Fertility history and health in later life: a record linkage study in England and Wales

https://doi.org/10.1016/j.socscimed.2004.11.046Get rights and content

Abstract

Women born at different periods within the 20th century in England and Wales have followed varying fertility pathways with large changes in, for example, proportions having no children or only one child. Among the consequences of these changes may be effects on women's health later in life. Links between fertility histories and later health and mortality have been investigated in several studies, but in many of these socio-economic characteristics have not been allowed for, even though there are socio-economic differences in both fertility and mortality patterns and results are conflicting. Here we analyse associations between the fertility histories of women born 1911–1940 in England and Wales and their mortality and health status after age 50. We used data from the Office for National Statistics Longitudinal Study; a record linkage study of approximately 1% of the population initially based on those enumerated in the 1971 Census of England and Wales. We used survival analysis to investigate the effects of parity, short birth intervals, and timing of fertility on mortality from age 50 to the end of 2000, controlling for a range of relevant socio-demographic characteristics. For survivors to 1991, we additionally used logistic regression to model probability of having a limiting long-term illness in 1991. We found that nulliparous women and women with five or more children had significantly higher mortality than other women, and that in the oldest groups women with just one child also had raised mortality. Women who had been teenage mothers had higher mortality and higher odds of poor health than other parous women. Mothers with short birth intervals, including mothers of twins, also had elevated risks in some cohorts. Late childbearing (after age 39) was associated with lower mortality.

Personal demographic history is an important factor to consider in analyses of health and mortality variations in later life. More research is needed to further elucidate causal pathways.

Introduction

The importance of life course influences on health and mortality differentials in later life is increasingly recognised, both because past experiences exert strong influences on current circumstances and because accumulated stresses may have long-term influences (Wadsworth, 1997; Blane, 1999; Kuh & Hardy, 2002). Parenthood represents a major domain of most people's lives with substantial short and long-term implications. Apart from the physiological and psychological effects of pregnancy and childbirth, the health of both women and men may be influenced by stresses, role changes and changes in allocation of personal and family resources associated with childrearing and by the emotional and social support benefits of parenthood. Whether the long-term balance of these influences has negative or positive effects on health in later life is likely to depend both on the fertility pathways followed and on the context in which these take place. In this paper we analyse associations between the fertility histories of women in England and Wales born between 1911 and 1940 and their mortality and health status after the age of 50.

There are a number of mechanisms which may underlie associations between fertility histories and health in later life. These include direct biological effects, selection effects, and indirect effects such as the relative costs and benefits of childrearing.

The direct effects of fertility on the later life health of women include the physiological consequences of pregnancy and childbirth. Apart from the risks of childbirth itself, childbearing in historical populations and contemporary less developed countries, has been associated with risks of both short and long term morbidity (Riley, 2003; Menken, Duffy, & Kuhn, 2003). Additionally, there are well-established linkages between some aspects of fertility patterns and certain diseases, most notably breast cancer (associated with nulliparity and delayed motherhood, Madigan, Ziegler, Benichou, Byrne, & Hoover, 1995). More recently reproductive histories have been linked with a range of later life chronic diseases, although it is unclear whether reproductive characteristics are risk factors for chronic diseases in their own right or whether they ‘merely reflect earlier exposures in childhood and adolescence that are the direct determinants of chronic disease’(Rich-Edwards, 2002).

Selection effects are undoubtedly important as socio-economic characteristics are strongly associated with nuptiality and fertility patterns on the one hand and with health and mortality on the other, although both the strength and the direction of these associations have varied over time (Ni Bhrolchain, 1993; Donkin, Goldblatt, & Lynch, 2002). Possible ‘reverse causation’ effects on fertility, including timing of motherhood, need consideration as well. Poor health may restrict opportunities for marriage and parenthood and decisions about subsequent childbearing (Kiernan (1988), Kiernan (1989)). Ability to conceive and deliver a child after the age of 40 may be an indicator of ‘slower’ ageing and overall better health (Snowdon et al., 1989) and in some populations variations in teenage fecundity, influenced by childhood nutritional status, may also be relevant. The effect of these selective processes will certainly have varied over time and between social groups. For example, a comparative study of teenage motherhood based on analyses of the 1946 and 1958 British birth cohort studies showed that teenage mothers in the later cohort, when teen childbearing was rarer, had more unfavourable antecedent characteristics than in the 1946 cohort (Maughan & Lindelow, 1997). Availability of medical care and access to modern methods of birth control are also likely to have resulted in important differences in the operation of these selective processes.

