AMERICAN JOURNAL OF HUMAN BIOLOGY 21:541–547 (2009)
Original Research Article
Women’s Fertility and Mortality in Late Mid Life: A Comparison
of Three Contemporary Populations
EMILY GRUNDY*
Centre for Population Studies, London School of Hygiene and Tropical Medicine, London WC1A 3DP, United Kingdom
ABSTRACT
Evolutionary theory suggests a trade-off between reproduction and somatic maintenance implying a
negative relationship between parity and longevity, at least in natural fertility populations. In populations in which
fertility control is usual, there are also a number of mechanisms that may link reproductive careers and later mortality,
but evidence of associations between women’s fertility patterns and their later life health has been judged inconclusive
due to varying controls for socio-economic characteristics and marital status. Here, we build on three recent studies
that followed a common framework to investigate associations between women’s parity and timing of first and last birth
with mortality in late middle age in three contemporary developed counties, Norway, England and Wales, and the USA.
Data were drawn from whole population registers (Norway); a large census-based record linkage study (England and
Wales), and a nationally representative survey linked to death records (USA). Results show that teenage childbirth was
associated with higher mortality risks in late middle age in all three countries. Risks of death were significantly raised
among nulliparous women in Norway and England and Wales, and also raised (although not significantly so) for childless US women. However, although higher parity was associated with a slight mortality disadvantage in England and
Wales and the USA, the reverse seemed the case in Norway. These finding suggest that in populations in which fertility
control is usual, contextual factors influencing the relative costs and benefits of childbearing may influence associations
between fertility histories and later mortality. Am. J. Hum. Biol. 21:541–547, 2009.
' 2009 Wiley-Liss, Inc.
A growing body of research has shown associations
between one important aspect of women’s lives—fertility
history—and their postreproductive mortality but our
understanding of underlying mechanisms, which in some
cases may be offsetting, is still limited (Beral, 1985;
Doblhammer, 2000; Friedlander, 1996; Hurt et al., 2006;
Kvale et al., 1994; Lund et al., 1990). Theories from evolutionary biology suggest trade-offs between reproduction
and investment in somatic maintenance that would imply
inverse associations between fertility and longevity
(Gagnon et al., 2008; Kirkwood and Rose, 1991; Smith
et al., 2002). However, results from studies of natural fertility populations are not all consistent with this theory and,
as discussed elsewhere in this volume (Gagnon et al., 2009),
their interpretation is complicated by data comparability
issues and varying allowance for selective influences. In
contemporary developed societies in which deliberate control of fertility is widespread, understanding of possible
mechanisms linking fertility and later mortality is arguably
even more complex. In both ‘‘natural’’ and ‘‘Malthusian’’
populations, a range of biosocial factors influence fecundity
and fertility, including opportunities for and timing of marriage, duration of marriage (largely mortality driven), and
nutritional, biological or behavioral factors influencing
fecundity, such as breastfeeding practices. In societies in
which birth control is usual, individual or couple-based
decision-making about the timing and number of births is
also important and will be influenced by the perceived costs
and benefits of childbearing and rearing. These are likely to
vary with context, as may the implications of fertility
patterns for women’s later life health and mortality.
FERTILITY PATTERNS, HEALTH, AND MORTALITY
Specific physiological mechanisms may link fertility
patterns and subsequent health and disease. For example,
C 2009
V
Wiley-Liss, Inc.
pregnancy, childbirth, and lactation are associated with
hormonal and other physiological changes that may
affect risks of developing cancers of the female reproductive system and also other cancers (Kabat et al., 2007;
Russo and Russo, 2007; Salehi et al., 2008). Epidemiological studies show an inverse relationship between
parity and the incidence of breast, ovarian and uterine
cancer with later childbearing conferring additional
risks of breast cancer (Lochen and Lund, 1997; Russo
and Russo, 2007; Salehi et al., 2008). Pregnancy is also
a state of relative insulin resistance, and repeated pregnancies may result in permanent deficiencies in lipid
and glucose metabolism and higher risks of diabetes and
cardiovascular disease (Kvale et al., 1994; Lawlor et al.,
2003). Pregnancy additionally involves calcium loss and
other nutritional challenges, oxidative stress, and
reduced immunological resistance to infectious diseases
(Christensen et al., 1998), although the implications of
these for later health and mortality may be less relevant
in well-nourished populations.
