Domains of Childhood Disadvantage and Functional Limitation Trajectories Among Midlife Men and Women in China

Abstract

Objective: To examine the impacts of four childhood disadvantage domains—parental behavior, hunger, health, and socioeconomic status (SES)—on functional limitation trajectories among midlife adults in China. Method: Data (2011-2015) from the China Health and Retirement Longitudinal Study (N = 8,646) were used to examine the associations between different domains of childhood disadvantage and functional limitation trajectories among adults aged 45 to 64. Results: Adverse parental behavior was not associated with baseline functional limitation but was associated with steeper trajectories for men. Health, hunger, and SES were associated with more functional limitations at baseline, and SES with steeper trajectories for men and women. Adulthood SES largely accounted for the associations between childhood SES and functional limitations for men. Discussion: It is important to examine multiple domains of childhood conditions because the type and magnitude of disadvantage may influence functional limitations in different ways among men and women in China.

Keywords

childhood adversity China disability social inequality

Introduction

The linkages between poor early-life conditions and worse health later in life have been well documented (Barker, 1997; Elo & Preston, 1992). Older adults who experienced poor early-life conditions are more likely to suffer from chronic conditions such as diabetes and hypertension (Blackwell, Hayward, & Crimmins, 2001), have a higher probability of engaging in unhealthy behaviors such as smoking (Non et al., 2014), and report more issues with functional mobility (Henretta & McCrory, 2016; Montez & Hayward, 2014). Functional mobility, which includes the accomplishment of daily tasks such as climbing stairs and carrying groceries, is a key dimension of health that reflects older adults’ ability to live independent lives (Verbrugge & Jette, 1994). Poor early-life conditions have been linked to a higher likelihood of onset of functional limitations, as well as faster rates of functional decline in developed countries worldwide (Haas, 2008; Haas, Oi, & Zhou, 2017).

Two limitations of the literature on childhood origins of later-life functional mobility merit attention. First, less is known about the influence of early-life conditions on functional limitation trajectories in less developed countries such as China. As less developed countries undergo epidemiological transitions, it is crucial to study how poor conditions experienced earlier in the life course may contribute to their increasing physical disability burdens. The majority of China’s current older individuals suffered from hunger, wars, infectious diseases, and poverty in their childhoods that may have had particularly deleterious consequences for their health (Guilmoto & Jones, 2016). Therefore, associations between childhood conditions and health may be stronger in the Chinese context compared with older adults in more developed countries. Moreover, from a demographic perspective, there are implications for population aging. China is one of the most rapidly aging countries in the world, with the number of adults 65 years and older expected to increase from 100 to 330 million by 2050 (United Nations, 2017). Thus, examining the childhood origins of health inequalities among China’s population is vital for assessing the need for and allocation of caregiving and health care services.

Second, questions remain about to operationalize poor early-life conditions. Some studies operationalize disadvantage in childhood as events (e.g., parental divorce; Umberson, Williams, Thomas, Liu, & Thomeer, 2014), whereas others examine experiences (e.g., child abuse; Yang et al., 2013), conditions (e.g., low socioeconomic status [SES]; Montez & Hayward, 2014), or health problems (e.g., childhood infectious disease; Margolis, 2010). Recently, results from several U.S. studies have made the case for simultaneously looking at clusters or “domains” of childhood disadvantage (Ferraro, Schafer, & Wilkinson, 2016; Kemp, Ferraro, Morton, & Mustillo, 2018; N. R. Smith et al., 2016). For example, Kemp and colleagues (2018) examined multiple domains of childhood disadvantage, including infectious diseases, SES, and parental behavior, on cancer risk among older adults. Heterogeneity in their findings by childhood domain underscores how various types of childhood disadvantage may shape later-life health in different ways (Kemp et al., 2018).

The present study seeks to address these limitations in the literature using three waves (2011-2015) of data from the China Health and Retirement Longitudinal Study (CHARLS) to explore how four clusters of childhood disadvantage shape functional mobility trajectories among midlife adults. Midlife is an important age group for study because previous research has shown that midlife functional limitations predict more severe disability and extensive health care use later in life (Stenholm, Guralnik, Bandinelli, & Ferrucci, 2013). In addition to contextualizing the analyses and results in the Chinese context, the findings provide new insight into the contributions of four types of childhood disadvantage to later-life functional mobility.

Life Course Pathways Linking Childhood Conditions and Adult Health

The life course theoretical framework conceptualizes how early-life events, beginning in the prenatal period and extending into early adulthood, have long-lasting impacts on adult health (Ben-Shlomo & Kuh, 2002). This framework proposes three hypotheses linking childhood conditions to later-life health: the sensitive period model, the accumulation of risks model, and the pathways model. The sensitive period model posits that exposures during sensitive periods of development (e.g., childhood) result in biological imprinting or scarring that increases vulnerability to health problems later in life (Ben-Shlomo & Kuh, 2002). For example, going hungry in childhood may negatively affect bone health, increasing the risk of osteoporotic fractures and susceptibility to functional limitations later in life (Langley-Evans, 2015). The accumulation of risks model hypothesizes that individuals accumulate experiences throughout the life course that then manifest in poor or good health later in life (Umberson et al., 2014). Finally, according to the pathways model, childhood conditions may influence adult health risks indirectly, such that early-life experiences place individuals on divergent trajectories that differentially expose them to conditions and stressors that influence health (Pudrovska & Anikputa, 2014). This model predicts that the impacts of early-life experiences on later-life health are mediated through conditions such as SES attainment in adulthood. In sum, while these three life course models are not mutually exclusive and difficult to disentangle empirically (Hallqvist, Lynch, Bartley, Lang, & Blane, 2004), they provide a useful framework for conceptualizing the pathways through which childhood conditions may affect health.

Clusters of Childhood Disadvantage and Health

Differentiating between types of childhood disadvantage is one issue that has received less attention in the literature linking early-life conditions to later-life health. Studies have typically examined one or two specific types of conditions or a sum of indicators, paying less attention to comparing different types of exposures (Friedman, Montez, Sheehan, Guenewald, & Seeman, 2015). For example, a study using The Irish LongituDinal Study on Ageing (TILDA) found that while low childhood SES was indirectly related to physical functioning via adulthood SES, poor childhood health was directly associated with worse physical functioning, net of adult SES (Henretta & McCrory, 2016). Similarly, in the context of a less developed country, Huang, Soldo, and Elo (2011) found that while childhood health and hunger were directly associated with higher odds of developing a lower-body functional limitation among older adults in Mexico, childhood SES was indirectly related to physical health mainly through its impact on SES in adulthood. Types of exposures in extant studies are limited but suggest that associations with health may not be uniform across all types of exposures in childhood.

