The data used in this study was obtained from the 2014 and 2018 China Health and Retirement Longitudinal Study (CHARLS).

CHARLS is a large-scale national follow-up survey designed and implemented by the Social Science Research Center of Peking University for middle-aged and older adults (45 years old and above). CHARLS conducted its national baseline survey in 2011, covering 150 county-level units and 450 village-level units across 28 provinces (autonomous regions and municipalities), with approximately 17,000 individuals from 10,000 households. Follow-up surveys were conducted in 2013, 2015, 2018, and 2020. By 2020, the sample covered 19,000 respondents [14]. Additionally, a life course survey was conducted retrospectively in 2014. CHARLS adopted a multi-stage sampling method with probability proportional to scale. The respondents were interviewed face-to-face in their homes via computer-assisted personal interviewing technology [15]. CHARLS data can be accessed through its official website (http://charls.pku.edu.cn/). The CHARLS study received approval from the Biomedical Ethics Review Committee of Peking University (IRB00001052-11015). Each participant provided written informed consent, and all procedures were conducted according to relevant guidelines and regulations [16]. The 2018 survey covered domains such as household conditions, health status and functioning, cognition and depression, healthcare and insurance, and socioeconomic status. The 2014 life course survey collected data on major areas of individual life histories, such as growth, education, marriage, childbirth, and employment, providing information on ACEs for this study [17].

This study used the 2018 dataset, matching with the 2014 data using personal ID, and included older adults aged 60 years and above, totaling 8,657 samples. The detailed procedures for sample selection are as follows (Figure 1).

This study aims to investigate the association between ACEs and the incidence of depressive disorders in old age (60 years and above) and further analyze the moderating effect of ACEs on the relationship between later-life disability and depression. Among the variables used in this study, apart from the current household assets as a numerical variable, all other variables are categorical.

In this study, the dependent variable is “whether one has depression,” measured using the Center for Epidemiological Studies Depression Scale Revised (CESD-R-10). This scale is a self-report measure of depression, with the total score calculated by summing 10 items. Samples with more than two missing items were excluded. Any score equal to or above 10 is considered indicative of depression [18]. The CESD is primarily used in epidemiological research to identify individuals who may have depressive symptoms, rather than as a clinical diagnostic tool.

The core explanatory variable in this study is “whether one has had ACEs,” which includes three dimensions: early family environment, early peer relationships, and early community environment. The variables measuring the early family environment include: 1) “whether one experienced stressful family events or environments in early life,” including the death or divorce of biological parents before age 18, long-term parental illness, severe mental illness or physical disability in parents, and adverse behaviors of parents such as alcoholism, smoking, gambling, drug abuse, unstable employment, lying, frequent fights, involvement in criminal activities, arrest or imprisonment and 2) “relationship with parents,” represented by the combined scores of the relationship with the father and the relationship with the mother, divided into tertiles. The variable measuring the early peer relationships is the frequency of bullying by neighboring children, referred to as peer bullying frequency. The variable measuring early community environment is the perceived safety of the community.

Based on the model of social determinants of health and existing research findings, control variables in this study include socio-demographic characteristics [19,20], socioeconomic status [20], medical insurance [21], lifestyle, and health status [20,22,23]. Socio-demographic variables include gender, age, and marital status. Age is divided into three stages: 60-69 years as younger elderly, 70-79 years as middle-aged elderly, and 80 years and above as older elderly. Marital status is categorized into two groups: those who are separated or have no spouse, and those who are married and not separated. Socioeconomic status is measured by current household assets, including cash, electronic money, deposits in financial institutions, bonds, stocks, funds, assets in others’ names but owned by the respondents, housing provident funds, unreturned capital, unpaid wages, and other debts owed by others. The total value of household assets was calculated by summing these values, with specific amounts used if provided, or average values for ranges. Anomalous values were processed. Medical insurance is measured by whether the respondent holds medical insurance, including urban employee insurance, urban and rural resident insurance, public medical care, medical assistance, commercial insurance, major illness insurance for urban unemployed residents, long-term care insurance, and supplementary insurance, with holding any of these considered as having insurance. Lifestyle variables include current social activities [24], at least 10 minutes of exercise per week [25], and current smoking or drinking status [20]. The Physical Self-Maintenance Scale (PSMS) is commonly used to assess disability status [26]. A disability is any condition of the body or mind (impairment) that makes it more difficult for the person with the condition to do certain activities (activity limitation) and interact with the world around them (participation restrictions) [27]. According to the Americans with Disabilities Act (ADA), a person is considered to have a disability if they have a physical or mental impairment that substantially limits one or more major life activities [28]. PSMS includes six items: dressing, bathing, eating, getting in and out of bed, using the toilet, and controlling bladder and bowel functions. Each item is rated on a four-level scale: no difficulty, some difficulty but can still complete, difficulty requiring assistance, and unable to complete. If any item is rated as requiring assistance or unable to complete, the individual is classified as disabled [29]. The basic characteristics of the main variables for the depression and non-depression groups are shown in Supplementary Table 1.

