## Abstract

This paper examines how the timing of childhood exposure to armed conflict influences both the magnitude of the impact it has on later-life health and the pathways through which those impacts manifest. Utilizing the Survey of Health and Retirement in Europe, we examine cohorts of children during World War II. We find that cohorts born during the war show the largest negative effects of exposure on health in later life. The pathways also vary the timing of exposure. Consistent with a latent critical period process, children born during the war experienced increased risk of poor health and illness in childhood, as well as adult cardiometabolic conditions and poor functional health. Conversely, cohorts born before the war experienced more indirect pathways consistent with cumulative disadvantage processes and institutional breakdown. These pathways include stunted socioeconomic attainment, increased risk behaviors, and poorer mental health. Overall, this study emphasizes that the timing of exposure is critical to understanding the long-term health effects of war.

The developmental impact of a succession of life transitions or events is contingent on when they occur in a person's life.

## Introduction

The life course has become a dominant theoretical paradigm with which to understand individual and population health. A growing body of research taking a life course perspective—alternatively known as the long arm of childhood, early-life origins of disease, and the developmental origins of health and disease—has shown that early-life exposure to adverse social and environmental conditions can have a dramatic impact on multiple dimensions of health across the life course (Haas 2008; Hayward and Gorman 2004). This work has examined a wide array of early-life conditions, including exposure to armed conflict.

Growing up amid conflict is associated with an increased likelihood of developing a wide array of health conditions, including cardiovascular disease, stroke, diabetes, mental health problems, and low self-rated health (Akbulut-Yuksel 2017; Akresh et al. 2012; Alastalo et al. 2009; Kesternich et al. 2014, 2015; Lindeboom and Ewijk 2015). Studies have found that childhood exposure to contexts of war and conflict also adversely impacts cognitive development and human capital accumulation through the interruption of schooling (Chamarbagwala and Morán 2011; Ichino and Winter–Ebmer 2004; Kesternich et al. 2014; Shemyakina 2011). This conflict-induced stunting of socioeconomic attainment has further negative implications for later-life health.

The literature has been careful to characterize war as a broad contextual phenomenon extending far beyond individual exposure to physical violence and affecting the life course much like natural disasters (Elliott 2015; Howell and Elliot 2019; Williams 2013, 2015). Indeed, pronounced violence tends to overshadow other factors, such as institutional and infrastructural breakdowns and contexts of poverty and stress. As such, wars and armed conflict can have enduring impacts on health by severely shifting the quantity and composition of nutritional resources and by limiting people's ability to work and to access markets, schooling, and other resources (Akresh et al. 2012; Akresh et al. 2011; Austin and McKinney 2012). War also induces displacement and often results in substantial destruction of property and family/household structures (Shemyakina 2011; Williams 2013, 2015). Acknowledging the broader social and institutional impacts of war expands the array of pathways through which it can impact life course health trajectories. In addition to direct physical insults, war can initiate deleterious health issues through increased social instability and disruption of crucial social institutions, thus harming physical, cognitive, and psychosocial development and stunting socioeconomic attainment (Lee 2014).

To date, research has underexplored two crucial and interrelated aspects of conflict's long-term impact on health. The first is whether war has heterogeneous effects depending on the timing of childhood exposure. Childhood is a period of life characterized by critical developmental changes, resulting in significant variation in vulnerability to health insults across the childhood years. The second is a lack of clarity in the processes and mechanisms through which exposure to armed conflict adversely impacts adult health. Although scholars have explored some potential pathways (e.g., reduced educational attainment and income, and increased mental health risk), these efforts have been disconnected. The examination of war has largely not been formulated within a life course theoretic framework of early-life adversity, socioeconomic attainment, and lifelong health trajectories. Moreover, the relative importance of various pathways has not been examined. Furthermore, we propose that these two issues may be interrelated—that timing of exposure not only may matter for the magnitude of its impact but also may differentiate the pathways through which it manifests.

By adopting a life course perspective focused on European cohorts who were children during World War II, the present study aims to shed light on the differential timing of early-life exposure to armed conflict and later-life health as well as the pathways through which these insults unfold. The conflict, spanning 1939–1945, resulted in catastrophic losses to population, large-scale displacement, destruction of towns and cities, and fundamental political shifts, with social instability, poverty, and hunger commonplace during and after the war. Furthermore, children were forced to witness the horrors of battles and bombings in which civilian populations were frequent targets.

## The Intersection of War and Developmental Critical Periods

Health trajectories emerge at the social-biological interface (Bartley 2016). By offering a set of organizing principles that describe the etiologic implications of socially embedded developmental and biological changes, the life course approach provides a theoretical framework for understanding that interface (Elder et al. 2003). A core principle of life course theory posits that the impact that events, exposures, and transitions have on life trajectories depends on their timing within the life course (Elder 1998). Timing matters because important developmental changes occur during specific periods of life. The nature of change during these periods may be biological—as with puberty or early neurocognitive development—or social—as during the transition to adulthood. During these critical/sensitive periods, adverse exposures or events may fundamentally alter developmental outcomes in ways that similar exposures occurring outside those periods would not.

An important thread within critical period research has investigated how material and nutritional deprivation during early life can lead to permanent physiological adaptations to the structure and functioning of important body systems, including permanently elevated blood pressure and dysregulation of glucose and lipid metabolism (Kuh and Ben Shlomo 2004; Lindeboom and Ewijk 2015; Scholte et al. 2015). Such physiological adaptations result in higher propensities for developing cardiovascular and cerebrovascular disease, Type 2 diabetes, and respiratory diseases. Beyond these physiological effects, evidence suggests childhood exposure to conflict also inhibits cognitive development, reduces labor market earnings, increases risk of unemployment, and psychologically scars individuals, increasing their risk for later-life mental health problems, such as depression, anxiety, and schizophrenia (Chapman et al. 2004; Gould et al. 2015; Sixsmith et al. 2014; Summerfield 2000). Again, empirical work suggests that prenatal and neonatal stages as well as early childhood are decisive life periods that are particularly sensitive to adverse exposure and events that may hinder development and create inequalities in later life (Brandt et al. 2012; Ferraro 2011; Haas 2008; Kuh and Ben Shlomo 2004).

## Heterogeneous Pathways: Critical Periods and Cumulative Risks

Emphasizing the timing of exposure as a source of heterogeneity introduces the possibility for differential pathways through which war may affect later-life health. Life course theory also provides a set of theoretical processes that explain how insults in early life influence lifelong health trajectories. The critical periods and the cumulative risk processes provide insight into how the timing of exposure to war can create heterogeneous pathways to poor health in later life.

