A cross-sectional survey study was conducted in Hangzhou city, Zhejiang Province, China in October 2017. The online questionnaire was hosted by the online questionnaire distribution company “Athena”. Ethical approval was obtained from the Ethics Committee of the China Pharmaceutical University (Project number: CPU2018016).
Sampling
The Yangtze River Delta is a region in China with a relatively developed economy, high medical standards. By the end of 2016, there were 31,548 healthcare institutions in Zhejiang Province, which is the highest number of medical institutions at province level. Zhejiang Province was, therefore, selected to be the research area of this study. According to the "2016 Zhejiang Health and Family Planning Yearbook"47, Hangzhou being the provincial capital had 244 hospitals in which there were 34,832 practicing (assisting) physicians. It is the city with the largest number of hospitals and the largest number of physicians in Zhejiang Province. Therefore, Hangzhou was chosen as the sampling city.
Stratified sampling was conducted as followed: firstly, the healthcare institutions were categorized into 3 categories based on the “Hospital Grading Standards” and in consultation with the Hospital Grading Information System in China: basic hospitals, secondary hospitals, and tertiary hospitals (and their overall scales were rated as small, medium and large respectively). Due to their small scale and the small number of healthcare responsibilities, basic hospitals were excluded from this study. Furthermore, for the secondary hospitals and tertiary hospitals, the number of hospitals at each grade was multiplied by the proportion of in the total number of hospitals to yield the hospital sampling number. For instance, there were 15 tertiary hospital and 12 secondary hospitals in Hangzhou, so the hospital sampling number would be 8 and 5 respectively (see Equations in the Supplementary Files).
Finally, with convenient sampling, a total of 13 hospitals were selected as the primary sampling sites. Within each primary sampling site, sampling was primarily conducted at the key departments, and covered at least half of the departments in each sampling hospital.
In each hospital, doctors were randomly selected by volunteering research assistants who were pharmacy students and had received sufficient research training for the task. Their aim was to randomly selected and collected information from at least 5 doctors from each department.
Data collection
Before answering the questionnaire, research assistants explained to the participants the research objective and topics, and only provided assistance in answering the questionnaire whenever requested by the participants. All of the research data were collected by team-designed software and were processed into identifiable electronic data. As part of the quality control of the survey, the research team had developed a set of guidelines to help secure good sampling procedure. The survey team was trained prior and during the fieldwork to enhance their adherence to the practice standards prescribed in the guideline. Daily supervision of the survey team and close audit of data entered in the survey software were also performed to identify and resolve any problematic items in the questionnaire and any problems during survey implementation. Systematic check of data quality at different stages of survey study was also conducted to estimate the completeness of the survey study and the frequency of missing data. Collectively, these actions were set to help ensure the quality of data collected in this survey study.
Questionnaire design
The questionnaire used in this study consisted of 3 parts and was developed in consultation with previous research work. A sample of the questionnaire is provided in the Supplementary Document.
Part 1 - Dependent variable – Presenteeism
Presenteeism is the dependent variable in this study. In order to find out if the participant had practiced presenteeism, the frequency of presenteeism was asked. Referring to the measurement used by scholars such as Johns, Cocker et al, Karanika-Murray et al, a one-question measurement was used to measure the act of presenteeism.1,48,49 Participants were asked “Has it happened over the previous 12 months that you have gone to work despite feeling that you should have taken sick leave because of your state of health?” and had to choose one of the four answers: “never”, “once”, “2-5 times” and “more than 5 times”. Choosing “never” or “once” was considered “no” to presenteeism, and choosing twice or more often was considered “yes” to presenteeism.7,45 This measurement has been widely used in previous studies and its reliability has been demonstrated.1,48,49
Part 2 - Independent variable – Perceived availability of social support
Interpersonal Support Evaluation List-12 was used to measure 3 aspects of social support including appraisal, belonging and tangible.27 Interpersonal Support Evaluation List-12 originated from the Interpersonal Support Evaluation List developed by Sheldon Cohen in 198550 and has been used repeatedly adopted in previous studies.51-54 For each of the 3 included aspects, there were 4 related questions in the questionnaire. Each of the question had four possible answers: “highly disagree”, “possibly disagree”, “possibly agree”, and “highly agree”, with each answer worth 0, 1, 2 and 3 points respectively except for 6 of the questions which had reverse score. The sum of the score of each questions related to each aspect was calculated separately, and totally to give a total score of perceived availability of social support. The higher the score, the better availability of social support perceived by the participants. The scale has been tested and shown to have good reliability.55,56
Part 3 - Control variables – demographic information and factors contributing to presenteeism
In addition to the perceived availability of social support that might affect physicians' act of presenteeism, this study also took into consideration the impact of other possible influencing factors and treated them as control variables. According to Cocker et al, factors which might contribute to presenteeism fell into one of the 4 categories: socio-demographic factors, health factors, financial factors and work-related factors.48 Socio-demographic factors had 6 items: age, gender, marital status, pregnancy for female participant or for the spouse of male participant, number of children, highest education level). Health factors mainly concerned the history of chronic diseases.15,57 Financial factors mainly included the presence of a reward system for full attendance7,58 and the participant’s monthly salary.57,59 Work-related factors included the number of years the participant had worked at the current hospital12,16,60,61, their position ranking62, their job title16,60, weekly work hours7, substitute availability57,59, and their superior’s leadership type63. All of these information was collected in Part 1. Substitute availability at work was measured using the method developed by Aronsson and Gustafsson.57 The type of superior’s leadership was measured based on the leadership theory developed by Lewin.64
The questionnaire design was initially assessed by 5 experienced researchers in public health or occupational health to ensure the theoretical construct was appropriately represented in the questionnaire. To ensure face validity of the questionnaire, these researchers were also asked to evaluate if the questions in the questionnaire would allow reasonable and operational measurements of the dependent variable, independent variables and control variables mentioned above. They were also asked to comment on the face validity of the translation to Chinese. Based on the researchers’ feedback, we revised the translation of two items in the multi-item scales about the perceived availability of social support (Interpersonal Support Evaluation List-12 in Question 2 of the questionnaire) to improve clarity. In a pilot study, the questionnaire was further tested by 8 doctors not included in the sample for readability, clarity and comprehensiveness of the questions. They all came to an agreement that the questions were straight forward and easy to understand. Cronbach’s alpha were also measured to determine the reliability of the multi-item scales related to the perceived availability of social support.
Statistical analysis
Data collected was organized using Excel initially and data analysis was performed using Stata 14.0. The logit model was used to analyze the association between presenteeism and perceived availability of social support first. In order to more accurately evaluate the impact of perceived availability of social support on presenteeism, the Logit model was used again to further analyze the association of each dimension of perceived availability of social support on presenteeism. Multicollinearity was then checked on each of the 3 dimensions of perceived availability of social support. Appraisal support was set as the dependent variable, and belonging support and tangible support as independent variables. Multiple linear regression was then used to explain the relationship between the dependent variable and the two independent variables. Correlation coefficient was then used to determine the direction and strength of the relationship between the variables. The association between the variables was found to be weak, so the 3 variables were used as separate variables in replacement of the overall perceived availability of social support while the control variables remained unchanged. The Logit model was then used again to test the association between each of the 3 dimensions of perceived availability of social support and presenteeism separately. The correlation was found to be strong, so all 3 dimensions were integrated in the Logit model again which was then run 3 times to analyze the association of each dimension with presenteeism.