Life-course pathways from childhood socioeconomic status to type 2 diabetes in mid-late Chinese adulthood

The relationship between childhood socioeconomic status (SES) and type 2 diabetes (T2D) remains inconclusive, and the pathways and mechanisms driving this relationship have yet to be clarified. This study aimed to examine the pathways linking childhood SES to T2D prevalence in mid-late adulthood in a low- and middle-income country. The incidence of T2D diagnosed in mid-late Chinese adulthood was assessed using self-reports from the Health and Retirement Longitudinal Study (CHARLS). Childhood SES was assessed by the education, occupation, survivorship of the parents and the financial situation of the family, whereas adulthood SES was assessed by education and wage. This study performed structural equation modeling to clarify the direct and indirect pathways from childhood SES to T2D via childhood health, childhood food shortage, adulthood SES and physical activity. A total of 15,132 participants were included, and the prevalence of T2D was 5.24%. This study found that childhood SES was directly associated with T2D in mid-late adulthood, the probability of developing T2D increased by 9.20% of the standard deviation for each decrease in standard deviation in childhood SES. Childhood SES was indirectly associated with T2D via adulthood SES, physical activity, childhood health and food shortage. Adulthood SES and physical activity mainly mediated the indirect pathway from childhood SES and T2D. This study showed direct and indirect pathways from disadvantaged childhood SES to increased risk of T2D in mid-late Chinese adulthood. Childhood SES, adulthood SES, physical activity, childhood health and food shortage were identified as life-course interventional targets that should be considered in the development of effective strategies to reduce the burden of T2D and SES-related health inequities in childhood.

life-course theoretical models are not mutually exclusive and can be used to evaluate the magnitude of pathways from childhood SES to T2D 8 .
Childhood SES has particular implications not only for understanding the current situation but also for inferring future health outcomes of patients in LMICs, who are more likely to be affected by their childhood experiences, compared with their counterparts in HICs 30 . China has undergone considerable SES changes for decades 40 , which may influence mid-late Chinese adulthood that is rendered vulnerable by disadvantaged childhood SES 25 . Chinese adults are characterized by disadvantaged living conditions, widespread social unrest, prolonged exposure to hunger, and infectious diseases in childhood that constitute health challenges 40 . Exposure to disadvantaged childhood SES may have shaped the life-course health trajectories of Chinese adults in different ways 25,41 . The following were observed in Chinese adults: disadvantaged childhood SES increased the risk of cognitive disorders 40 and depression 26 ; disadvantaged childhood health significantly predicted cognitive and physical impairment 42 ; and childhood food shortage was associated with late-life glucose abnormalities 23 .
Taken together, a comprehensive understanding of pathways linking childhood SES to T2D in mid-late Chinese adulthood is crucial to identify potential interventional targets, thereby reducing the risk for late-life T2D and health inequities from childhood 43 . The conceptual frameworks ( Fig. 1) based on life-course theoretical models combined with a literature review were as follows: (1) a direct pathway may exist from childhood SES to T2D in mid-late Chinese adulthood; (2) Childhood food shortage and health may mediate the pathways from childhood SES to T2D; (3) Adulthood SES and lifestyle behaviors may mediate the indirect pathways from childhood SES to T2D.

Methods
Study design and participants. The data used in this study were from the 2014 and 2015 waves of the China Health and Retirement Longitudinal Study (CHARLS), a nationally representative survey among community-dwelling Chinese adults aged 45 and older 44 . CHARLS adopted a three-stage stratified probability-proportional-to-size sampling framework involving 10,803 households from 450 villages/residential communities in 28 provinces across mainland China 44 . CHARLS was approved by the Ethical Review Committee of Peking University and informed consent was obtained from each participant at the time of participation 44 . All methods were carried out in accordance with relevant guidelines and regulations. The 2014 wave was a life history survey of 20,543 participants and collected retrospective data on childhood SES (the education, occupation, survivorship of the parents and financial situation of the family), childhood health and food shortage 25 ; the 2015 wave was a regular follow-up survey of 21,095 participants and collected data on adulthood SES (education and wage), lifestyle behaviors (drinking, smoking and physical activity), and health outcomes 28 . Participants in the 2014 and 2015 waves were combined based on IDs to trace their information on childhood SES, childhood food shortage, childhood health, adulthood SES, lifestyle behaviors, and T2D incidence. Participants aged 45 or younger or those with missing information on T2D and childhood SES were excluded. This study ultimately included 15,132 participants in the current analyses.
