Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Examining the effect of fertility on lifespan using data from the Swedish Twin Registry
(USC Thesis Other)
Examining the effect of fertility on lifespan using data from the Swedish Twin Registry
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
EXAMINING THE EFFECT OF FERTILITY ON LIFESPAN USING DATA FROM
THE SWEDISH TWIN REGISTRY
by
Elizabeth Chereji
A Thesis Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
MASTER OF ARTS
(PSYCHOLOGY)
December 2009
Copyright 2009 Elizabeth Chereji
ii
Table of Contents
List of Tables iii
Abstract iv
Introduction 1
Evolution of Senescence: Early theories 1
Application of the Disposable Soma
Theory in Research 5
Gender and Disposable Soma Theory 7
Possible Confounding Factors 9
Specific Aims 11
Method 12
Study Participants 12
Variables of Interest 15
Statistical Methods 17
Analysis Plan 18
Results 20
Descriptive Statistics 20
Figure 1a: Individual-Level Survival plot 25
Individual-Level Survival Analysis 25
Co-twin Survival Analysis 32
Discussion 33
Main Findings 33
Possible Mechanisms Underlying Main
Findings 34
Comparing Results to Past Studies 38
Strengths and Limitations 39
Future Directions 41
References 42
iii
List of Tables
Table 1: Lifespan and Parenting Status Descriptive
Statistics for Individuals 21
Table 2: Lifespan and Parenting Status Descriptive
Statistics for Twins 23
Table 3: Differences in Last Age and Number of
Children Within Pairs Where Both Twins Have
Offspring 24
Table 4: Survival Analysis at the Individual Level 26
Table 5: Survival Analysis at the Individual Level,
Including Birth Year 29
Table 6: Survival Analysis at the Individual Level,
Including Environmental Covariates 30
Table 7: Parameter Estimates, Standard Error, and
Confidence Intervals for Variables in Final Model 31
Table 8: Co-twin Survival Analysis 33
iv
ABSTRACT
The disposable soma theory suggests that reproduction occurs at the expense of
physiological resource maintenance (Kirkwood, 1977). Elaborating on this theory, some
researchers have proposed that a negative relationship exists between longevity and
reproduction (e.g. Westendorp & Kirkwood, 1998). Studies examining this relationship
were often restricted to isolated groups that may be considered questionable
representations of the general population (Westendorp & Kirkwood, 1998, Korpelainen,
2000). The current study examined whether the number of children a woman bears during
her lifetime decreases her lifespan, as posited by the disposable soma theory. Female and
male twin pairs from the Swedish Twin Registry were analyzed using a co-twin control
model and survival analysis. Twin data helps elucidate whether this phenomenon is
accounted for by familial factors or non-shared environmental factors. Based on the
results of the survival analyses, there were main effects of birth year, gender, and having
any children in predicting lifespan. More specifically, females, individuals who had
children, and individuals born in later cohorts versus earlier cohorts had increased
chances of survival. Also, birth year accounted for part of the association between having
any children and lifespan. The results are consistent with a causal interpretation at the
individual level, as the inclusion of co-twin information did not increase the predictability
of survival. The evidence in this study does not support the disposable soma theory and
sheds light on other factors that might affect lifespan.
1
INTRODUCTION
Evolution of Senescence: Early theories
The study of the evolution of aging has been the subject of interest to scientists
who question the mechanisms underlying the process of senescence. One of the first
scientists to link evolution and senescence was August Weismann, who speculated that
senescence removes older organisms in order to allow resources for the younger
generations (Weismann, 1898, Kirkwood & Rose, 1991). This suggests the existence of
an aging mechanism, which regulates the onset of senescence. Weismann’s theory has
largely been dismissed as simplistic and lacking a definitive mechanism that perpetuates
the aging process (Kirkwood & Holliday, 1979). While Weismann’s theory is no longer
accepted, the value of evolution in the context of aging has continued to fuel theories.
As evident in Weismann’s theory, Darwinian natural selection is the primary
force behind many theories related to the evolution of senescence. At its core, natural
selection drives the adaptation and survival of a species. In other words, reproduction
perpetuates the survival of a species and is more vital to survival than aging past the
reproductive time period. When considering natural selection in the context of
senescence, it is important to remember that factors that influence survival past the
reproductive years are irrelevant (Kirkwood & Rose, 1991).
While natural selection may be the fundamental basis for some of the most
renowned evolutionary theories of aging, these theories differ with regards to the specific
mechanisms causing senescence. Mutation accumulation theory and antagonistic
pleiotropy theory are two of the dominant evolutionary theories pertaining to possible
2
mechanisms that introduce senescence into adulthood. Both of these theories are based
on approaches from the field of population genetics.
Peter Medawar developed the mutation accumulation theory as a response to
earlier theories related to evolution in senescence. According to Medawar, harmful
genetic mutations aggregate throughout an organism’s lifetime. He attributes aging to the
cumulative phenotypic expression of the harmful genes. These genetic mutations do not
harm the organism in its youth. Medawar explained that, in younger members of a
species, natural selection is related to the preservation and promotion of the expression of
genes beneficial to reproductive capacities. However, the accumulation of harmful
genetic mutations over a lifetime begins to affect the organism as the force of natural
selection ceases to favor postreproductive organisms. The actions of such harmful genetic
mutations, left unchecked and unregulated later in life, lead to senescence and the various
pathologies that accompany aging (Medawar 1952, Charlesworth, 2000). While many
organisms face mortality even before the onset of this aging process (due to predation,
disease, etc.), those who live beyond their reproductive capacities are expected to
experience the effects of mutation accumulation (Charlesworth, 2001).
George Williams expanded on Medawar’s mutation accumulation theory by
proposing the existence of pleiotropic genes favored by natural selection. According to
the antagonistic pleiotrophy theory, a specific gene can be beneficial early in life but
detrimental later in life (Williams, 1957). A possible example of antagonistic pleiotropy
is the production of ROS (reactive oxygen species) by Nox enzymes. In early life, Nox-
derived ROS is beneficial to a variety of biological functions, including immunity,
3
biochemical reactions, and signal transduction. However, in later life, detrimental
chronic conditions are associated with ROS, such as atherosclerosis, hypertension, and
diabetic nephropathy (Lambeth, 2007).
Natural selection can favor pleiotropic genes because of the benefits provided in
early life. Additionally, the advantageous actions may benefit organisms early in life
while the negative effects surface during the postreproductive portion of life. As a result,
the same genes that benefit survival and reproduction in earlier years can lead to
senescence and deterioration in later years (Williams 1957, Kirkwood & Holliday, 1979).
In contrast to mutation accumulation theory, beneficial and harmful elements are
incorporated into the same genes rather than separate genes. Some consider antagonistic
pleiotropy theory to be more credible than mutation accumulation theory because it
accounts for beneficial and harmful effects (Kirkwood & Holliday, 1979, Kirkwood &
Rose, 1991).
The disposable soma theory, proposed by Thomas Kirkwood, approaches the
concept of longevity from a physiological ecology perspective, derived from life history
theory (Kirkwood, 1977, Kirkwood & Rose, 1991). The disposable soma theory pertains
to the problem of optimal resource allocation among tasks that constantly demand
energy. Organisms take energy in from the environment and expend it on specific
demands. Among such demands are reproduction, somatic maintenance and growth. Due
to the finite nature of the organism’s metabolic resources, it must balance resource
allocation among individual tasks. As a result, resources invested in one demand cannot
be devoted to other important demands (Kirkwood & Holliday, 1979, Kirkwood & Rose,
4
1991). Multiple resource demands result in resource allocation to somatic maintenance
in a manner that will allow the organism to survive through its natural life (specifically,
survival in the wild) without investing enough to live indefinitely. Specifically, natural
selection favors resource allocation to reproduction. In essence, this can be viewed as a
trade-off between somatic maintenance and reproduction (Le Bourg, 2001). The limited
investment in somatic maintenance results in the eventual molecular and cellular damage
that leads to aging (Drenos & Kirkwood, 2004).
Similarities and differences exist among the disposable soma theory and the
theories of aging previously mentioned. At a basic level, they share the concepts of
deterioration and molecular decline as an age-related mechanism. In addition, the
importance of reproduction is evident across the theories. Parallels have been drawn
between the disposable soma theory and antagonistic pleiotropy theory, as the disposable
soma theory also relies on a beneficial and deleterious dichotomy of the same mechanism
(i.e. devoting limited resources to the soma benefits reproduction but later leads to aging)
(Kirkwood & Holliday, 1979). However, disposable soma theory is unique in the manner
it incorporates physiological ecology as an important theoretical component. The optimal
allocation approach of the disposable soma theory places an emphasis on resource
distribution and the physiological consequences of balancing resources. This theoretical
emphasis differs from the population genetics perspective used in mutation accumulation
and antagonistic pleiotropy (Le Bourg, 2001). Also, the other theories’ emphasis on
population genetics is best understood as a species-level phenomenon, whereas the
disposable soma theory focuses on individual organisms. While the disposable soma
5
theory is concerned with the individual, natural selection is still at the center of this
mechanism. It is most beneficial for organisms to strategically allocate resources in order
to maximize fitness, as they are subject to natural selection (Kirkwood & Rose, 1991).
Organisms face the dilemma of devoting energy either to reproduction or somatic
maintenance. According to the disposable soma theory, organisms that reproduce
repeatedly dedicate fewer resources to somatic maintenance. The lack of resources given
to the soma leads to cellular deterioration, which propels the process of aging. The
disposable soma theory has been applied to organisms in the wild as well as humans. The
next section will explore the research relevant to the disposable soma theory in non-
human organisms. Then, research in humans will be described. It is worth noting that the
majority of the studies use females as the primary gender of interest, since the
physiological investments made while bearing offspring are considered most
representative of resource optimality and physiological ecology. Issues related to gender
and the disposable soma theory will be discussed in further detail later.
