Studying setting
Anhui is located east of central China, and its economic development is at a medium level across the country. Each year, 22,635 primary health institutions in Anhui provide outpatient and inpatient services, accounting for 61.27% and 20.90%, respectively, which are higher than the national average of 54.12% and 18.21% [14].
This province was selected for the following reasons: (1) Anhui has been at the forefront of China in the comprehensive reform of primary medical institutions making its primary health care institutions a government priority. (2) Considering feasibility, Anhui guaranteed the compliance of participants. (3) Anhui has a large number of primary healthcare workers.
Participants and data collection
We used a cross-sectional survey design. Primary medical staff in this study refers to the medical staff in community health service centers (CHC), community health service stations (CHS), township health centers, village clinics, and outpatient departments. According to the regional characteristics, Anhui is divided into three regions: Northern Anhui, Central Anhui, and Southern Anhui. We used random sampling to select a district and a county from southern and central Anhui and one district and two counties from northern Anhui (northern Anhui has a larger population). We used a series of questionnaires to collect data, and the respondents filled out the questionnaires anonymously and voluntarily. The questionnaires in each area were collected by special investigators and withdrawn immediately after completion. If the respondent answered regularly to the questionnaires or filled in the questionnaires with incomplete content, it would be regarded as invalid and eliminated.
Before conducting the survey, we held several investigator trainings and set inclusion and exclusion criteria (survey subject must be over 20 years old, held their position for one year, occupation of medical staff is limited to doctors and nurses, pharmacists, administrative staff, etc.). We also held a series of on-site, preliminary investigations before official investigation.
We recovered 1,300 questionnaires, of which 1,152 were valid, making the effective recovery rate 88.62% (1,152/1,300).
Measures
Demographic characteristics questionnaire
First, we used the general characteristics questionnaire, which we independently designed based on relevant literature and expert consultation, including three parts:(1)General demographic characteristics: gender (male, female), age (20–30 years old, 31–40 years old, 41–50 years old, over 50 years old), professional title (primary, intermediate and above), education (secondary; technical school and below, college; undergraduate and above), working years (1–10 years, 11–20 years, 20 years or more), marital status (married, other) (2) Job characteristics: Monthly income (CNY 3000 and below, higher than CNY 3000), occupations (doctor, nurse, pharmacist and administrative staff), work unit (CHC, CHS, township health center, village clinic, outpatient department) (3) Regional characteristics: The city or county (district) where the respondents are located.
PCQ-24
Psychological capital is measured using the psychological capital scale (PCQ-24) compiled by Luthans et al. [15], with a total of 24 items. It measures psychological capital from four dimensions: self-efficacy, hope, resilience, and optimism. Points are graded from 1 to 7. A score of 124 indicates extremely high psychological capital; above 100, is high level of psychological capital; above 80, the psychological capital is a medium level; below 80, it is necessary to strengthen and train psychological capital. The reliability of the scale is Cronbach's α coefficient of 0.97, and the results of confirmatory factor analysis showed the scale has good reliability.
PSSS
We used Perceived Social Support Scale (PSSS) compiled by Zaimet et al. [16] in 1987 and revised by Chinese scholar Wang XD [17]. The scale consists of 12 items, divided into two dimensions: in-family support (items 3, 4, 8, 11), and out-of-family support (the remaining items). Points 1 to 7 are used for scoring. Scores between 12–36 are considered low support state; between 37–60 points are an intermediate support state; between 61–84 points are a high support state. The reliability of the scale is Cronbach’s α coefficient is 0.94
Job burnout scale
We used the Chinese Maslach Burnout Inventory (CMBI) of Li YX et al. [18] for scoring. It is revised based on MBI (Maslach Burnout Inventory) questionnaire [19]. There are 15 questions in the questionnaire, with five questions in each three dimensions: emotional exhaustion, disintegration of personality, and reduction in sense of achievement. It uses Likert's 7-point scoring method, 1 means "strongly disagree" and 7 means "strongly agree." The scale divides the level of burnout into four levels through three critical values: 25 points for emotional exhaustion, 11 points for disintegration of personality, and 16 points for reduced sense of achievement. When all three dimensions are less than the critical value, it is zero burnout; when any of the three dimensions is higher than the critical value, it is mild burnout; otherwise, it is moderate or severe burnout. The internal consistency test of the scale showed Cronbach’s α coefficient is 0.767, indicating good reliability.
Turnover intention scale
The scale of turnover intention was translated and revised by Michael J E et al. [20]. The scale includes a total of six items, divided into three dimensions: possibility of quitting current job (turnover intention I, items 1 and 6), motivation to find other jobs (turnover intention II, items 2 and 3), and obtaining external possibility of work (turnover intention III, items 4 and 5). It uses reverse scoring on a scale of 1 to 4. If the score is higher, the turnover intention is higher. The reliability of the scale is Cronbach's α coefficient of 0.80.
Statistical Analyses
We used Epi Data 3.1 for database building, and researchers double-entered data and performed error detection, SPSS 20.0 (IBM Corp, Armonk, NY, USA) for statistical analysis of data. Count data is described by composition ratio and measurement data is described by mean and standard deviation (M ± SD).
For univariate analysis, we conducted an independent sample t-test for binary variables (gender, professional titles, marital status, monthly income). Then, we divided the age of primary medical staff into four groups: 20–30 years old, 31–40 years old, 41–50 years old, 50 years and older; divided the working years into three groups: 1–10 years, 10–20 years, more than 20 years, which changes the age and working years of primary medical staff from continuous variables to categorical variables. We performed one-way analysis of variance (ANOVA) on multiple categorical variables (age, education, working years, employment agency, occupation, region). In addition, we used Pearson correlation analysis to explore the correlation between turnover intention and the scores of psychological capital, social support, and job burnout among them, the test level α = 0.05.
In multivariate analysis, we used multiple linear regression models to set dummy variables for categorical variables and set a control for each survey item, thereby testing its association with other items under standard and non-standard coefficients, the test level α = 0.05.