The association analysis was conducted using logistic regression models. In the section on moderating effect analysis, this study first illustrates the results graphically to preliminarily explore the possible differences between groups, employing binary logistic regression to report the marginal effects of each interaction group and holding all variables except the main variables and interaction terms at their mean values. Secondly, a multiplicative interaction effect method based on logistic regression models was employed to rigorously calculate and analyze the moderating effect of ACEs on the relationship between later-life disability and depression. The calculation for the multiplicative interaction effect is as follows:

Taking the interaction between disability status and the relationship with parents as an example, the multiplicative interaction effect is interpreted as the ratio of the odds ratio (OR_ab) of depression among the disabled elderly with poor parental relationships (using the non-disabled counterparts as the reference group) to the odds ratio (OR_cd) of depression among the disabled elderly with good parental relationships (also using the non-disabled group as the reference): ORabORcd.

This ratio reflects whether the impact of disability on depression differs depending on the quality of early-life parent-child relationship, i.e., whether a significant moderation effect of early-life parent-child relationship exists on the relationship between disability status and depression in later life. An illustration of the multiplicative interaction effect method is shown in Figure 2 (data are hypothetical). Stata 18.0 (Stata Corp.) was used to conduct the above statistical analyses, with a significance level set at 0.05.

Among older adults aged 60 years and above, the proportion of those with depressive disorders (41.69%) is lower than that of those without depressive disorders (58.31%). Older adults who had experienced stressful family events (42.95%), reported poor parental relationships (47.51%), were sometimes or often subjected to peer bullying (49.50%), or had lived in less safe community environments (52.62%) during childhood demonstrated a higher prevalence of depression in later life. Conversely, those who had not experienced stressful family events (38.07%), reported good or moderate parental relationships (38.89% and 42.95%), had never or rarely been bullied by peers (40.50%), or had lived in safe communities (40.36%) during childhood exhibited a lower prevalence of depression in later life (Table 1).

Further regression analysis reveals that, after adjusting for individuals’ sociodemographic characteristics, socioeconomic status, lifestyle behaviors, and health conditions (including gender, age, marital status, household assets, medical insurance coverage, social activity participation, physical exercise, smoking and drinking behaviors, and disability status), compared to older adults who had good relationships with their parents during childhood, those with moderate or poor relationships had 14.0% and 43.8% higher odds of developing depression, respectively. Compared to older adults who had not experienced stressful family events during childhood, those who had experienced such events had 21.4% higher odds of developing depression. Compared to older adults who lived in relatively safe communities during childhood, those who lived in less safe communities had 28.4% higher odds of developing depression. Compared to older adults who were never or rarely bullied by peers during childhood, those who were sometimes or often bullied had 41.5% higher odds of developing depression. These regression results suggest a positive association between ACEs and depression in old age (Table 2).

Existing research suggests that disability is a risk factor for depression among older adults [6,7], and the likelihood of depression increases with the severity of disability [30]. Previous data analysis in this study indicated a positive correlation between ACEs and depression in old age. Furthermore, after disaggregating older adults by disability status, this study finds that, among those with disability, older adults who had experienced stressful family events or lived in less safe communities in childhood show a greater difference in the proportion of individuals with depression compared to those without depression (40.92% and 67.74%) than those who had not experienced these ACEs (31.50% and 36.38%). However, among those with disability, older adults who had a poor relationship with their parents or who had been sometimes or often bullied by peers in childhood show a smaller difference in the proportion of individuals with depression compared to those without depression (39.02% and 26.82%) than those who had not experienced these ACEs (40.06% and 41.14%) (Table 3). This suggests that ACEs, particularly peer bullying, may exert a potential moderating effect on the association between later-life disability and the likelihood of depression.

To further examine the significance of the moderating effect of ACEs, this study first presents the results graphically to explore the potential differences between groups. Figure 3 illustrates the marginal effects of disability on depression among older adults with varying ACEs. Across all ACEs groups, the marginal effect of disability on depression (solid line) is greater compared to non-disability (dashed line). However, the marginal effects of disability on depression appear relatively smaller among older adults who experienced stressful family events, were sometimes or often bullied by peers, or had moderate to poor relationships with their parents during childhood, compared to those without such ACEs. In addition, a multiplicative interaction effect method based on logistic regression models was employed to rigorously assess the moderating role of ACEs in the relationship between later-life disability and depression. The results show that the association between disability and depression is significantly weaker among older adults who were sometimes or often bullied by peers during childhood than among those who were never or rarely bullied (OR=0.410; 95% confidence interval, 0.204-0.826), suggesting that ACEs significantly moderate the relationship between disability and depression in old age (Table 4).