Socioeconomic status and depressive symptoms in older people with the mediation role of social support: A population‐based longitudinal study

1 INTRODUCTION

Depression is a major mental health problem contributing to disability and the burden of disease globally (Mathers & Loncar, 2006; Whiteford et al., 2015). According to data issued by the World Health Organization data in 2017, more than 300 million people worldwide suffered from depression, with the prevalence rising by 18.4% over the past decade. In China, 54 million people suffered from depression, and the annual incidence had reached 4.2% (World Health Organization, 2017). Reports indicated that the incidence of depression increased by age and was particularly high among middle-aged and elderly people (Liu et al., 2018). A nationally representative study showed that 30% of men and 43% of women aged 45 and over in China suffered depressive symptoms in 2011 and 2012 (Lei et al., 2014). With the increasing aging of China (UNFPA, 2020), the number of people suffering from depression is expected to rise dramatically in the middle-aged and elderly. Individuals with depression have a 20-time higher risks for suicide compared to that of general population (Osby et al., 2001), as well as a 60% increased risk of cardiovascular diseases (Barth et al., 2004). Furthermore, depression is associated with the loss in productivity and unemployment, placing a heavy burden on families and the whole society. For middle-aged and older people, their body functions continue to deteriorate, and they may also face the risk of losing independence, financial troubles, and the coexistence of multiple chronic diseases (Boss & Seegmiller, 1981; Sivertsen et al., 2015).

The association between socioeconomic status (SES) and depressive symptoms has been reported in several studies (Freeman et al., 2016; Lotfaliany et al., 2019; Ng et al., 2014; Quashie et al., 2021; Wang et al., 2019). Lower SES may reduce the capacity to manage stress, and the long time exposure to the social stressors may result in higher depressive symptoms (Gallo & Matthews, 2003; Turner et al., 1995). However, the studies conducted among the older Chinese had been dubbed the “causation-selection” issue, with most of the studies cross-section designed or relative short time following up (Hu et al., 2019; Lei et al., 2014; Ruiz et al., 2019). On the other hand, the criticism of the previous work on SES and depressive symptoms includes the lack of control for confounding by sensory function (Harrison et al., 2019; Rong et al., 2020), housing conditions (Fang et al., 2019) and childhood deprivation (Ruiz et al., 2019) may partially explain the relationship.

Social support defined as “support accessible to an individual through social ties to other individuals, groups, and the larger community” (Fang et al., 2019), is negatively correlated with depressive symptoms (Faramarzi et al., 2015; Hu et al., 2018; Zhong et al., 2020). People with more social support tend to have more interpersonal resources when dealing with daily stress and are more likely to recover from stressful conditions (Mcdougall et al., 2016; Wang et al., 2016). Many studies confirmed that the distribution of social support was unequal among different socioeconomic classes and was more common in higher social classes (Matthews et al., 1999; Mickelson & Kubzansky, 2003). It has been found that social support mediates socioeconomic differences in health (Geckova et al., 2003; Matthews et al., 2008). However, studies with prospective design examining the role of social support in the relationship between SES and depression are quite scarce, particularly for the middle and low-income countries.

This study aimed to prospectively examine the association between SES and the risks of developing depressive symptoms during 7-year follow up from 2011 to 2018 in an older Chinese population. With few studies have examined the mechanisms and pathways through which SES impacts the prevalence of depressive symptoms, this study explored the mediation effect of social support on the relation between SES and depressive symptoms. We hypothesized that lower SES increased the risk of developing depressive symptoms and social support mediated the relationship between SES and depressive symptoms.

2 STUDY POPULATION AND METHODS 2.1 Data and study sample

The data for this study were derived from the China Health and Retirement Longitudinal Study (CHARLS), a nationally representative sample of Chinese adults aged 45 years or older and their spouses. The 2011 CHARLS baseline survey used multistage probability sampling to select 150 county-level units in which 450 communities were chosen with 28 provinces in China. The interviewers were trained by Peking University before the survey and the participants were interviewed face-to-face in each household. A more detailed description about the study design and sampling procedure can be found in the cohort profile of CHARLS (Zhao et al., 2014). The data used in this study were followed up from the baseline survey to the latest CHARLS wave (2018). The 2011 baseline survey data included a total of 10,257 households and 17,708 individuals over 45. We excluded participants with missing data in depression (n = 1671), having depressive symptoms at baseline (n = 4504), those lost in follow-up (n = 3250) and with missing critical data (n = 2606) from the analyses. In the end, the study included a total of 5677 participants over 45 years old without depressive symptoms at baseline and with completed 7-year follow-up. For the nonresponse group, the likelihood of nonresponses appeared to be uncorrelated with demographic, socioeconomic characteristics (Zhao et al., 2014). We further confirmed that there was no obvious correlation between social support and the likelihood of nonresponses. Figure 1 described the detailed information of the study flow chart.

image

Flowchart of the study sample

The CHARLS study was ethically approved by the Peking University institutional review board (Zhao et al., 2014). This study was a secondary analysis of the CHARLS data sets and was waived from the ethical approval. The longitudinal study obtained written informed consent from all participants and was conducted following the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline (von Elm et al., 2007).