As discussed earlier, exposures to adversity during developmentally critical or sensitive periods can permanently alter important outcomes, ultimately resulting in a wide variety of negative consequences across the life course (Ferraro 2011; Kuh and Ben Shlomo 2004). One of the most cited bodies of research supporting critical period effects is the fetal origins hypothesis, which suggests that nutritional deprivation in utero alters the process of organ and tissue formation (Barker 1990). When faced with resource deprivation, the fetus adapts its growth and development to prioritize structures and tissues that maximize its chance of survival within the womb. The period of gestation through the first few years of life has been identified as particularly vulnerable to shocks that trigger physiological adaptations (e.g., altering lipid and glucose metabolism) (Gluckman et al. 2008; Lumey et al. 2011; Wang et al. 2018). However, such adaptations may be disadvantageous in later life. This is particularly the case when the child, who experienced in utero and neonatal deprivation and whose body is biologically “programmed” to expect an environment of scarcity, grows up in a world without resource deprivation, thereby creating a phenotype-environment mismatch. This mismatch can lead to rapid rebound growth in the period after age 4, increasing the risk of obesity and cardiometabolic diseases in mid- to late adulthood (Dietz 1997). As such, one of the key features of critical period effects is that they frequently remain latent during a large part of the life course. Critical periods also imply that these negative exposures have direct effects on health in late life, which are largely robust to amelioration from improvements in social conditions.

Armed conflict often creates circumstances of severe material and nutritional deprivation that can trigger physiological adaptations. For example, in utero nutritional deprivation resulting from the Dutch hunger winter at the end of World War II created substantial cardiovascular risk for the cohorts exposed, independent of subsequent socioeconomic factors (Ekamper et al. 2014). Given both the theoretical rationale and prior empirical literature, individuals exposed to armed conflict during the earliest years of life may be at particular risk via processes of physiological adaptation.

Cumulative risk processes reflect another core insight of the life course perspective—that individual outcomes at any particular point can be understood only within the context of the cumulative impact of prior lived experience. Individual health trajectories result from the accumulation of salubrious inputs and noxious risks deriving from social, environmental, and behavioral exposures over the life course. Armed conflict can have important implications for risk accumulation. The substantial familial and social dislocation and institutional breakdown that often accompanies conflict can have serious detrimental impacts on myriad social processes that shape life trajectories and, ultimately, health. For example, war-induced experience of parental death, financial hardship, and psychological stress may create additional accumulation of risk for poor physical and psychological health. It may also induce unhealthy coping behaviors, such as smoking and excessive alcohol consumption. Similarly, stunting of socioeconomic attainment may occur due to the curtailment or disruption of schooling or the necessity for older children to enter the labor market to help support the family (Shemyakina 2011). In turn, reduced socioeconomic attainment substantially increases the risk of poor health throughout adulthood (Kitagawa and Hauser 1973; Link and Phelan 1995).

There is reason to suspect that the timing of war exposure may also influence cumulative risk processes and may do so differentially across developmental domains. Developmental outcomes (e.g., physical, cognitive, socioemotional) have heterogenous normative trajectories with unique critical/sensitive periods that span somewhat different stages of childhood. Therefore, disruptions occurring at various points in childhood are likely to have heterogeneous consequences across developmental outcomes. As discussed earlier, the literature on cardiometabolic critical periods emphasizes the prenatal period through age 4. The temporary closing of schools is unlikely to have much, if any, impact on children who are not yet of school age. Therefore, we might not expect exposure to armed conflict in the first five years of life to have the same adverse impact on educational and occupational outcomes as we might for school-aged children and adolescents. Conversely, adolescence is a critical period for mental health and the risk of depression. Depression risk rises substantially beginning from age 10 to 12 (Kessler et al. 2001), with the peak increase in depressive symptoms occurring at age 13 for girls and 16 for boys (Kwong et al. 2019), and the peak incidence of depression occurring between ages 15 and 18 (Hankin et al. 1998). Thus, war exposure may be particularly noxious for mental health if it occurs during the adolescent years.

## The Present Study

By using the experience of European children during World War II as a case study, we build on the growing body of scholarship on the long-term impact of exposure to armed conflict. We focus on three key empirical questions related to the timing of exposure. First, given the well-established literature on critical periods that points to childhood as being especially sensitive to adverse environmental exposures, does the effect of armed conflict on later-life health vary by age of exposure?

Second, through what pathways does exposure to war impact later-life health? Figure 1 presents potential pathways by which childhood exposure to war may influence later-life health. The pathways in block A represent childhood circumstances that may have a lasting impact on health. These include direct, immediate impacts on health in the form of injury/illness that persists across the life course. It also includes indirect pathways that may increase lifetime accumulation of risk, such as the death of a parent or nutritional or financial hardship and stress. Block B represents indirect adult sociobehavioral pathways, including increased risk of smoking and excessive drinking as well as the negative health impacts of reduced human capital accumulation and socioeconomic attainment, which may result from the broader context of social instability, dislocation, and institutional breakdown created by war. Block C represents the long-term increase in risk for a variety of more proximal cardiometabolic, muscular-skeletal/functional, and psychological conditions. Thus, childhood exposure to war may increase the risk of poor health in later life by increasing the risk of immediate injury/illness, parental death, and hardship in childhood (block A); through adverse impacts on socioeconomic attainment and increased exposure to unhealthy behaviors in adulthood (block B); or through increasing the risk of adult chronic disease, physical disability, and psychological distress (block C).

Finally, does the relative importance of each pathway vary depending on the timing of exposure? Given that developmental domains have somewhat unique childhood trajectories and attendant critical/sensitive periods, it seems reasonable to suspect that war exposure occurring at different periods of childhood may manifest its impact on later-life health through heterogeneous pathways. We examine whether such differences are apparent in these cohorts.

## Methods

### Data

Analyses for this study draw on data from the Survey of Health, Ageing and Retirement in Europe (SHARE) release 7.0. The SHARE provides detailed life histories for Europeans ages 50 and older alongside a wide range of health and socioeconomic characteristics in later life (Börsch-Supan and Jürges 2005). We focus on participants of Waves 3 and 7 (SHARELIFE), which captured detailed residential and living condition histories of more than 140,000 respondents since their birth. We constrain our sample to those with potential childhood exposure to armed conflict during World War II. First, we exclude respondents in countries without war exposure. Second, we exclude individuals born after 1945. Finally, we exclude individuals who were adults for the duration of the war. These exclusion criteria result in an analytic sample of 28,057 individuals.