Recently, several studies on the United States have simultaneously examined the influences of clusters or “domains” of childhood disadvantage on later-life health (Ferraro et al., 2016; Kemp et al., 2018; N. R. Smith et al., 2016). Using the National Survey of Midlife Development in the United States (MIDUS), Ferraro and colleagues (2016) found that the domains of low childhood SES and childhood physical abuse were associated with baseline number of chronic conditions, as well as the development of new chronic conditions over time. Similarly, Williams, Kemp, Ferraro, and Mustillo (2018) found that risky parental behavior, low SES, childhood physical impairments, and risky adolescent behavior all reduced the odds of being disease free later in life among older U.S. adults. Although most of the childhood domains appeared to operate indirectly through adulthood SES, there was a direct association between risky parental behaviors and lower odds of being disease free (Williams et al., 2018). These results suggest that adulthood SES may be in the pathway between childhood conditions and health for some childhood domains, but not others. Beyond looking at disease, a recent study analyzed the connection between childhood domains of disadvantage and handgrip strength using the U.S. Health and Retirement Survey (HRS), finding that multiple domains were related to baseline handgrip strength (N. R. Smith et al., 2016). Interestingly, only the childhood physical impairment domain was related to handgrip strength over time, and only for men, suggesting differences by both childhood domain and sex (N. R. Smith et al., 2016).

The foundation for simultaneously studying multiple domains of childhood disadvantage rests on cumulative inequality theory, which was developed from the life course framework (Ferraro & Shippee, 2009). According to this theory, it is important to study different childhood domains because disadvantage in one life domain may spill over to others (Ferraro & Shippee, 2009). For example, some studies examine whether poor health during childhood influences health later in life, whereas others focus on the impacts of family financial hardship. On one hand, poor health in childhood may contribute to or worsen family finances (i.e., cost of seeing a doctor). On the other hand, family financial hardship may lead to or worsen a child’s health issue (i.e., unable to afford medication). Furthermore, cumulative inequality theory posits that the magnitude of disadvantage within each domain is also related to health (Ferraro & Morton, 2016). Evidence for a dose–response relationship between childhood disadvantage and later-life health has been documented, where the risk of disease and disability increases for each additional disadvantage reported (Friedman et al., 2015). However, the type of childhood disadvantage and response may not be comparable. For instance, children experiencing physical abuse and those living with the aftermath of a severe illness likely contend with diverse challenges that may prompt different biological and psychosocial responses. As such, Kemp and colleagues (2018) contend that the next step for research on the childhood origins of later-life health is to examine accumulated childhood disadvantage within and across domains.

The Case of China

Childhood disadvantage has been shown to be a salient predictor of health among older adults in China (Shen & Zeng, 2014; Wen & Gu, 2011). Poor early-life conditions have been linked to cognitive impairment (Zhang, Liu, Li, & Xu, 2017), higher risk of mortality (Shen & Zeng, 2014), and poorer self-rated health (Wen & Gu, 2011). However, less is known about the relationship between childhood disadvantage and functional mobility in China. The extant research has largely been cross-sectional or with short-term follow-up, and few studies have examined trajectories of functional mobility over time. Results from the studies that have used longitudinal data to examine functional mobility in China are mixed. Wen and Gu (2011), using 3-year follow-up (2005) data from the Chinese Longitudinal Healthy Longevity Survey (CLHLS), documented that high childhood SES (having both parents alive at age 10) was associated with lower odds of having activities of daily living (ADLs) limitation among adults 65 years and older. In contrast, Han and Shibusawa (2015), also using the CLHLS (1998-2005), found that higher childhood SES was not associated with baseline risk of having an ADL, nor change over time. However, their childhood SES measure was a continuous variable comprising five indicators: born in an urban area, obtained adequate medical services, never went to bed hungry, both parents alive at age 10, and father was in white-collar occupation (Han & Shibusawa, 2015). The inconsistencies in the results of these studies may be due to differences in their respective childhood SES measures. Han and Shibusawa’s (2015) childhood SES measure included different types of childhood disadvantage that may have had differential impacts on later-life functional mobility. Given these mixed results, it would be useful to examine accumulated childhood disadvantage within and across multiple domains to parse out the contribution of each type of childhood disadvantage to functional mobility in the Chinese context.

The Present Study

Contribution and Research Questions

In sum, childhood disadvantage has been linked to worse functional mobility among older adults in more developed countries. Still, it remains unclear how childhood disadvantage influences functional mobility among older adults in China. Part of this ambiguity stems from questions about how to operationalize childhood disadvantage in studies linking early-life conditions to later-life health. The present study contributes to the literature on childhood disadvantage and functional mobility in four ways. First, this study focuses on a less developed country with a rapidly aging population that is increasingly in need of health care and caregiving resources. Chinese older adults, compared with their counterparts in the United States, are characterized by their poor early-life circumstances including impoverished living conditions, persistent hunger, and heightened exposure to infectious diseases that may have had long-lasting consequences for their health (McEniry & McDermott, 2015). Therefore, the impact of childhood disadvantage on later-life health may be stronger in the Chinese context. Second, focusing on midlife adults captures a younger age group than previously studied, providing insight into how childhood disadvantage may influence functional mobility at younger ages. Third, few studies have examined functional mobility over time in China, including the possibility that childhood disadvantage may contribute to widening or diminishing health disparities as respondents’ age. Finally, little is known about how different domains of childhood disadvantage may affect functional health in the Chinese context. CHARLS includes an extensive set of questions on childhood conditions including those regarding parental alcohol abuse and childhood physical abuse that have not previously been asked on surveys of older adults in China. Applying the approach of recent U.S. studies operationalizing childhood disadvantage as several domains (i.e., Ferraro et al., 2016; Kemp et al., 2018; N. R. Smith et al., 2016), the present study therefore seeks to examine the independent contributions of four different childhood domains of disadvantage to midlife functional health in China. Specifically, this study addresses the following research questions:

Research Question 1: Which types of childhood disadvantage are associated with baseline functional limitations among men and women in China?

Research Question 2: Does adulthood SES account for any of the associations between the different types of childhood disadvantage and baseline functional limitations among men and women in China?

Research Question 3: Do the relationships between the different types of childhood disadvantage and functional limitations change as people age?

Data and Method

Data

Data were obtained from Wave 1 (2011), Wave 2 (2013), and Wave 3 (2015) of the CHARLS, which is a nationally representative sample of Chinese residents aged 45 and older (see Zhao, Hu, Smith, Strauss, & Yang, 2014). Adopting a multistage stratified Probability Proportional to Size (PPS) sampling, the baseline survey included about 10,000 households and 17,500 individuals living in 150 counties/districts and 450 villages/residential communities in 28 provinces. Between the 2013 and 2015 core waves, CHARLS fielded a special Life History Survey in 2014 for the entire sample. The purpose of this survey was to gather retrospective information about respondents’ health, family, and living conditions during childhood. The childhood conditions’ measures were obtained from this special survey.