Apart from these effects directly related to pregnancy
and childbirth, women’s health may be influenced by a
range of other positive and negative stresses and supports
associated with parenthood. On the negative side, mother-
Contract grant sponsor: UK Economic and Social Research Council;
Contract grant numbers: RES-000-0394, RES-348-25-0004; Contract grant
sponsors: Norwegian Centre for Advanced Study, Centre for Longitudinal
Study Information and User Support service (CeLSIUS).
*Correspondence to: Emily Grundy, Centre for Population Studies,
London School of Hygiene and Tropical Medicine, 49-50 Bedford Square,
London WC1A 3DP, UK. E-mail: emily.grundy@lshtm.ac.uk
Received 8 February 2009; Accepted 27 March 2009
DOI 10.1002/ajhb.20953
Published online 5 May 2009 in Wiley InterScience (www.interscience.
wiley.com).
542
E. GRUNDY
hood involves considerable economic outlays (Joshi, 2002),
which might be expected to have adverse effects especially
for those on low incomes, and the potential for role overload and stress related to children (D’Elio et al., 1997;
Elstad, 1996). More positively, motherhood may provide
an incentive to adopt healthier behaviors, provide
enhanced opportunities for social interaction, and pave
the way for social support from children in later life, all of
which should be health enhancing (Antonucci et al., 2003;
Kendig et al., 2007). The relative balance of negative and
positive effects on health is likely to vary according to the
context and particular fertility pathways. Lone mothers in
Britain, for example, have been found to have higher risks
of depression than childless women while the mental
health of married mothers is better than that of the childless (Harrison et al., 1999).
In addition to physiological and biosocial pathways linking fertility histories and mortality, a range of selective
influences and possible confounding factors also need consideration. Antecedent serious health problems or disabilities, for example, may preclude or reduce chances of partnership and parenthood, as well as increasing risks of
death (Kiernan, 1989). Health-related behaviors may
themselves influence fecundity and fertility, and both fertility and mortality vary considerably by indicators of
socio-economic status such as education (Kravdal and
Rindfuss, 2008; Ni Brolchain, 1993). Lower parental socioeconomic status, experience of family disruption, lower
educational attainment, and high propensities for risk
taking have been shown to be associated with earlier
entry into sexual activity, partnership, and parenthood
(Maughan and Lindelow, 1997); which in turn is associated with higher overall parity; these and other early life
characteristics are also associated with mortality differentials. Late childbearing, conversely, may be indicative of a
slower pace of ageing influencing fecundity (Cooper et al.,
2000; Jacobsen et al., 2003) and, perhaps more pertinently
in contracepting populations, decisions about having children at relatively late ages may by influenced by perceptions of health.
Most previous studies of contemporary developed country populations suggest a J- or U-shaped relationship
between parity and women’s postreproductive mortality
with higher risks for the childless (and in some studies
those with only one child) than for mothers of two
children and also excess mortality for high parity women
(Doblhammer, 2000; Kvale et al., 1994). However, as with
studies of natural fertility populations, there are inconsistencies in results and the authors of a systematic review
of studies up to 2005 considered that the evidence was far
from conclusive due to varying control for socio-economic
factors and marital status and differences in the parity
groups compared (Hurt et al., 2006). Several studies
including controls for socio-economic and marital status
have been reported subsequently, these include three, of
England and Wales, Norway, and the USA (Grundy and
Kravdal, 2008; Grundy and Tomassini, 2005; Henretta,
2007), conducted as part of two linked projects in which
common definitions and analytical approaches were used.
In this article, we present comparative results from these
studies in order to explore possible contextual effects on
associations between reproductive behavior and later
mortality in contemporary developed societies. We first
briefly consider the demographic and social policy context
in the three countries.
American Journal of Human Biology
FERTILITY AND MORTALITY TRENDS
As shown in Figure 1, the three countries considered
have experienced broadly similar trends in fertility and in
female life expectancy over the past century. All had falling fertility in the first decades of the twentieth century,
reaching a low point in the 1930s, followed by a post-war
baby boom and then a return to low fertility in the 1980s
and 1990s. However, the USA post-war baby boom started
earlier and was more substantial, both in terms of level
and in duration, than in either European country and for
most of this period Norwegian fertility rates were slightly
higher than those in England and Wales. Trends in female
mortality have also been similar, although Norwegian
women have generally enjoyed slightly higher life expectancy at age 10 than the other groups, particularly nonWhite US females (Fig. 2).
