age poses greater physical challenges than does childbirth when younger, selection processes are magnified in the group of older first-time mothers. WHI participants who survived childbirth at an older age were likely to have more physical reserve and greater potential longevity than women who delivered their first child when they were younger.

Shadyab et al. acknowledge that women who survived childbirth at an older age are likely to have been healthier to start with than were women who delivered at earlier ages, but they do not specify the implications. Another recent study, which reported that women who had their last child at age 33 years or older had longer leukocyte telomeres than did women delivering their last child before age 29 years, was more explicit. It concluded, “extended maternal age at last childbirth may be a marker for longevity.”^4(p497) A similar conclusion is implicit, but not highlighted, in the current study.

Shadyab et al. consider several confounders that could account for the association between age at first childbirth and survival to age 90 years. At study entry, later-delivering women had characteristics, such as more education and higher income that contributed to both better health and longevity. Adjusted analyses controlled for these and other covariates, but the measures may have been insufficient to capture the full impact of their joint association with health status and with timing of first birth, leaving residual confounding.

IMPLICATIONS FOR RESEARCH AND THE PUBLIC

It is unclear what can be concluded from the findings of this study, since the relationship of age at first childbirth and old age longevity appears to represent an epiphenomenon rather than a meaningful association. Shadyab et al. point to the need for further research to help women better plan their pregnancies. A different research design will be needed. To assess the impact of the timing of their first delivery on longevity, women need to know their overall chances of reaching age 90 years if their first birth occurs earlier or later in their reproductive years. It is not very helpful for a woman to know her chances conditional on whether she survives to age 70 years or beyond. Data sources other than the WHI can enable better tests of the life-course consequences of having a child at an older age. Sources include national health registries that span the full life course for the entire population of a given country. As one example, Cohen et al.⁵ linked data from the Danish Medical Birth Registry, the National Patient Registry, and the Register of Causes of Death to test the hypothesis that mothers giving birth to offspring with congenital abnormalities would have greater mortality risk both in the first 10 years following the birth and over a longer time span. Although there are limitations in generalizing from other countries to US populations, such data could determine whether there is a unique relationship of childbearing age and survival that emerges only late in life.

Population registries that cover long time periods could also assess the robustness of findings over time. Women in the WHI gave birth during an unusual historical period. Typical participants would have been 25 years old between 1943 and 1948—during World War II and the start of the baby boom. Unique selection effects operating during that time, in addition to differences in medical practices, may limit the transportability of findings.

The implications of this study for public health are constrained by biases introduced by faulty sample selection, immortal time bias, and residual confounding. Surviving first childbirth at a relatively older age is more likely to be an indicator of health than a contributor to it. The authors acknowledge, “Our findings do not imply that intentionally delaying childbearing will increase the likelihood of living to age 90 years and do not support delaying child-bearing, given the complications associated with older maternal age.”^1(p119) Rather, their research points more strongly to the potential value of understanding factors associated with surviving a first pregnancy at a relatively older age.