Parenting experiences more generally may have consequences for health. Parenting has many benefits, but also costs. On the positive side parenthood may enhance social networks, provide a positive social role and a potential important source of social support, especially in the longer term (Grundy & Shelton, 2001). The costs of parenthood include reduced opportunities for fulfilment of other roles, the stresses associated with rearing children, and substantial economic costs (Graham, 1984; Joshi, 2002). The impact of these stresses is likely to vary with socio-economic circumstances. For example, research has shown high rates of depression and stress in lone mothers but not in married mothers whose mental and general health is better than that of non-mothers (Harrison, Barrow, Gask, & Creed, 1999). Certain fertility pathways, particularly early childbearing and closely spaced births, may increase both physiological and social and psychological stresses and are known to increase risks of partnership disruption which itself may have negative consequences for health (Murphy, 1984).

Finally some evolutionary theories of ageing posit a negative relationship between fertility and longevity reflecting ‘trade offs’ in investment in younger-age reproductive ‘fitness’ and somatic maintenance in the post-reproductive period (Kirkwood & Rose, 1991).

Given the complexity of these hypothesised links between fertility and health, it is perhaps not surprising that results from empirical studies are in some respects confusing or conflicting. These ambiguities may also reflect the range of methods and populations studied and inability in many of them to control for important interactions with socio-economic and socio-demographic characteristics. Several studies of historical populations have reported links between parity, especially high parity, and later life mortality. Westendorp and Kirkwood (1998) reported a negative association between number of children borne and mortality after the age of 60 for both female and male members of the English aristocracy, although concerns about the quality of the data used in this analysis have been raised (Gavrilov & Gavrilova, 2002). An analysis of the mortality of Southern Californians born 1880–1929 also showed a negative relationship between parity and survival among women, but not men (Friedlander, 1996). A study of Danish twins born 1893–1923 used dentition at age 73 or over as the outcome measure (tooth loss being a well established indicator of cumulative nutritional strain) and found a positive association between parity and tooth loss—those with more children had fewer remaining teeth—and an interaction with socio-economic status (Christensen, Gaist, Jeune, & Vaupel, 1998). Other historical studies, however, have found multiparous women had better post menopausal survival than women with no or only one child or that the direction of the association depended on whether duration of marriage was taken into account (Lycett, Dunbar, & Voland, 2000; Muller, Chiou, Carey, & Wang, 2002).

Results from studies of more recent populations are also sometimes conflicting. Some report higher mortality among parous than nulliparous women; others the reverse (Beral, 1985; Green, Beral, & Moser, 1988). Several studies suggest nulliparity and high parity are associated with higher risk (Kvale, Heuch, & Nilssen, 1994). These, include research by Doblhammer (2000), whose analysis was based on Office for National Statistics (ONS) Longitudinal Study data for England and Wales and the 1981 Austrian census with linked 1 year follow up of deaths. However, this study was restricted to ever-married women (as were several of the others cited above) and the only socio-economic indicator allowed for was presence/absence of higher-level educational qualifications which is a poor discriminator of socio-economic status in the current older female population (Grundy and Holt, 2000). Moreover only three parity groups were contrasted; the nulliparous, those with one or two children, and mothers of three or more children. Kington, Lillard, and Rogowski (1997) in a study of a contemporary US population, which they reported was the first in which associations between reproductive history and health, rather than mortality, were examined found that higher parity women aged 50 and over had worse health compared with nulliparous or low parity (1–2) women but emphasised that their findings needed verification using other data. Grundy and Holt (2000) using data from the British ONS Retirement Surveys, found that parity was associated with health and disability in early old age, but sample size was relatively small and findings were suggestive rather than conclusive.