The three countries also differ with regard to welfare
state arrangements and social policies, which may be important determinants of health and health inequalities in
advanced industrial societies as they mediate the extent,
and impact, of socio-economic status on health (Bambra,
2006). Such policies may also have an influence on the
relative costs and benefits of childbearing and childrearing (Aassve et al., 2005; Curtis and Phipps, 2004), and so
potentially on the association between fertility pathways
and later mortality. Norway is one of the countries identified by Esping-Andersen (1990) as constituting the ‘‘social
democratic’’ type of regime with a generous welfare system underpinned by a commitment to gender equality and
individual entitlement (Rønsen, 2004). The USA belongs
to the liberal tradition with a greater emphasis on means
tested benefits for the most marginalized; the UK lies
somewhat in between with more supports for families
(including free health care) than the US, but fewer than
in Norway (Esping-Andersen, 1990). These differences
would suggest that the ‘‘costs’’ of childbearing and rearing
in Norway would be the lowest, and those in the USA the
highest among the countries considered here.
DATA AND METHODS
Published and unpublished results from three data sets
are used to compare associations between women’s fertility and mortality in late mid life. These data sets are population register data for the whole Norwegian population; a
record linkage study of 1% of the population of England
and Wales [the Office for National Statistics Longitudinal
Study (ONS LS)]; and the US Health and Retirement Survey (HRS), a nationally representative sample of the older
United States population linked to mortality records.
Norway
The Norwegian data come principally from the Norwegian Central Population Register, which was established
drawing on the 1960 Census and subsequently has been
continuously updated. All Norwegian residents are
included in the register and are assigned a personal identification number used in all dealings with official agencies and many commercial ones. A wide range of other
registers based on the same identification number are also
maintained, including registers of level of education and
of mortality (Longva et al., 1998). For women born
between 1935 and 1958, maternity histories can be
assembled as parents’ identification numbers have been
543
WOMEN’S FERTILITY AND MORTALITY IN LATE MID LIFE
Fig. 1.
Fertility (total fertility rates) in Norway, England and Wales, and the USA from 1911–1915 to 1991–1995.
Fig. 2. Trends in female further life expectancy at age 10, Norway, England and Wales and the USA, 1850–1999. Sources: Human Mortality
Database (HMD; http://www.mortality.org); Carter et al. (2006).
recorded at registration of all births since 1965, when
those included in this analysis were aged 7–30. Earlier
births to the oldest members of the study can be captured
through linkage of parent–child information from the
1970 (and to some extent the 1960) census undertaken by
Statistics Norway. It is possible that some births prior to
1965 to cohort members old enough to have given birth
before then (i.e., themselves born before 1950) would have
been missed if the children were put up for adoption or
died and so were not living with their mothers in 1960 or
1970, but such cases would be few in number. Further
details of the data have been reported elsewhere (Grundy
and Kravdal, 2008). Mortality in the analyses presented
here was restricted to ages above 45 (i.e., no earlier than
1980), when women had largely completed their childbearing, and below 68, the age of the oldest cohort at the
end of follow-up in 2003. In the period considered, fewer
than 5% of men and 3% of women died before age 45 so
these survivors constitute the vast majority of their respective birth cohorts (Statistics Norway, 2006). 744,784
women contributed 23,241 deaths and 7.20 million person-years of follow-up.
American Journal of Human Biology
544
E. GRUNDY
England and Wales
The ONS LS is a record linkage study of 1% of the population initially based on those enumerated in the 1971
Census of England and Wales. Sample members were
traced in the National Health Services Central Register
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 immigrants, but these analyses are based
on follow-up of women from the initial 1971 sample who
were then aged 31–40 (birth cohorts 1931–1940). The
1971 Census included questions on the marital and fertility histories of ever-married women. As only ever-married
women were asked to provide fertility history data and
they were only asked to record births within marriage,
data on nonmarital births prior to 1971 is lacking. However, for cohorts born between 1920 and 1940, nonmarital
fertility generally accounted for only 4–6% of all births in
England and Wales, so the effect of this omission is relatively slight. Subsequent births to sample mothers (of all
marital statuses) are captured through linkage-to-birth
registration records. Some births may have been missed
due to linkage failure, estimated to be around 8–9%
(Hattersley and Creeser, 1995), but comparisons with
other sources of data on completed parity, reported elsewhere, show a good match (Grundy and Tomassini, 2005).