Application of the Disposable Soma Theory in Research
Research studies in non-human organisms have analyzed the relationship between
reproduction and longevity. Studies using Drosophila melanogaster (fruitflies) have
found that increased reproductive output is accompanied by shorter lifespan. Similarly,
conditions that restricted reproduction led to a negative relationship between fecundity
and longevity in fruitflies (Fowler & Partidge, 1989, Tartar, Chien, & Priest, 2001).
Additional research has been conducted to examine this association in mammals. One
study observed mice inhabiting dangerous environments in the wild. In risky
6
environments where a species is subject to the dangers of predation, reproduction must
occur quickly before they die. Consistent with this concept, this species of mice have a
brief natural lifespan and produce several offspring. The rapid rate of reproduction,
accompanied by accelerated senescence, may support a trade-off in favor of reproduction.
Rather than heavily investing resources into the soma, the mice invest resources into the
reproduction of several offspring (Kirkwood & Holliday, 1979). The trade-off between
life and reproduction seen in non-human species can be considered reflective of the
principles proposed by the disposable soma theory (Kirkwood & Rose, 1991). Non-
human studies on longevity and reproduction compliment human research as a way to
generalize this evolutionary phenomenon across species.
In analyzing the relationship between fertility and longevity, investigators have
turned to data from historic human populations. Many of these studies utilize
information on contained aristocratic or rural groups reproducing in natural conditions
(i.e. without contraception). A study coauthored by Thomas Kirkwood examined British
aristocrats between the 8
th
and 19
th
century. According to the results, females who had
survived beyond the age of menopause (60 years and over) demonstrated a negative
correlation between lifespan and number of children. The investigators viewed the results
as potential support for the trade-off described by the disposable soma theory
(Westendorp & Kirkwood, 1998). A study by Doblhammer and Oeppen (2003) also used
the British aristocracy to study the longevity-reproduction trade-off. However, this study
used a more specific group within the population. Some of the restrictions included
individuals who were born between 1603 and 1959, died after the age of 50, and married
7
only once. This study found a positive relationship between fertility and mortality only
when controlling for specific confounds (Doblhammer & Oppen, 2003).
Studies of reproductive trade-offs have also been conducted using data on rural
populations. A study on rural families in Finland during the 1700s and 1800s found a
negative relationship between longevity and reproduction in mothers (Korpelainen,
2000). Another study used a more modern sample of Old Order Amish individuals born
between 1749 and 1912. While this sample may consist of more modern individuals than
the historic populations previously described, the investigators believed that the OOA’s
natural fertility conditions and large family sizes made this population ideal for studying
longevity and reproduction trade-offs. Longevity was reduced only in mothers who had
more than 14 children, with a loss of approximately 4.01 years after each subsequent
child (McArdle, 2006).
Gender and Disposable Soma Theory
In their study of the disposable soma theory, Westendorp & Kirkwood (1998)
found a trade-off between longevity and reproduction in males similar to that found in
women. However, subsequent studies both opposing and defending the existence of
trade-offs in humans have discounted this trend in males. The role of males in
reproduction has been minimal, according to research on reproductive success and
longevity in both genders (e.g. Doblhammer & Oeppen, 2003). One study concluded that
men’s reproductive fitness improves with increased longevity and found little evidence of
any physical impact due to reproductive investment (Korpelainen, 2000). Other
8
researchers have simply limited their investigation to females (Doblhammer, 2000,
Lycett, Dunbar, & Voland, 2000, Helle, Lummaa, & Jokela, 2004).
While females have been the main gender of interest in investigating the trade-off
between longevity and reproduction, males have also received some attention. Although
males do not contribute to embryonic development or childbirth, Vinogradov (1998)
claims they devote energy to frequent mating that might be used for somatic
maintenance. Specifically, he framed males and females as participants in r-K selection.
The r-selection strategy, which is beneficial for populations living in risky environments,
involves higher rates of reproduction and a shorter lifespan. In contrast, K selection is
beneficial in more predictable environments and involves the production of fewer
offspring and a longer lifespan (MacArthur & Wilson, 1967, Gadgil & Solbrig, 1972).
These designations are usually used for entire species, but some sociobiologists have
used them as an analogy for the sex differences in human mating. Because males mate
frequently and die earlier than females, males are designated as r-strategists and females
are K-strategists (Vinogradov, 1998). As a result, Vinogradov (1998) claims that males
also experience a trade-off in reproduction and longevity, but in a different manner
compared to females. Trade-offs have also been found both mothers and fathers in a
recent study utilizing the Utah Population Database (Penn & Smith, 2007, Le Bourg,
2007), but the trend in males was eventually reported as a result of socioeconomic factors
due to higher economic costs incurred by parity (Le Bourg 2007). Currently, the
existence of a physiological cost of reproduction in men is difficult to conceptualize and
is not strongly supported (Le Bourg, 2007).
9
Possible Confounding Factors
While the link between longevity and fertility is the primary relationship of
interest in the research mentioned thus far, possibly confounding variables may also
influence this relationship.
Confounding variables include both biological and psychosocial factors. Stress-
related biological factors associated with rearing more offspring is one possibly
confounding variable. Research outside of fertility and longevity has suggested that stress
can have deleterious effects on health (Lovallo, 2005). A related issue is stress resistance.
Organisms that have better stress resistance appear to live longer and overcome more
environmental stresses (Harshman, Moore, Sty, & Magwire, 1999, Harshman & Haberer,
2000). The act of mating alone has also been investigated as a potential confound. In
fruitflies, females that mated frequently were found to die at higher rates compared to
those that did not mate as frequently. Suggested causes for this phenomenon include
mechanical injury, parasitic transfer, or adverse effects associated with sperm (Fowler &
Partridge, 1989). Molecules in seminal fluid have also been identified as potential causes
of early death in female fruitflies that mated frequently. Exposure to male fruitfly seminal
fluid products has been found to increase female fruitfly death rate (Chapman, Liddle,
Kalb, et al., 1995). These studies suggest that the act of mating, rather than reproduction
itself, can be associated with costs that limit lifespan.
Socioeconomic status is a psychosocial factor that can affect the observed
relationship between longevity and fertility. A disparity in death rates appears to exist
among SES groups. Individuals with lower levels of education and lower incomes can
10
experience higher mortality rates than those from higher SES backgrounds (Pappas,
Queen, Hadden, & Fisher, 1993). Socioeconomic status can also affect access to health
care for individuals with low incomes (Fiscella, Franks, Gold, & Clancy, 2000). The
number of children in a family is also related to SES, in part because increased number of
children can place a greater financial burden on a family (Thomson, Hanson, &
McLanahan, 1994, Downey, 1995).
Even though confounds were not addressed in every study related to longevity
and fertility, some included such variables in their analyses in an attempt to rule out their
influence. Specific demographic features have been identified as possible psychosocial
confounds among the groups. Marriage is one demographic variable that is believed to be
a confound. Marriage may lead to economic gains, which can affect some of the factors
associating SES with childrearing (Gavrilov & Gavrilova, 1999, Doblhammer & Oeppen,
2003). One study using data from 153 countries accounted for several potentially
confounding variables, including religion, ethnic group, and geography (Thomas,
Teriokhin, Renaud, et al., 2000). In some studies, socioeconomic status has also been a
controlled (Lycett, Dunbar, & Voland, 2000, Thomas, Teriokhin, Renaud, et al., 2000). It
has been suggested that the relationship between longevity and fertility increases with
increasing levels of poverty (Lycett, Dunbar, & Voland, 2000).
Health is another variable that has been included in analyses to monitor its
influence on longevity and fertility. One study examined health (specifically, frailty) as a
factor that led to death in mothers. They concluded that a strong relationship between
11
parity and mortality exists only when controlling for health (Doblhammer & Oeppen,
2003).
Some researchers used populations where certain confounding variables would
not present problems in studying longevity and reproduction. The use of historic data
from the British aristocracy and European nobility has been defended as reliable and
possessing few confounds related to education and income, since every member exceeds
a specific economic threshold (Doblhammer, 2003, Gavrilov & Gavriloa, 2004).
Specific Aims
Theories about the evolution of senescence have searched for ways to explain
aging from an evolutionary theory standpoint. While other theories are based on
population genetics, the disposable soma theory addresses aging from a resource
optimality perspective derived from physiological ecology. The disposable soma theory
of aging sparked a question investigated in animal and human studies: does reproduction
occur at the expense of somatic maintenance? More specifically, does increased fertility
decrease longevity?
The present study aimed to examine the same questions addressed by the earlier
studies. The association between number of offspring and longevity was analyzed using
twin data from the Swedish Twin Registry. Females were the primary gender of interest,
as past research suggests that they are most affected by physiological tradeoffs. Male
twin pairs were also included in order to test whether the phenomenon is observed
exclusively in females. Using twin data rather than singleton data provides valuable
information regarding the possible influence of familial factors (genetic and
12
environmental) versus unshared environmental factors. Considering the threat of
confounding variables to the observed relationship between longevity and fertility,
certain covariates (e.g., socioeconomic status, environment) were included in the
analyses.
Based on the existing literature, I predicted that the proposed analyses would find
an association between number of offspring and decreased lifespan. In addition, the
association between offspring and lifespan was expected to be found exclusively in
females and was not expected to be accounted for by confounding variables. In the twin
analyses, this association was expected to be attributable to individual-specific
environmental factors rather than familial factors. An association between lifespan and
fertility due to individual-specific factors would be consistent with the disposable soma
theory.