Sample

From the baseline sample of 17,708, proxy respondents, spouse respondents younger than 45 years old, and respondents older than 64 were excluded, leaving with a sample of 12,527 respondents aged 45 to 64 at baseline who had complete information on the functional mobility questions. Between baseline (2011) and Wave 3 (2015), 296 respondents died and 3,007 were lost to follow-up. After further excluding respondents with missing data on key predictor variables, the analytic sample was 8,646 participants. Those respondents who died or were lost to follow-up between waves had more chronic conditions, higher body mass index (BMI), lower education, and lower monthly household expenditure per capita compared with those who remained in the sample.

Measures

The variables comprising the childhood disadvantage domains were obtained from the special Life History Survey fielded in 2014. Education, household per capita expenditure, obesity, smoking behavior, chronic conditions, and urban residency were measured at Wave 1. Functional mobility was measured at each core wave (2011, 2013, and 2015). Independent and control variables were mean centered to facilitate the interpretation of findings (Singer & Willet, 2003).

Dependent Variable

Functional limitation represents a measure of later-life health status that is of notable empirical and theoretical importance because it plays a role in connecting disease pathology to disability within the disablement process (Verbrugge & Jette, 1994). There are several possibilities to consider in measuring functional limitations. The functional limitation questions, which roughly correspond to the Nagi functional mobility questions, were used because there may be less cultural bias in using them compared with ADLs (Hardy, Acciai, & Reyes, 2014; Nagi, 1976). Therefore, focusing on functional limitations allows the results of the present study to be contextualized within the larger empirical literature on aging and disability worldwide (Haas et al., 2017). The measure of functional limitations assessed restrictions in basic physical functions and fine motor skills (Guralnik & Ferrucci, 2003). At each wave, respondents were asked whether they had any difficulty (yes/no): (a) walking 1 km; (b) getting up from a chair after sitting for long periods; (c) climbing several flights of stairs without resting; (d) stooping, kneeling, or crouching; (e) reaching or extending your arms above shoulder level; (f) lifting or carrying weights more than 10 lb/5 kg, such as a bag of groceries; and (g) picking up a small coin from a table. The items were summed into a count of functional limitations (range 0-7; Osman & Walsemann, 2017). Internal consistency (Cronbach’s α) was 0.749 for all items combined at baseline.

Independent Variables

The independent variables of interest include four domains of childhood disadvantage: (a) adverse parental behavior, (b) family SES, (c) childhood health, and (d) childhood hunger. Each domain comprises multiple indicators with each indicator coded as 1 if the respondent reported the event or condition, and 0 otherwise. The four domains were created as count variables, informed by previous studies (Ferraro et al., 2016; Kemp et al., 2018; N. R. Smith et al., 2016), correlation matrices, polychoric factor analyses, and the China-specific context. The adverse parental behavior, SES, and health domains were top coded at 2+ and treated as categorical variables with 0 as the reference category. The childhood hunger domain was not top coded at 2+ because it comprised three categories (0, 1, 2) where the highest value was 2.

Adverse parental behavior has three indicators. Respondents were asked whether they had a parent or guardian who abused alcohol, whether they had a parent or guardian who smoked, and whether they were physically abused by a parent or guardian during childhood. Previous studies on the United States have linked these types of parental behaviors to a higher risk of health problems later in life (Greenfield & Marks, 2009; Lee, Tsenkova, & Carr, 2014). Childhood SES comprises three indicators that capture information about the SES of the participant when she or he was growing up: whether both parents were illiterate, whether both parents were employed in agricultural occupations, and perception of family SES (1 = much worse off than their neighbors during childhood). Previous studies on China have shown parental literacy to be a robust predictor of childhood SES (Shen & Zeng, 2014; Wen & Gu, 2011) The literacy and occupation of one parent were used in the case where the respondent reported information for only one parent. Childhood health comprises three indicators: lacking a usual source of medical care in childhood, being confined to bed for a month or more, and fair/poor self-rated childhood health. Previous research has suggested that general assessment of childhood health has reasonably good reliability and validity (Haas, 2007).

Childhood hunger was chosen because it has been linked to a wide variety of adult health outcomes, including functional limitations, in less developed countries (Huang et al., 2011). It was treated as a separate domain because previous studies have recommended including sources of ill health in childhood as separate variables because health insults may affect children in diverse ways (Currie & Vogl, 2013). Moreover, hunger was widespread in China during the middle of the 20th century due to the Great Leap Forward Famine (1952-1962) and widespread food shortages at the beginning of the Cultural Revolution (1966-1976) (Guilmoto & Jones, 2016). Respondents were asked whether they ever went hungry in childhood. Those who responded yes were asked to choose during which ages they experienced hunger in childhood: 6 to 11 years old and/or 12 to 17 years old. Each of these age ranges were coded as binary variables, resulting in a variable ranging from 0 (experienced no hunger in childhood) to 2 (experienced hunger during both age ranges in childhood).

Adulthood SES

All adult SES variables were measured at baseline in 2011. Three measures of adult SES were included in the analyses: (a) education, (b) urban residency, and (c) monthly household expenditure per capita. Education was measured as three categories: (a) no formal education, (b) elementary school, and (c) middle school or more. CHARLS includes education as a categorical variable as opposed to years of schooling. No formal education includes illiterate respondents, literate respondents who did not finish primary school, and those who received private tutoring.

Urban residency was a binary variable equal to 1 if the respondent lives in an urban area and 0 if the respondent lives in a rural area. Urban residents in China tend to have better access to health care, higher standards of living, and overall higher SES than rural residents (Wen & Gu, 2011). Because older Chinese adults tend to reside in multigenerational households, personal income is typically not a good measure of SES. The household member most knowledgeable about expenses was asked to report monthly expenditure on a range of goods and services. In contrast to asset-based wealth indices (Filmer & Pritchett, 2001), household expenditure is generally viewed as a short-term measure for evaluating current household financial resources (Lei, Smith, Sun, & Zhao, 2014). Average expenditure per capita was then calculated as the sum of all the expenditure aggregates divided by the number of household members.

Additional Covariates

Other covariates were measured at baseline in 2011 and included age, sex, health behaviors (smoking and obesity), and chronic conditions. Smoking behavior was measured by a dichotomous variable equal to 1 if the respondent was a current smoker. BMI was measured using height and weight. Respondents were categorized as obese (≥30 BMI) and not obese (<30 BMI). Other specifications of BMI yielded similar results. A number of physician-diagnosed chronic health conditions were reported by respondents. These included hypertension, dyslipidemia, diabetes, cancer, chronic lung conditions, liver disease, heart disease, kidney disease, and digestive disease (scale 0-9).