### Measurement

#### Main Dependent Variable: Adult Self-rated Health

This ordinal measure is a self-assessed summary of the respondent's overall health status at ages 50 and older. The most commonly used measure of health has been deemed a valid measure of overall health that successfully predicts objectively measured health conditions as well as mortality (Idler and Benyamini 1997; Jylhä 2009). Self-rated health also provides a generalized measure of overall health that is likely to capture a wide range of both health conditions (physical, functional, and psychological) and underlying mechanisms (social and biological). We dichotomize the variable so that it expresses positive health (1 = excellent/very good; 0 = good/fair/poor). We tested a variety of alternative specifications (e.g., ordinal), and the pattern of results was substantively the same across specifications.

#### Childhood Conflict Exposure

One contribution of this study is to provide an improved measure of European childhood exposure to contexts of conflict brought forth by World War II. This measure builds on the work of Kesternich et al. (2014) and combines historical and military data, provided by Ellis (1993) and the U. S. Military Academy (Cheng 2004), with SHARELIFE's residential history, enabling us to determine the amount of time each respondent resided in a region that featured conflict during the war.1 Following previous research, this measure of exposure is best understood as a contextual variable that proxies for regional social impact at the second level of the Nomenclature of Territorial Units for Statistics (NUTS2) rather than as an individual measure of conflict exposure (Kesternich et al. 2014). Specifically, we map the timing and location of hostilities across NUTS2 regional units to each respondent's residential history during the war, including the movement of front lines, pockets of conflict, sieges, and bombings. The resulting variable is measured as the number of childhood weeks the respondent resided in an area that hosted conflict during World War II. This variable ranges from 0 to 20 weeks of exposure. We further collapse this into a three-category ordinal variable that compares no exposure (0 weeks), low exposure (1–7 weeks of exposure), and high exposure (8–20 weeks of exposure). To examine the robustness of this specification, we tested it against a variety of alternatives, including linear, quadratic, and a linear spline. The specification that best fits the data is the ordinal variable. In the following section, we provide an example of the coding procedure. A full explanation of the measure and justification for the specification can be found in the online appendix.

Figure 2 presents the movement of Allied forces across southern Italy beginning in Sicily (the ITG1 NUTS2 region) from July 10, 1943, through September 9, 1943. British and U.S. forces entered Sicily from Avola, Pachino, Gela, and Licata. The arrows indicate the British and U.S. advances throughout Sicily, where German and Italian forces were progressively pushed back across the island and eventually on to the Italian peninsula. The first battles occurred on July 10, 1943, in the points of entry, and concluded with German and Italian retreats in the north of Sicily near Scaletta on September 3, 1943. Between July 10 and September 3, Sicily harbored several battlefields. As such, in our coding scheme, those respondents who resided in Sicily in 1943 are coded as having been exposed to 54 days of conflict. Given that there is some discrepancy in the exact dates of the advances of the troops depending on the source used, we round days of exposure to weeks of exposure. Hence, instead of 54 days of exposure, individuals residing in Sicily are assigned to have experienced eight weeks of exposure.

The U.S. 7th Army entered Calabria (ITF6) on August 17, followed by the British on September 3. Allied forces then moved upward, with the goal of connecting with the U.S. 5th Army invading Salerno (ITF3) on September 9. German and Italian forces left few troops to aid the fallback of their main forces. As such, the U.S. 7th Army encountered little resistance. Individuals residing in Calabria at that point were exposed to only a few weeks of combat. By September 9, the British 8th Army proceeded to Taranto (ITF5) and Brindisi (ITF4), encountering very little combat resistance that totaled approximately one week of exposure.

#### Childhood Health and War-Related Hardships

Childhood health is a variable that retrospectively asks the respondents to self-rate their health during childhood years (from birth to age 15). Values range from 1 = poor to 5 = excellent. Previous research has demonstrated the reliability and validity of this measure (Haas 2007; Haas and Bishop 2010; Smith 2009). We also include a dichotomous measure of childhood illness indicating whether, as a result of a health condition, the respondent had ever (1) spent a month or more in the hospital, (2) been confined to bed or home for a month or more, or (3) missed school for a month or more during childhood. Respondents who answered affirmatively to any of these questions are assigned a value of 1. To examine the broader impact of wartime exposure on childhood circumstances, we include a count measure of the number of hardships (hunger, financial hardship, stress) ranging from 0 to 3, as well as an indicator for whether the respondent experienced parental death. These measures correspond to the period during the war and the immediate postwar period (1939–1950).

#### Adult Socioeconomic Position and Health Behaviors

Midlife socioeconomic status (SES) is measured by educational attainment and occupation at age 35. Education is measured as years of completed schooling and ranges from 0 to 25. Occupation is measured using the International Standard Classification of Occupations 1988 (ISCO 88). The ISCO 88 classification provides an ordinal classification of occupational attainment divided into nine categories, where higher values represent occupations with higher levels of prestige, skill complexity, and educational requirements (Ganzeboom and Treiman 1996). Behavioral health risks are measured by two indicators: (1) daily smoker, a dichotomous indicator designating those who ever smoked daily; and (2) daily drinker, based on an item that asks whether the respondent consumes alcohol every day or five/six days per week.

#### Cardiometabolic, Muscular-Skeletal, and Psychological Conditions

Respondents were asked whether they had ever been diagnosed with a series of health conditions and diseases. We include four indicator variables for the presence of cardiovascular disease, stroke, diabetes, and obesity. We also include two measures of muscular-skeletal/physical functional health. Any functional limitation is set to 1 if the respondent reported having difficulty with any of a series of physical tasks (e.g., climbing a flight of stairs, moving large objects, walking a few blocks) and 0 otherwise. Maximum grip strength is measured in kilograms using a handheld dynamometer. We also include two measures of psychological health. The first is a dichotomous indicator for whether the respondent has ever been diagnosed or treated for an affective or emotional disorder. The second is an indicator of current depression based on the EURO-D scale. Respondents were assigned a value of 1 if they reported experiencing 3 or more of 14 symptoms (Prince et al. 1999).

#### Childhood Socioeconomic Controls

We utilize three measures to capture childhood SES: (1) father's/main breadwinner's occupation at age 10, again standardized using ISCO 88; (2) the number of books in the household at age 10; and (3) the number of features in the household in which they lived at age 10 (e.g., central heating, inside toilet, hot running water). These variables have been shown to be valid indicators of childhood SES (Kesternich et al. 2014; Mazzonna 2012). All analyses control for gender.