METHOD
Study Participants
The Swedish Twin Registry (STR), established in the late 1950’s, contains data
on various health factors, psychosocial information, and demographic factors collected
from twins in Sweden through a series of questionnaires and interviews. The Swedish
Twin Registry is maintained at the Karolinska Institutet in Stockholm, Sweden, where it
is regularly matched to other databases to obtain and update information. The Swedish
Twin Registry consists of three cohorts differentiated by year of birth: 1886-1925, 1926-
1958, and 1959-1990. Birth year data for the earliest cohort was obtained from parishes
13
in Sweden, while subsequent cohorts utilized national birth record registers. Data
collection methods and the information obtained varied with each cohort.
The current study used data collected from the earliest cohort born between 1886
and 1925 (STR 1) to ensure that participants reached their postreproductive years. Data
was collected from members of the old cohort through questionnaires mailed in 1960-
1961, 1963, 1967, and a questionnaire given to just a portion of the cohort in 1970. The
data obtained from this cohort includes demographics, medical information, and lifestyle
information. Data obtained from the 1967 (Q67) and 1970 (Q70) questionnaires were
used in the present study.
The participants in the analyses reported here were drawn from a larger sample of
21,587 individuals from the Swedish Twin Registry. This sample consists of individuals
from male twin pairs (N=9491) and from female twin pairs (N=12096). Among the
females 3977 are from MZ pairs, 7661 are DZ, and 458 have unidentified zygosities.
Among the males 3236 individuals are MZ, 5845 are DZ, and 410 have unidentified
zygosities.
A cutoff for birth year was implemented to ensure that participants were alive to
complete the 1967 questionnaire. This cutoff also eliminated possible survival effects, as
individuals in earlier birth cohorts who survived long enough to complete the
questionnaire may not represent survival in the rest of the sample. By setting 1901 as the
cutoff birth year, 3,867 individuals born between 1886 and 1900 were excluded from the
study. A known current age or age at death was also required in order to examine the
relationship between number of children and age. If age was unavailable due to
14
incomplete data, then these individuals were treated as censored. A total of 21 twins
were censored, as their survival status (i.e., dead or alive) and the possible year of death
remain unknown. In this sample, 87.90% of the males and 78.22% of the females are
deceased.
Information on offspring was combined from the 1967 and 1970 questionnaire to
update existing values and replace missing data. For those who completed both
questionnaires, the response for the number of children from the 1967 questionnaire was
subtracted from the 1970 questionnaire response to calculate discrepancies. If the
discrepancy was a difference of less than 3 children, then the larger of the two values was
selected for analysis. If the difference was equal to or larger than 3, the individuals were
not included in the analysis, as such a large change in number of children between the
questionnaires was deemed unlikely (N=18). Within-individual responses to the 1967
and 1970 questionnaires for number of children (r=0.95) and whether or not the
individual had children (r=0.97) were highly correlated, suggesting good reliability. Few
individuals were excluded due to values that were not linked to valid codes in the
codebook (N=2). Individuals who indicated they adopted children were also eliminated,
as the items in the questionnaire do not clearly identify whether adoptive parents have
ever given birth to biological children (N=5). If information related to whether or not the
individual had children was missing in both Q1967 and Q1970, then the observation was
excluded (N=2052). After excluding the aforementioned participants from the original
sample, the sample size was reduced to N=15643. The sample consists of 6819
individuals from male twin pairs and 8824 from female twin pairs. Among the males,
15
2287 individuals are MZ, 4266 are DZ, and 266 have unidentified zygosities. Among
the females, 2933 are MZ, 5572 are DZ, and 319 have unidentified zygosities.
Variables of interest
In the old cohort, information on zygosity was collected by sending the 1961
questionnaire to like-sexed pairs. Zygosity was determined according to the response to
the question, “During childhood, were you and your twin partner as similar as cherries
(i.e. two peas in a pod) or not more alike than siblings in general?” If both twins agreed
with the first phrase, they were classified as monozygotic (MZ). If both twins described
themselves as different, they were defined as dizygotic (DZ). In twin pairs where one
member did not respond or both members did not agree, zygosity was designated as “not
determined” (XZ). Some of these pairs have been assigned to zygosity based on
responses to subsequent questionnaires and interviews. When compared to the analyses
of blood markers, the responses to this question correctly determined zygosity in 95% of
the twin pairs. The cohort born in 1886-1925 primarily consists of same-sexed pairs.
Information on unlike-sexed twin pairs born between 1906 and 1925 has been obtained
recently, but these pairs were not included in Q67 and Q70. The pair analyses in the
current study include MZ and DZ same sex pairs.
In the current study, the number of children is the primary independent variable
predicted to affect lifespan Number of offspring was identified for each member of a twin
pair, for both genders and both zygosities. As previously mentioned, number of children
was determined using information from items in the 1967 and 1970 STR questionnaires.
When collected, the data were coded to range from 0 to 7, with individuals having more
16
than 7 children assigned a value of 7. Adopted children were coded ‘8.’ The number of
adopted children was not specified in Q67 or Q70.
Survival age was determined using information from the Cause of Death Registry.
Members from twin pairs who have died were identified and their age at death was
included in the analysis. Individuals not included in the Cause of Death Registry were
presumed alive, and their age was calculated using their age as of the last Cause of Death
Registry update. Twins who died because of an accident were identified and were
included in the analysis, but their age at death was treated as censored in analyses.
Each individual’s birth year was included in the analysis as a covariate. This
information is relevant for its possible role as a confound, as individuals from different
birth cohorts may display differences in lifespan. For example, individuals born earlier
may not have benefited from health care that was as sophisticated as services available to
the younger cohorts.
Marital status was included as a covariate in the analyses by grouping together
individuals who were ever married (married, divorced, remarried, widowed) compared to
those who identified as unmarried. Married individuals may differ from unmarried
individuals in ways that ultimately effect health and lifespan, including the increased
availability of wealth and other resources.
Considering that socioeconomic status may be important in understanding the
stresses involved with raising children and its effect on longevity, it was included as a
covariate in the analysis. Father’s occupation was used as a proxy for socioeconomic
status and was obtained through Q67. The questionnaire inquired about father’s
17
occupation status in an open-ended question format and then coded the response
according to lower, middle, or upper classes. Additionally, urban or rural living
environment during adulthood was also included as a covariate. Environment was
distinguished based on rural, town, or urban residency in Q67 and Q70.
Statistical Methods
The analysis employed survival analysis and the co-twin control method in
analyzing data from the twins, both as individuals and pairs.
Survival analysis is a time-to-event analysis method. In the current study, the
time (in years) to death is the dependent variable of interest. Survival distributions serve
as a descriptive representation of survival trends. Various predictors were entered in the
survival analysis in order to determine which variables increased predictability of
lifespan. Both individual-level and pair-level survival analyses were conducted. Details
about variables and models tested in the survival analyses will be discussed later in this
paper.
The genetic and environmental similarities within twin pairs make twin data a
valuable resource. For the purpose of the present study, the co-twin control method was
employed. The co-twin control method applies a case-control design to twin pairs, where
one twin is treated as the case and the other is treated as a control. The co-twin control
method is often used in twin pairs that do not share a disorder, as it permits the study of
the association between a hypothesized risk factor and outcome while controlling for
genetics and early environmental effects that are unmeasured and shared by the twins. In
contrast to the case-control design often employed by the co-twin control method,
18
discordance in the present study was determined by the degree of difference in number
of offspring in pair -level analyses. Comparing differences in lifespan in twin pairs with
differences in number of children helped reduce nuisance variation, elucidated possible
familial factors that may account for associations between offspring variables and
lifespan found at the individual level, and helped determine whether increasing number
of children decreased lifespan. Monozygotic twins provide information on genetic factors
that might contribute to the lifespan-offspring association. Monozygotic twins that were
reared together share all their genes and common environmental factors (i.e., aspects of
the environment shared by siblings within a family). In contrast, dizygotic twins that
were reared together share all of their common environmental factors and (on average)
half their genes. Comparisons between monozygotic and dizygotic twin pairs can help
clarify whether differences in number of offspring lead to differences in lifespan rather
than familial factors accounting for the observed associations.
Analysis Plan
The first step of the analysis was to examine the association between offspring
and lifespan using the data from twins, treated as individuals (ignoring the pair status).
This allowed comparisons to results in the literature using other samples. Three separate
sets of survival analysis models were tested. In the first set of models, gender, having any
children and number of children served as the independent variables expected to affect
lifespan. The second set of models included the aforementioned independent variables
and birth year was included as a covariate. Interaction terms composed of gender with
each of the offspring variables were also added to these models. The sex * number of
19
offspring and sex*anykids interactions are vital, as the disposable soma theory
hypothesizes that the offspring-lifespan association is stronger in females than males. In
the third set of models, environment and paternal socioeconomic status covariates were
added to models consisting of birth year and gender to determine if SES and rural or
urban habitat affect lifespan. Variables were added to models in a sequential manner and
changes in the chi-square statistic were calculated in order to compare improvements in
fit among nested models. A model excluding all predictors and covariates served as the
baseline for each separate set of analyses.
The second step in the analysis combined a co-twin control design and a survival
analysis to examine possible common environmental or genetic factors that might
account for part of any associations observed at the phenotypic level. This analysis used a
“double entry” data set that included each twin and information about his/her co-twin.
Similar to the survival analysis at the individual level, information related to having any
offspring, served as the primary independent variables predicting lifespan. In contrast to
the phenotypic analyses, co-twin information was added as a covariate. Co-twin
covariates include whether the co-twin had any children and co-twin number of children.
The addition of co-twin offspring variables provided insight into possible common
environmental factors that might affect lifespan. Interaction terms composed of zygosity
and each co-twin offspring variable were added to the survival analyses to determine if
genetic factors accounted for part of the associations between the independent variables
and lifespan at the individual level. If the mechanism is consistent with the disposable
soma theory, the co-twin variables besides co-twin age should not be relevant. More
20
specifically, if the number of offspring is a direct factor influencing lifespan, then
including co-twin information would not increase the predictability of survival when
added to the survival analysis. In contrast, these factors should increase the prediction of
lifespan if the mechanism is attributable to a familial (genetic or environmental) factor.