Analytic Strategy

Frequencies for the childhood disadvantage domains and functional limitations were first described by sex. Descriptive statistics are presented at baseline for all variables used in the analyses. Correlations among the four types of childhood disadvantage were low and diagnostic tests revealed no issues with multicollinearity. Growth curve models were estimated to examine the relationships between the childhood disadvantage domains and functional limitations at baseline and over time, along with the set of covariates identified above. Growth curve modeling is designed to analyze the trajectories of repeated measures in longitudinal data (Singer & Willet, 2003). Growth curves are characterized by a fixed part which contains average effects for the intercept (initial status) and slope (rate of change over time) and a random part which contains individual differences (variance) in the intercept, slope, and the within-person residual. Positive effects on the intercept reflect higher baseline levels of limitation, whereas negative ones reflect lower baseline levels. Those covariates with positive effects on the slope terms are associated with steeper increases in functional limitations over age. Those covariates with negative effects on the slopes are associated with more gradual increases. Finally, different covariance structures were tested and the unstructured covariance matrix was selected, which allowed a correlation between the random slope and random intercept.

In these analyses, time was defined in terms of chronological age (as opposed to study duration) because convergence or divergence over age was an interest of the study. Models were estimated separately for men and women because preliminary analyses showed significant interactions between sex and childhood domains on functional limitations. For men and women, two nested models were estimated that jointly included all four childhood domains of disadvantage. Model 1 included the four childhood domains, demographics, health behaviors, and chronic conditions. Model 2 included everything in Model 1 in addition to the adult SES measures. This modeling strategy allowed for an examination of whether the adult SES variables accounted for the relationships between the childhood domains and functional limitations. Post-estimation Wald tests were used to assess whether there were significant differences between the coefficients within each childhood domain (e.g., one disadvantage vs. 2+ disadvantages). All models controlled for chronic conditions and health behaviors because CHARLS does not ask about the order of occurrence of chronic conditions, health behaviors, or functional limitations. For example, chronic conditions may be in pathway between childhood disadvantage and functional limitations. However, it is also possible that functional limitations may be in the pathway between childhood disadvantage and chronic conditions.

The grand mean age for all sample members, which is equal to 55, was used to center age. An age-squared term was used to test for nonlinearity in the relationship between age and functional limitations but was not significant and therefore not included in the models. Interaction terms between age and each childhood domain were included in all models to examine whether childhood disadvantages were associated with widening or diminishing disparities in functional mobility over time. Adulthood SES variables and all of the other covariates were included as time-invariant measures. The results presented are based on complete case analysis. All models were estimated using Stata 14.

Results

Descriptive Statistics

Table 1 presents descriptive statistics of the full sample and separately for women and men. Women had more functional limitations at baseline as well as in subsequent waves compared with men (p < .001). All types of childhood disadvantage differed by sex. About 16% of women and 25% of men experienced two or more adverse parental behaviors. Nearly equal percentages of men and women experienced two or more indicators of socioeconomic disadvantage in childhood. Approximately 7% of women and 6% of men experienced two or more childhood health problems, and about one third of the female and male samples experienced hunger in age ranges of 6 to 11 and 12 to 17 in childhood. Differences between women and men were also observed for age, education, health behaviors, and chronic conditions.

Table 1.

Descriptive Statistics by Sex From the China Health and Retirement Longitudinal Study (2011-2015).

	Full sample (N = 8,646)	Women (N = 4,669)	Men (N = 3,977)	χ²/t test
	% or M (SE)	% or M (SE)	% or M (SE)	χ²/t test
Functional limitations				***
2011	1.14 (1.49)	1.37 (1.56)	0.87 (1.34)
2013	1.28 (1.54)	1.55 (1.61)	0.95 (1.37)
2015	1.38 (1.62)	1.70 (1.71)	1.00 (1.42)
Demographics
Age (years)	54.62 (5.69)	54.32 (5.69)	54.97 (5.67)	***
Childhood disadvantage domains
Adverse parental behavior				***
0	35.00	39.02	30.27
1	45.11	45.21	44.98
2+	19.89	15.76	24.74
Socioeconomic status
0	14.53	15.38	13.53	*
1	35.23	35.60	34.80
2+	50.24	49.03	51.67
Health				***
0	67.77	65.13	70.86
1	25.41	27.50	22.96
2+	6.82	7.37	6.19
Hunger				***
0	35.89	38.66	32.64
1	36.57	34.74	38.72
2	27.54	26.60	28.64
Adult socioeconomic status
Education				***
No formal education	43.27	56.91	27.26
Elementary school	21.39	17.63	25.80
Middle school or more	35.35	25.47	46.94
Household monthly PCE^a	8,247.77 (8,411.03)	8,161.39 (8,466.92)	8,349.18 (8,344.86)
Residence
Rural	65.04	64.62	65.53
Urban	34.96	35.38	34.47
Health behaviors and chronic conditions
Current smoker	30.55	5.31	60.17	***
Obese (BMI ≥30)	19.73	21.20	18.00	***
Number of chronic conditions	1.37 (1.38)	1.45 (1.41)	1.28 (1.34)	***

Note. All variables are from baseline except for function limitations. PCE = per capita expenditure; BMI = body mass index.

Household monthly per capita expenditure (in Chinese Yuan).

p < .05, **p < .01, ***p < .001.

Growth Curve Model Results

Table 2 presents results from growth curve models showing the associations between the four domains of childhood disadvantage and midlife functional limitations for Chinese men and women. The first half of the table shows the influence of each childhood domain on the baseline level of functional limitation. Positive coefficients indicate higher baseline levels of functional limitation, whereas negative coefficients reflect lower baseline levels of functional limitation. The second half of the table shows interactions between each childhood domain and age. Positive coefficients for these age interactions reflect faster increases in functional limitations with age, whereas negative coefficients indicate slower increases in functional limitations with age. Model 1 included the four domains of childhood disadvantage, age, health behaviors, and chronic conditions, as well as interactions between each childhood domain and age. Model 2 further included education, monthly household expenditure per capita, and urban residency to examine whether adulthood SES attenuated any of the associations between the childhood domains and functional limitations. In Model 1, for the female sample, experiencing one (b = 0.19, p < .001) or 2+ SES disadvantages (b = 0.32, p < .001) was associated with more baseline functional limitations compared with women who experienced no SES disadvantages in childhood. Furthermore, one (b = 0.22, p < .001) or 2+ childhood health disadvantages (b = 0.48, p < .001) were associated with more functional limitations at baseline. Experiencing hunger during one (b = 0.20, p < .001) or two (b = 0.19, p < .001) age groups in childhood was associated with more baseline functional limitations. There was one significant multiplicative interaction with age for women: experiencing 2+ SES disadvantages in childhood was associated with a faster rate of accumulation, suggesting that childhood SES inequalities in functional limitations widen over age for midlife women (b = 0.02, p = .023).