2.2 Depressive symptoms

In the CHARLS survey, depressive symptoms were measured using a short form of the Center for Epidemiological Studies Depression Scale (CESD-10), which contains 10 questions: (1) troubled by trivial matters, (2) troubled by inattention, (3) feeling frustrated, (4) laborious to do anything, (5) full of hope, (6) feeling fear; (7) poor sleep; (8) feeling happy; (9) feeling lonely, and (10) unable to continue living. Depressive symptoms over the past week were measured from 0 (little or no time [<1 day]) to 3 (most or all time [5–7 days]). Before summing the project scores, question 5 and 8 scores need to be converted. The CESD-10, a useful mental health measure for the older people, has demonstrated satisfactory content and temporal validity and reliability among Chinese middle-aged and older adults (Boey, 1999; Yang et al., 2015). The total score of CESD-10 ranges from 0 to 30. A higher score indicates more depressive symptoms and previous studies show that a score of 12 or more have reasonable levels of sensitivity (0.76) and specificity (0.55) among older Chinese (Cheng & Chan, 2005; Li et al., 2015). We used 12 as the cut point for having elevated depressive symptoms (=1) and not depressed (=0; Chen & Mui, 2014; Li et al., 2019).

2.3 Socioeconomic status

We measured the SES by the participant's educational level and household annual income, as they have strong theoretical associations with depression (Lorant et al., 2003; Miech & Shanahan, 2000). Education level was classified as illiterate, primary school and below, and junior school and above. Quintiles for household annual income were used with household annual income ≤3000 RMB coded level 1, 3000−10,000 RMB coded level 2, 10,000−24,000 RMB coded level 3, 24,000−36,000 RMB coded level 4, >36,000 coded level 5. To examine the mediation effect of SES and depressive symptoms through social support, a composite score of SES was constructed by multiplying the education level and household annual income level (Katsarou et al., 2010).

2.4 Social support

A comprehensive index including family support, community support and public support were used to measure social support in our study (Hu et al., 2018; Zurlo et al., 2014). Family support referred to whether the participants live with a spouse (0 = no; 1 = yes), whether the participants weekly contact with their children in person (0 = no; 1 = yes), and whether the participants weekly contact with their children by phone or email (0 = no; 1 = yes). Community support included community infrastructure and resources, which was measured by the following questions: whether the village/community has outside exercising facilities (0 = no; 1 = yes); whether the village/community has dancing team or other exercise organizations (0 = no; 1 = yes); whether the village/community has organizations for helping the elderly and the handicapped (0 = no; 1 = yes); whether the village/community has activity center for the elderly (0 = no; 1 = yes); whether the village/community has elderly association (0 = no; 1 = yes). Public support was proxied by social security and welfare for the older people, referring to whether the participants have pensions (0 = no; 1 = yes), and whether the participants have medical insurance or welfare (0 = no; 1 = yes). By summing the 3 aspects of social support, the comprehensive index ranges from 0 to 10, with higher score reflecting higher level of social support.

2.5 Covariates

Information about other sociodemographic status, health behavior, health condition, household condition, and childhood deprivation of the participants were controlled in our study, including age, gender (male or female), smoking status (current, former of never), alcohol drinker (current, former, or never), number of non-communicable diseases (0, 1–2, or ≥3), self-assessment health (excellent, very good, good, moderate, or bad), body mass index (≤19, 20–24, 25–28, or ≥29), hearing function (good or poor). Activities of daily living (ADLs) were measured using a six-item summary in CHARLS that included bathing, dressing, eating, getting in/out of bed, using the toilet and controlling urination with 0 for no difficulty and 1 for having difficulty in each activity (Zhou et al., 2020). Non-communicable diseases included hypertension, dyslipidemia, diabetes, cancer, chronic lung diseases, liver diseases, heart diseases, stroke, kidney diseases, digestive diseases, psychiatric diseases, memory-related diseases, arthritis and asthma. The household condition was measured by whether the residence has a telephone connection. Childhood deprivation referred to whether the participants ever lacked enough food to eat in the CHARLS.

