Design
This study adopted a multi-centre cross-sectional study design, using a self-reported questionnaire.
Sample And Setting
In 2022, data on hospital staff from six provinces, municipalities, and autonomous regions of China were collected, and doctors, nurses, and administrative staff from these hospitals were included in the study. Inclusion criteria for sample selection were: (a) regular employees; (b) having at least 6 months of hospital work experience; (c) any place of birth; (d) any position; (e) accessible during data collection. With the support of all central units participating in this study and informed consent of all respondents, a total of 1034 questionnaires were collected and included in statistical analysis. People with fewer than 6 months of hospital work experience were excluded because they may not have had a good understanding of the job.
Patient And Public Involvement
No patients were involved in this study.
Measures
The questionnaire used in this study includes four parts: demographic information survey, silence behaviour scale, psychological empowerment scale, and PSC scale, all of which are classic scales.
The questionnaire was used to collect the basic characteristics of respondents, including region, sex, age, education level, length of service, marital status, current position, employment form, and current position.
Silence behaviour was measured using a 12-item scale developed by Zheng et al. (2008)24. The scale has been improved and optimized on the basis of a scale developed by Van Dyne et al. (2003)25. The scale includes three dimensions: acquiescent silence, defensive silence, and indifferent silence, using a five-point Likert scale. The internal consistency reliability (Cronbach's alpha coefficient) and construct validity of the scale were tested, which were 0.938 and 0.923 (Table 1).
The level of psychological empowerment in respondents was measured using a 12-item scale developed by Spreitzer et al. (1995)26. The scale includes four dimensions: meaning, competence, self-determination, and impact. The Cronbach's alpha coefficient and construct validity of the scale were 0.938 and 0.907, respectively (Table 1).
The Hospital Survey of Patients’ Safety Culture (HSOPSC) tool was developed by healthcare research and quality agencies to assess a healthcare facility's culture of patient safety. The HSOPSC is composed of 42 questions scored with a five-point Likert scale measuring 12 PSC components. The Cronbach's alpha coefficient and construct validity calculated in this study were 0.846 and 0.938, respectively (Table 1).
Data Collection
Data were collected from June 2022 to August 2022, as guided by the questionnaire prepared for this study. A paper-based data collection method was initiated to achieve high response rates. We coordinated with the hospital management departments where each centre was located to develop a data collection plan. The hospital where each centre was located had a primary contact responsible for data collection for that centre. The contact person of each centre was responsible for sealing the collected answer sheets and mailing them to the researchers.
Statistical Analysis
Analyses were performed using R 4.2.1 and IBM SPSS Aoms24 (IBM Corp., Armonk, NY, USA).
We used descriptive statistics to broadly describe the demographic characteristics of respondents. Positive responses were counted for each item in the scale to determine the percentage of positive scores for each dimension. Dimensions rated as positive by 75% or more of respondents were considered strengths, and dimensions rated as positive by 50% or fewer respondents were considered weakness27.
The mean and standard deviation for each dimension were also calculated. Baseline characteristics of participants were expressed as frequencies and percentages in categorical data.
Correlation analysis was used to verify the correlation between silence behaviour, psychological empowerment, PSC, and each dimension.
Mediation analysis was used to determine the direct and indirect effects of silence behaviour on PSC through psychological empowerment using the following two pathways:
Path a: Y = cX + e1
Path b: Y = c’X + bM + e2
Silence behaviour, psychological empowerment, and PSC mean were considered independent variables, mediating variables, and dependent variables, respectively. Silence behaviour first fit the PSC. Then, the second equation was constructed by fitting psychological empowerment and silence behaviour in PSC.
A mediation effect was then estimated using the product-of-coefficient method 28–29. A bootstrap analysis with 1000 replications was then applied to estimate average mediation effects30. The mediation effect was estimated for each bootstrap, averaged across 1000 replications, and its corresponding 95% confidence interval (CI) was determined using a bias-corrected bootstrap technique.
P values less than 0.05 indicated statistical significance for all analyses.