The literature implies a 'diminished entelechy' hypothesis in which birth cohorts subjected to relatively many or relatively virulent environmental insults early in life do not realize their otherwise expected lifespan. No direct test of this hypothesis appears in literature.

We test the hypothesis directly by measuring the association between mortality in the first 5 years and life expectancy at age 5 for male and female cohorts born in Sweden (1751-1912), Denmark (1835-1913), and England and Wales (1841-1912). The methods control for trends, seasonal cycles, and other forms of autocorrelation that could induce spurious associations.

Our results support the hypothesis in that life expectancy at age 5 fell below the values expected from history in cohorts in which child mortality before age 5 increased over its expected value. We find no evidence for culling effects in which a cohort remaining after suffering relatively many environmental insults may be smaller but hardier than expected.

These findings converge with individual-level studies and suggest that suffering relatively virulent or many environmental insults during childhood reduces the subsequent lifespan of birth cohorts.

Social position and early life environment consistently correlate with health and disease in human populations and animal models. Social epidemiology has succeeded at describing relationships between social factors and health outcomes; however, fundamental questions of etiology and causation persist. Studies on the health effects of relative social position face complex challenges including the clustering of risk factors, the inability to adequately control confounders, questions of temporality, and difficulty in measuring complicated social environments.

One central hypothesis concerning the link between social status and health focuses on differential exposure and response to stressors, suggesting that the stress of low social status throughout the lifecourse chronically activates biological stress responses, increasing risk of disease. Insight into the biological underpinnings of these relationships in humans can be furthered by innovative approaches to social epidemiology, including the application of carefully designed animal studies.

In Chapter 1, the relationship between social status and biological and behavioral outcomes is examined in group-housed male laboratory rats. Rats matched on weight and maternal behavior at weaning form a social hierarchy in their homecage without intervention. Results demonstrate a social gradient in endocrine response to acute stress, exploratory behaviors, and cognitive performance. Subordinate rats showed a blunted corticosterone response to acute stress, lowest levels of exploratory behavior, and poorest cognitive performance in the homecage. Stress of social subordination likely causes the observed differences. This study improves on animal models of social hierarchy by fully characterizing and controlling for early life environment. Results demonstrate that laboratory rats can be used as a model for social hierarchy in humans.

Chapter 2 examines the interaction between early life experience and adult social position, comparing laboratory rats and humans. For the human study, college students reported on their current subjective social status and family home ownership during their childhood. The rat study used the model of social hierarchy described in Chapter 1. An interaction between early life experience and adult social status was identified in both species. Rats and humans who experienced early life adversity (low levels of maternal care in rats, low childhood socioeconomic status in humans) represented both the highest and lowest levels of IL-6 in adulthood, depending on their social status as young adults. Therefore, early adversity may not have a monotonically negative effect on later life health, but appears to alter responsiveness to later exposures.

Chapter 3 examined the formation of social hierarchies within cages of group-housed male laboratory rats. Rats were housed together at weaning and competed for access to chocolate and water within their homecage throughout the juvenile, adolescent, and adult periods. Late adolescence (postnatal day 45) emerged as a crucial time for hierarchy development. All cages showed a distinct hierarchy at 45 days of age, and half of the groups maintained stable hierarchies from postnatal day 45 into adulthood. Results suggest that careful consideration of developmental windows would improve studies of social status in animal models.

This approach improves on translational research by providing an animal model specifically designed to address public health questions. Results of these studies suggest that an animal model of social hierarchy can inform questions from social epidemiology and further the understanding of relationships between social experience and health in human populations.

This dissertation examines maternity leave policies, utilization and consequences in three separate, but related, papers. In the Introduction, I describe the ways state-level leave policies influence, but do not perfectly predict, utilization and then discuss the heterogeneity of women’s experiences during pregnancy and post-partum, and how that heterogeneity informs interpretation of results in this field.

In Chapter 2 (“Laboring until labor: the prevalence and correlates of antenatal maternity leave in the United States”), I use a national survey of English-speaking women to examine maternity leave taken in the final weeks of pregnancy. I describe individual-, employer-, and policy-level correlates of antenatal leave (ANL), focusing in particular on variation in state temporary disability insurance (TDI) laws. I find that two-thirds of women in this sample stopped working more than a week before their due date, and that state policies significantly influenced leave-taking, even after controlling for characteristics of women and their jobs. While certain individual-level characteristics of women’s work were important, employer policies were not significantly associated with ANL use or duration. The literature does not yet include a national analysis of antenatal leave and its correlates. This paper fills that gap and sets up the following chapter on the consequences of taking ANL.