### Statistical Analysis

The analysis proceeds in three stages. The first stage estimates the association between childhood conflict exposure and positive self-rated health in later life, overall and then separately for each cohort group. The experience of war is a complex phenomenon creating multifaceted cascades of physical, nutritional, cognitive, socioemotional, socioeconomic, and behavioral impacts and adaptations. Each of these developmental domains has its own age-structured normative trajectory as well as its own attendant critical/sensitive periods, which may or may not align neatly with those of the others. Therefore, our strategy for identifying timing effects is to isolate specific cohort groups who experienced the war at different ages. Our goal in specifying cohort groups is to maximize our ability to compare as many key developmental stages as possible (prenatal, early childhood/preschool, early school years, early adolescence, and late adolescence) while also being cognizant of sample size limitations and statistical power. We compare five cohort groups representing those born during the war (1939–1945), and those who were aged 0–4 (born in 1934–1938), 5–9 (born in 1929–1933), 10–14 (born in 1924–1928), and 15+ (born in 1921–1923) at the start of the war. We estimate a logistic regression model with the following specification:
$Yic=α+β1(Conflict exposure)i+β(Xi)+εi,$
(1)

where positive self-rated health ($Yic$) for individual i from cohort group c is a function of the intercept ($α$), exposure to conflict, a vector of individual covariates (Xi) (i.e., gender, country fixed effects), and an error term ($εi$). To test for differential impact by timing, we estimate the model separately by cohort group.

The second stage of the analysis investigates whether exposure to conflict during different periods of childhood is associated with the different hypothesized mediators. For this, we estimate models with each potential mediator as an outcome variable. We estimate these models separately by cohort to test whether the association with conflict exposure varies depending on the timing of exposure.

Finally, we estimate formal mediation models using the approach developed by Breen and colleagues (Breen et al. 2013) (hereafter, KHB), which allows for the assessment of the proportion mediated via specific pathways in discreet outcome models. Furthermore, the KHB estimation provides a formal test for the difference in coefficient change (i.e., the Sobel test). The approach follows the same logic as stepwise regression, with the main difference residing in the KHB calculations; hence, the KHB mediation procedure can be expressed as follows:
$Yic= α+ β1(Conflict exposure)i+β2(Block A)i+β(Xi)+εi.$
(2)
$Yic=α+β1(Conflict exposure)i+β2(Block B)i+β(Xi)+εi.$
(3)
$Yic=α+β1(Conflict exposure)i+β2(Block C)i+β(Xi)+εi.$
(4)

In Eq. (1), we set the baseline model for each cohort. In subsequent models, we independently introduce the mediators of interest to test whether they mediate the effect of conflict exposure on self-rated health, as expressed in Eqs. (2), (3), and (4). Block A of Figure 1 is represented by $β2(Block A)i$, which represents the impact of conflict exposure as mediated through poor childhood circumstances (poor health, illness, parental death, and hardship). Block B is represented by $β2(Block B)i$, a set of indicators of education, midlife occupation, and health risk behaviors. Block C is represented by $β2(Block C)i$, a set of indicators for cardiometabolic conditions, physical health, and mental health at ages 50 and older. Each of these blocks provides potential insights as to whether the processes by which exposure to war influences later-life health are direct or indirect, reflect the accumulation of risk or latent critical period effects, and operate through social/behavioral, biological, or psychological processes. For example, if mediation largely occurs through block A with little through blocks B or C, then it suggests that the impact of warfare mostly matters because of the direct consequences it has on children's health and well-being (parental death, hardship, and deprivation) at the time of exposure with less potential for the amelioration of subsequent influences across the life course. However, if mediation largely occurs through block B, then indirect, cumulative risk-oriented, sociobehavioral processes predominate. If block C predominates with little input from blocks A and B, then latent, critical period effects are most important, with permanent alterations of cardiometabolic, physical, or psychological functioning that manifest disease many years later.

## Results

Descriptive results are presented in Table 1. Conflict exposure was widespread among European children during World War II. Approximately 63% of the sample lived in areas with direct exposure to conflict during some portion of their childhood, with an average length of exposure of 4.6 weeks among those exposed. The exposed group overall had worse self-rated health in later life compared with the unexposed group. Whereas 11.5% of the unexposed group rated their health as excellent or very good, 9.2% of the exposed group did so. Additionally, the prevalence of cardiovascular disease, obesity, functional limitations, depression, and affective or emotional disorders was higher among the exposed. The remaining covariates are overall balanced, except for childhood health, years of education, and midlife occupation.

Table 2 presents logistic regression estimates for the association between childhood exposure to conflict on positive self-rated health in late adulthood. The estimate for the full sample shows that moderate exposure to conflict (one to seven weeks) reduced the odds of having positive self-rated health by approximately 17% (1 – e–0.183). Those exposed to eight weeks or more had approximately 27% lower odds (1 – e–0.306 ) of having positive self-rated health compared with those not exposed. When disaggregated by cohort, conflict exposure is significantly associated with reduced later-life health only for those cohorts born during the war and those who were under age 10 in 1939. For those who were adolescents at the start of the war, the association between conflict exposure and health is not statistically significant. Note that sample sizes for the older cohorts in our sample are substantially smaller, increasing the risk of a false negative finding due to a lack of statistical power. Those born during the war were vulnerable to moderate levels of exposure, whereas those under age 10 in 1939 show negative effects only if they were exposed for eight weeks or more. Those aged 0–4 in 1939 and exposed for eight weeks or more had 23% lower odds (1 – e–0.265) of positive self-rated health than their cohort peers who were not exposed, whereas those aged 5–9 had 44% lower odds (1 – e–0.596) of positive self-rated health.

The cohort patterns can be best seen in the predicted probabilities of positive self-rated health presented in Figure 3. Those born in 1939–1945 show an immediate decline in self-rated health, from about 16% for those with no exposure to about 13% for one to seven weeks of exposure, with a smaller decline observed for those with more severe exposure. Conversely, those who were under age 4 in 1939 show declines only after severe exposure (more than eight weeks).

Tables 3 and 4 present estimates of the association between conflict exposure and potential mediating pathways. Because we do not observe significant associations with health for those aged 10 and older at the start of the war, we exclude those cohorts from the rest of the analysis.2Table 3 presents regression estimates for childhood circumstances, midlife SES, and behavioral risks (blocks A and B from Figure 1). Those born during the war and exposed to conflict had significantly lower odds of reporting positive childhood health and significantly increased odds of a serious childhood illness compared with their unexposed peers. Conflict exposure was not associated with reduced early-life health among the other cohorts. Neither parental death nor hardship was more frequent for those exposed to the conflict within any of the cohorts.

Different patterns also emerge for behavioral and socioeconomic pathways (block B), depending on the timing of exposure. Those born during the war show no signs of an exposure-related increase in risky behaviors or reduced socioeconomic attainment. Conversely, among those who were school-aged at the start of the war (5–9) or who became so during the conflict (0–4) experienced reduced educational attainment when exposed to severe levels of conflict. Severe conflict exposure was associated with 0.32 and 0.58 fewer years of schooling for those aged 0–4 and 5–9, respectively. Exposure was also associated with increased risk of smoking and drinking for those who were young children (aged 0–4) during the war and with drinking for those who were school-aged children (aged 5–9). Conflict exposure was not associated with reduced occupational attainment for any cohort.