As mentioned earlier, evidence that the offspring-lifespan association is due to individual
factors would be consistent with the disposable soma theory.
RESULTS
Descriptive Statistics
Table 1 displays descriptive information at the individual level, including mean
number of children among those with children, mean age at last contact, percentage of the
sample that was childless, and percentage that was dead at last contact. Individuals were
grouped by gender (regardless of zygosity) and gender by zygosity.
Among the males in the sample (including those of unknown zygosities), 24.67%
were childless. Among males who had children, the average number of offspring was
2.24 (SD=1.20). Approximately and 87.90% were deceased. For descriptive purposes,
the age at death information was combined with the age of living twins at the last registry
update to yield a last contact age. In males, this had a mean of 78.10 years (SD=9.65).
Among the monozygotic males, 22.96% were childless and the average number of
children among monozygotic male parents was 2.20 (SD=1.16). Approximately 88.09%
were dead and the average last age was 78.49 years (SD=9.69). In dizygotic males,
25.50% were childless, while dizygotic male twins had average of 2.25 children
(SD=1.21). Approximately 87.79% of the dizygotic males were deceased and the
21
Table 1: Lifespan and Parenting Status Descriptive Statistics for Individuals
81.86(9.06) 77.96 2.32(1.29) 22.92 5572 Female DZ
81.97(9.24) 78.10 2.31(1.29) 21.58 2933 Female MZ
81.88(9.10) 78.22 2.32(1.30) 22.46 8824 All Females
77.95(9.56) 87.79 2.25(1.21) 25.50 4266 Male DZ
78.49(9.69) 88.09 2.20(1.16) 22.96 2287 Males MZ
78.10 (9.65) 87.90 2.24(1.20) 24.67 6819 All Males
Mean age at death or
last contact(SD)
% Dead Mean number
of Children
(SD)
% childless N [individual]
81.86(9.06) 77.96 2.32(1.29) 22.92 5572 Female DZ
81.97(9.24) 78.10 2.31(1.29) 21.58 2933 Female MZ
81.88(9.10) 78.22 2.32(1.30) 22.46 8824 All Females
77.95(9.56) 87.79 2.25(1.21) 25.50 4266 Male DZ
78.49(9.69) 88.09 2.20(1.16) 22.96 2287 Males MZ
78.10 (9.65) 87.90 2.24(1.20) 24.67 6819 All Males
Mean age at death or
last contact(SD)
% Dead Mean number
of Children
(SD)
% childless N [individual]
-Total N= 15643
-Notes: 266 Males and 319 Females missing zygosity information are included in the all male/all female
categories.
average last known age was 77.95 years (SD=9.56). With regard to all the females in the
sample, and the percent childless was 22.46% and the average number of children among
mothers was 2.32 (SD=1.30). The percent of all females who were dead was 78.22% and
the average age at death or last contact was 81.88 years (SD=9.10). Among monozygotic
females, 21.58% were childless and the mean number of children was 2.31 (SD=1.29).
Approximately 78.10% of the monozygotic females were dead and the mean last age was
81.97.0 years (SD=9.24). Similarly, 22.92% were childless among dizygotic females and
those that had children had an average of 2.32 children (SD=1.29). Approximately
77.96% of the females in the sample were deceased and the average last recorded age was
81.86 years (SD=9.06).
Table 2 presents pair-wise information on the age at last contact, within-pair
differences in last known age, and within-pair correlations of last age. Pairs were
22
grouped based on whether each twin had a child, to examine similarities in last age
among twins concordant and discordant for having children. The Pearson correlation
coefficients were used to index the association between the last recorded ages in pairs
similar or discordant for having children. Based on the disposable soma theory, females
with children are expected to have shorter lifespans than individuals without children.
Therefore, more similarity in last recorded age would be expected in female pairs where
neither twin had children or where both twins had children, compared to female pairs
discordant for having children. These patterns are not expected in males and should not
vary with zygosity if the disposable soma theory is accurate. In contrast to the
hypotheses based on the disposable soma theory, the correlations comparing last recorded
age within twin pairs were modest across gender and twin pair type. Twin pair
resemblance for last age was strongest among monozygotic pairs, particularly among
males.
Table 3 shows within-pair differences in last age and within pair differences in
number of offspring, restricted to pairs where both twins had offspring. Among all the
males in the sample, the average within-pair differences in last recorded age was 9.22
years (SD=7.92) and the average within-pair difference in number of offspring was 1.14
(SD=1.10). In monozygotic males, the within-pair difference in last known age was 8.49
(SD=7.37) and the average within-pair difference in number of offspring was
2.19 (SD=1.15). Similarly, average within-pair differences in last age in dizygotic males
was 9.62 (SD=8.14) and the average difference in number of offspring was 2.25
23
Table 2: Lifespan and Parenting Status Descriptive Statistics for Twins
0.14 1.69(11.67) 82.51 80.82
(8.58) (9.17)
754 DZ Female
0.30 0.79(11.04) 82.03 81.24
(9.28) (9.42)
328 MZ Female
0.19 1.39(11.43) 82.39 81.00
(8.78) (9.18)
1123 All female
0.16 1.68(12.25) 78.27 76.59
(9.25) (9.67)
655 DZ Male
0.40 2.32(10.49) 79.63 77.31
(9.23) (9.89)
270 MZ Male
0.22 1.64(11.89) 78.54 76.90
(9.34) (9.72)
964 All Male
Twin 1 – Twin 2 Twin1 Twin 2 Twin 1 had offspring,
Twin 2 no
offspring
0.12 -0.14(12.02) 78.44(9.59) 1615 DZ Female
0.29 -0.07(10.88) 78.78(9.68) 932 MZ Female
0.18 -0.04(11.62) 81.97(9.05) 2642 All female
0.13 0.48(12.65) 78.44(9.59) 1117 DZ Male
0.33 0.41(11.22) 78.78(9.68) 686 MZ Male
0.21 0.46(12.17) 78.54(9.68) 1872 All Male
Both twins had
offspsring
0.08 0.86(12.63) 81.98(9.33) 229 DZ Female
0.30 1.76(11.59) 82.15(9.87) 136 MZ Female
0.16 1.15(12.25) 82.00(9.51) 377 All female
0.08 -0.36(12.95) 77.25(9.54) 165 DZ Male
0.37 -0.60(10.44) 77.08(9.33) 109 MZ Male
0.18 -0.35(12.05) 77.21(9.43) 286 All Male
Neither twin had
offspring
Within-pair correlation
of
last age
Within-pair difference
in last age
M years (SD)
Age in years at death or
last contact
M (SD)
Number of Pairs Pair Type
0.14 1.69(11.67) 82.51 80.82
(8.58) (9.17)
754 DZ Female
0.30 0.79(11.04) 82.03 81.24
(9.28) (9.42)
328 MZ Female
0.19 1.39(11.43) 82.39 81.00
(8.78) (9.18)
1123 All female
0.16 1.68(12.25) 78.27 76.59
(9.25) (9.67)
655 DZ Male
0.40 2.32(10.49) 79.63 77.31
(9.23) (9.89)
270 MZ Male
0.22 1.64(11.89) 78.54 76.90
(9.34) (9.72)
964 All Male
Twin 1 – Twin 2 Twin1 Twin 2 Twin 1 had offspring,
Twin 2 no
offspring
0.12 -0.14(12.02) 78.44(9.59) 1615 DZ Female
0.29 -0.07(10.88) 78.78(9.68) 932 MZ Female
0.18 -0.04(11.62) 81.97(9.05) 2642 All female
0.13 0.48(12.65) 78.44(9.59) 1117 DZ Male
0.33 0.41(11.22) 78.78(9.68) 686 MZ Male
0.21 0.46(12.17) 78.54(9.68) 1872 All Male
Both twins had
offspsring
0.08 0.86(12.63) 81.98(9.33) 229 DZ Female
0.30 1.76(11.59) 82.15(9.87) 136 MZ Female
0.16 1.15(12.25) 82.00(9.51) 377 All female
0.08 -0.36(12.95) 77.25(9.54) 165 DZ Male
0.37 -0.60(10.44) 77.08(9.33) 109 MZ Male
0.18 -0.35(12.05) 77.21(9.43) 286 All Male
Neither twin had
offspring
Within-pair correlation
of
last age
Within-pair difference
in last age
M years (SD)
Age in years at death or
last contact
M (SD)
Number of Pairs Pair Type
-N=7264 pairs-Pair excluded if pair type not available: Males=500, Females=583;
-Pair excluded if zygosity is unknown: Males=96, Females=131
(SD=1.20). Among all females in the sample, the average within-pair difference in last
known age was 8.29 (SD=8.16) and the average difference in number of offspring was
1.22 (SD=1.16). Monozygotic females had an average difference in last age of 7.48
24
(SD=7.94) and an average difference in number of offspring of 2.36 (SD=1.30). The
average difference in last age in dizygotic females was 8.74 (SD=8.28) and the average
Table 3: Differences in Last Age and Number of Children Within Pairs Where Both Twins Have Offspring
-0.01 2.30(1.29) 8.74(8.28) 1598 DZ Female
0.001 2.36(1.30) 7.48(7.94) 922 MZ Female
-0.01 1.22(1.16) 8.29(8.16) 2614 All female
-0.005 2.25(1.20) 9.62(8.14) 1103 DZ Male
0.03 2.19(1.15) 8.49(7.37) 680 MZ Male
0.01 1.14(1.10) 9.22(7.92) 1851 All Male
Correlation between within-
pair difference in last
age and within-pair
difference in number
of offspring
Within pair difference
in number of
offspring
M (SD)
Within-pair difference
in last age
M (SD)
Number of Pairs
-0.01 2.30(1.29) 8.74(8.28) 1598 DZ Female
0.001 2.36(1.30) 7.48(7.94) 922 MZ Female
-0.01 1.22(1.16) 8.29(8.16) 2614 All female
-0.005 2.25(1.20) 9.62(8.14) 1103 DZ Male
0.03 2.19(1.15) 8.49(7.37) 680 MZ Male
0.01 1.14(1.10) 9.22(7.92) 1851 All Male
Correlation between within-
pair difference in last
age and within-pair
difference in number
of offspring
Within pair difference
in number of
offspring
M (SD)
Within-pair difference
in last age
M (SD)
Number of Pairs
-Note: Only pairs with available data indicating number of children and last age were included.