Table 2.

Growth Curve Model Estimates for Childhood Disadvantage and Functional Limitations.

	Women (N = 4,669)		Men (N = 3,977)
	Model 1	Model 2	Model 1	Model 2
	Estimate (SE)	Estimate (SE)	Estimate (SE)	Estimate (SE)
Fixed effects
Intercept (at age 55)	1.10 (0.06)***	1.35 (0.06)***	0.67 (0.06)***	1.04 (0.07)***
Types of childhood disadvantage
Adverse parental behaviors (reference: none)
1 adverse parental behavior	−0.06 (0.04)	−0.05 (0.04)	0.01 (0.04)	0.03 (0.04)
2+ adverse parental behaviors	−0.10 (0.05)	−0.09 (0.05)	0.06 (0.05)	0.08 (0.05)
Socioeconomic disadvantages (reference: none)
1 SES	0.19 (0.05)***	0.07 (0.06)	0.15 (0.05)***	0.05 (0.06)
2+ SES	0.32 (0.05)***	0.15 (0.05)**	0.25 (0.05)***	0.10 (0.05)
Health disadvantages (reference: none)
1 health	0.22 (0.04)***	0.18 (0.04)***	0.16 (0.04)***	0.12 (0.04)**
2+ health	0.48 (0.07)***	0.41 (0.07)***	0.50 (0.07)***	0.45 (0.07)***
Hunger (reference: none)
1 hunger	0.20 (0.04)***	0.18 (0.04)***	0.04 (0.04)	0.04 (0.04)
2 hunger	0.19 (0.05) ***	0.16 (0.05)***	0.10 (0.05)*	0.06 (0.05)
Age	0.01 (0.01)	0.01 (0.01)	−0.01 (0.01)	−0.02 (0.01)
Adverse parental behaviors (reference: none)
1 adverse parental behavior × Age	0.00 (0.01)	0.00 (0.01)	0.02 (0.01)**	0.02 (0.01)**
2+ adverse parental behavior × Age	−0.00 (0.01)	−0.01 (0.01)	0.02 (0.01)**	0.02 (0.01)**
Socioeconomic disadvantages (reference: none)
1 SES × Age	0.01 (0.01)	0.01 (0.01)	0.02 (0.01)*	0.02 (0.01)*
2+ SES × Age	0.02 (0.01)*	0.02 (0.01)*	0.02 (0.01)	0.01 (0.01)
Health disadvantages (reference: none)
1 health × Age	0.00 (0.01)	−0.00 (0.01)	0.01 (0.01)	0.01 (0.01)
2+ health × Age	−0.00 (0.01)	−0.00 (0.01)	−0.02 (0.01)	−0.02 (0.01)
Hunger (reference: none)
1 hunger × Age	−0.01 (0.01)	−0.00 (0.01)	−0.00 (0.01)	−0.00 (0.01)
2 hunger × Age	0.01 (0.01)	0.01 (0.01)	0.00 (0.01)	0.00 (0.01)
Adult socioeconomic status
Education (reference: no formal education)
Elementary school		−0.11 (0.05)*		−0.15 (0.05)**
Middle school or more		−0.18 (0.05)***		−0.32 (0.04)***
Household per capita monthly expenditure^a		−0.00 (0.00)***		−0.00 (0.00)***
Urban residency (reference: rural)		−0.16 (0.04)***		−0.10 (0.04)**
Health behaviors and chronic conditions
Current smoker	0.17 (0.08)*	0.15 (0.08)	−0.04 (0.04)	−0.05 (0.03)
Obese (BMI ≥30)	0.17 (0.04)***	0.21 (0.04)***	0.09 (0.04)*	0.13 (0.04)**
Number of chronic conditions	0.37 (0.01)***	0.38 (0.01)***	0.29 (0.01)***	0.29 (0.01)***
Goodness of fit
Log likelihood	−24,981.66	−24,927.73	−19,254.40	−19,201.34
AIC	50,013.32	49,913.45	38,558.79	38,460.68
BIC	50,202.00	50,132.32	38,743.47	38,674.90

Note. China Health and Retirement Longitudinal Study, 2011 to 2015. SES = socioeconomic status; BMI = body mass index; AIC = Akaike information criterion; BIC = Bayesian information criterion.

Per capita expenditure is b = −0.0000159 for women and b = −0.0000126 for men.

p < .05. **p < .01. ***p < .001.

With the introduction of the adult SES measures in Model 2, the association between one SES disadvantage in childhood and baseline functional limitation was no longer significant (b = 0.07, p = .207). There remained a significant relationship between experiencing 2+ childhood SES disadvantages and more functional limitations at baseline (b = 0.15, p = .005), suggesting that higher SES in adulthood may help compensate for some SES disadvantage in childhood for women. The remaining significant childhood disadvantage domain coefficients barely changed with the introduction of the adulthood SES variables, indicating that childhood health, hunger, and 2+ SES disadvantages were directly associated with more baseline functional limitations. A post-estimation Wald test (p = .002) revealed that the coefficients for one and 2+ health disadvantages were significantly different, suggesting a dose–response relationship between childhood health disadvantage and functional limitations. However, the Wald test (p = .628) for childhood hunger showed no significant difference between experiencing hunger during one or two age groups. Interestingly, the interaction between age and 2+ SES disadvantages in childhood remained significant (b = 0.02, p = .032), signifying that achieved SES in adulthood was not strong enough to offset the influence of poor childhood SES on widening age disparities in functional mobility. Higher education, more monthly household per capita expenditure, and urban residency were associated with fewer baseline functional limitations. Obesity and chronic conditions were associated with more baseline functional limitations.

Similar to the female sample, Model 1 showed that men who experienced one (b = 0.15, p < .001) or 2+ (b = 0.25, p < .001) childhood SES disadvantages were more likely to have a higher number of baseline functional limitations compared with men who experienced no childhood SES disadvantages. One (b = 0.16, p < .001) and 2+ childhood health disadvantages (b = 0.50, p < .001) were also associated with more functional limitations at baseline for men. However, only those men who experienced hunger during both age groups in childhood had an increased risk of more functional limitations at baseline (b = 0.10, p = .041).

Interestingly, adverse parental behavior was not associated with baseline functional limitations for men or women. However, there were significant multiplicative interactions between age and one (b = 0.02, p = .002) and 2+ adverse parental behaviors (b = 0.02, p = .005), suggesting faster functional decline for men who experienced at least one adverse parental behavior in childhood. Experiencing one SES disadvantage in childhood was also associated with steeper accumulation of functional limitations over age for men (b = 0.02, p < .001). Taken together, these results suggest that adverse parental behaviors and childhood SES disadvantage contribute to widening disparities in functional mobility among midlife men in China.