2.6 Statistical analysis

Firstly, descriptive statistical analysis and the Pearson χ2 test were used to analyze the difference between the respondents who developed depressive symptoms during the seven-year follow up and those who had no depressive symptoms by the end of the seven-year follow up. To model the effect of SES on the incidence of depressive symptoms, cox proportional hazards regression was used to fit the longitudinal data. We calculated the follow-up time as the number of months from the wave 1 interview date to follow-up interview date where the first diagnosis of depressive symptoms occurred, or censoring (the date of the participant's last interview) took place. We fitted an unadjusted model and gradually adjusted for social support, demographic characteristics, health behaviors, health conditions, childhood deprivation, hearing function and telephone connection. The proportional hazards (PH) assumption was checked using statistical tests based on the scaled Schoenfeld residual, and all the models meet the assumption with P > 0.05. Structural equation model (SEM) is designed to test more intricate mediation models in a single analysis and can be used to extend a mediation process to multiple independent variables or outcomes (MacKinnon, 2008). We examined the mediation effect of social support on the relationship between SES and depressive symptoms through the generalize structural equation model after controlling for the all the covariates (Angkananard et al., 2019; Gunzler et al., 2013). We further evaluated whether the mediation effect differed between male and female.

To address concerns about selection bias, we conducted sensitivity analyses. The analyses were conducted as follows: (1) participants with missing income variable were included in model 1 to model 4 with a number of 6627 recipients; (2) binary mediation analysis with bootstrap standard errors and confidence intervals was used to explore the mediation effect of SES and depressive symptoms through social support. Individual weights in the CHARLS data were applied in the regression analysis to take into account selection probability, non-response patterns and post-stratification factors (Liu et al., 2017). We considered a two-sided.

P < 0.05 statistically significant. The software Stata version 14.1 for Mac (Stata Corp) was used for statistical analysis.

3 RESULTS

Table 1 described the incidence of depressive symptoms after seven years among the CHARLS participants and their subgroups who were free of depression at baseline and the relationship between depressive symptoms and baseline characteristics in participants without depressive symptoms after 7 years follow-up. Among 5677 middle-aged and elderly patients who were followed up, 2398 (42.2%) developed depressive symptoms after 7 years. The incidence of depressive symptoms was statistically significantly different among subgroups. For example, the rate was higher in female (50.1%) than male (34.1%), participants with lower household annual income or lower level of education, those with more difficulties in daily activity, those with three or more noncommunicable diseases (22.5%) compared to those with no diseases (13.8%), participants who had childhood deprivation (44.4%) compared to those with no deprivation (37.3%), and those who had poor hearing function (46.0%) compared to respondents with good hearing function (38.6%) (Table 1).

TABLE 1. Seven-year incidence of depressive symptoms among study participants free of disease at baseline and the association between sociodemographic and incidence Variables N Depressive symptoms n (%) p-value (χ2 test) Total 5677 2398 (42.2) Household income (Yuan) <0.001 ≤3000 1006 516 (51.3) 3000–10,000 926 463 (50.0) 10,000–24,000 1159 498 (43.0) 24,000–36,000 771 282 (36.6) >36,000 1815 639 (35.2) Education <0.001 Illiterate 1301 757 (58.2) Primary school and below 2312 1022 (44.2) Junior high school and above 2064 619 (30.0) Age <0.001 <60 3622 1462 (40.4) 60–69 1575 695 (44.1) 70–79 439 213 (48.5) >80 41 28 (68.3) Gender <0.001 Female 2879 1438 (50.1) Male 2807 960 (34.1) Smoking <0.001 Current smoking 1804 669 (37.1) Former smoking 451 159 (35.3) Never smoking 3422 1570 (45.9) Drinking <0.001 Current drinking 1966 685 (34.8) Former drinking 403 189 (46.9) Never drinking 3308 1524 (42.2) Number of NCDs <0.001 0 1939 659 (34.0) 1–2 2718 1202 (44.2) ≥3 812 440 (54.2) Missing 208 97 (46.6) Self-assessment health <0.001 Excellent 43 12 (27.9) Very good 427 120 (28.1) Good 850 266 (31.3) Moderate 2079 865 (41.6) Bad 634 384 (60.6) Missing 1644 751 (45.7) ADL score <0.001 0 5117 2064 (40.3) 1 332 181 (54.5) 2–6 172 128 (74.4) Missing 56 25 (44.6) BMI <0.001 ≤19 531 270 (50.9) 20–24 2582 1097 (42.5) 25–28 1440 581 (40.4) ≥29 516 214 (41.5) Missing 608 236 (38.8) Childhood deprivation <0.001 No 1702 634 (37.3) Yes 3975 1764 (44.4) Hearing impairment <0.001 Good 2873 1109 (38.6) Poor 2804 1289 (46.0) Telephone connectivity <0.001 No 2780 1259 (45.3) Yes 2897 1139 (39.3) Abbreviations: ADL, activity of daily living; BMI, body mass index; NCD, non-communicable diseases.