USA
US data are drawn from the first five waves of the
Health and Retirement Study collected between 1992 and
2000. The sample used here consists of the original HRS
cohort born between 1931 and 1941. Data are linked to
the National Death Index (NDI) up to 2000, providing
date of death for deceased respondents. In addition, the
study conducts a proxy interview after a respondent’s
death and collects date of death. The sample is limited to
female age-eligible respondents born between 1931 and
1941, who were first interviewed in 1992 and were also in
the 1994 round when questions on numbers and ages of
living children were fielded. The fertility data used thus
relate to living children rather than children ever born.
Further details have been reported in Henretta (2007)
and Henretta et al. (2008).
The three samples thus differ somewhat in construction
and also considerably in sample size, and so statistical
power.
Variables used in the analysis
Fertility history. Dummy variables were used to identify
those with 0, 1, 2 (reference group), 3, 4 or 5 or more
births. In Norway and England and Wales these refer to
children ever-born, in the USA to living children in 1994.
Two additional dummy variables identified those with a
birth before age 20 and those with a birth at age 40 or
older.
Covariates. The three data sources do not provide the
same information on socio-demographic characteristics,
but covariates were chosen to provide as near as possible
equivalent indicators of socio-demographic circumstances.
These were: age in single years (all countries), marital staAmerican Journal of Human Biology
tus (time varying in Norway; last census/survey point in
England and Wales and the USA with time varying indicators of widowhood), and socio-economic status. Norwegian
data on socio-economic status came from detailed time
varying information on the educational qualifications of
women studied, and their spouses if relevant, distinguishing five levels of educational attainment. In England and
Wales, indicators comprised housing tenure and household access to a car at baseline (1971); possession of a
higher level educational qualification (gained at around or
above age 18) and occupational social class of self and/or
husband. In the USA years of education, race/ethnicity,
and log net worth comprised the indicators used.
Methods. Postreproductive mortality risks were estimated using discrete time event history models for deaths
observed between 1980 and 2003 at ages 45–68 (Norway);
deaths observed between 1980 and 2000 at ages 50–69
(England and Wales); and deaths observed between 1994
and 2000 at ages 53–69 (USA). Following standard procedures, (Allison, 1984), a series of observations for discrete
time periods (in this case 1 year) were created for each
person from the relevant starting point until the event of
interest (death) or censorship (at the end of the follow-up
period or exit through emigration or loss to follow-up).
Observations for all discrete time periods are then pooled.
Logistic models were then fitted using the Proc Logistic
procedure in the SAS software package.
RESULTS
Table 1 presents descriptive information on the fertility
characteristics of the women in the three studies. Overall
fertility was highest in the US sample that included a
much higher proportion (21%) of women with five or more
children than in either England and Wales (7%) or Norway (4%). US women were also the least likely to be childless (8% compared with 15% in England and Wales and
11% in Norway). In the USA women with two or three children were equally numerous, whereas in both England
and Wales and Norway those with two children comprised
the largest group.
Table 2 shows mortality risks in the three populations
by parity and, for the parous, additionally by experience
of a teenage or a late birth (at age 40 or greater). Results
are from fully adjusted models including controls for the
socio-demographic characteristics discussed above. In the
analyses of mortality among all women (top half of the
table), results for nulliparous and low parity women were
broadly similar in the three populations. In both Norway
and England and Wales, nulliparous women had significantly raised mortality risks; in the USA the odds ratio for
nulliparous women was also raised but did not reach
conventional levels of statistical significance. In Norway
odds of death were raised for mothers of one child; again
odds ratios in the USA were similar but not statistically
significant.