-N=49 pairs were excluded because the data indicated they had children but number of children
were not available.
-Absolute values of differences in last age and mean number of offspring were taken prior to calculating
means
within-pair difference in number of offspring was 2.30 (SD=1.29). Correlations between
within-pair differences in last recorded age and within-pair differences in number of
offspring were also calculated. The disposable soma theory predicts that discrepancies in
the number of children between twins in a pair would lead to within-pair differences in
survival, particularly among women. However, the observed correlations between
within-pair differences in the number of offspring and differences in last recorded age
were low, ranging from -0.001 to 0.03.
Figure 1a presents the survival distributions based on gender and having any
children. Lifespan is shorter among males compared to females, as the male survival
plots are parallel and shifted to the left of the female survival plots. The distributions
25
also indicate that childless individuals have shorter lifespans than individuals with
children within the same gender. Additional survival distributions were generated for
number of children, but these plots indicated that there were no differences in lifespan
based on number of offspring. Based on the disposable soma theory, lifespan should
Figure 1a: Individual-Level Survival Plot
0
0.25
0.50
0.75
1
0 40 20 60 80 100
Age in Years
Proportion Surviving
Women no children
Women with children
Men , with children
Men, no children
decrease with increasing number of children only among females. In contrast to the
disposable soma theory, lifespan does not vary with number of children in either gender
in survival distributions stratified by number of children.
Individual-Level Survival Analyses
The results from models testing the association between reproduction and
lifespan at the individual level are shown in Table 4. The analyses in this table include
gender, having any children, number of children, and interaction between gender and the
26
offspring variables. A baseline model excluding all predictors was created. Gender was
added to the baseline model to estimate differences in lifespan between males and
females. The estimate for gender was 0.16, indicating that females are 17% more likely to
Table 4: Survival Analysis at the Individual Level
1 0.06 6 9. sex, anykids, numkids
1 0.18 7 8. sex, numkids, numkids*sex
1 7.54 5 7. sex, numkids
1 19.2 5 6. sex anykids
--- --- --- 5. sex
1 1.04 3 4. sex, anykids, dsex*anykids
1 19.16 2 3. sex, anykids
1 85.38 1 2. sex
--- --- --- 1. (baseline--no predictors)
∆df ∆χ2 Compare to
model #
Model
1 0.06 6 9. sex, anykids, numkids
1 0.18 7 8. sex, numkids, numkids*sex
1 7.54 5 7. sex, numkids
1 19.2 5 6. sex anykids
--- --- --- 5. sex
1 1.04 3 4. sex, anykids, dsex*anykids
1 19.16 2 3. sex, anykids
1 85.38 1 2. sex
--- --- --- 1. (baseline--no predictors)
∆df ∆χ2 Compare to
model #
Model
-For models 1 - 4, N=15538
-For models 5-9, N=15453, excluding individuals where there is a value for having any children
(“anykids”) but a missing value for number of offspring (“numkids”)
survive any year compared to males (Model 2 vs 1, Δχ²=85.38, Δdf=1). Information on
whether individuals had any children was added to the baseline model to estimate the
effect of reproduction on lifespan. This resulted in improved model fit (Model 3 vs 2,
Δχ²=19.16, Δdf=1), and the value of the parameter estimate indicated that having any
children increases lifespan. Since the disposable soma theory posits that fertility
decreases lifespan only in females, the interaction between gender and having any
children was added and resulted in minimal improvement in model fit compared to the
main effect model and was not statistically significant (Model 4 vs 3, Δχ²=1.04, Δdf=1).
To test the hypothesis that lifespan decreases with more offspring, the number of
27
offspring variable was added to the baseline model. This second series of models
(Model 5 - 9) is based on a slightly smaller number of participants of N=15453 compared
to N=15538 individuals used in models 1-4, because some individuals were missing data
on how many children they had. The addition of the number of offspring predictor
improved the model fit compared to the baseline model (Model 7 vs 5, Δχ²=7.54, Δdf=1),
but was not comparable to the improvement in model fit seen in the main effect model
that included having any children (Model 3 vs 2, Δχ²=19.16, Δdf=1). Adding the
interaction between gender and number of children to Model 7 tested the hypothesis that
an increase in the number of children would decrease lifespan more in females than in
males. This did not significantly improve the overall model fit (Model 8 vs 7, Δχ²=0.18,
Δdf=1). A main effect model (9) including both having any children and number of
offspring was fit to determine if number of offspring increased the predictability of
survival beyond that from having any children alone, but the fit was not significantly
better (Model 9 vs 6, Δχ²=0.06, Δdf=1).
Table 5 displays the results of survival analysis including birth year as a covariate
to explore if any of the associations in Table 4 are due to differences among birth cohorts
in fertility and longevity. The addition of birth year led to considerable improvement in
model fit relative to the baseline model with no predictors (Model 2 vs 1, Δχ²=419.32,
Δdf=1). The estimate for birth year was 0.03, indicating that individuals born in a
subsequent year were 3% more likely to survive to a particular year than an individual
born in the prior year. The addition of gender to the birth year model improved model fit
relative to birth year alone (Model 3 vs 2, Δχ²=90.82, Δdf=1). To estimate the effect of
28
reproduction on lifespan, information on whether individuals had any children was
added to the birth year and gender model This resulted in improved model fit (Model 4
vs 3, Δχ²=7.48, Δdf=1), though the addition of birth year led to a decrease in chi-square
change relative to Model 3 in Table 4. The interaction between gender and number of
children to Model 5 did not significantly improve the overall model fit (Model 5 vs 4,
Δχ²=0, Δdf=1). Model 6 - 10 is based on 15453 participants compared to 15538
individuals used in models 1-5 because of missing data on number of children. There was
only a slight improvement in model fit after including number of children (Model 8 vs. 6,
Δχ²=2.68, Δdf=1). The inclusion of birth year did not diminish the association between
number of children and lifespan. The chi-square change generated after adding number of
offspring to gender and birth year (Model 8 vs. 6, Δχ²=2.68, Δdf=1) was less than half the
change in model fit when adding information on having any children to gender and birth
year (Model 4 vs 3, Δχ²=7.48, Δdf=1), indicating that there is still a stronger association
between having any children and lifespan compared to the association between number of
children and lifespan. The addition of the interaction between number of children and
having any children was minimal (Model 10 vs. 9, Δχ²=0.02, Δdf=1). Also, the main
effect model including birth year, gender, information on having any children, and
number of children resulted in a negligible change model fit (Model 10 vs. 7, Δχ²=0.06,
Δdf=1).
Though several participants were missing marital status (N=1973), survival
analyses were conducted to briefly test the effect of marriage. To do this, chi-square
change was calculated by comparing a model including birth year, gender, and marital
29
Table 5: Survival Analysis at the Individual Level, Including Birth Year
1 0.06 7 10. Birthyr, sex, anykids, numkids
1 0.02 6 9. Birthyr, sex, numkids,
numkids*sex
1 2.68 6 8. Birthyr, sex, numkids
1 7.40 6 7. Birthyr, sex, anykids
--- --- --- 6. birthyr, sex
1 0 4 5. Birthyr, sex, anykids,
dsex*anykids
1 7.48 3 4. Birthyr, sex, anykids
1 90.82 2 3. Birthyr, sex
1 419.32 1 2. birthyr
--- --- --- 1. (baseline--no predictors)
∆df ∆χ2 Compare to
model #
Model
1 0.06 7 10. Birthyr, sex, anykids, numkids
1 0.02 6 9. Birthyr, sex, numkids,
numkids*sex
1 2.68 6 8. Birthyr, sex, numkids
1 7.40 6 7. Birthyr, sex, anykids
--- --- --- 6. birthyr, sex
1 0 4 5. Birthyr, sex, anykids,
dsex*anykids
1 7.48 3 4. Birthyr, sex, anykids
1 90.82 2 3. Birthyr, sex
1 419.32 1 2. birthyr
--- --- --- 1. (baseline--no predictors)
∆df ∆χ2 Compare to
model #
Model
-For models 1 - 5, N=15538
-For models 6-10, N=15453, excluding individuals where there is a value for having any children
(“anykids”) but a missing value for number of offspring (“numkids”)
status to an identical model with the addition of having any children. The model with the
addition of having any children compared to the birth year, gender, and marital status
model resulted in a seemingly small improvement in model fit (Δχ²=1.86). While this
change in model fit may seem minimal, it is worth noting that marital status and having
any children are highly correlated (males: r=0.91; females: r=0.85), indicating that the
results of survival analyses including marital status do not discredit the effect of having
children on lifespan.