Table 2 showed the results of cox proportional hazards regressions. Participants with higher SES had a significantly lower risk of developing depressive symptoms in unadjusted model or models gradually adjusting for covariates during the 7-year follow up. After adjusting for all potential confounders (in model 4), compared to the level 1 household annual income, participants with level 4 and level 5 household annual income had an estimate of 16% (HR = 0.84, 95% CI, 0.71–0.99) and 20% (HR = 0.80, 95% CI, 0.70–0.92) reduction in risk for elevated depressive model, respectively. Participants with primary school and below education and junior high school and above education were independently associated with a 17% and 41% reduced risk for developing elevated depressive symptoms (HR = 0.83, 95% CI, 0.75–0.92; HR = 0.59, 95% CI 0.52–0.69). Social support was negatively associated with depressive symptoms in all the models. With one score increase of social support, the risk of elevated depressive symptoms reduced 2% (HR = 0.98, 95% CI, 0.95–1.00) in model 4. We further explored the individual relationship of each type of social support with the risk of depressive symptoms. After seven-year follow up, every increase in community support reduced the likelihood of elevated depressive symptoms by 3% (HR = 0.97, 95% CI 0.95–1.00, Table S1).

TABLE 2. Hazard ratio of elevated depressive symptoms (95% CI) Variables Model 1 Model 2 Model 3 Model 4 Household income (≤3000) 3000–10,000 0.95(0.84–1.08) 0.95(0.84–1.08) 0.98(0.86–1.11) 1.00(0.88–1.14) 10,000–24,000 0.85(0.75–0.97)** 0.87(0.76–0.99)* 0.89(0.78–1.02) 0.91(0.79–1.03) 24,000–36,000 0.76(0.65–0.90)*** 0.79(0.67–0.92)** 0.81(0.69–0.96)** 0.84(0.71–0.99)* >36,000 0.71(0.62–0.81)*** 0.74(0.65–0.85)*** 0.76(0.67–0.87)*** 0.80(0.70–0.92)*** Education (illiterate) Primary school and below 0.72(0.65–0.80)*** 0.73(0.66–0.81)*** 0.82(0.74–0.91)*** 0.83(0.75–0.92)*** Junior high school and above 0.47(0.41–0.53)*** 0.48(0.42–0.55)*** 0.60(0.53–0.67)*** 0.59(0.52–0.69)*** Social support - 0.97(0.94–0.98)* 0.97(0.95–1.00)* 0.98(0.96–1.00)* Gender (female) Male - - 0.73(0.63–0.84)*** 0.70(0.61–0.81)*** Smoking (current smoking) Former smoking - - 0.80(0.66–0.98)* 0.78(0.64–0.96)* Never smoking - - 0.98(0.86–1.11) 0.99(0.86–1.12) Drinking (current drinking) Former drinking - - 1.12(0.94–1.33) 1.12(0.94–1.33) Never drinking - - 1.10(0.96–1.25) 1.08(0.95–1.24) Age (<60) 60–69 - - 0.94(0.85–1.04) 0.92(0.83–1.02) 70–79 - - 0.99(0.83–1.17) 1.04(0.88–1.23) >80 - - 1.33(0.90–1.95) 1.46(0.99–2.14) Number of NCDs (0) 1–2 - - 1.30(1.15–1.46)*** 1.29(1.15–1.45)*** ≥3 - - 1.61(1.38–1.86)*** 1.59(1.36–1.85)*** Missing - - 1.18(0.93–1.49) 1.16(0.92–1.46) Self-assessment health (excellent) Very good - - 0.84(0.45–1.56) 0.78(0.41–1.49) Good - - 0.93(0.51–1.70) 0.85(0.45–1.60) Moderate - - 1.16(0.64–2.11) 1.05(0.56–1.96) Bad - - 1.61(0.88–2.95) 1.46(0.78––2.75) Missing - - 1.08(0.59–1.96) 0.98(0.52–1.83) ADL score (0) 1 - - 1.11(0.95–1.31) 1.09(0.93–1.29) 2–6 - - 1.59(1.32–1.90)*** 1.57(1.31–1.88)*** Missing - - 1.23(0.70–2.16) 1.26(0.72–2.19) BMI (≤19) 20–24 0.89(0.78–1.0

留言 (0)

沒有登入
gif