Measurements. The incidence of T2D was assessed based on responses to the question: "Have you been diagnosed with T2D by a doctor?", which had two options ("Yes" or "No").
Seven variables were used to assess childhood SES: the financial situation of the family, education, occupation and survivorship of the father and the mother 12,25 . The financial situation of the family was assessed based on the question "When you were a child under 17 years old, how was the family's financial situation?" and then  www.nature.com/scientificreports/ categorized into "a lot better off than them", "somewhat better off than them","same as them", "somewhat worse off than them" and "a lot worse off than them". This study classified the financial situation of the family as "not worse" ("a lot better off than them","somewhat better off than them" and "same as them") and "worse" ("somewhat worse off than them" and "a lot worse off than them"). The survivorship of the father and the mother were dichotomized into "Yes" or "No" in response to the question "Was your father (mother) alive in your childhood?". China being a largely agricultural society in the early twentieth century, the general education level was very low 19 , more than four-fifths (83.02%) of the mothers of the participants did not receive education, about 90% of the parents of the participants were farmers. Accordingly, the education of the mother and father were accordingly dichotomized into "received some education" or "no education, " and the occupation of the father and the mother were dichotomized into "non-agricultural" or "agricultural" 11,25 . Childhood food shortage was assessed based on the question: "When you were a child before age 17 was there ever a time when your family did not have enough food to eat?" and dichotomized into "Yes" or "No" 26 .
Childhood health was self-reported based on the question: "How would you evaluate your health during childhood, up to and including age 15?" and categorized into "much healthier", "somewhat healthier", "fair", "somewhat less healthy" and "much less healthy". This study classified childhood health as "healthy" ("much healthier" and "somewhat healthier"), "fair" or "unhealthy" ("somewhat less healthy" and "much less healthy") 26 .
Adulthood SES was assessed based on education and wage 14 . The education was grouped into four categories "illiterate", "primary school", "middle school" or "high school and above". The wage was determined on the question "Have you received wage last year?" and dichotomized into "Yes" or "No".
Lifestyle behaviors included smoking status, drinking status and physical activity. Currently drinking status was classified into "Yes" or "No" based on the question "Did you drink any alcoholic beverages in the past year?"; currently smoking status was classified into "Yes" or "No" based on the question "Did you have the habit of smoking cigarettes/smoking a pipe/chewing tobacco now?". Physical activity was determined by the responses to the question "During a usual week, did you do any physical activities for at least 10 min continuously?" and dichotomized into "Yes" or "No".
Statistical analyses. Descriptive analyses and Spearman correlation analysis of variables were conducted using STATA 14.0, and the calculations and estimates of the SEM models were performed using MPlus7.4. No significant difference in the distribution of covariates was found between the final analysis sample and those with missing data in T2D and childhood SES. Frequencies (percentages) and mean values (standard deviation, SD) for the categorical and continuous variables were calculated. For descriptive statistics, Chi-squared tests and Fisher's Exact tests were used to compare the categorical variables including gender, childhood SES, adulthood SES and lifestyle behaviors between participants with and without T2D. Spearman correlation analyses were performed to explore the correlation between childhood SES and T2D in mid-late adulthood. Based on the conceptual framework, this study developed the analytical model through the SEM approach to explore direct and indirect pathways between childhood SES and T2D. Confirmatory factor analysis was used to construct childhood SES and adulthood SES, and acceptable factor loadings were determined. Multiple modification indices were used to adjust and modify the SEM model to realize the best-fit model 45 . These indices included the comparative fit index (CFI), goodness-of-fit index (GFI), adjusted goodness-of-fit index (AGFI), standardized root mean square residual (SRMR), and the root-mean-square error of approximation (RMSEA). CFI ≥ 0.80, RMSEA and SRMR ≤ 0.06, GFI and AGFI > 0.90, indicating an acceptable model fitness 46 . Although χ 2 values should be reported as one of the fit indices, they were highly sensitive to large sample sizes and thus were excluded 47 . The best-fitting SEM model included childhood health, childhood food shortage, adulthood SES including education and wage, and physical activity as mediating variables in indirect pathways from childhood SES to T2D. Bias-corrected bootstrapping (2000 bootstrap samples) was then used to test the statistical significance of the direct and indirect effects of each pathway in the model. The total effects were estimated using the sum of the direct and indirect effects, mathematically expressed as 48 : c = c′ + ab, where c = total effect, c′ = direct effect, ab = indirect effect. P < 0.05 indicated statistical significance.