Table 4 presents regression estimates for the effects of combat exposure on cardiometabolic, muscular-skeletal/functional health, and psychological conditions at ages 50 and older. Exposure in utero or during the first years of life is associated with an increased risk of developing cardiometabolic conditions (i.e., cardiovascular disease, obesity, and diabetes). Conversely, exposure at later ages shows no association with an increased risk of developing any of these conditions. Regarding muscular-skeletal/functional health, exposure for all three cohorts is associated with a higher risk of having any functional limitation as well as reduced grip strength. There is some evidence of an age gradient in the severity of effects. Those exposed to eight weeks or more of combat while in utero or in their first years of life have grip strength that is 1.87 kilograms lower than their unexposed peers. Yet, older cohorts with similar exposure have a grip strength that is, respectively, 1.54 and 1.4 kilograms lower than their unexposed peers. Those born during the war show no increase in risk for either of the two measures of mental health. Conversely, those exposed in early childhood (aged 0–4 in 1939) had approximately 30% higher odds of screening positive for current depression compared with their unexposed peers. Additionally, those exposed to 8–20 weeks in late childhood (aged 5–9 in 1939) are twice as likely (e0.697 = 2) to have ever been diagnosed with an affective or emotional disorder in their life compared with their unexposed peers.

Finally, Table 5 provides the KHB decomposition estimates of the association between conflict exposure and positive self-rated health and formal tests of mediation. The estimates shown are for mediation analyses solely for high levels of exposure (8 to 20 weeks of exposure) because statistically significant effects for older cohorts were observed for only more prolonged exposure. Panel A presents a comparison of coefficients between the baseline model without any mediators and conditional models (with mediators). Additionally, it provides an estimate of the difference in coefficients and tests whether these differences are statistically significant (Sobel test). Panel B provides the KHB calculations of the percentage mediation attributed to each variable. All estimates are calculated separately for each cohort.

Columns 1–3 present estimates for cohorts born during the war. Column 1 examines mediation via childhood circumstances. The difference between the baseline and the conditional model is small but statistically significant. Hence, the drop in the coefficient of high exposure to conflict when we introduce the childhood circumstances mediators is of a factor of just 0.051 log odds. The percentage mediated is also low: only about 5% of the association between high combat exposure and self-rated health is mediated through these variables. Column 2 presents mediation via socioeconomic attainment and health behaviors. The difference in coefficients from the baseline to the conditional model is not statistically significant, indicating that no mediation occurs via these pathways for those born during the war. Column 3 shows mediation estimates via cardiometabolic, muscular-skeletal/functional, and mental health conditions at ages 50 and older. The reduction in the coefficient when these mediators are introduced is by a factor of 0.117 log odds. In terms of percentage mediated, as a total, block C mediates approximately 38% of the association. The main mediators are cardiovascular disease, functional limitations, and grip strength. These findings are consistent with a largely direct, although latent, effect of war exposure.

Columns 4–6 test mediation for the cohorts aged 0–4 in 1939. Column 4 presents mediation via childhood circumstances. The difference in model estimates is not statistically significant, indicating no mediation via childhood circumstances for these cohorts. However, mediation occurs via indirect socioeconomic and behavioral pathways. As can be seen in column 5, the coefficient declines significantly from the baseline to the conditional model. Most of that mediation occurs via education. Approximately 6% of the effect of being exposed to eight or more weeks of combat is due to reduced education. This mediation, although statistically significant, is still modest, indicating that most of the effect of conflict is not accounted for by these mediators. As seen in column 6, mediation via proximal cardiometabolic, muscular-skeletal/functional, and mental health is quite substantial, accounting for approximately 27% of the association. However, unlike with the youngest cohorts, no mediation occurs through cardiometabolic pathways. Rather, functional limitation and current depressive symptoms account for the largest share of the association for these cohorts.

Columns 7–9 present mediation for cohorts aged 5–9 in 1939. The estimated effect of conflict exposure is not statistically significant for any of the blocks of mediators. When we look at the percentage reduced via each variable, however, we see that approximately 4% is due to education; 8%, to current depressive symptoms; and 16%, to lifetime affective/emotional disorders. Despite a lack of statistical significance in the mediation analysis for older cohorts, the size and the direction of the mediation effects reveal patterns similar to those for cohorts aged 5–9 in 1939. Mediation analyses for oldest cohorts (presented in Table A7 of the online appendix) show patterns that mostly align with conflict exposure operating through interruptions in education and cardiovascular disease for those aged 10–14 as well as through problem drinking and affective and emotional disorders for those aged 15 and older.

## Discussion

According to the Uppsala Conflict Data Program, the number of wars and armed conflicts has risen substantially in the last few decades (Pettersson and Öberg 2020). This increase is concentrated in particular areas of Africa, Asia, and the Middle East. As such, exposure to contexts of armed conflict in childhood is not uncommon, among current aging European populations, like those examined here, or among current and recent cohorts of children in various parts of the world. Parallel to this trend, the literature linking war and health has increased substantially in recent years. This growing body of research has shown that childhood exposure to armed conflict has persistent negative effects across the life course (Akbulut-Yuksel 2017; Islam et al. 2017; Kesternich et al. 2015; Lindeboom and Ewijk 2015). However, as shown here, a life course perspective emphasizing the differential timing of exposure demonstrates that such adverse health effects can be quite heterogeneous. The present study utilized data for European children during World War II to demonstrate that those exposed at earlier ages—particularly those born during the war and were thus in utero or in the first years of life at the time of exposure—were more vulnerable to negative consequences of conflict. On the other hand, cohorts of young children during the war show negative effects only if they were exposed for prolonged durations, whereas those who experienced the war as adolescents appear to show no long-term health effects from conflict exposure.