With the introduction of the adulthood SES measures in Model 2, the associations between one (b = 0.05, p = .29) and 2+ (b = 0.10, p = .07) SES disadvantages in childhood and baseline functional limitations were no longer significant for men, suggesting that higher SES in adulthood may attenuate the negative impacts of childhood SES on functional health. Furthermore, experiencing hunger during two age groups in childhood was no longer significant (b = 0.06, p = .19), indicating that adulthood SES may be in the pathway between childhood hunger and midlife functional mobility for men. However, the associations between one (b = 0.12, p = .005) and 2+ childhood health disadvantages (b = 0.45, p < .001) remained statistically significant, suggesting a direct relationship between childhood health and worse functional mobility at midlife. Moreover, a post-estimation Wald test (p < .001) revealed that the coefficients for one and 2+ health disadvantages were significantly different, suggesting a dose–response relationship between childhood health disadvantage and functional limitations. The interactions between age and adverse parental behaviors and one childhood SES disadvantage also remained significant in Model 2. Interestingly, adulthood SES appeared to attenuate the associations between childhood SES and baseline level of functional limitations, but not the association between childhood SES and steeper accumulation of functional limitations over age. Similar to women, higher education, more monthly household per capita expenditure, and urban residency were all associated with fewer functional limitations at baseline for men. Being obese and more chronic conditions were both associated with more baseline functional limitations.

Discussion

Adult health is shaped by events that occur across the life course. Experiencing risky parental behaviors, prolonged hunger, poor health, and growing up in a socioeconomically disadvantaged family can impart strong and enduring impacts on health later in life, often irrespective of adult circumstances. However, considerably less attention has been paid to determining how best to operationalize childhood experiences, especially in the context of less developed countries such as China. Using the first longitudinal survey with a nationally representative sample of adults 45 years and older in China, the present study addressed this gap by examining the influence of four domains of childhood disadvantage on functional mobility. Results highlighted the importance of examining accumulated childhood disadvantage within and across domains in China. Childhood disadvantage was associated with baseline and rate of change in functional mobility, but the relationships were distinct by type, magnitude of disadvantage (e.g., one or two or more in a domain), and sex.

The first research question sought to identify which domains of childhood disadvantage were associated with baseline number of functional limitations among midlife adults in China. In the full models, experiencing one or 2+ health disadvantages in childhood were associated with more baseline functional limitations for both men and women, even after adjustment for adult SES, health behaviors, and chronic conditions. Moreover, there was evidence of a dose–response relationship for childhood health disadvantage where each additional disadvantage was significantly associated with more baseline functional limitations. These findings may reflect biological scarring, in which poor health in childhood is indicative of physiological changes to the body (Barker, 1997). These results also support several previous studies in more developed countries that found direct associations between childhood health and worse physical functioning later in life (Henretta & McCrory, 2016; Montez & Hayward, 2014).

Experiencing hunger during one or two age groups in childhood was associated with more baseline functional limitations for women, but not men. This result was surprising given that Wen and Gu (2011) did not find a significant association between childhood hunger and having an ADL among Chinese adults 65 years and older. Inconsistencies in these results may reflect the living conditions during different time periods of Chinese history. The respondents in Wen and Gu’s (2011) study were children during the first half of the 20th century, whereas respondents in the present study were children during the 1950s to 1970s. That childhood hunger was found to be associated with functional mobility in the present study may reflect biological scarring incurred during the Great Leap Forward Famine (1952-1962) and the beginning of the Cultural Revolution (1966-1976) when collectivization of farms led to widespread food shortages (Zhang et al., 2017). Furthermore, while Huang and colleagues (2011) did find that childhood hunger was associated with more functional limitations among older adults in Mexico, they did not find evidence of gender differences. The case of China may be different due to entrenched patriarchy when respondents were children where sons were more likely to receive more and better food compared with daughters (Croll, 1983).

The second research question sought to address whether adulthood SES would account for any of the associations between the different types of childhood disadvantage and baseline functional limitations. Experiencing one or 2+ SES disadvantages in childhood was associated with more baseline functional limitations for both men and women. Although the measurement of childhood SES in Smith and colleagues (2016) article comprised different elements compared with how childhood SES was measured in the present study, the authors also reported that more disadvantageous childhood SES events were associated with worse physical functioning (handgrip strength) among older U.S. adults. However, after controlling for adult SES, the associations between one SES event and 2+ SES and baseline functional limitations were no longer significant for Chinese men, suggesting that adult SES largely accounted for these relationships. These results are in line with previous studies, in both developed and less developed countries, showing that high SES in adulthood may ameliorate the impacts of poor SES in childhood (Montez & Hayward, 2014; Shen & Zeng, 2014). In particular, this result is consistent with Shen and Zeng’s (2014) study that found that childhood SES operated indirectly through adult SES to affect health and mortality among adults 80 years and older in China.

However, while adulthood SES appeared to attenuate the coefficient for experiencing one childhood SES disadvantage for women, the association between 2+ SES disadvantages and baseline functional limitations remained statistically significant. These findings may suggest the importance of accumulated disadvantage and a possible threshold effect for women: 2+ childhood SES disadvantages may be the tipping point leading to poor health.

Taken together, the sex differences in the results regarding the childhood hunger and SES domains may reflect a deeply rooted gender stratification system in China where women were disadvantaged throughout their lives compared with men (Croll, 1983). In traditional Chinese communities, there were strong economic incentives to favor boys at the expense of girls not only in their education and occupation outcomes but also in their nutrition and health care (Lei et al., 2014). These economic incentives were largely driven by Confucian filial piety beliefs that a son should support his parents in their old age (Shi, 2018). Thus, while high achieved SES in adulthood may have had the power to attenuate the impact of childhood SES disadvantages for midlife men, it likely had less of an impact for ameliorating this hardship for women.

The third research question addressed whether the relationships between the childhood disadvantage domains and functional limitations changed over time. Results revealed that childhood SES disadvantage (2+ for women and one for men) was associated with widening disparities in functional mobility. Although the size of some of the coefficients for these age interactions was small, they are in line with those found in other studies (Haas, 2008; Haas et al., 2017). Interestingly, while adulthood SES appeared to mediate the influence of childhood SES disadvantage on baseline functional mobility for men, it did not attenuate the negative influence of childhood SES on the functional limitation trajectories. These results were inconsistent with previous studies in several developed countries that found that childhood SES disadvantage did not influence trajectories of functional limitations over time (Haas, 2008; Haas et al., 2017). Haas and colleagues (2017) found that the impact of childhood SES (father’s education and occupation) on trajectories of functional limitations was largely mediated by SES attainment in adulthood in 12 high-income countries. Furthermore, Han and Shibusawa (2015) did not find an association between childhood SES and ADL trajectories among Chinese adults 65 years and older. Perhaps, conceptualizing childhood SES disadvantage as accumulation within a domain revealed associations over time among the most disadvantaged respondents (i.e., women who experienced 2+ childhood SES disadvantages). It is also possible that, as discussed earlier, these results reflect biological scarring incurred while living through the tumultuous Cultural Revolution in China.