Ethical approval and consent to participate. CHARLS was approved by the Ethical Review Committee of Peking University and informed consent was obtained from each participant at the time of participation. All methods were carried out in accordance with relevant guidelines and regulations. Table 1 presents the descriptive statistics and comparison by T2D for the analyzed variables. T2D incidence was reported by 15,132 participants (7011 males and 8121 females), with the average age of 60.56 ± 10.15 years. A total of 793 participants were diagnosed with T2D by the doctor (5.24%). For the childhood SES variables, 2533 (16.75%) and 4224 (27.91%) participants reported the survivorship of their fathers and mothers during childhood, 7974 (52.52%) and 12,562 (83.02%) participants reported their fathers and mothers did not receive education, 13,890 (91.79%) and 12,622 (83.41%) participants reported the occupation of their fathers and mothers were farmers, and 6120 participants reported the financial situation of the family as worse. Of 10,207 participants (67.45%) experienced childhood food shortage, 13,073 (86.39%) reported childhood health status as poor/fair. Individuals whose father/mother did not receive education were more likely to develop T2D than those whose father/mother received some education (p < 0.05). Individuals who reported the survivorship of father/mother during childhood were less likely to develop T2D than those who reported death of father/mother (p < 0.05). Individuals who experienced childhood food shortage were more likely to develop T2D than those who did not experience food shortage (p < 0.05). Individuals with lower education and did not receive wage were more likely to develop T2D than those with higher education and received wage (p < 0.01). Individuals who did not do www.nature.com/scientificreports/ physical activity were more likely to develop T2D than those did physical activity, and female were more likely to develop T2D than males (p < 0.05). Table 2 presents the correlations between variables of childhood SES, childhood food shortage, childhood health, adulthood SES, physical activity and T2D. T2D was positively correlated with the survivorship of the father (r = 0.058, p < 0.05) and mother (r = 0.057, p < 0.05), receiving no education of the father (r = 0.022, p < 0.05) and mother (r = 0.019, p < 0.05). Childhood food shortage was positively correlated with higher risk for T2D (r = 0.018, p < 0.05). Figure 2 presents the standardized path estimates of the pathways from childhood SES to late-life T2D, Tables 3 and 4 present the standardized estimates of the direct, indirect, and total effects of childhood SES on late-life T2D prevalence as well as the specific effects through multiple pathways of childhood health, childhood food shortage, adulthood SES and physical activity. Childhood SES exerted significant direct (β = 0.083, p < 0.05), indirect (β = 0.009, p < 0.05) and total (β = 0.092, p < 0.05) effects on T2D; 90.21% of the total

SEM results.
The survivorship of the father (1)

Discussion
To the best of our knowledge, this study is the first to examine the potential pathways from childhood SES to T2D by using a nationally representative sample and the SEM approach in mid-late Chinese adulthood, verifying the existence of life-course theoretical models in an LMIC. This study showed significant direct and indirect pathways from childhood SES to T2D. Disadvantaged childhood SES was indirectly associated with T2D via adulthood SES, physical activity, childhood health and childhood food shortage, which were identified as valuable life-course intervention targets to reduce the risk for T2D. Both direct and indirect pathways from childhood SES to T2D may provide important arguments for policymakers to improve childhood SES, consequently reducing health inequities in early life. Consistent with the conceptual framework, this study confirmed that disadvantaged childhood SES was directly associated with a higher risk for T2D in mid-late Chinese adulthood, the finding supported the evidence from HICs 21 . A Finnish population-based study showed that individuals with disadvantaged childhood SES were more likely to develop late-life T2D 49 ; a similar association was reported in Americans 50 . Consistent evidence from Latin America, the Caribbean, and the United States revealed that disadvantaged childhood SES increased the risk for late-life T2D 18,51 . The direct effect of childhood SES on T2D indicated that childhood was a critical period across the life course and supported the critical period model. Disadvantaged childhood SES activated and modified inflammation processes and hypothalamic-pituitary-adrenal regulations 52 , which may facilitate the development of T2D 50 . Disadvantaged childhood SES limited the opportunities for obtaining sufficient physical, psychosocial, and medical resources and increased the risk for low-quality parental interactions and adverse stimulations 53 . These adverse experiences constitute adverse interpersonal experiences and social exclusion 26 , which may be contributory to the risk for T2D 54 .
