2.1 Study design
The data presented in this report was based on the baseline sample of an ongoing observational prospective cohort study investigating mental health problem among older PLWH in Sichuan province, China (the Sichuan Older HIV-infected Cohort Study, SOHICS). Questions asked by this study were parts of the cross-sectional baseline survey of the cohort, which was conducted from November 2018 to February 2019.
2.2 Data collection
As of 2018, there were 183 districts/counties in Sichuan. A stratified multi-stage cluster sampling design was used to select the study area of the SOHICS. First, two districts/counties were randomly selected from five districts/counties with the highest HIV prevalence among the elderly in Sichuan in 2018. Then, all sub-district (i.e., communities) or sub-county units (i.e., towns) within two selected districts/counties were contacted for participating in the study. According to the basic information system for AIDS Prevention and Treatment, all eligible subjects in those participating units were screened and invited to participate in the survey. The inclusion criteria for SOHICS participants were 1) receiving confirmatory HIV diagnosis, 2) at the age of 50 years or older at the time of diagnosis, 3) living in Sichuan for more than five years, and 4) receiving care and/or treatment in the local township health centers. Participants were excluded if they were found to have major psychiatric illness (e.g., schizophrenia and bipolar disorder) from their medical records, or were unable to communicate with the interviewers (Figure 1).
According to the inclusion criteria and exclusion criteria, the participants were screened in the basic information system for AIDS Prevention and Treatment. Medical staff in township health centers phoned and briefed prospective participants about the study and confirmed their eligibility to participate in the study. Guarantees were made on anonymity and their right to quit at any time without being questioned, and that refusal would not affect their right to use any service in township health centers. Those who were interested in participating in the study were invited to pay a visit to their local township health centers. On site, trained interviewers briefed them again about the study and obtained their written informed consent. Ethics approval was obtained from the Ethics Committee West China School of Public Health and West China Fourth Hospital, Sichuan University.
A panel consisting of four epidemiologists, one health psychologist, and two staff in township health centers designed a questionnaire. It was tested and validated by conducting an anonymous face-to-face interview for 50 participants (25 randomly selected from each districts/counties) in private rooms in their local health centers. Their feedback was used to revise and finalize the questionnaire after panel discussion. Based on the pilot results, slight changes were made to improve readability of the questionnaire. No major change (e.g., removal of items) was made. Participants of the pilot study did not take part in the actual survey. This validated questionnaire was then used to aid with a 20-minute anonymous face-to-face interview for all in private rooms in the same health centers. Participants were asked about their socio-demographic information, including age, sex, place of residence, ethnicity, education level, marital status, number of children, employment status, monthly personal income, duration of out migrating for work, and infection of spouse with HIV. Disease-related characteristics were extracted from their medical records, including route of HIV transmission, time since diagnosis, duration on ART, stage of HIV infection, and their most recent CD4 cell counts.
2.3 Sample size planning of the SOHICS
Assuming the prevalence of probable depression/anxiety is 50% at the end of the follow-up period (12 months after baseline survey), the sample size of 400 older PLWH will confine the 95% confidence interval within +/- 4.9%. Given the assumption that prevalence of probable depression/anxiety in the reference group (e.g., those without a risk factor at baseline) to be 20-40% at Month 12, the sample size could detect the smallest odds ratios of 1.76 between those with and without such risk factor at baseline, with a statistical power of 0.8 and an alpha value of 0.05. Assuming the loss-to-follow-up rate to be 20% at Month 12, 500 older PLWH need to be recruited at baseline (PASS 11.0; NCSS; Kaysville; the United States).
2.3.1 Social Capital
The measurements of social capital used in this study were adapted from two scales in a validated Chinese version of Health-related Social Capital Measurement , i.e., the individual and family (IF) social capital scale and the community and society (CS) social capital scale. The IF scale had seven items. They were: 1) ‘You have many close contacts’, 2) ‘You have many social interaction with people other than your family members in the past month’, 3) ‘You always trust people who have social interaction with you’, 4) ‘You are satisfied with your marriage’, 5) ‘You always received emotional/financial/instrumental support from your spouse’, 6) ‘You always received emotional/financial/instrumental support from your relatives’, and 7) ‘You always received emotional/financial/instrumental from your close contacts in the last year’.