In addition to examining the differential impact of conflict, the present study also focused on how the timing of exposure shapes the pathways through which it impacts self-rated health in late life. We examined the extent to which the impact of conflict exposure operated via mechanisms consistent with latent critical period or cumulative risk pathways and whether this depended on age at exposure. The results suggest that exposure in the earliest ages is most consistent with latent critical period processes. Cohorts born during the war were more likely to experience poor health and illness in childhood as a result of conflict exposure. They were also at significantly higher risk for a range of cardiometabolic conditions and reduced muscular-skeletal/functional health. The increase in cardiometabolic risk is consistent with the large literature on early-life critical periods that highlights the in utero and very early childhood environment (Gluckman et al. 2008). We found little evidence to indicate that those cohorts experienced socioeconomic stunting. In addition, war exposure did not appear to have strong effects on their risk-taking behaviors in adulthood or mental health. It may be the case that the youngest children were simply too young to be cognizant of the trauma around them relative to their older peers. They had also not yet entered school and were thus not subject to institutional breakdowns in schooling. Conversely, the results for preschool cohorts (aged 0–4 in 1939) and early primary school–aged children (aged 5–9 in 1939) are more consistent with cumulative risk processes. For these cohorts, exposure to conflict was not associated with childhood health or illness, nor was it associated with increased cardiometabolic risk. However, conflict had substantial negative effects on their educational attainment and increased their propensity toward risky behaviors. The socioeconomic impacts in particular are consistent with the substantial institutional breakdown that frequently accompanies armed conflict. World War II brought widespread destruction of both physical and institutional infrastructure as well as economic and sociopolitical upheaval. We also found evidence that war exposure for those under age 10 in 1939 increased the risk of muscular-skeletal/functional and mental health problems. The added psychological vulnerability may owe to the fact that these children were mostly old enough to be aware of the conflict and destruction around them yet not necessarily old enough to have the required cognitive and emotional skills to adequately cope with such intense stress and trauma. In addition, many of those aged 5–9 at the start of the war were likely to have experienced the trauma of war during early adolescence, a critical period in which the risk of depression and related psychological conditions begins to rise substantially (Kessler et al. 2001).

The present study has implications for broader research on the long-term impact of childhood conditions and the developmental origins of health and disease. The results also complement the long-standing life course literature on the disruptive impacts of war. Whereas much of the now-classic work in life course research has explored the ways in which World War II shaped the lives of the men who fought it (Elder 1986, 1987), the present study expanded the scope to examine the life course impacts of the war on their children and the pathways through which their health trajectories were transformed. As with prior research on adults, we found that the timing of exposure matters in terms of both the magnitude of its impact and the pathways and mechanisms through which it manifests. Among the youngest children, conflict exposure is experienced largely as a physiologically scarring process, whereas school-aged children experience war as an important disruptive event that permanently alters socioeconomic trajectories and creates lasting psychological vulnerabilities. The present study is a first step toward understanding the heterogeneous effects of war exposure. Our results suggest that future research should engage in deeper explorations within specific developmental domains to examine how differential timing of war exposure may impact domain-specific outcome trajectories.

The results also have implications for potential policy interventions designed to mitigate the impact of childhood exposure to conflict. The broader literature on the long-term impacts of early-life conditions demonstrates direct impacts of childhood health insults and socioeconomic deprivation on physical and functional health in later life and additional indirect impacts via subsequent socioeconomic attainment and labor market outcomes in adulthood (Ferraro 2011; Ferraro and Kelley-Moore 2003; Haas 2006; Haas et al. 2011). That work would imply that interventions could target both the prevention of the initial childhood exposure and policies in adolescence and adulthood to ameliorate their long-term impacts. The results of the present study suggest that when it comes to armed conflict, different pathways—and therefore different types of interventions—are likely to be required depending on the timing of exposure. For the youngest victims, adverse health impacts associated with armed conflict may be less sensitive to the ameliorative effect of improvement in subsequent social and economic conditions. Rather, focusing on the conflict-induced illness and injuries in childhood will be key, along with recognizing that the bulk of adverse health impacts are not likely to manifest for several decades. Conversely, for school-aged children, reducing the disruptive impact of institutional breakdown on socioeconomic outcomes and dealing with psychological trauma will be critical. In any case, the results suggest that policy interventions for those exposed to conflict are likely to be most impactful if they directly target the physical, functional, and psychological impairments that manifest as a result of childhood exposure.

The present study has some limitations. First, the data do not provide a direct measure of individual exposure to armed conflict. As a proxy, we used a contextual approach to measure conflict exposure, a common approach in the literature (Akbulut-Yuksel 2014; Kesternich et al. 2014; Lindeboom and Ewijk 2015). We could determine only that the respondent lived in an area at a particular time with a documented proximity to the conflict. However, given that war's impacts extend well beyond individual exposure to violence, this approach also has important advantages in that it likely captures a broader swath of the population exposed to the adverse consequences of warfare that may be obscured by individual-focused exposure measures. However, to the extent that war creates broader social and economic hardships and institutional breakdown beyond the specific regions that experienced documented hostilities, our analyses likely underestimated the true effect of war on children.

Another limitation to this study is the potential impacts of mortality and migration selection. SHARELIFE captures only those respondents who survived to at least age 50 in 2009 and were residing in Europe. For those interviewed in Wave 7, SHARELIFE captured only those aged 50 or older in 2017. Individuals who died or who permanently fled Europe during or after the war will not be in the sample. Our estimates may be subject to selection effects due to migration patterns in response to combat exposure. If those most harmed by war exposure were most likely to migrate out of Europe permanently or to underreport wartime residence in combat-afflicted areas, then our results will underestimate the true impact of war. Conversely, if those most affected by exposure were the least likely to migrate, our results will be biased upward.

In addition, individuals most exposed to conflict and hardship during the war may have been less likely to have survived to be included in the sample. Hence, individuals that SHARE comprises might be systematically different from the underlying exposed population. Such differential mortality selection is of particular importance to the validity of the results vis-à-vis the timing of exposure. If among respondents who were exposed as older children, those with the worst health outcomes were most likely to have died—leaving behind only the healthiest of the exposed—then the cohort patterns we observed could be artifactual. However, we do not think this is the case. Although we could not directly observe the exposure status of those cohort members who died before SHARE enrollment, we were able to test for mortality selection effects indirectly (see the online appendix). If differential mortality among the exposed were operating, then we should see evidence of this in the sample. Specifically, this must manifest itself in war exposure rates across cohorts in which the rate of exposure declines substantially with increasing age. We found no such pattern. Rates of exposure were consistent across the cohorts we examined, as was the level of exposure among the exposed. We also tested for differential mortality selection since enrollment in SHARE. We found that among the SHARE respondents who died since enrollment, rates of war exposure were very similar for those born during the war and for those aged 15 or older at the start of the war. Given the lack of differential mortality, we believe that the cohort patterns highlighted in this study are very unlikely to result from differential mortality selection. If anything, we believe that our results are likely to be underestimates of the true associations.