In accordance with the previous study 3 , the current study showed that childhood SES was indirectly associated with T2D in mid-late Chinese adulthood via adulthood SES, supporting the chains-of-risk and accumulative risk models. The finding highlighted the importance of T2D-related SES pathways and indicated that advantaged adulthood SES could compensate for the consequences of disadvantaged childhood SES in the risk for T2D 21 . Individuals with disadvantaged childhood SES, which is relatively stable across the life course, tend to acquire limited access to high-quality education, obtain few job opportunities to positions requiring higher education, and experience poor living conditions, their disadvantaged SES may be retained in adulthood 55 . The level of education and wage in adulthood played important mediating effects in the relationship between childhood SES and late-life health 15 . Disadvantaged adulthood SES was closely related to poor health literacy, which is associated with poor late-life health outcomes, and individuals with inadequate literacy were less likely to acquire preventive information on T2D 56 . Individuals with disadvantaged adulthood SES could have a higher risk for social isolation, less social support, and increased stress, which are associated with a higher risk for glucometabolic impairment, prompting the development of T2D 56 .
The current study clarified the indirect pathway from disadvantaged childhood SES to T2D through limited physical activity, supporting the chains-of-risk and accumulative risk models 3 , which agreed with previous studies involving middle-aged Americans 57 and older Englishmen 8 . Inadequate studies have been reported on the pathway from childhood SES to T2D mediated by physical activity in adulthood. Meanwhile, the present study provided novel findings regarding this association and indicated that the differences in lifestyle behaviors, particularly physical activity, had existed in childhood and persisted into adulthood 35 . The underlying reasons for the mediating role of physical activity in the pathway from childhood SES to T2D are complex 35 . Disadvantaged childhood SES is associated with poor health consciousness, negative health beliefs, and suboptimal lifestyle behaviors, which may affect internal control and access to health-related information on T2D 21 . Existing cultural norms and the provision of healthy products for individuals with disadvantaged SES are associated with the acquisition of exercise facilities 58 , which may influence the adoption of healthy behaviors, particularly physical activity, in childhood and the development of T2D 57 . Individuals with disadvantaged childhood SES have inferior knowledge and skills necessary to develop healthy lifestyle behaviors, they may learn and accept fewer intervention programs for T2D prevention 58 .
The current study showed the sequential mediating effects of childhood food shortage and childhood health in indirect pathways from childhood SES to T2D. Experiencing food shortage and having poor health during childhood were associated with poor cognitive performance, delayed school attendance, dropping out of school, decreased schooling completion, and subsequent lower wages in adulthood 59 . Childhood health may drain socioeconomic resources preventing investment in healthcare and limiting the capacity to acquire skills necessary to shape adulthood SES 60 , which is associated with an increased risk for T2D 56 . The mild effects of childhood health and food shortage mediated the indirect pathways between childhood SES and the risk for T2D. Improving these aspects, which may influence health, education, and economic benefits in adulthood, may not preclude the achievement of progress 59 . The present study also found that females were more likely to develop late-life T2D, which was consistent with an Indian study reporting a higher prevalence of T2D in females than in males 61 .