Results for higher parity women, however, differ. In
England and Wales the mortality of women with five or
more births was raised, odds ratios for high parity US
women were also raised but did not quite reach conventional levels of statistical significance (P < 0.07). In both
England and Wales and the USA, odds ratios for women
with four births (and in the case of the USA, also women
545
WOMEN’S FERTILITY AND MORTALITY IN LATE MID LIFE
TABLE 1. Distribution of samples by fertility indicators and birth
cohort, England and Wales, Norway, and the USA
England and Wales
Norway
USAa
Birth cohort
All women
TFR
Parity 0 (%)
Parity 1
Parity 2
Parity 3
Parity 4
Parity 51
Parous women
Birth before 20
Birth after age 39
N
1931–1940
2.2
15.2
13.7
33.4
21.1
9.6
7.2
1935–1958
2.3
11.3
11.4
37.3
26.1
9.7
4.2
1931–1941
3.2
8.3
8.8
23.1
23.1
15.8
20.9
7.5
3.1
26,436
13.5
3.1
7.2 Mb
23.0
3.7
4,335
a
Children alive in 1994.
Person years.
Sources: Analysis of ONS LS (Grundy and Tomassini, 2005); Norwegian Registry
data (Grundy and Kravdal, 2008) and US HRS (produced by John Henretta).
b
parous women and including the birth timing variables,
associations between high parity and mortality appear
slightly different than in the analyses including nulliparous women. In Norway, the overall trend of lower mortality with higher parity appears similar in models including
and excluding the nulliparous, but in the model restricted
to the parous this negative association between high
parity (five or more children) and mortality reached conventional levels of statistical significance, in short the
results tend to be slightly stronger than in the model
including all women. In England and Wales and the USA,
the reverse seems to be the case: restricting the model to
mothers and including birth timing variables resulted in
the positive (worse) association between high parity and
mortality ceasing to be significant, that is, the effect of
including the indicators of birth timing and excluding the
nulliparous seems to be to attenuate the association
between higher parity and mortality seem in the model
including all women.
DISCUSSION
TABLE 2. Results from discrete time event history models of mortality
in late mid life by fertility characteristics, women in Norway, England
and Wales, and the USA
All women
0
1
2 (ref)
3
4
51
Parous women
1
2 (ref.)
3
4
51
Birth before 20
Birth after 39
Number of deaths
England and
Wales deaths
1980–2000 at
ages 50–69, birth
cohort 1931–1940
Norway deaths
1980–2003
at ages 45–68,
birth cohort
1935–1958
USA deaths
1994–2000
at ages 53–69,
birth cohort
1931–1941
Odds ratio
1.28**
1.10
1.00
1.01
1.11
1.25**
Odds ratio
1.50**
1.31**
1.00
0.95*
0.95*
0.94
Odds ratio
1.47
1.34
1.00
1.21
1.29
1.41y
1.12
1.00
0.99
1.05
1.12
1.30**
0.94
2,212
1.37**
1.00
0.92**
0.90
0.89**
1.21**
0.86**
23,241
1.46
1.00
1.09
1.04
1.05
1.57**
0.73
329
y
P < 0.10; *P < 0.05; **P < 0.01.
Sources: Analysis of ONS LS (Grundy and Tomassini, 2005; unpublished analyses); Norwegian Registry data (Grundy and Kravdal, 2008) and US HRS (produced by John Henretta).
with three births) were above 1, although again these differences were not statistically significant. Results for Norwegian women, in contrast, showed significantly lower
mortality for those with three or four births compared
with mothers of two—although differences in odds ratios
were not large—and a ratio below 1 for women with five or
more births.
Results of analyses restricted to parous women and
including dummy variables indicating experience of a
teenage and/or a late birth (as well as the socio-demographic controls), are shown in the lower panel of Table 2.