In Chapter 3 (“Antenatal maternity leave and childbirth using the First Baby Study: a propensity score analysis”), I use survey data from a prospective cohort in Pennsylvania to test whether women who take maternity leave at the end of pregnancy have better labor and delivery outcomes. In this sample of women giving birth for the first time, fully half of the sample did not stop working before delivery. Using propensity score matching, I find that women who did stop working at least two days prior to delivery experienced more negative delivery outcomes, including an increased likelihood of unplanned Cesarean section. This paper highlights the strong selection into leave-taking, particularly in a context of limited leave availability.

In Chapter 4 (“The impact of California’s Paid Family Leave law on maternal time use”), I shift focus to postnatal maternity leave and use the American Time Use Survey, a nationally-representative dataset collected by the U.S. Census Bureau, to examine whether the country’s first paid family leave (PFL) law increased the amount of time exposed women spent on childcare and decreased their time in paid work. Using a difference-in-difference-in-difference approach with variation in state, time, and age of youngest child, I find that after PFL, women in California significantly increased the time they spent with children in their care and slightly reduced their time spent working. Exploiting a natural experiment, I am able to avoid some of the selection concerns present in the previous chapter, but the daily diary nature of the data do not illuminate whether time use changed due to leave-taking, schedule changes, or some other factor.

The results of these studies will inform future research on maternity leave and maternal health, and guide policymaking with regards to targeting and promoting maternity leave policies.

The critical and sensitive periods model, a key component of the life course health and development (LCHD) framework, describes windows of growth, development, or change during which exposures can permanently affect the structure or function of the body and influence its trajectories for health. Research on critical and sensitive periods has traditionally focused on the impact of early life influences on later life health. This focus on infancy and youth is due to the rapid pace at which development and growth occurs, which can make individuals particularly susceptible to the influences of risk and protective factors.

Perimenopause—the transition period bridging women’s reproductive and post-reproductive years—is another window, however, in which rapid physiological changes occur, yet minimal research has investigated this time as a critical or sensitive period. The three papers included in this dissertation investigate the health of women at mid-life and explores how their experiences during the menopausal transition affect their chronic disease risk and longevity.

The first paper, a scoping review, compares the physiological similarities between perimenopause and puberty and their respective associations with some of the most prevalent chronic conditions in the United States. The second paper, a longitudinal cohort analysis of women undergoing the menopausal transition, explores whether perimenopause represents a sensitive window for stress responsivity. Using survey data from the Study of Women’s Health Across the Nation (SWAN), a national study of 3,300 women at midlife, the relation between psychological stress, perimenopause, and fibrinogen—a biomarker for systemic inflammation—is examined. The third paper, a time series analysis of population cohorts in France and England in Wales during the 19th Century, tests whether the age range at which perimenopause occurs at the population level is a sensitive window for longevity. The relation between cohort mortality at ages 45-49 and life expectancy at age 60, specifically, is examined using data from the Human Mortality Database.

The primary findings from these three papers are as follows: 1) Perimenopause shares many physiological similarities with puberty, a well-documented sensitive window for a number of health outcomes. The interaction of certain behaviors or exposures with the significant hormonal shifts during perimenopause, moreover, appears associated with risk for mood disorders, metabolic-related morbidities, autoimmune diseases, cardiovascular disease, cancer, musculoskeletal disorders, and premature mortality. 2) Women’s perceptions of stress appear to change during the course of the menopausal transition; however, such changes are not associated with adverse physiological effects, as measured through changes in fibrinogen. 3) Adverse environmental conditions during perimenopause appear significantly related to decreased longevity among women in France and England and Wales during the 19th Century.

Background and objectives

Methodology

Results

Conclusions and implications

Lay summary

Background. Social epidemiology has become increasingly interested in the effect of early-life experiences on health outcomes in adulthood. It is now widely recognized that early-life adversity (such as being abused or neglected) is an important risk factor for adulthood depression and other psychological disorders. Research emerging from the developmental programming paradigm posits that suboptimal early-life environments can encode vulnerability to later-life psychological impairment through permanent alterations in biological systems, particularly the neuroendocrine and immune systems. However, a subset of persons exposed to early-life adversity appear to be psychologically resilient, experiencing no or very minimal mental health problems later in life. It is a critical goal in epidemiology to ascertain whether these distinct life-course psychological trajectories are mediated by distinct biological characteristics. In this dissertation, I used data from a unique cohort study to explore the relationship between participants' early adversity exposure, their longitudinal history of psychological distress, and their neuroendocrine and immunological stress reactivity systems.