The CS scale also had seven items. They were: 1) ‘You frequently participated in activities organized by community organizations in the last year’, 2) ‘You always received support from community organizations in the last year’, 3) ‘You trust health organizations (i.e., hospitals and centers for diseases control and prevention) very much’, 4) ‘You trust community organizations very much’, 5) ‘You trust other governmental organizations very much’, 6) ‘You agree with the statement that hardworking people will be rewarded by the society’, and 7) ‘Do you agree with the statement that talented people will be recognized by the society’.
Response categories for both scales ranged from 1 (strongly disagree) to 5 (strongly agree), with a higher total score indicating the stronger social capital.
2.3.2 Mental health
We focused on two major types of mental health problems among older PLWH: depression and anxiety . Depressive symptoms were measured using the Center for Epidemiologic Studies Depression (CES-D)-10 scale, which was constructed on the basis of self-reported responses to ten questions on whether, over the past week, participants had experienced symptoms associated with depression, such as worrying, sleeping difficulty and difficulty relaxing (Appendix A). The CES-D-10 scale has a good test-retest reliability and predictive validity when compared with the original version of CES-D with twenty items , also with high sensitivity (97–100%) and specificity (84–93%) for screening major depression in middle-aged and older adults . Responses were reported using a four-point Likert scale ranging from 0 (rarely or none of the time) to 3 (most or all of the time). The total score ranged from 0 to 30, with higher scores indicating more severe depressive symptoms. A cut-off score of 10 was used to define the presence of probable depression, which has been validated owning to minimizing false-positive results with little loss of sensitivity .
The Generalized Anxiety Disorder Scale (GAD-7) was used to screen participants for probable anxiety . It recorded how often participants have suffered from seven problems over the past two weeks (Appendix B). Responses were also reported using a four-point Likert scale ranging from 0 (not at all) to 3 (nearly every day). The total score ranged from 0 to 21, with higher scores indicating more severe anxiety symptoms. A cut-off score of 5 was suggested to define the presence of probable anxiety .
Both the CES-D-10 scale and the GAD-7 scale were validated in Chinese population [40, 41]. Both scales have been widely used in studies targeting PLWH and older adults in China [42-45].
Covariates considered in this study included age, sex (male and female), place of residence (rural and urban), ethnicity (Han and minority), education level (illiterates, primary school, junior high school, and senior high school or above), marital status (unmarried, married and living with spouse, married but not living with spouse, and divorced and widowed), employment status (employed, retired, and unemployed), duration of out-migrating for work (none, ≤10 years and >10 years), monthly personal income (none, <1,000 yuan, 1,000-1,999 yuan, and ≥2,000 yuan) (one yuan was equivalent to about 0.14 US dollars at the time of conducting this study), having an HIV-infected spouse (yes, no, and do not have spouse), number of children (0, 1, 2, and ≥3), route of HIV transmission (sexual behaviour with spouse, sexual behaviour with a non-spouse opposite-sex partner, sexual behaviour with a same-sex partner, and blood transfusion), time since diagnosis (<1 years, 1-3 years, and >3 years), duration on ART (≤2 years and >2 years), stage of HIV infection (HIV, AIDS, and missing), CD4 cell counts (<200, 200-350, 351-500, >500, and missing).
2.4 Statistical Analyses
Presence of probable depression (CES-D-10 score ≥10) and probable anxiety (GAD-7 score ≥5) were used as dependent variables. In line with many published studies, we dichotomized scores of the CES-D-10 and GAD-7 and used logistic regression models to investigate factors associated with probable depression and probable anxiety [46-48]. Two-level logistic regression models (level 1: towns/communities, level 2: individual older PLWH) were fit to analyze factors associated with the dependent variables. Random intercept models were used to allow the intercept of the regression model to vary across communities/towns, which could account for intra-correlated nested data. Two-level logistic regression models are commonly used in studies using similar sampling method as ours [49, 50]. Univariate logistic models were first used to examine the significance of the association between each covariate and outcome variable. Two sets of multivariate two-level logistic regression models were included in this study. The first set of multivariate logistic regression models controlled age and sex, as their relationships with mental health status are well established in previous studies . Covariates with p<0.10 in univariate analysis were adjusted in the second set of multivariate logistic regression models; similar approach was used in many published studies [52, 53].
Descriptive statistics, reliability analysis, and logistic regression embedded in the SPSS version 23.0 for Windows (SPSS, Inc, Chicago, IL, the United States) was used for data analysis, with p<0.05 considered as statistically significant in final models. In addition, Cronbach’s alpha, ranging from 0 to 1, was used to assess the reliability of social capital scales. A Cronbach’s alpha of 0.6 or greater was considered acceptable [54, 55].