Another limitation is that although we hypothesized specific latent biological pathways (e.g., programming of cardiometabolic function) through which exposure to war has an impact on later-life health, we were not able to observe these directly. We were able to use only proxy downstream sequela of these pathways (cardiovascular disease, stroke, diabetes, and obesity). In addition, with the exception of the Baltic states, SHARE does not include territories of the former Soviet Union (e.g., Ukraine, Belarus, Western Russia), where some of the most severe fighting and devastation of civilian populations occurred. Finally, our estimates pool over a wide array of European contexts, each with its own unique wartime experience and prewar and postwar social, economic, and health/welfare state structures. The impact of childhood war exposure may vary across country contexts. Unfortunately, because of sample size limitations, the SHARE data are not well suited to examine such differences. Future research would be wise to examine how local context shapes the impact of armed conflict on the lives of those who experience it.

## Acknowledgments

We thank Jonathan Daw, Joeun Kim, and Iris Kesternich and colleagues for their support and their helpful insight on various drafts of the manuscript. Previous versions of this article were presented at the annual meetings of the Population Association of America and the American Sociological Association.

## Notes

1

We thank Kesternich and colleagues for providing programming code for the replication exercise of their prior work (Kesternich et al. 2014). This assistance was crucial in the construction of our revised measure.

2

Estimates of the impact of combat exposure on the mediators and the formal mediation analysis for cohorts aged 10 and older at the start of the war are presented in Tables A5–A7 of the online appendix.