Mothers with a birth before age 20 had higher mortality
than other mothers in all three countries whereas late
childbearing was associated with odds ratios below 1 in all
three countries, but only in Norway was this association
statistically significant. In these analyses restricted to
There are a number of limitations to this comparison,
although efforts were made to derive equivalent indicators
of socio-economic status and the fertility variables were
defined in the same way, there are nevertheless considerable differences in data design and size. Additionally,
deaths were observed at a relatively early stage of postreproductive life and associations with mortality in later old
age might differ [although in the England and Wales
study, mortality patterns were also examined in older
cohorts and were found to be similar to those shown here
(Grundy and Tomassini, 2005)]. However structured comparisons, even those with some limitations as this one,
have the potential to shed light on a topic that has been
confused by the number of studies with apparently conflicting results, as illustrated in the article by Gagnon
et al. also included in this volume. In contrast to the
Gagnon article, however, these results from three contemporary populations in which birth control is usual, are
suggestive both of interesting similarities across the populations considered, and some intriguing differences. First,
nulliparous women seem to be disadvantaged in all three
populations. This may partly reflect various selective
influences, but it is interesting that patterns are similar
despite the variations in the proportions of nulliparous
women in the three data sources. In the USA, only 8% of
the women in the sample were childless compared with
11% of the Norwegian women and 15% of women in England and Wales, in short childless women in the US represent a more selected group. This would suggest that they
might include a larger representation of women with particularly unfavorable characteristics (such as childhood
poor health and/or infecundity) likely to be independently
associated with later mortality risks. In short, if prior
selection effects largely account for the higher mortality of
childless women, we might expect the differential to
greater in the US than in England and Wales (where
childless women accounted for a larger proportion of the
cohort) in much the same way as excess mortality of
never-married people has been shown to be highest in
populations in which the proportions of never-married are
lowest (Hu and Goldman, 1990). Although the odds ratio
of mortality for childless women was indeed higher in the
US than the equivalent figure for women in England and
American Journal of Human Biology
546
E. GRUNDY
Wales (although no higher than for Norwegian women), it
was not significantly raised and there is no clear picture
of differentiation in effect. Further investigation using a
larger US data set, and data from a wider range of
countries varying in the proportions childless, would allow
further investigation of this. Second, teenage motherhood
is also associated with raised mortality risks in later life
in all three populations. Unmeasured early life disadvantages influencing both fertility and mortality are likely to
be important but, as in the case of the comparison of nulliparous women, it is notable that this pattern is evident
throughout despite the very large variations in the proportions of women who had had a teenage birth (23% of
US women compared with 8% of women in England and
Wales). It is interesting that Gagnon et al. (2009) find that
in the two earlier of the natural fertility populations they
consider, Old Quebec and Saguenay-Lac-St-Jean, the
association between age at first birth and mortality is in
the opposite direction, that is, a later age at first birth
increases mortality risk. This may reflect both different
pathways to and different consequences of early motherhood. In modern populations, it is known that teenage
motherhood is associated with early life disadvantages
[although it is still associated with poorer later health
even when these are controlled, as illustrated by Henretta
et al. (2008)]. In preindustrial rural populations, by contrast, it may be that healthier more advantaged women are
able to marry earlier and also conceive sooner because of
nutritional advantages (Gagnon et al., 2008). Moreover in
industrialized societies, but not in preindustrial farm-based
communities, part of the negative effect of early parenthood
may arise from disrupted career progression with consequent negative effects on later socio-economic status.
Late motherhood was associated with odds of death
below 1 in all populations, but only in the large Norwegian
data set (which included the lowest proportion of mothers
with a birth at ages 40 or above) was this statistically significant. Selective factors are likely to be important here
too (in all three populations, the proportion of women having a birth at age 40 or above was very low), but it has
also been suggested that having children who are still
young in later life may have health-promoting effects
(Perls et al., 1997; Yi and Vaupel, 2004).
In terms of differences, the most interesting is the difference between England and Wales and Norway in the
association between high parity and later mortality. In
Norway, there appears to be no high parity ‘‘penalty’’ and
indeed mothers of three or four children had lower mortality than the reference group of mothers of two (although
the difference was not great). Other Scandinavian studies
have also failed to find mortality disadvantages even
among very high parity women. Thus, a Finnish study of
high women with five or more children reported below
average all cause mortality and found no mortality disadvantage even for women who had had ten or more births
(Hinkula et al., 2006). In England and Wales (and in the
USA), by contrast, women with five or more births had
raised mortality. This difference might reflect some variation in the composition of the high parity groups not
adequately controlled for using the covariates available.
The fact that the association between high parity and
higher mortality weakened (to nonsignificance) in
England and Wales but, if anything, appeared slightly
stronger in Norway in models restricted to parous women
and including birth timing variables lends some support
American Journal of Human Biology
to this possibility as an early start to childbearing is positively associated with overall parity. However, it is also
possible that the more generous supports available for
parents in Norway tip the balance between the costs and
benefits of childbearing toward the benefits. In short,
associations between fertility and later mortality in modern populations in which birth control is usual, may
depend on context and reflect a range of biosocial, rather
than strictly biological, factors. Further investigations
using both large scale demographic data sets and also
data sets including behavioral and biomarker data might
shed further light on this. The first could be used to see
whether associations between parity and mortality vary
between countries with different social policy regimes, but
more detailed data are needed to elucidate biosocial pathways linking reproduction and health. It would be interesting, for example, to examine variations by parity in
indicators such as cortisol levels in different populations
and also to investigate whether associations between
reproductive pathways and health-related behaviors vary.