Timing of childbearing, as well as number of children, has been associated with health in later life in some studies. Very early childbearing has been linked with a range of negative outcomes, including poorer health (Maughan & Lindelow, 1997) and, in poorer countries, impaired growth (Riley, 1994). Few studies have looked beyond health in early or mid life, although Grundy and Holt (2000) found that self reported health among women aged 55–74 was negatively associated with birth before age 23 (taking account of socio-economic circumstances) and in Doblhammer's analysis ever married women who gave birth before 20 had a higher mortality risk at ages 50–85 than other parous women. The possible effects of late childbearing on health and mortality at older ages are more controversial. In the study of English aristocrats referred to above, women who gave birth after the age of 40 had better survival after age 60 (Westendorp & Kirkwood, 1998); a number of other studies have reported similar findings (Perls, Alpert, & Fretts, 1997). This includes a recent study of very old women and men in China in which childbearing after age 35 or 40, particularly having several children after these ages, was positively associated with both good health and survival from ages 80–85 to very old age (Yi & Vaupel, 2004). In contrast, a recent study from the US reported associations between late childbearing (first birth after age 35) and a range of adverse health indicators after the age of 50 including diabetes, hypertension, congestive heart failure, poor dental health, vision difficulties and impaired physical mobility (Alonso, 2002). Consistent with this, a long-term follow up of college women in the US found that women who gave birth after the age of 40 had the highest mortality risk (Cooper, Baird, Weinberg, Ephross, & Sandler, 2000).

One further aspect of fertility history which may be associated with health in later life is birth spacing. In the developing country context, it has been hypothesised that closely spaced births may be associated with a risk of maternal depletion and consequent poorer health, although evidence on this is inconclusive (Menken et al., 2003). Longer term implications of short birth intervals have not, however, been examined.

Improving our knowledge and understanding of associations between fertility histories and health and mortality in mid and later life is particularly important in the current context of population ageing and substantial changes in partnership and parenting histories over the course of the 20th century in England and Wales and elsewhere. Period total fertility rates (TFRs) in England and Wales fell to a low point of 1.8 in the second half of the 1930s before rising to a peak of 2.9 in 1964 and then falling again to a second low point in the second half of the 1970s (OPCS, 1987). A slight ‘recovery’ in fertility in the early 1980s was followed by a further fall and in 2001 the TFR fell to the lowest level ever recorded (1.64). Viewed from a cohort perspective, there have been large changes in the proportions never-marrying, remaining childless and, among parous women, in distribution by family size and in timing of fertility. Thus some 15% of women born in 1900 were never-married by age 50 and results from the 1951 census suggest that as many as a third of women in these very early twentieth century cohorts remained childless (Werner, 1987; Grundy, 1996). Among those born in 1935, by contrast, fewer than 5% were never-married by age 50 and only 11% remained childless. Changes in the timing of fertility have also been substantial. Age specific fertility rates for women aged 40 and over exceeded those of teenagers, often substantially, until the 1940s. By 1971, however, teenage age specific fertility rates (51 births per 1000) were five times higher than those of the 40 and over group (9 per 1000) (Werner, 1987). More recent cohorts show a ‘return’ both to later ages of childbearing and a high prevalence of childlessness, but in the context of more diverse, and disrupted, partnership arrangements.

These historical changes mean that there are considerable differences in the fertility histories of women who are now of advanced age, in early old age or mid life. The consequences of these variations for support in old age are recognised to be important. In this paper we investigate whether there are also consequences for health status in later life, which is a major influence on needs for support. We use data from a nationally representative record linkage study from England and Wales, the Office for National Statistics Longitudinal Study (ONS LS) to analyse associations between fertility history and an indicator of health status in mid or later life and mortality after age 50 among women born between 1911 and 1940. As reported above, this data set was used by Doblhammer (2000) to investigate the mortality of 56,000 ever-married women aged 40–59 in 1971 during the period 1971–1996. Our analysis extends this work in several important ways. Firstly, we include more adequate control for socio-economic characteristics by using measures of housing tenure, car access and occupationally based social class, as well as the educational qualification indicator used in Doblhammer's analysis (which is of very limited use as 92% of the women in the cohorts considered in Doblhammer's analysis fell into the ‘basic’ qualification group); we also control both for baseline marital status and for subsequent widowhood. Secondly we include all women, not just those who were ever-married in 1971 and so our analysis of associations between nulliparity and later health compares all nulliparous women with all parous women, rather than just nulliparous ever-married women with parous ever-married women. Thirdly, by making use of linked birth data we extend the analysis to all women born between 1911 and 1940 and alive in 1971; use of linked birth data also improves capture of births after 40 among the earlier cohorts. Fourthly, we examine the possible effect of short birth intervals on later life mortality and health and analyse associations between parity and mortality in greater detail, rather than using aggregated categories. Thus we compare women with 0, 1, 2, 3, 4, and 5 or more births, rather than collapsing parous women into those with one or two births and those with three or more. Fifthly, as well as analysing associations between teenage and late (>39) childbearing, we also examine variations according to age at first and last birth across the distribution. Finally, we include presence of limiting long-standing illness in 1991 as an outcome measure, as well as mortality 1971–2000.

