Design and participants
This cross-sectional study was conducted in the Mental Health Center of Shandong Province, Jinan, China, and participants were enrolled using a convenience sampling strategy. Participation in the research was voluntary. Participants informed about the feasibility nature of the study. All the participants provided informed verbal consents before the study began.
Inclusion criteria for caregivers were: (1) aged 18 years and older; (2) serving as the primary caregiver (self-reported) for the patient diagnosed of schizophrenia by psychiatrists using the Diagnostic and Statistical Manual of Mental Disorders-IV-text revision or the International Classification of Disease-10; (3) reporting no experience of other traumatic events, e.g., divorce, serious illness in the past three months. Exclusion criteria were: (1) taking care of 2 or more family members either with physical or mental illness simultaneously; (2) having a history of mental illness; (3) refusing to participate.
Measures
Basic characteristics
We used a self-administered questionnaire to collect basic characteristics from primary caregivers of schizophrenic patients. The information collected from primary caregivers included gender, age, educational background, per capita income per month, and their relationship with the patients in care. Patients’ characteristics of gender, age, enroll in the public health insurance or not, and times of hospitalization were retrieved from their medical charts.
Response variable: PTG
PTG was assessed with the Chinese version Posttraumatic Growth Inventory (PTGI) [19]. The Chinese version scale has 21 items that are distributed into five domains, i.e., personal strength (4 items), new possibilities (5 items), relating to others (7 items), appreciation of life (3 items), and spiritual change (2 items). The scaling scheme of each item ranges from 0 = not at all to 5 = very strongly degree, with the higher score representing the greater PTG. Good psychometric characteristics of the Chinese version PTGI have been confirmed [19], and in this study, the Cronbach’s alpha coefficients of five domains range from 0.75 to 0.85.
Explanatory variable: Perceived social support
Perceived social support was assessed with the Chinese version perceived social support scale [20]. The Chinese version scale has 12 items aiming to assess how individuals perceive social support from their families (4 items), friends (4 items), and significant others (4 items). Each item is graded on a 7-point Likert format (1 = very strongly disagree to 7 = very strongly agree), with the higher score representing the stronger perceived social support. Good psychometric characteristics of this scale have been confirmed in Chinese populations [20, 21], and the Cronbach’s alpha coefficients of three domains range from 0.88 to 0.93 in this study.
Mediating variables
Resilience:
Resilience was evaluated with the Chinese version Connor-Davidson Resilience Scale (CD-RISC) [22]. The Chinese version scale has 25 items that are distributed into three domains: strength (8 items), optimism (4 items), and tenacity (13 items). Each item is scored from 0 = not at all to 4 = all the time, with the higher score indicating the greater resilience. Studies have confirmed the satisfactory psychometric characteristic of Chinese version CR-RISC [22, 23], and in this study, the Cronbach’s alpha coefficients of three domains range from 0.80 to 0.89.
Coping styles:
The Simplified Coping Style Questionnaire [24] was used to evaluate caregivers’ coping styles. This scale consists of 20 items which are divided into two subscales: positive coping style (12 items) and negative coping style (8 items). Each item is graded on a 4-point scaling scheme ranging from 0 = never to 3 = always. The higher sum score of each subscale presents the more frequent usage of that coping style. The good reliability and construct validity have been confirmed in the Chinese population [24, 25], and the Cronbach’s alpha coefficient was 0.90 for positive coping style and 0.76 for negative coping style in this study.
Data collection
Data were collected from October 2018 to January 2019. Trained students in our research team reserved the bright, spacious and quiet room in the center to recruit eligible caregivers, introduce our study, obtain dual informed consent (i.e., their own participation and our review of patients’ medical chart) and distribute the questionnaires. Additional supports for participants with presbyopia and illiterate participants would be provided when needed. The questionnaire took approximately 30 minutes to complete. Although no formal sample size determination was done, this study did follow the rule of thumb when applying structural equation modeling to run analysis, i.e., the sample size should be at least 200.
Statistical analysis
Descriptive statistics were used to summarize basic characteristics about the primary caregivers and patients in care, and average scores of domains/subscales for four main concepts (i.e., PTG, perceived social support, resilience and coping styles) reported by primary caregivers were tabulated. Student’s t-test and analysis of variance were applied where appropriate to test the associations between basic characteristics of primary caregivers and their PTG. Distributions of items in scales assessing four main concepts were tested using Shapiro-Wilk statistic in IBM® SPSS® Statistics (version 26, IBM Corp., Armonk, NY), and the estimator was finalized as the maximum likelihood estimation given the normal distribution inferred by -3.0 to 3.0 for skewness and -8.0 to 8.0 for kurtosis of this sample. Model testing was performed with Amos 23.0.
We used the domain-representative approach to get item parcels for three multi-domain concepts, i.e., perceived social support, resilience and PTG and random assignment approach to get item parcels for positive coping style and negative coping style [26]. We used path analysis approach to evaluate the construct validity across concepts and then tested the structural model after the validation of the measurement model. The significant basic characteristics of caregivers were controlled as covariates in the structural model.
We used Comparative Fit Index (CFI) ≥ 0.90, normed fit index (NFI) ≥ 0.90, Root Mean Square Error of Approximation (RMSEA) ≤ 0.08, and Standardized Root Mean Square Residual (SRMR) ≤ 0.05, along with the c2 and RMSEA 90% Confidence Interval (CI) to assess the goodness of fit for all models. We used 95% bootstrap CI to evaluate the significance of overall indirect effect and separated indirect effect, and the latter was obtained by running “User-defined estimands” in Amos. Two-sided p-value < 0.05 was considered statistically significant.