Further comparative approaches using common frameworks and definitions are also important to enable the
best to be made of the data currently available.
ACKNOWLEDGMENTS
Results in this paper draw on collaborative work with
Cecilia Tomassini and John Henretta and Oystein Kravdal. Analysis of data from the Office for National Statistics
Longitudinal Study of England and Wales was supported
by the Centre for Longitudinal Study Information and
User Support service (CeLSIUS) at the London School of
Hygiene & Tropical Medicine, UK. The paper was presented at the IUSSP International Seminar on Trade-offs
in female life histories: Raising new questions in an integrative framework, Bristol UK 23–25 July 2008 and has
benefited from comments made then and by the referees
to whom the author is very grateful.
LITERATURE CITED
Aassve A, Mazzuco S, Mencarini L. 2005. Childbearing and well-being: a
comparative analysis of European welfare regimes. J Eur Soc Policy
15:283–289.
Allison PD. 1984. Event history analysis: regression for longitudinal event
data. Newbury Park, CA: Sage Publications.
Antonucci TC, Arjouch KJ, Janevic MR. 2003. The effect of social relations
with children on the education-health link in men and women aged 40
and over. Soc Sci Med 56:949–960.
Bambra C. 2006. Health status and the worlds of welfare. Soc Policy Soc
5:53–62.
Beral V. 1985. Long term effects of childbearing on health. J Epidemiol
Community Health 39:343–346.
Carter SB, Gartner SS, Haines MR, Olmstead AL, Sutch R, Wright G, editors. 2006. Historical statistics of the United States. New York: Cambridge University Press.
Christensen K, Gaist D, Jeune B, Vaupel J. 1998. A tooth per child? Lancet
352:204.
Cooper G, Baird D, Weinberg C, Ephross S, Sandler D. 2000. Age at menopause and childbearing patterns in relation to mortality. Am J Epidemiol 151:620–623.
Curtis L, Phipps S. 2004. Social transfers and the health status of mothers
in Norway and Canada. Soc Sci Med 58:2499–2507.
D’Elio M, Ness R, Matthews K, Kuller L. 1997. Are life stress and social
support related to parity in women? Behav Med 23:87–94.
Doblhammer G. 2000. Reproductive history and mortality later in life: a
comparative study of England and Wales and Austria. Popul Stud
54:169–176.
Elstad JI. 1996. Inequalities in health related to women’s marital, parental
and employment status—a comparison between the early 1970s and the
late 1980s, Norway. Soc Sci Med 42:75–89.
WOMEN’S FERTILITY AND MORTALITY IN LATE MID LIFE
Esping-Andersen G. 1990. Three worlds of welfare capitalism. Cambridge:
Polity Press.
Friedlander N. 1996. The relation of lifetime reproduction to survivorship
in women and men: a prospective study. Am J Hum Biol 8:771–783.
Gagnon A, Mazan R, Desjardins B, Smith K. 2008. Post reproductive
survival in a natural fertility population. In: Bengtsson T, Mineau GP,
editors. Kinship and demographic behaviour in the past. Dordrecht:
Springer. p 225–241.
Gagnon A, Smith K, Tremblay M, Vézina H, Paré PP, Desjardins B. 2009.
Is there a trade-off between fertility and longevity? A comparative study
of three large historical demographic databases accounting for mortality
selection. Am J Hum Biol. DOI 10.1002/ajhb20893.
Grundy E, Kravdal Ø. 2008. Reproductive history and mortality in late
middle age among Norwegian men and women. Am J Epidemiol
167:271–279.
Grundy E, Tomassini C. 2005. Fertility history and health in later life: a
record linkage study in England and Wales. Soc Sci Med 61:217–228.
Harrison J, Barrow S, Gask L, Creed F. 1999. Social determinants of GHQ
score by postal survey. J Public Health Med 21:283–288.
Hattersley L, Creeser R. 1995. Longitudinal study 1971–1991. History,
organisation and quality of data. London: HMSO.