The specific research questions addressed are whether early childbearing and closely spaced births are associated with worse health outcomes in mid and later life, even after control for socio-economic characteristics; whether overall parity is associated with later life health outcomes (again allowing for socio-economic characteristics); whether late childbearing has positive or negative associations with later health and mortality and whether these effects vary between cohorts.

Section snippets

Data and methods

The ONS LS is a record linkage study of approximately 1% of the population initially based on those enumerated in the 1971 Census of England and Wales (approximately 500,000 people). Sample members were traced in the National Health Services Central Register (NHSCR) and record linkage used to add information from subsequent censuses and from vital registration, including births to sample mothers, death of spouse, and death. The LS has been maintained through the addition of 1% of new births and

Fertility characteristics and differentials in survival

In this section, we consider the fertility characteristics of the cohorts of interest and, in order to assess any bias arising from linkage failure and the other problems referred to above, make some comparisons with results from other sources. We then present the results of survival analysis undertaken for each cohort separately and then for a pooled cohort observed from ages 50–69. Table 3 shows indicators of the fertility histories of the three cohorts considered and also the odds ratios for

Discussion

This analysis of data from a large nationally representative study of women in England and Wales lends support to hypotheses of links between fertility histories and health in mid and later life. In analyses including controls for indicators of socio-economic position and socio-demographic status, we found higher risks of adverse outcomes among nulliparous than parous women; higher risks associated with high parity and closely spaced births; and associations between teenage childbearing and

Acknowledgements

The work reported here was supported by the Economic and Social Research Council, UK, Grant Reference Res-000-0394. We are grateful to the Office for National Statistics for access to data from the ONS Longitudinal Study; the project was approved by the Longitudinal Study Research Board.

References (43)

  • K. Christensen et al.

    A tooth per child?

    Lancet

    (1998)
  • E. Grundy et al.

    Adult life experiences and health in early old age in Great Britain

    Social Science & Medicine

    (2000)
  • J.H. Hibbard et al.

    Effects of domestic and occupational roles on morbidity and mortality

    Social Science & Medicine

    (1991)
  • P.K. Adelman

    Multiple roles and physical health among older adults

    Research on Aging

    (1994)
  • A. Alonso

    Long term health consequences of delayed childbirthNHANES III

    Women's Health Issues

    (2002)
  • V. Beral

    Long term effects of childbearing on health

    Journal of Epidemiology and Community Health

    (1985)
  • D. Blane
  • G. Cooper et al.

    Age at menopause and childbearing patterns in relation to mortality

    American Journal of Epidemiology

    (2000)
  • G. Doblhammer

    Reproductive history and mortality later in lifea comparative study of England and Wales and Austria

    Population Studies

    (2000)
  • A. Donkin et al.

    Inequalities in life expectancy by social class 1972–1999

    Health Statistics Quarterly

    (2002)
  • N. Friedlander

    The relation of lifetime reproduction to survivorship in women and mena prospective study

    American Journal of Human Biology

    (1996)
  • L.A. Gavrilov et al.

    Evolutionary theories of aging and longevity

    The Scientific World Journal

    (2002)
  • H. Graham

    Women, health and the family

    (1984)
  • A. Green et al.

    Mortality in women in relation to their childbearing history

    British Medical Journal

    (1988)
  • E. Grundy et al.

    Contact between adult children and their parents in Great Britain 1986–1999

    Environment and Planning A

    (2001)
  • E.M.D. Grundy

    Population reviewthe population aged 60 and over

    Population Trends

    (1996)
  • J. Harrison et al.

    Social determinants of GHQ score by postal survey

    Journal of Public Health Medicine

    (1999)
  • L. Hattersley et al.

    Longitudinal Study 1971–1991. History, organisation and quality of data

    (1995)
  • H. Joshi

    Production, reproduction and educationwomen, children and work

    Population and Development Review

    (2002)
  • K. Kiernan

    Who remains celibate?

    Journal of Biosocial Science

    (1988)
  • K. Kiernan

    Who remains childless?

    Journal of Biosocial Science

    (1989)
  • Cited by (155)

    View all citing articles on Scopus
    View full text