Abstract
Childbearing has been increasingly delayed in Western countries. As older mothers are likely to be advantaged over their younger counterparts, the demographic literature has tended to view this demographic trend as potentially beneficial for child well-being. Conversely, less attention has been given to medical studies showing that giving birth at advanced ages is associated with health risks for children. This article uses data from the Millennium Cohort Study (UK) and ordinary least squares and logistic regression models to compare cognitive and behavioral outcomes, and obesity at age 5 for first-born children by maternal age at first birth. On one hand, the findings suggest that giving birth to the first child at ages 30–39 compared to ages 23–29 is positively associated with children’s cognitive and behavioral outcomes and not associated with obesity. On the other hand, delaying the first birth to ages 40 and above is not associated with children’s cognitive and behavioral outcomes and is associated with increased risk of obesity. Although the results are unable to support the argument that this occurs because of the health risks associated with giving birth at advanced maternal ages, they suggest that there is a need to more closely investigate the potential trade-offs involved when births are delayed toward older maternal ages.
Notes
1 Child well-being, when used in general terms throughout this article, refers to different dimensions of well-being that are associated with development in childhood as well as well-being later in life. These dimensions of well-being would most likely reflect cognitive, socio-emotional, and health outcomes.
2 The authors, using the MCS, analyze the association between maternal age and children’s cognitive and behavioral outcomes. The older age group comprises all mothers who gave birth at ages 30 and above since the aim of the paper is to establish whether there is an inherent disadvantage to having a young mother. The focus is therefore on young maternal ages at first birth rather than older maternal ages, as in the present study.
3 Using contemporary data, Savage et al. (Citation2013) show that children of mothers who gave birth at ages 30–35 and above were at lower risk of obesity than children of younger mothers, but the sample was small (n = 277), while in contrast Sutcliffe et al. (Citation2012), using data from the MCS, show that children of mothers aged 33 and above are more likely to be overweight. In both studies, first births and mothers aged 40 and above were not analyzed separately. Myrskylä and Fenelon (Citation2012) show that children of mothers aged 35 and above have worse health in adulthood than children of younger mothers. Their analyses are based on cohorts born in the 1950s, when the selection into childbearing at an older maternal age was different from what we observe in contemporary developed contexts.
4 For example, having a second- or third-order birth at an older age does not necessarily reflect the same level of resource accumulation that it does in having a first-order birth at an older age. The health processes might also differ when looking at higher-order births to the extent that the health of the mother might deteriorate after she has a first- or higher-order birth.
5 The analyses have been replicated using similar outcomes measured when the cohort children were aged 3 and 7. The results are qualitatively similar to those presented in the article.
6 The SDQ consists of the main respondent’s report of 25 items, grouped into 5 categories, which measure the child’s conduct problems, hyperactivity, emotional symptoms, and pro-social behavior.
7 In families with twin cohort children, only one child (i.e., the first one reported in the data) per family has been included in the analyses.
8 This categorization is based on a derived variable in the dataset that groups respondents according to National Vocational Qualifications (NVQ), which include both academic and vocational qualifications. To ease interpretation of these categories for a non–UK audience, the categories are coded based on the international ISCED qualification levels.
9 As a robustness check, models have also been estimated controlling for income and family structure at age 5, and the results are essentially unchanged.
10 Controlling for whether other languages (in addition to English) are spoken at home does not change the results substantively.
11 In line with the descriptive analyses, the 35–39 age coefficient is smaller than the 30–34 one, but the difference between the two coefficients is small and not significant.
12 In line with the descriptive analyses shown in , the 35–39 coefficient is larger than the 30–34 one, but the difference is not significant.
13 The two variables are significantly correlated (p < .000), but controlling for them separately doesn’t change the results.
14 It is not possible to look at very low birth weight as an indicator of child health. Its prevalence is too low, and the small sample of mothers giving birth in the oldest age groups make the analyses unfeasible.