Finally, results suggest that adverse parental behaviors in childhood may shape functional limitation trajectories among men. This was an interesting result in the Chinese context because the adverse parental behaviors domain comprised questions that have not been asked in surveys on older adults in China. However, studies from the United States have linked childhood physical abuse to worse physical functioning later in life, largely through enduring forms of bodily harm and by raising the risk poor health behaviors (Greenfield & Marks, 2009; Lee et al., 2014). Furthermore, parental smoking and alcohol abuse may influence children’s health through environmental exposure to tobacco smoke (i.e., secondhand smoke), modeling a lifestyle of smoking and drinking, and perhaps by biologically inducing symptoms of early nicotine dependence in the womb. Indeed, there are substantial differences in smoking behavior and alcohol consumption in China, with Chinese men smoking and drinking at vastly higher rates than women (J. P. Smith, Strauss, & Zhao, 2014). It is possible that men who saw these poor health behaviors as children were more likely to develop poor health behaviors that subsequently affected their functional health later in life. Future research is needed to examine the potential pathways through which adverse parental behavior may influence health in China.

Limitations

This study has several limitations. First, response bias due to using retrospectively reported childhood conditions may have resulted in underestimation of the impacts of childhood conditions on health. Although some studies (Haas, 2007; J. P. Smith, 2009) have found evidence that retrospective childhood health reports are reasonably reliable, other studies (i.e., Batty, Lawlor, Macintyre, Clark, & Leon, 2005) have concluded that retrospective reporting may underestimate the real effects. Second, trajectories of functional limitations were only observed for 4 years (2011-2015). It is quite possible that a longer time period is needed to observe change. Third, mortality selection may have biased the estimates downward. The fact that the adults in this sample survived into middle age suggests that their health was particularly robust. Therefore, childhood disadvantages may have had less of an impact on their health compared with those who succumbed due to poor health earlier in the life course. Fourth, there was notable attrition (approximately 30%) due to lost to follow-up between waves that may have biased the estimates downward. Fifth, unfortunately, formally testing mediation was among the childhood disadvantage domains was beyond the scope of this article due to the coarseness of when the childhood events occurred. In CHARLS, questions are only asked about known disadvantage experienced before age 16 with the exception of the questions about childhood hunger. Finally, these results may not be generalizable to other less developed countries. China has a unique history of war, famine, and rapid development. Older adults in other less developed countries likely experienced different types and severity of childhood disadvantage that may have shaped their health in different ways.

Conclusion

This study contributed to the growing body of literature investigating how different types of childhood disadvantage affect physical functioning among older adults in less developed countries. Findings highlighted how childhood health, hunger, SES, and parental behaviors endured to influence functional mobility among midlife Chinese adults, with notable differences by domain and sex. These results have important implications for China’s increasingly dire eldercare problem. Although sons and their wives are the typical caregivers of elderly parents in China, the sheer size of the aging population and erosion of traditional norms of filial piety that dictate familial caregiving has raised alarms regarding who will provide care (Shi, 2018). As findings from the present study demonstrate, care may be especially needed for older women who experienced multiple SES disadvantages in childhood, as well as older men who grew up in households with parental smoking, alcoholism, and/or physical abuse. Future research should continue to explore the relationship between different types of early-life conditions and later-life health, especially in less developed countries where aging cohorts are contributing to increasing disability burdens that may be due, in part, to disadvantage experienced earlier in the life course.

Footnotes

Declaration of Conflicting Interests

The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author disclosed reciept of the following financial support for the research, authorship, and/or publication of this article: This material is based upon work supported by the National Science Foundation Graduate Research Fellowship Program under Grant No. DGE-1845298. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author and do not necessarily reflect the views of the National Science Foundation.

ORCID iD

Morgan E. Peele

References

Barker

D. J.

(1997). Maternal nutrition, fetal nutrition, and disease in later life. Nutrition, 13, 807-813.

Batty

G. D.

Lawlor

D. A.

Macintyre

Clark

Leon

D. A.

(2005). Accuracy of adults’ recall of childhood social class: Findings from the Aberdeen children of the 1950s study. Journal of Epidemiology & Community Health, 59, 898-903.

Ben-Shlomo

Kuh

(2002). A life course approach to chronic disease epidemiology: Conceptual models, empirical challenges and interdisciplinary perspectives. International Journal of Epidemiology, 31, 285-293.

Blackwell

D. L.

Hayward

M. D.

Crimmins

E. M.

(2001). Does childhood health affect chronic morbidity in later life? Social Science & Medicine, 52, 1269-1284.

Croll

(1983). Chinese women since Mao. London, England: Zed Books.

Currie

Vogl

(2013). Early-life health and adult circumstance in developing countries. Annual Review Economics, 5(1), 1-36.

Elo

I. T.

Preston

S. H.

(1992). Effects of early-life conditions on adult mortality: A review. Population Index, 58, 186-212.

Ferraro

K. F.

Morton

P. M.

(2016). What do we mean by accumulation? Advancing conceptual precision for a core idea in gerontology. The Journals of Gerontology, Series B: Psychological Sciences & Social Sciences, 73, 269-278.

Ferraro

K. F.

Schafer

M. H.

Wilkinson

L. R.

(2016). Childhood disadvantage and health problems in middle and later life: Early imprints on physical health? American Sociological Review, 81, 107-133.

10.

Ferraro

K. F.

Shippee

T. P.

(2009). Aging and cumulative inequality: How does inequality get under the skin? The Gerontologist, 49, 333-343.

11.

Filmer

Pritchett

L. H.

(2001). Estimating wealth effects without expenditure data—Or tears: An application to educational enrollments in states of India. Demography, 38, 115-132.

12.

Friedman

E. M.

Montez

J. K.

Sheehan

C. M.

Guenewald

T. L.

Seeman

T. E.

(2015). Childhood adversities and adult cardiometabolic health: Does the quantity, timing, and type of adversity matter? Journal of Aging and Health, 27, 1311-1338.

13.

Greenfield

E. A.

Marks

N. F.

(2009). Profiles of physical and psychological violence in childhood as a risk factor for poorer adult health: evidence from the 1995-2005 national survey of midlife in the United States. Journal of Aging and Health, 21, 943-966.

14.

Guilmoto

C. Z.

Jones

G. W.

(Eds.). (2016). Contemporary demographic transformations in China, India and Indonesia. Dordrecht, The Netherlands: Springer.

15.

Guralnik

J. M.

Ferrucci

(2003). Assessing the building blocks of function: Utilizing measures of functional limitation. American Journal of Preventive Medicine, 25, 112-121.