## References

Akbulut-Yuksel, M. (
2014
).
Children of war: The long-run effects of large-scale physical destruction and warfare on children
.
Journal of Human Resources
,
49
,
634
662
.
Akbulut-Yuksel, M. (
2017
).
War during childhood: The long run effects of warfare on health
.
Journal of Health Economics
,
53
,
117
130
.
Akresh, R., Lucchetti, L., & Thirumurthy, H. (
2012
).
Wars and child health: Evidence from the Eritrean–Ethiopian conflict
.
Journal of Development Economics
,
99
,
330
340
.
Akresh, R., Verwimp, P., & Bundervoet, T. (
2011
).
Civil war, crop failure, and child stunting in Rwanda
.
Economic Development and Cultural Change
,
59
,
777
810
.
Alastalo, H., Räikkönen, K., Pesonen, A. K., Osmond, C., Barker, D. J. P., Kajantie, E., . . . Eriksson, J. G. (
2009
).
Cardiovascular health of Finnish war evacuees 60 years later
.
Annals of Medicine
,
41
,
66
72
.
Austin, K. F., & McKinney, L. A. (
2012
).
Disease, war, hunger, and deprivation: A cross-national investigation of the determinants of life expectancy in less-developed and sub-Saharan African nations
.
Sociological Perspectives
,
55
,
421
447
.
Barker, D. J. (
1990
).
The fetal and infant origins of adult disease
.
BMJ
,
301
,
1111
. https://doi.org/10.1136/bmj.301.6761.1111
Bartley, M. (
2016
).
Health inequality: An introduction to concepts, theories and methods
(2nd ed.).
Cambridge, UK
:
Polity Press
.
Börsch-Supan, A., & Jürges, H. (Eds.). (
2005
).
The Survey of Health, Aging, and Retirement in Europe—Methodology
.
Mannheim, Germany
:
Mannheim Research Institute for the Economics of Aging
.
Brandt, M., Deindl, C., & Hank, K. (
2012
).
Tracing the origins of successful aging: The role of childhood conditions and social inequality in explaining later life health
.
Social Science & Medicine
,
74
,
1418
1425
.
Breen, R., Karlson, K. B., & Holm, A. (
2013
).
Total, direct, and indirect effects in logit and probit models
.
Sociological Methods & Research
,
42
,
164
191
.
Chamarbagwala, R., & Morán, H. E. (
2011
).
The human capital consequences of civil war: Evidence from Guatemala
.
Journal of Development Economics
,
94
,
41
61
.
Chapman, D. P., Whitfield, C. L., Felitti, V. J., Dube, S. R., Edwards, V. J., & Anda, R. F. (
2004
).
.
Journal of Affective Disorders
,
82
,
217
225
.
Cheng, C. P. (
2004
).
World War II database
Dietz, W. H. (
1997
).
Periods of risk in childhood for the development of adult obesity—What do we need to learn?
Journal of Nutrition
,
127
,
1884S
1886S
.
Ekamper, P., van Poppel, F., Stein, A. D., & Lumey, L. H. (
2014
).
Independent and additive association of prenatal famine exposure and intermediary life conditions with adult mortality between age 18–63 years
.
Social Science & Medicine
,
119
,
232
239
.
Elder, G. H.Jr. (
1986
).
Military times and turning points in men's lives
.
Developmental Psychology
,
22
,
233
245
.
Elder, G. H.Jr. (
1987
).
War mobilization and the life course: A cohort of World War II veterans
.
Sociological Forum
,
2
,
449
472
.
Elder, G. H.Jr. (
1998
).
The life course as developmental theory
.
Child Development
,
69
,
1
12
.
Elder, G. H., Kirkpatrick Johnson, M.Jr., & Crosnoe, R. (
2003
).
The emergence and development of life course theory
. In Mortimer, J. T. & Shanahan, M. J. (Eds.),
Handbook of the life course, handbooks of sociology and social research
(pp.
3
19
).
Boston, MA
:
Springer
.
Elliott, J. R. (
2015
).
Natural hazards and residential mobility: General patterns and racially unequal outcomes in the United States
.
Social Forces
,
93
,
1723
1747
.
Ellis, J. (
1993
).
World War II: A statistical survey—The essential facts and figures for all the combatants
.
New York, NY
:
Facts on File
.
Ferraro, K. F. (
2011
).
Health and aging: Early origins, persistent inequalities?
In Settersten, R. A. & Angel, J. L. (Eds.),
Handbook of sociology of aging, handbooks of sociology and social research
(pp.
465
475
).
New York, NY
:
Springer
.
Ferraro, K. F., & Kelley-Moore, J. A. (
2003
).
Cumulative disadvantage and health: Long-term consequences of obesity?
American Sociological Review
,
68
,
707
729
.
Ganzeboom, H. B. G., & Treiman, D. J. (
1996
).
Internationally comparable measures of occupational status for the 1988 international standard classification of occupations
.
Social Science Research
,
25
,
201
239
.
Gluckman, P. D., Hanson, M. A., Cooper, C., & Thornburg, K. L. (
2008
).
Effect of in utero and early-life conditions on adult health and disease
.
New England Journal of Medicine
,
359
,
61
73
.
Gould, C. E., Rideaux, T., Spira, A. P., & Beaudreau, S. A. (
2015
).
Depression and anxiety symptoms in male veterans and non-veterans: The Health and Retirement Study
.
International Journal of Geriatric Psychiatry
,
30
,
623
630
.
Haas, S. A. (
2006
).
Health selection and the process of social stratification: The effect of childhood health on socioeconomic attainment
.
Journal of Health and Social Behavior
,
47
,
339
354
.
Haas, S. A. (
2007
).
The long-term effects of poor childhood health: An assessment and application of retrospective reports
.
Demography
,
44
,
113
135
.
Haas, S. A. (
2008
).
Trajectories of functional health: The ‘long arm’ of childhood health and socioeconomic factors
.
Social Science & Medicine
,
66
,
849
861
.
Haas, S. A., & Bishop, N. J. (
2010
).
What do retrospective subjective reports of childhood health capture? Evidence from the Wisconsin Longitudinal Study
.
Research on Aging
,
32
,
698
714
.
Haas, S. A., Glymour, M. M., & Berkman, L. F. (
2011
).
Childhood health and labor market inequality over the life course
.
Journal of Health and Social Behavior
,
52
,
298
313
.
Hankin, B. L., Abramson, L. Y., Moffitt, T. E., Silva, P. A., McGee, R., & Angell, K. E. (
1998
).
Development of depression from preadolescence to young adulthood: Emerging gender differences in a 10-year longitudinal study
.
Journal of Abnormal Psychology
,
107
,
128
140
.
Hayward, M. D., & Gorman, B. K. (
2004
).
The long arm of childhood: The influence of early-life social conditions on men's mortality
.
Demography
,
41
,
87
107
.
Howell, J., & Elliott, J. R. (
2019
).
Damages done: The longitudinal impacts of natural hazards on wealth inequality in the United States
.
Social Problems
,
66
,
448
467
.
Ichino, A., & Winter–Ebmer, R. (
2004
).
The long–run educational cost of World War II
.
Journal of Labor Economics
,
22
,
57
87
.
Idler, E. L., & Benyamini, Y. (
1997
).
Self-rated health and mortality: A review of twenty-seven community studies
.
Journal of Health and Social Behavior
,
38
,
21
37
.
Islam, A., Raschky, P., & Smyth, R. (
2017
).
The long-term health effects of mass political violence: Evidence from China's cultural revolution
.
Social Indicators Research
,
132
,
257
272
.
Jylhä, M. (
2009
).
What is self-rated health and why does it predict mortality? Towards a unified conceptual model
.
Social Science & Medicine
,
69
,
307
316
.
Kessler, R. C., Avenevoli, S., & Merikangas, K. R. (
2001
).
Mood disorders in children and adolescents: An epidemiologic perspective
.
Biological Psychiatry
,
49
,
1002
1014
.
Kesternich, I., Siflinger, B., Smith, J. P., & Winter, J. K. (
2014
).
The effects of World War II on economic and health outcomes across Europe
.
Review of Economics and Statistics
,
96
,
103
118
.
Kesternich, I., Siflinger, B., Smith, J. P., & Winter, J. K. (
2015
).
Individual behaviour as a pathway between early-life shocks and adult health: Evidence from hunger episodes in post-war Germany
.
Economic Journal
,
125
,
F372
F393
.
Kitagawa, E. M., & Hauser, P. M. (
1973
).
Differential mortality in the United States: A study in socio-economic epidemiology
.
Cambridge, MA
:
Harvard University Press
.
Kuh, D., & Ben Shlomo, Y. (
2004
).
A life course approach to chronic disease epidemiology
.
Oxford, UK
:
Oxford University Press
.
Kwong, A. S., Manley, D., Timpson, N. J., Pearson, R. M., Heron, J., Sallis, H., . . . Leckie, G. (
2019
).
Identifying critical points of trajectories of depressive symptoms from childhood to young adulthood
.
,
48
,
815
827
.
Lee, C. (
2014
).
In utero exposure to the Korean War and its long-term effects on socioeconomic and health outcomes
.
Journal of Health Economics
,
33
,
76
93
.
Lindeboom, M., & Ewijk, R. (
2015
).
Babies of the war: The effect of war exposure early in life on mortality throughout life
.
Biodemography and Social Biology
,
61
,
167
186
.
Link, B. G., & Phelan, J. (
1995
).
Social conditions as fundamental causes of disease
.
Journal of Health and Social Behavior
,
35
(
Extra issue
),
80
94
.
Lumey, L. H., Stein, A. D., & Susser, E. (
2011
).
.
Annual Review of Public Health
,
32
,
237
262
.
Mazzonna, F. (
2012
).
The long-lasting effects of family background: A European cross-country comparison
(MEA Discussion Paper No. 245-11).
Rochester, NY
:
Social Science Research Network
.
Pettersson, T., & Öberg, M. (
2020
).
Organized violence, 1989–2019
.
Journal of Peace Research
,
57
,
597
613
.
Prince, M. J., Reischies, F., Beekman, A. T. F., Fuhrer, R., Jonker, C., Kivela, S. L., . . . Copeland, J. R. M. (
1999
).
Development of the EURO-D scale—A European Union initiative to compare symptoms of depression in 14 European centres
.
British Journal of Psychiatry
,
174
,
330
338
Scholte, R. S., van den Berg, G. J., & Lindeboom, M. (
2015
).
Long-run effects of gestation during the Dutch Hunger Winter famine on labor market and hospitalization outcomes
.
Journal of Health Economics
,
39
,
17
30
.
Shemyakina, O. (
2011
).
The effect of armed conflict on accumulation of schooling: Results from Tajikistan
.
Journal of Development Economics
,
95
,
186
200
.
Sixsmith, J., Sixsmith, A., Callender, M., & Corr, S. (
2014
).
Wartime experiences and their implications for the everyday lives of older people
.
Ageing & Society
,
34
,
1457
1481
.
Smith, J. P. (
2009
).
Reconstructing childhood health histories
.
Demography
,
46
,
387
403
.
Summerfield, D. (
2000
).
War and mental health: A brief overview
.
BMJ
,
321
,
232
235
.
Wang, G., Bartell, T. R., & Wang, X. (
2018
).
Preconception and prenatal factors and metabolic risk
. In Halfon, N., Forrest, C. B., Lerner, R. M., & Faustman, E. M. (Eds.),
Handbook of life course health development
(pp.
47
59
).
Cham, Switzerland
:
Springer International Publishing
.
Williams, N. E. (
2013
).
How community organizations moderate the effect of armed conflict on migration in Nepal
.
Population Studies
,
67
,
353
369
.
Williams, N. E. (
2015
).
Mixed and complex mixed migration during armed conflict: Multidimensional empirical evidence from Nepal
.
International Journal of Sociology
,
45
,
44
63
.
This is an open access article distributed under the terms of a Creative Commons license (CC BY-NC-ND 4.0).