Interventions from the life course perspective. The pathways from childhood SES to T2D indicated that early-life intervention is crucial to human development and promotes avoidance and resilience against T2D development 10 . Exposure to early-life disadvantages is irreversible and weakly supported by an emphasis on investment in childhood 62 . Interventions to promote resilience in children with disadvantaged SES contribute to the development of optimal mental and physical health, equipping children better and earlier to endure, adapt to, and thrive after exposure to childhood disadvantages. Increasing focus on the multidimensional interventional approach, which encompasses education, healthcare, food availability, and physical activity, to prevent T2D are highly needed 43 . Current substantial investment in education and improvement in the school system could be leveraged to formulate optimal childhood health interventions 62  www.nature.com/scientificreports/ hood health and food availability as high priorities in T2D prevention, education, and SES agenda to alleviate the negative effects of disadvantaged childhood SES 59 . The school system in LMICs with limited healthcare coverage may provide a cost-efficient platform for health promotion and the delivery of healthcare resources and services for children with disadvantaged SES 63 . School-based interventions should combine teachings in health education both inside and outside classrooms, including health curricula incorporating families and communities; these supplements, which may facilitate the whole life-course health and reduce the risk for T2D 62,63 . Schoolbased programs, such as meals, vaccination, and annual deworming programs, should be conducted to promote school participation, improve childhood health and food availability, which is a cost-effective intervention for females to provide intergenerational benefits 62 . Consequently, far-reaching effects are exhibited on reducing the risk for T2D 13 . Evidence suggests that substantial returns on investment in education and employment 62 and multi-sectorial strategies should be targeted at providing educational opportunities and vocational training during adulthood 11 . Expanding access to higher education and offering more employment opportunities for females may reap substantial benefits, including the provision of comprehensive health knowledge and the development of optimal lifestyle behaviors 62 , as well as narrow the SES disparities in behaviors 35 . Community-, school-and family-based health-promoting interventions for lifestyle behaviors (such as physical activity), targeted at children with disadvantaged SES, may prevent the risk of T2D 36 .

Strengths and limitations.
This study has several strengths. To the best of our knowledge, this is the first study to examine the direct and indirect pathways from childhood SES to T2D in a LIMC, using the largest nationally representative sample in mid-late Chinese adulthood. Moreover, this study innovatively used the SEM approach to disentangle the direct and indirect pathways from childhood SES to T2D, and identified the mediating effects of disadvantaged childhood health, food shortage, disadvantaged adulthood SES and limited physical activity. SEM addresses methodological challenges in estimating direct and indirect effects via multiple pathways, and is valuable in examining multiple complex intercorrelations, identifying mediating variables and understanding pathways to propose targeted interventions in the critical life period 46 .
The findings in this study should be interpreted cautiously for several limitations. First, this cross-sectional study is limited to determining the definitive causal relationship between childhood SES and T2D. Although the retrospective information was reported as reasonably reliable 64 , measurements of childhood conditions were based on retrospective self-reports with ineluctable recall bias. Second, this study was limited by the available variables and responses coded in the existing CHARLS database, such as the categorization of occupation and education of parents, smoking status and drinking status. Potential variables, such as dietary habits, history of smoking, and drinking status that are associated with T2D might not be included in the database, limiting the conclusion of pathways from childhood SES to T2D. This study attempted to adjust biomarkers, including the body mass index, blood pressure and heart rate, drinking and smoking status in the SEM model; however, the results showed a poor modeling fit and insignificant associations. Finally, individuals suffering from extremely disadvantaged childhood SES and childhood health might not have reached 45 years old, underestimating the relationship and bias in the pathways between childhood SES and T2D 26 .

Conclusion
This study is thus far the first study to clarify the direct and indirect pathways from childhood SES to T2D in midlate Chinese adulthood. Childhood health and food shortage, adulthood SES, and physical activity mediated the indirect pathways. The findings in this study showed that disadvantaged childhood SES may propel individuals on unhealthy trajectories, such as T2D, and suggested that attention be particularly directed toward preventing T2D and thereby reduce health disparities on individuals with disadvantaged childhood SES. This study identified adulthood SES, physical activity, childhood health and food shortage, which could be targeted interventions to reduce the risk of T2D in mid-late adulthood caused by disadvantaged childhood SES. Future birth-cohort studies with a national representative sample in LMICs are recommended to explore the underlying pathways between childhood SES and late-life health outcomes, as well as to design and modify targeted interventions to reduce health inequities in childhood.

Data availability
The data that support the findings of this study are available from the China Health and Retirement Longitudinal Study (CHARLS), but restrictions apply to the availability of these data, which were used under license for the current study. Data are however available from the authors upon reasonable request and with permission of CHARLS research team. The authors had no special access privileges in accessing data from CHARLS.