Table 6 shows the results of survival analysis including birth year, gender,
father’s socioeconomic status and rural or urban environment. For the survival analyses,
father’s socioeconomic status was coded based on upper, middle , or lower class. For the
environment variables, rural and urban environment variables were added separately to
30
the analyses. Rural and urban environment were coded based on dichotomous “yes” or
“no” responses, where “no” on both variables would indicate that the individual’s
environment was a small town or municipality. This set of analyses tested whether
socioeconomic status and environment were relevant covariates that account for some of
the associations seen in earlier analyses. These analyses were conducted separately
because these variables were available for only N=10033 of the total sample. Similar to
prior analyses (Table 5), both birth year and gender were associated with lifespan (Model
2 vs. 1, Δχ²=297.78, Δdf=1; Model 3 vs 2, Δχ²=63.02, Δdf=1). To test if living in a rural
environment affects lifespan, information on living in a rural environment was added to
the model with birth year and gender. The addition of this covariate did not improve
model fit (Model 4 vs. 3, Δχ²=0, Δdf=1). Similarly, the addition of information
on living in an urban environment did not improve model fit, (Model 5 vs. 4, Δχ²=0,
Δdf=1). Father’s socioeconomic status was also added to the model to test for its
association with lifespan and resulted in minimal improvement in model fit (Model 6 vs.
5, Δχ²=2.64, Δdf=1).
Table 6: Survival Analysis at the Individual Level, Including Environmental Covariates
1 2.64 5 6. Birthyr, sex, rural, city, dad_ses
1 0.14 4 5. Birthyr, sex, rural, city
1 0 3 4. Birthyr, sex, rural
1 63.02 2 3. Birthyr, sex
1 297.78 1 2. birthyr
-- -- --- 1. (baseline--no predictors)
∆df ∆χ2 Compare to model
#
Model
1 2.64 5 6. Birthyr, sex, rural, city, dad_ses
1 0.14 4 5. Birthyr, sex, rural, city
1 0 3 4. Birthyr, sex, rural
1 63.02 2 3. Birthyr, sex
1 297.78 1 2. birthyr
-- -- --- 1. (baseline--no predictors)
∆df ∆χ2 Compare to model
#
Model
-For all models, N=10033, excluding individuals who do not have a value for father’s socioeconomic status
and environment (rural or urban)
31
The diminished effect of having any children on lifespan after taking birth year
into account may indicate that birth year accounts for most of the association between
lifespan and having children. Based on the survival analyses at the individual level, the
model including birth year, gender and having any children (Table 5, Model 4) was
selected as the final model. Estimates obtained from the final model can be found in
Table 7. Odds ratios were calculated based on the parameter estimates for gender
(b=0.17, SE=0.02, p<.0001) and having any children (b=0.06, SE=0.02, p<0.006). Males
without children were used as the reference group for comparisons in survival. Based on
these estimates, females with children are 26% more likely to survive in a particular year
than males without children. Males with children were 6% more likely to survive than
men without children, while females without children are 19% more likely to survive
than childless men. With regard to birth year, the analyses indicate individuals born in a
later year were 3% more likely to survive a particular year than those born earlier
(b=0.03, SE=0.001, p<0.0001).
Table 7: Parameter Estimates, Standard Error, and Confidence Intervals for Variables in Final Model
-Model 4 in Table 5 selected as final model
Odds ratios- Men without children= 1.00, Women with children= 1.26, Men with children=1.06
Women without children= 1.19
0.02-0.10 0.02 0.06 Any children
0.13-0.20 0.02 0.17 Sex (female=1)
0.02-0.03 0.001 0.03 Birth year
4.04-4.14 0.03 4.09 intercept
95% Confidence
Interval
Standard Error Parameter
estimates
0.02-0.10 0.02 0.06 Any children
0.13-0.20 0.02 0.17 Sex (female=1)
0.02-0.03 0.001 0.03 Birth year
4.04-4.14 0.03 4.09 intercept
95% Confidence
Interval
Standard Error Parameter
estimates
32
Co-twin Survival Analyses
Table 8 displays the results of the survival analyses that include co-twin data. Co-
twin data were analyzed to investigate genetic and environmental sources that may
explain the association between having any children and lifespan. Co-twin survival
analyses were conducted separately for males and females. Model improvement was
assessed through comparison to a model consisting of birth year and having any children
at the individual level. Based on the disposable soma theory, the addition of co-twin
variables is not expected to improve the model fit, as lifespan should be dependent on
individual factors rather than familial factors.
Information on the co-twin bearing any offspring was added to Model 3 to
determine if common environmental factors account for part of the association between
fertility and lifespan, but model fit did not improve for either gender (Model 4 vs. 3,
females: Δχ²=0.2, Δdf=1; males: Δχ²=0.46, Δdf=1 ). The interaction between the co-twin
having any children and zygosity was added to the model with birth year and having any
children to test whether part of the association between lifespan and fertility is
due to genetic factors. This did not improve model fit relative to the baseline models in
either gender ( Model 5 vs. 3, females: Δχ²=0.04, Δdf=1; females: Δχ²=0.6, Δdf=1 ).
Similar procedures were applied to a second model (model #4 in Table 8 ) composed of
the number of offspring to determine whether common environmental or genetic factors
partially account for the association between lifespan and number of offspring. The
addition of co-twin number of offspring did not improve the predictability of survival in
males or females (Model 8 vs. 7, females: Δχ²=0.06, Δdf=1; males: Δχ²=0, Δdf=1 ).
33
Table 8: Co-twin Survival Analysis
0.02 7 0.14 7 1 9. birthyr, # offspring, co-twin #
offspring * zygosity
0 7 0.06 7 1 8. birthyr , # offspring, co-twin #
offspring
1.1 6 0.7 6 1 7. birthyr , # offspring
--- -- --- -- -- 6. Birthyr (baseline)
0.16 3 0.04 3 1 5. birthyr, anychild, anychild_co *
zygosity
0.46 3 0.2 3 1 4. birthyr, anychild, anychild_co
3.42 2 2.47 2 1 3. birthyr, anychild
74.08 1 361.14 1 1 2. birthyr
--- --- --- --- --- 1.(BASELINE)
∆χ2 Compare to
model #
∆χ2 Compare to
model #
∆df Model
Males Females
0.02 7 0.14 7 1 9. birthyr, # offspring, co-twin #
offspring * zygosity
0 7 0.06 7 1 8. birthyr , # offspring, co-twin #
offspring
1.1 6 0.7 6 1 7. birthyr , # offspring
--- -- --- -- -- 6. Birthyr (baseline)
0.16 3 0.04 3 1 5. birthyr, anychild, anychild_co *
zygosity
0.46 3 0.2 3 1 4. birthyr, anychild, anychild_co
3.42 2 2.47 2 1 3. birthyr, anychild
74.08 1 361.14 1 1 2. birthyr
--- --- --- --- --- 1.(BASELINE)
∆χ2 Compare to
model #
∆χ2 Compare to
model #
∆df Model
Males Females
-For models 1 - 5, N=7951 for females and N=5979 for males
-For models 6-9, N=7867 for females and N=5921 for males, excluding individuals where there is a value
for having any children (“anykids”) but a missing value for number of offspring (“numkids”)
Similarly, the addition of the interaction between co-twin number of offspring and
zygosity did not improve model fit in males or females (Model 9 vs. 7 , females:
Δχ²=0.14, Δdf=1; males: Δχ²=0.02, Δdf=1 ).
DISCUSSION
Main Findings
Contrary to the hypotheses and the disposable soma theory, in this sample, both
males and females who have children lived longer than individuals who do not have
children. Females with children lived the longest, while males without children had the
shortest lifespans. There was a main effect of birth year, with individuals from later birth
cohorts living longer than those from earlier cohorts. Also, birth year accounted for some
34
of the association between lifespan and having any children after being included as a
covariate. However, socioeconomic status and rural or urban environment did not affect
the associations found between lifespan and having children. The effect of marriage
could not be differentiated from having any children, as marital status and offspring
information were highly correlated. The association between fertility and lifespan was
due to individual specific factors rather than familial factors. The lack of model
improvement after adding co-twin information indicates that the association is not due to
common environmental factors. Similarly, the lack of improvement after adding the
interaction between the co-twin having any children and zygosity indicates that genes do
not account for the association. The results are consistent with a causal interpretation at
the individual level, as adjustments in family fertility did not increase the predictability of
survival.
Possible Mechanisms Underlying Main Findings
The increased chances of survival among females in this study may be attributable
to common trends in female longevity compared to males. In virtually every country,
females generally live longer than males (Johansson, 1989, Austad, 2006). While this
trend is somewhat universal, the exact mechanisms contributing to female longevity
remain unknown. Estrogen has been identified as a potential protective biological factor
in women, as this hormone is believed to raise high-density lipoprotein and lowers low-
density lipoprotein while testosterone appears to increase levels of harmful low-density
lipoprotein. As a result, hormone-based sex differences in lipoprotein levels are believed
to decrease the chances of death related to atherosclerosis in females relative to men
35
(Hazzard, 1985, Hazzard & Appelbaum-Bowden, 1990). Similarly, estrogen has also
been found to lower mortality in postmenopausal women receiving estrogen therapy
(Paganini-Hill, Corrada, & Kawas, 2006). Another possible reason for lower life
expectancy among men includes higher rates of dangerous behaviors and events,
including alcohol and tobacco use, suicide, homicide, and more hazardous jobs (Waldron,
1976). From an evolutionary perspective, lifespan in females is relevant not only for
reproductive purposes but also to rear offspring into adulthood. As such, menopause may
be understood as the transition from reproductive fitness to time devoted to increasing
the probability of survival in children and grandchildren (Perls & Fretts, 2001, Shanley,
et al., 2007).
Possible benefits of having children may account for the increased chances of
survival among individuals with children. The family system becomes particularly vital
in old age, as many elderly individuals depend on family for care (O’Neill, 1991). Adult
children are often the primary source of care, providing physical care, economic
resources, and social support. The responsibilities often fulfilled by adult children are
important for survival and well-being (O’Neill, 1991). Without adult children to care for
them during late life, childless elderly individuals may not have the resources and support
to survive as long as older adults with a stronger family support system. In addition,
higher overall life satisfaction among parents compared to the childless may contribute to
differences in lifespan, though research findings are mixed. In general, higher levels of
life satisfaction are associated with longer lifespan, particularly among males (Deeg &
van Zonnenfeld, 1989, Koivumaa-Honkanen et al., 2000). Some research has suggested
36
that childless individuals report lower levels of life satisfaction than individual with
children, while other research suggests that other variables, such as marriage, religiosity,
and social support, determine whether childless women experience lower levels of life
satisfaction (Beckman & Houser, 1982).