16.

Haas

S. A.

(2007). The long-term effects of poor childhood health: An assessment and application of retrospective reports. Demography, 44, 113-135.

17.

Haas

S. A.

(2008). Trajectories of functional health: The “long arm” of childhood health and socioeconomic factors. Social Science & Medicine, 66, 849-861.

18.

Haas

S. A.

Zhou

(2017). The life course, cohort dynamics, and international differences in aging trajectories. Demography, 54, 2043-2071.

19.

Hallqvist

Lynch

Bartley

Lang

Blane

(2004). Can we disentangle life course processes of accumulation, critical period and social mobility? An analysis of disadvantaged socio-economic positions and myocardial infarction in the Stockholm Heart Epidemiology Program. Social Science & Medicine, 58, 1555-1562.

20.

Han

W. J.

Shibusawa

(2015). Trajectory of physical health, cognitive status, and psychological well-being among Chinese elderly. Archives of Gerontology and Geriatrics, 60, 168-177.

21.

Hardy

M. A.

Acciai

Reyes

A. M.

(2014). How health conditions translate into self-ratings: A comparative study of older adults across Europe. Journal of Health and Social Behavior, 55, 320-341.

22.

Henretta

J. C.

McCrory

(2016). Childhood circumstances and mid-life functional mobility. Journal of Aging and Health, 28, 440-459.

23.

Huang

Soldo

B. J.

Elo

I. T.

(2011). Do early-life conditions predict functional health status in adulthood? The case of Mexico. Social Science & Medicine, 72, 100-107.

24.

Kemp

B. R.

Ferraro

K. F.

Morton

P. M.

Mustillo

S. A.

(2018). Early origins of adult cancer risk among men and women: Influence of childhood misfortune? Journal of Aging and Health, 30, 140-163.

25.

Langley-Evans

S. C.

(2015). Nutrition in early life and the programming of adult disease: A review. Journal of Human Nutrition and Dietetics, 28, 1-14.

26.

Lee

Tsenkova

Carr

(2014). Childhood trauma and metabolic syndrome in men and women. Social Science & Medicine, 105, 122-130.

27.

Lei

Smith

J. P.

Sun

Zhao

(2014). Gender differences in cognition in China and reasons for change over time: Evidence from CHARLS. The Journal of the Economics of Ageing, 4, 46-55.

28.

Margolis

(2010). Childhood morbidity and health in early adulthood: Life course linkages in a high morbidity context. Advances in Life Course Research, 15, 132-146.

29.

McEniry

McDermott

(2015). Early-life conditions, rapid demographic changes, and older adult health in the developing world. Biodemography and Social Biology, 61, 147-166.

30.

Montez

J. K.

Hayward

M. D.

(2014). Cumulative childhood adversity, educational attainment, and active life expectancy among US adults. Demography, 51, 413-435.

31.

Nagi

S. Z.

(1976). An epidemiology of disability among adults in the United States. The Milbank Memorial Fund Quarterly. Health and Society, 54, 439-467.

32.

Non

A. L.

Rewak

Kawachi

Gilman

S. E.

Loucks

E. B.

Appleton

A. A.

Kubzansky

L. D.

(2014). Childhood social disadvantage, cardiometabolic risk, and chronic disease in adulthood. American Journal of Epidemiology, 180, 263-271.

33.

Osman

Walsemann

K. M.

(2017). Ethnic disparities in the onset and progression of functional limitations among Israeli middle-aged and older adults: Socioeconomic and health-related pathways. The Journals of Gerontology: Series B, 72, 140-150.

34.

Pudrovska

Anikputa

(2014). Early-life socioeconomic status and mortality in later life: An integration of four life-course mechanisms. The Journals of Gerontology Series B: Psychological Sciences and Social Sciences, 69, 451-460.

35.

Shen

Zeng

(2014). Direct and indirect effects of childhood conditions on survival and health among male and female elderly in China. Social Science & Medicine, 119, 207-214.

36.

Shi

(2018). Preparing for an “insured” old age: Insurance purchase and self-support in old age in rural China. Journal of Cross-Cultural Gerontology, 33, 183-195.

37.

Singer

J. D.

Willet

J. B.

(2003). Applied longitudinal data analysis. New York, NY: Oxford University Press.

38.

Smith

J. P.

(2009). Reconstructing childhood health histories. Demography, 46, 387-403.

39.

Smith

J. P.

Strauss

Zhao

(2014). Healthy aging in China. The Journal of the Economics of Ageing, 4, 37-43.

40.

Smith

N. R.

Ferraro

K. F.

Kemp

B. R.

Morton

P. M.

Mustillo

S. A.

Angel

J. L.

(2016). Childhood misfortune and handgrip strength among black, white, and Hispanic Americans. The Journals of Gerontology Series B: Psychological Sciences and Social Sciences, 74, 526-535.

41.

Stenholm

Guralnik

J. M.

Bandinelli

Ferrucci

(2013). The prognostic value of repeated measures of lower extremity performance: Should we measure more than once? The Journals of Gerontology Series A: Biomedical Sciences and Medical Sciences, 69, 894-899.

42.

Umberson

Williams

Thomas

P. A.

Liu

Thomeer

M. B.

(2014). Race, gender, and chains of disadvantage: Childhood adversity, social relationships, and health. Journal of Health and Social Behavior, 55, 20-38.

43.

United Nations. (2017). Revision of world population prospects. https://www.un.org/development/desa/publications/world-population-prospects-the-2017-revision.html

44.

Verbrugge

L. M.

Jette

A. M.

(1994). The disablement process. Social Science & Medicine, 38(1), 1-14.

45.

Wen

(2011). The effects of childhood, adult, and community socioeconomic conditions on health and mortality among older adults in China. Demography, 48, 153-181.

46.

Williams

M. M.

Kemp

B. R.

Ferraro

K. F.

Mustillo

S. A.

(2018). Avoiding the major causes of death: Does childhood misfortune reduce the likelihood of being disease free in later life? The Journals of Gerontology. Series B: Psychological Sciences and Social Sciences, 74, 170-180.

47.

Yang

B. Z.

Zhang

Weder

Douglas-Palumberi

Perepletchikova

. . . Kaufman

(2013). Child abuse and epigenetic mechanisms of disease risk. American Journal of Preventive Medicine, 44, 101-107.

48.

Zhang

Liu

(2017). The long arm of childhood in China: Early-life conditions and cognitive function among middle-aged and older adults. Journal of Aging and Health, 30, 1319-1344. doi:10.1177/0898264317715975.

49.

Zhao

Smith

J. P.

Strauss

Yang

(2014). Cohort profile: The China Health and Retirement Longitudinal Study (CHARLS). International Journal of Epidemiology, 43, 61-68.