Henretta J. 2007. Early childbearing history, marital status and women’s
health and mortality after age 50. J Health Soc Behav 48:254–266.
Henretta JC, Grundy EMD, Okell LC, Wadsworth MEJ. 2008. Early motherhood and mental health in midlife: a study of British and American
cohorts. Aging Ment Health 12:605–614.
Hinkula M, Kauppila A, Nãyhã S, Pukkala E. 2006. Cause-specific mortality
of grand multiparous women in Finland. Am J Epidemiol 163:367–373.
Hu YR, Goldman N. 1990. Mortality differentials by marital status: an
international comparison. Demography 27:233–250.
Hurt LS, Ronsmans C, Thomas SL. 2006. The effect of number of births on
women’s mortality: systematic review of the evidence for women who
have completed their childbearing. Popul Stud 60:55–71.
Jacobsen BK, Heuch I, Kvåle G. 2003. Age at natural menopause and all
cause mortality: a 37-year follow-up of 19,731 Norwegian women. Am J
Epidemiol 157:923–929.
Joshi H. 2002. Production, reproduction and education: women, children
and work. Popul Dev Rev 28:445–474.
Kabat GC, Miller AB, Rohan TE. 2007. Reproductive and hormonal factors
and risk of lung cancer in women: a prospective cohort study. Int J
Cancer 120:2214–2220.
547
Kendig H, Dykstra P, van Gaalen RI, Melkas T. 2007. Health of parents
and childless individuals. J Fam Issues 28:1457–1486.
Kiernan K. 1989. Who remains childless? J BioSoc Sci 21:387–398.
Kirkwood T, Rose M. 1991. Evolution of senescence: late survival sacrificed
for reproduction. Philos Trans R Soc Lond B Biol Sci 332:15–24.
Kravdal Ø, Rindfuss R. 2008. Changing relationships between education
and fertility—a study of women and men born 1940–64. Am Sociol Rev
73:854–873.
Kvåle G, Heuch I, Nilssen S. 1994. Parity in relation to mortality and
cancer incidence: a prospective study of Norwegian women. Int J Epidemiol 23:691–699.
Lawlor DA, Emberson JR, Ebrahim S, Whincup PH, Wannamethee SG,
Walker M, Smith GD. 2003. Is the association between parity and coronary
heart disease due to biological effects of pregnancy or adverse lifestyle risk
factors associated with childrearing? Circulation 107: 1260–1264.
Lochen ML, Lund E. 1997. Childbearing and mortality from cancer of the
corpus uteri. Acta Obstet Gynecol Scand 76:373–377.
Longva S, Thomsen I, Severeide P. 1998. Reducing costs of Censuses in
Norway through use of administrative records. Int Stat Rev 2:223–224.
Lund E, Arnesen E, Borgan JK. 1990. Pattern of childbearing and
mortality in married women—a national prospective study from
Norway. J Epidemiol Community Health 44:237–240.
Maughan B, Lindelow M. 1997. Secular change in psychosocial risks: the
case of teenage motherhood. Psychol Med 27:1129–1144.
Ni Brolchain M. 1993. New perspectives on fertility in Britain. Studies on
Medical and Population Subjects 55. London: HMSO.
Perls T, Alpert L, Fretts R. 1997. Middle aged mothers live longer. Nature
389:133.
Rønsen M. 2004. Fertility and family policy in Norway—a reflection on
trends and possible connections. Demogr Res 10:265–286.
Russo IH, Russo J. 2007. Primary prevention of breast cancer by hormoneinduced differentiation. Recent Results Cancer Res 174:111–130.
Salehi F, Dunfield L, Philips KP, Krewski D, Vanderhyden BC. 2008. Risk
factors for ovarian cancer: an overview with emphasis on hormonal
factors. J Toxicol Environ Health B Crit Rev 11:301–321.
Smith KR, Mineau GP, Bean LL. 2002. Fertility and post-reproductive
longevity. Soc Biol 49:185–205.
Statistics Norway. Statistical Yearbook of Norway 2006. Oslo, Statistics
Norway, 2006. Available at www.ssb.no/english/yearbook/2006.
Yi Z, Vaupel JW. 2004. Association of late childbearing with healthy
longevity among the oldest-old in China. Popul Stud 58:37–53.
American Journal of Human Biology