Though the results of the present study identify a main effect of birth year, it is
unclear what its exact role may be. A trend of increased life expectancy at birth over
time in the present study is consistent with existing Swedish demographic data (Statistics
Sweden, 2004). There is some possibility that birth year may be related to having
children, which in turn affects lifespan. While life expectancy in this birth cohort
increased over time, fertility rates have decreased (Statistics Sweden, 2004). However,
the results of the study do not indicate that number of children have a considerable effect
on lifespan. Another possible explanation for the increase in lifespan over time may be
the advances in medicine and health care. Demographic research points to improvements
in medical interventions as one important factor in increased life expectancy. Advances
in medicine in industrialized countries have increased survival among patients with
deadly diseases, including cancer and heart disease (Wilmoth, 2000). Increased
likelihood of survival in Sweden (as well as other countries) can be attributed to
improved chances of survival among younger individuals due to decreased mortality
caused by infectious disease, as well as the decreased mortality among the elderly due to
degenerative diseases (Wilmoth, 2000). In some developed countries, improvements in
scientific knowledge have also decreased disease-related mortality. For example, more
detailed understanding of the dangers of specific lifestyle factors have led to changes in
37
diet and substance use (e.g. alcohol and tobacco), decreasing the chances of death due to
unhealthy lifestyle (Crimmins, 1981, Wilmoth, 2000).
Considering the high correlation between having children and marriage, it is
possible that marital status may also be important. Among the childless in the present
study, 89% have never been married. Past research investigating differences in lifespan
among married versus unmarried individuals found increased survival among married
men and women (Rowland, 2007). Marriage appears to affect behaviors and resources
that may affect lifespan. Married couples appear to engage in fewer problem behaviors
than unmarried individuals. Reduction in smoking and problem drinking is particularly
pronounced in married men compared to unmarried men. Other risk-taking behaviors,
such as frequency of accidents and fighting, are lower among married men and women
(Waite, 1995). In general, marriage appears to promote self-regulation and provides
social support during stressful times (Umberson, 1987). Increased availability of material
resources among married couples, such as income, may also improve chances of survival
by increasing availability of medical care, nourishment, and providing a safer
environment (Waite, 1995).
With the percent childless reaching above 20% in both males and females, it is
possible that this particular birth cohort may be unique. Major historical events may help
explain the high rates of childlessness in this study. Economic depression in the 1930’s
may have interfered with marital stability and reduced resources available for child
rearing (Rowland, 2007). Such economic difficulties may have encouraged couples to
postpone childbearing. This delay may have increased the chances of permanent
38
childlessness. Additionally, economic difficulties may have led to instability and
marital conflict that could result in separation and subsequent childlessness. However,
power over childlessness and reasons to avoid having children may have changed over
time. With regard to generalizability in the present-day, changes in the effectiveness and
availability of birth control may affect reasons for childlessness. The reasons why
individuals did not have children during the 1920s-1950s may not be the same reason
why individuals do not have children now. Though there were relatively unreliable
contraceptive methods used as early as the late 1800’s, the birth control pill did not reach
Sweden until 1964 (Rengel, 2000), which is past the assumed reproductive period in the
present study. Beginning around the 1970’s, various communication mediums, including
newspapers and billboards, promoted awareness about personal responsibility with regard
to pregnancy and sexually transmitted diseases. Currently, contraceptive use is high as
various birth control devices are accessible to adults and teenagers alike. In the present
day, contraception is instrumental among individuals who voluntarily remain childless
for education and career-related reasons. With voluntary childlessness becoming more
common and culturally acceptable, it is uncertain whether a population assessed after the
advent of the birth control pill may present different outcomes (Rengel, 2000).
Comparing Results to Past Studies
In contrast to the findings in this study, previous studies have confirmed a
decrease in lifespan with increasing number of offspring. There are several possible
reasons why the results in this study differed from past findings. One possible
explanation may be the number of participants and the comprehensive data available for
39
this sample. Other studies (e.g., Westendorp & Kirkwood, 1998, Doblhammer & Oppen,
2003) used samples that did not include members of the general population, such as
British aristocrats. The use of such a restricted sample of individuals makes it difficult to
test specific covariates and compromises the generalizability of the results. The Swedish
Twin Registry provides a more representative sample with more detailed information on
specific covariates, such as socioeconomic status, birth year, environment, and marriage.
In addition, the quality of data solely based on church and community records (as was the
case in some of these studies) may not be as carefully monitored as the data used in the
present study. The Swedish twin data is collected based on the administration of
comprehensive questionnaires and selected information is regularly updated, such as the
causes of death. The use of survival analyses in this study may have provided more
accurate and detailed results. While some of the previous studies depended solely on
correlations and comparisons of mean ages at death, survival analyses in this study
provided detailed information on main effects, interactions, and the effects of specific
covariates. The participants, data quality, and analyses are likely reasons why the results
of the present study differed from past research.
Strengths and Limitations
With regard to the strengths of this study, it is the first to examine the disposable
soma theory using twin data. This study provides insight into familial and non-shared
environmental factors in a manner which, in turn, allows a more detailed evaluation of
causal associations. Also, the present study used a relatively modern sample. Past studies
have resorted to records on samples existing several centuries ago. This study also used a
40
large sample from the general population, whereas previous studies have relied on small
unrepresentative samples. In addition, sufficient power due to the large sample size is a
strength of this study. This ensures that the lack of evidence supporting the disposables
soma theory is not due to low power.
While the present study has its strengths, there are also some weaknesses. Even
though the Swedish sample may be more representative of the general population than a
sample of British aristocrats, the sample is almost entirely Caucasian. Such a
homogeneous racial composition may impact the generalizability of the results to other
populations. However, homogeneity could be considered beneficial, as heterogeneity
could introduce confounding variables that would require the inclusion of additional
covariates. Another weakness in this study is the information available on number of
children. Ideally, a thorough study of disposable soma theory would assess detailed
histories of miscarriages and abortions, in addition to number of offspring. The only
information available for this study was the reported number of children. Also, an item
inquiring about the number of children could be interpreted differently by some
participants. The items related to offspring in Q67 and Q70 simply ask if the participant
has children and how many they have. Therefore, individuals who adopted children
could be undifferentiated from the individuals who had biological offspring, unless they
volunteered information about the adoption. The number of offspring item does not
clarify if individuals had children who died, making it possible that some individuals
underreported number of children or were placed in the same category as the childless.
41
Future Directions
This study carefully took into account some of the shortcomings of past research
on the relationship between fertility and lifespan. Future studies investigating fertility and
longevity could add to the findings in the present study by including data on
miscarriages, abortions, and deceased children. By including this data in the analyses,
future studies can test the disposable soma theory in a manner that more accurately takes
into account the physiological resources invested in any pregnancy. Also, further
investigation of the disposable soma theory may want to include parent’s age at childbirth
and number of male and female infants in order to explore possible differences in
parent’s survival among these groups. While there are still some factors that are open to
question, present study has examined the question in a manner unseen in the previous
literature. The nature of this research question encompasses topics relevant to
evolutionary theory, biology, and psychology, and the results of this study will contribute
relevant knowledge to these fields of study.
42
References
Austad, S.N. (2006).Why women live longer than men: sex differences in longevity.
Gender Medicine, 3, 79-92.
Chapman, T., Liddle, L.F., Kalb, J.M., Wolfner, M.F., Partridge, L. (1995). Cost of
mating in Drosophila melanogaster females is mediated by male accessory gland
products. Nature, 373, 241-244.
Charlesworth, B. (2000). Fisher, Medawar, Hamilton and the evolution of aging.
Genetics, 156, 927-931.
Crimmins, E.M. (1981). The changing pattern of American mortality decline, 1940-77
and implications for the future. Population and Development Review, 7, 555-572.
Beckman, L.J., & Houser, B.B. (1982). The consequences of childlessness on social-
psychological well-being of older women. Journal of Gerontology, 37, 243-250.
Doblhammer, G. (2000). Reproductive history and mortality later in life: A comparative
study of England and Wales and Austria. Populations Studies, 54, 169-176.
Dobhammer, G., Oeppen, J. (2003). Reproduction and longevity among the British
peerage: the effect of frailty and health selection. Proceedings of the Royal
Society of London Biological Sciences, 270, 1541-1547.
Downey, D.B. (1995). When bigger is not better: family size, parental resources, and
children’s educational performance, American Sociological Review, 60, 746-761.
Drenos, F., Kirkwood, T.B.L. (2004). Modelling the disposable soma theory of aging.
Mechanisms of Ageing and Development, 126, 99-103.
Fiscella, K., Franks, P., Gold, M.R., Clancy, C.M. (2000). Addressing socioeconomic,
racial, and ethnic disparities in health care. The Journal of the American Medical
Association, 283, 2579-2584.
Fowler, K., Partridge, L. (1989). A cost of mating in female fruitflies. Nature, 338, 760-
761.
Gadgil, M., Solbrig, O.T. (1972). The Concept of r- and K- Selection: evidence from wild
flowers and some theoretical considerations. The American Naturalist, 106, 14-
31.
Gavrilov, L.A., Gavrilova, N.S. (2002). The reliability theory of aging and longevity.
Journal of Theoretical Biology, 213, 527-545.