Methods. The dataset comprised a subsample of the ongoing Whitehall II occupational epidemiologic cohort study. The subsample was a group of healthy older male and female Whitehall II members (n = 543) who participated in the 2008 Heart Scan Study (HSS), which was primarily designed to investigate the association between physiological reactivity to an experimental laboratory stressor and sub-clinical coronary artery calcification. The Heart Scan Study data were subsequently linked to six waves of the participants' previously collected Whitehall II study data (spanning years 1985-2004).

I defined exposure to early-life adversity using a composite variable derived from questions collected during both the Heart Scan Study and in one Whitehall II wave. Participants were categorized as exposed if they reported experiencing physical abuse, parental death, separation from their mother or time spent in an orphanage for 1+ years, serious familial mental illness or substance abuse, parental divorce, frequent parental conflict, and/or a harsh parenting style before the age of 16. Participants' history of psychological distress during adulthood was derived from their Whitehall II data, which included General Health Questionnaire-28 (a measure of psychiatric symptoms) scores at each wave. I used two of the physiological reactivity parameters measured in the HSS - cortisol (a key stress hormone) and interleukin-6 (an important inflammatory protein marker) - as my outcome variables.

First, I used piecewise multilevel growth curve modeling to examine whether study participants' joint early-life adversity and psychological history status - i.e., early-life adversity exposure (yes/no) and recurrent psychological symptoms (yes/no) - predicted mean differences in their cortisol response trajectories. Second, I used the mixture model technique of group-based trajectory modeling to identify whether distinct clusters of stress-reactivity patterning existed within cortisol and interleukin-6, and if so, whether the clustering pattern within on biomarker was associated with clustering in the other. Lastly, I used an extension of group-based trajectory modeling to determine whether early-life adversity and longitudinal history of psychological distress predicted study members' membership in those clusters.

Results. Using the piecewise multilevel growth curve analysis, I found that participants exposed to early-life adversity had dysregulated cortisol reactivity trajectories, but that the pattern of those trajectories differs based on their history of psychological distress during adulthood. Adversity-exposed adults with a history of recurrent psychological distress had blunted cortisol responses compared to non-adversity-exposed adults, as well as decreased overall cortisol output. In contrast, adversity-exposed adults with minimal psychological distress had elevated levels of cortisol at baseline and prolonged responses to stress. There was also some evidence that this association was modified by gender.

Using group-based trajectory modeling, I found that distinct clusters did appear to exist within both the cortisol and IL-6 data, although these results need to be interpreted with extreme caution due to statistical flaws in the modeling approach. The four cortisol clusters were distinguished from each other by both mean baseline cortisol levels as well as slightly different reactivity patterns during the experimental laboratory session; the clusters corresponded to a "low-declining" group, a "middle-flat" group, a "higher-flat" group, and finally a "high-declining" group. Within the IL-6 data, the fitted solution comprised five trajectory groups, all of which shared a similar growth pattern (steadily increasing) but had slightly different mean baseline values. The cortisol and IL-6 trajectory groups appeared to be moderately correlated with each other, such that "opposite" biomarker groups had a slight tendency to associate together; however, the principal finding from this dual trajectory analysis was that membership in a given cortisol or IL-6 group is not a very sensitive predictor of group membership in the other biomarker.

In the last piece of analysis, I found that adversity-exposed individuals who subsequently had recurring psychological distress symptoms were marginally more likely to fall into the lowest cortisol trajectory group. Otherwise, cortisol and IL-6 trajectory group membership was not significantly predicted by any combination of early adversity exposure and psychological distress history. Participants' age, body mass index, and socioeconomic status did predict both cortisol and IL-6 group memberships.

Conclusions. My findings partially support the hypothesis that different patterns of atypical stress reactivity - in this case, blunted vs. heightened cortisol responses - may be encoded in the aftermath of early-life adversity, and that these patterns in turn may promote divergent susceptibilities to lifetime psychological disorder. Inflammatory stress reactivity, as measured by interleukin-6, does not appear to play a role in this association. However, distinct clusters of stress reactivity patterning do appear to exist within cortisol and interleukin-6, and may be predicted by theoretically important sociodemographic variables, an observation that should be explored in future research. Although is unclear how much of the burden of depression/anxiety symptoms the observed cortisol patterns might explain among adversity-exposed populations, my findings do provide support for the notion that social experiences "get under the skin" and contribute to population ill-health. This work may lead to a better understanding of the biological sequelae of early-life adversity, and the effects those sequelae can have on psychological risk and resilience throughout the lifecourse.