43
Gavrilov, L.A., Gavrilova, N.S. (2004). Does exceptional human longevity come with a
high cost of Infertility?: Testing the evolutionary theories of aging. Annals of the
New York Academy of Sciences, 2004, 513-517.
Harshman, L.G., Haberer, B.A. (2000). Oxidative stress resistance: A Robust
correlated response to selection in extended longevity lines of drosophila
melanogaster. The Journal of Gerontology Series A: Biological Sciences and
Medical Sciences, 55, 415-417.
Harshman, L.G., Moore, K.M., Sty, M.A., Magwire, M.M. (1999). Stress resistance and
longevity in selected lines of Drosophila melanogaster. Neurobiology of Aging,
20, 521-529.
Hazzard, (1985). Atherogenesis: why women live longer than men. Geriatrics, 40, 42-
51.
Hazzard W.R., & Appelbaum-Bowden, D. (1990). Why women live longer than men: the
biologic mechanism of the sex differential in longevity. Transactions of the
American Clinical and Climatological Association, 101, 168-189.
Helle, S., Lummaa, V., Jokela, J. (2004). Are reproductive and somatic senescence
coupled in humans? Late, but not early, reproduction correlated with longevity in
historical Sami women. Proceedings of the Royal Society of London Biological
Sciences, 272, 29-37.
Johansson, S. (1989). Longevity in women. Cardiovascular Clinics, 19, 3-16.
Kirkwood, T.B.L. (1977). Evolution of ageing. Nature, 270, 301-304.
Kirkwood, T.B.L (2002). Evotion of ageing. Mechanisms of Ageing and Development,
123, 737-745.
Kirkwood, T.B.L, Holliday, F.R.S. (1979). The evolution of ageing and longevity.
Proceedings of the Royal Society of London Biological Sciences, 205, 531-546.
Kirkwood, T.B.L., Rose, M.R. (1991). Evolution of Senescence: late life sacrificed for
reproduction. Philosophical Transaction of the Royal Society of London
Biological Sciences, 332, 15-24.
Koivumaa-Honkanen, H., Honkanen, R., Viinamaki, H., Heikkila, K., Kaprio, J., &
Koskenvuo, M. (2000). Self-reported life satisfaction and 20-year mortaility in
healthy Finnish adults. American Journal of Epidemiology, 152, 983-991.
44
Korpelainen, H. (2000). Fitness, reproduction and longevity among European
aristocratic and rural Finnish families in the 1700s and 1800s. Proceedings of the
Royal Society of London Biological Sciences, 267, 1765-1770.
Lambeth, J.D. (2007). Regulation of Nox Duox enzymatic activity and expression.
Free Radical Biology and Medicine, 43, 319-331.
Le Bourg, E. (2001). A mini-review of the evolutionary theories of aging. Is it the time to
accept them? Demographic Research, 4, 1-28.
Le Bourg, E. (2007). Does reproduction decrease longevity in human beings?. Ageing
Research Reviews, 6, 141-149.
Lycett, J.E., Dunbar, R. I., Voland, E. (2000). Longevity and the costs of reproduction in
a historical human population. Proceedings of the Royal Society of
London Biological Sciences, 31, 31-35.
MacArthur, R.H., Wilson, E.O. (2001). The Theory of Island Biogeography. Princeton
University Press.
McArdle, P.F. (2006). Does Having Children Extend Life Span? A Genealogical Study of
Parity and Longevity in the Amish. The Journals of Gerontology SeriesA:
Biological Sciences and Medical Sciences, 61,190-195.
Medawar, P.B. (1952). An unsolved problem in biology. London: HK Lewis.
O’Neill, C. (1991). Home care of the elderly: A family perpective. Advances in Nursing
Science, 13, 28-37.
Paganini-Hill, A., Corrada, M. M., & Kawas, C.H. (2006). Increased longevity in older
users of postmenopausal estrogen therapy: the Leisure World Cohort Study.
Menopause, 13, 12-18.
Pappas, G., Queen, S., Hadden, W., Fisher, G. (1993). The increasing disparity in
mortality between socioeconomic groups in the United States, 1960 and 1986.
New England Journal of Medicine, 15, 103-109.
Penn, D.J., Smith, K.R. (2007). Differential fitness costs of reproduction between the
sexes. Proceedings of the National Academy of Sciences, 104, 553-558.
Perls, T.T., & Fretts, R.C. (2001). The evolution of menopause and human life span.
Annals of Human Biology, 28, 237-245.
45
Rengel, M. (2000). Encyclopedia of birth control. Santa Barbara, CA: Greenwood
Publishing Group, 221-234.
Rowland, D.T.(2007) Historical trends in childlessness. Journal of Family Issues, 28,
1311-1338.
Shanley, D.P., Sear., R., Mace, R., Kirkwood,T.B.L. (2007). Testing evolutionary
theories of menopause. Proceeds of the Royal Society Biological Sciences, 274,
2943-2949.
Statistics Sweden, (2004). Men and women in Sweden: facts and figures 2004.
www.scb.se.
Tartar, M., Chien, S.A., Priest, N.K. (2001). Negligible senescence during reproductive
dormancy in drosophila melanogaster. The American Naturalist, 158, 248-258.
Thomson, E., Hanson, T.L., McLanahan, S.S. (1994). Family structure and child well-
being: economic resources vs. parental behaviors. Social Forces, 73, 221-242.
Thomas, F., Teriokhin, A.T., Renaud, F., De Meeus, T., Guegan, J.F. (2000). Human
longevity at the cost of reproductive success: evidence form global data. Journal
of Evolutionary Biology, 13, 409-414.
Umberson, D. (1987). Family status and health behaviors: social control as a dimension
of social integration, Journal of health and Social Behavior, 28, 306-319.
Vinogradov, A.E. (1998). Male reproductive strategy and decreased longevity. Acta
Biotheoretica, 46, 157-160.
Waite, L.J. (1995). Does marriage matter? Demography, 32, 487-507.
Waldron, I. (1976). Why do women live longer than men? Journal of Human Stress, 2, 2-
13.
Weismann, A. (1891). Essays upon heredity and kindred biological problems. London:
Oxford University Press.
Westendorp, R.G.J., Kirkwood, T.B.L. (1998). Human longevity at the cost of
reproductive success. Nature, 396, 743-746.
46
Williams, G.C. (1957). Pleiotropy, natural selection, and the evolution of senescence.
Society for the Study of Evolution, 11, 398-411.
Wilmoth, J.R. (2000). Demography of longevity: past, present, and future
trends.Experimental Gerontology, 35, 9-10.
Abstract (if available)
Abstract
The disposable soma theory suggests that reproduction occurs at the expense of physiological resource maintenance (Kirkwood, 1977). Elaborating on this theory, some researchers have proposed that a negative relationship exists between longevity and reproduction (e.g. Westendorp & Kirkwood, 1998). Studies examining this relationship were often restricted to isolated groups that may be considered questionable representations of the general population (Westendorp & Kirkwood, 1998, Korpelainen, 2000). The current study examined whether the number of children a woman bears during her lifetime decreases her lifespan, as posited by the disposable soma theory. Female and male twin pairs from the Swedish Twin Registry were analyzed using a co-twin control model and survival analysis. Twin data helps elucidate whether this phenomenon is accounted for by familial factors or non-shared environmental factors. Based on the results of the survival analyses, there were main effects of birth year, gender, and having any children in predicting lifespan. More specifically, females, individuals who had children, and individuals born in later cohorts versus earlier cohorts had increased chances of survival. Also, birth year accounted for part of the association between having any children and lifespan. The results are consistent with a causal interpretation at the individual level, as the inclusion of co-twin information did not increase the predictability of survival. The evidence in this study does not support the disposable soma theory and sheds light on other factors that might affect lifespan.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Predicting cognitive decline and dementia in elderly twins from indicators of early life oral health
PDF
Drinking alcohol to improve mood partially mediates the relation between major depression and alcohol dependence
PDF
Sources of stability and change in the trajectory of openness to experience across the lifespan
PDF
The effects of familism and cultural justification on the mental and physical health of family caregivers
PDF
Effects of stress and the social environment on childhood asthma in the children' s health study
PDF
Marital conflict and parenting: moderating effects of ethnicity and parent sex
PDF
Decisions to ratify the Kyoto Protocol: a Latin American perspective on poliheuristic theory
PDF
Selected papers on the evaluation of healthcare costs of prematurity and necrotizing enterocolitis using large retrospective databases
PDF
Pregnancy in the time of COVID-19: effects on perinatal mental health, birth, and infant development
PDF
Effect of biomass fuel exposure on infant respiratory health outcomes in Bangladesh
PDF
Understanding the role of population and place in the dynamics of seasonal influenza outbreaks in the United States
Asset Metadata
Creator
Chereji, Elizabeth
(author)
Core Title
Examining the effect of fertility on lifespan using data from the Swedish Twin Registry
School
College of Letters, Arts and Sciences
Degree
Master of Arts
Degree Program
Psychology
Publication Date
11/01/2009
Defense Date
05/18/2009
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
aging,disposable soma theory,Evolution,lifespan,longevity,OAI-PMH Harvest
Place Name
Sweden
(countries)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Prescott, Carol A. (
committee chair
), Gatz, Margaret (
committee member
), Manis, Franklin R. (
committee member
)
Creator Email
chereji@usc.edu,elizabeth.chereji@gmail.com
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-m2702
Unique identifier
UC1291917
Identifier
etd-Chereji-3122 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-269883 (legacy record id),usctheses-m2702 (legacy record id)
Legacy Identifier
etd-Chereji-3122.pdf
Dmrecord
269883
Document Type
Thesis
Rights
Chereji, Elizabeth
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Repository Name
Libraries, University of Southern California
Repository Location
Los Angeles, California
Repository Email
cisadmin@lib.usc.edu
Tags
disposable soma theory
lifespan
longevity