Associations of Workplace Violence and Psychological Capital With

BMJ Open: first published as 10.1136/bmjopen-2018-024186 on 24 May 2019. Downloaded from PEER REVIEW HISTORY

BMJ Open publishes all reviews undertaken for accepted manuscripts. Reviewers are asked to complete a checklist review form (http://bmjopen.bmj.com/site/about/resources/checklist.pdf) and are provided with free text boxes to elaborate on their assessment. These free text comments are reproduced below.

ARTICLE DETAILS

TITLE (PROVISIONAL) Associations of workplace violence and psychological capital with depressive symptoms and burnout among doctors in Liaoning, China: a cross-sectional study AUTHORS sui, guoyuan; liu, guangcong; jia, lianqun; wang, lie; yang, guanlin

VERSION 1 - REVIEW

REVIEWER Shailesh Kumar Waikato Clinical Campus, University of Auckland Private bag 3200 Hamilton New Zealand REVIEW RETURNED 08-Aug-2018

GENERAL COMMENTS I enjoyed reading this paper. Authors have identified an important area and have done justice to the topic. The paper needs significant language review. sentences are often structured poorly and sometimes difficult to understand (e.e Page 3 line 8; page 4-lines 4- 5, 11-12 and 14-19). Page 5 why should only Chinese doctors be vulnerable - line 15 well-being spelling needs correction, I suggest authors should give a concise and clear definition of psychological http://bmjopen.bmj.com/ capital early in the paper and explain how they protected respondents' privacy and anonymity especially when they had such high response rates. I would also like to read a bit more about the impact of deleting two major items of physical violence and vicarious violence and its impact on the overall quality of the tool

REVIEWER Oriol Yuguero on September 28, 2021 by guest. Protected copyright. Institut de Recerca Biomedica de Lleida REVIEW RETURNED 16-Aug-2018

GENERAL COMMENTS I think it's a very interesting article. Perhaps a shorter introduction would be fine. As a non-Chinese reader, I would like to know the situation of the professionals that may induce aggression towards them. In the Methodology section I would reduce the explanation of the sample size calculation. We already deduce that the calculation has been made correctly. The results are interesting and I think they show a situation suffered by professionals in China. Perhaps it can help to establish improvements in the health system to improve the mental health of professionals and avoid aggressions to doctors.

REVIEWER Lynn Monrouxe Chang Gung Medical Education Research Centre (CG-MERC) Chang Gung Memorial Hospital, Taiwan. REVIEW RETURNED 17-Aug-2018

BMJ Open: first published as 10.1136/bmjopen-2018-024186 on 24 May 2019. Downloaded from GENERAL COMMENTS Thank you for giving me the opportunity to review the manuscript “Associations of workplace violence and psychological capital with depressive symptoms and burnout among doctors in liaoning, China: a cross-sectional study”. I have read it with great interest.

The subject of physician wellbeing is of great importance. However,

The Introduction does not quite set the study up adequately. I believe that some of this is to do with the English language that needs to be addressed, some to do with lack of references and some to do with the argument not building sufficiently.

Workplace violence was measured via a categorical scale but other measurements were (possibly) on a continuous scale – I confess that I am not a high-level statistician but mixing nominal data with ordinal/continuous data isn’t possible is it? I suggest someone who understands statistics gives a view here – I found the analysis and results section hard to follow.

In terms of the results section – some of my difficulty was with the reporting of the findings and the lack of any detail at all in the main body of the report – while I understand the authors wished to summarise their findings, it’s too brief and the reader is left wondering what was really found to be statistically significant.

In the discussion section - that “psychological capital mediated the effects of workplace violence on depressive symptoms and two dimensions of burnout” is, of course, important. But then the authors go on to say that Doctors who “suffered workplace violence would be more likely to experience lower levels of psychological capital, which in turn increased the levels of depressive symptoms and burnout.” doesn’t actually follow logically.

Following this the authors say “Therefore it may be a effective http://bmjopen.bmj.com/ strategy to develop some programs to increase psychological capital, thus decreasing doctors’ depressive symptoms and burnout in the long run” – again – this is too much of a claim based on the authors’ data and too simplistic to have ‘some programs to increase psychological capital’ (so many constructs involved) and then depressive symptoms and burnout are decreased – I can’t see this being the case as these are really complex mental illness phenomena. on September 28, 2021 by guest. Protected copyright.

Despite the authors admitting that their study is such that it is impossible to draw causal relationships – they continually attempt to do such a thing. Caution should be taken here.

The authors have filled in the STROBE-Vet which is an extension of the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) statement. However, their study is not an observational study in epidemiology (hence they responded ‘no’ to many items). I suggest they switch to using: Kelley K, Clark B, Brown V, Sitzia J. Good practice in the conduct and reporting of survey research. Int J Qual Health Care. 2003;15(3):261-266.

I suggest the authors fond someone to English Edit their work.

DETAILS OF ADDITIONAL CHANGES Abstract Please briefly define psychological capital as it is not clear. The design just says it’s a “cross sectional study” but of what kind? Many different qualitative and quantitative studies can be cross sectional – can the authors please be more specific. Also, the authors state “Psychological capital mediated the effects of workplace violence on depressive symptoms and burnout” – but in what way? For the BMJ Open: first published as 10.1136/bmjopen-2018-024186 on 24 May 2019. Downloaded from conclusion section it is going beyond the data to say that in China “most doctors encountered various types of workplace violence”: the authors are extrapolating their findings too much.

Strengths and limitations “This is the first study exploring mediating role of psychological capital in the associations of workplace violence with depressive symptoms and burnout.” Is not focused on the study methods/methodology so should be revised.

Introduction IBBN? Please spell out acronyms first.

“In China the ratio of doctor to general population (1:735) is lower than that in western countries (1:280-1:640)” – this range is really very large and maybe not significantly higher than the top end of 1:640. I think it is sufficient to give us the figures for China and emphasise that doctors are under strain due to insufficient human resources.

Lines 13 – 19 on Page 4 really needs some references.

Page 5: “sample size was relative small, and results were varied” – more information on this required.

Line 13; and lines 16-18 on page 5 – references please?

Line 4 on Page 6 – reference please?

VERSION 1 – AUTHOR RESPONSE http://bmjopen.bmj.com/

Reviewer: 1

Reviewer Name: Shailesh Kumar

Institution and Country: Waikato Clinical Campus, University of Auckland, Private bag 3200 Hamilton New Zealand

Q1: The paper needs significant language review. sentences are often structured poorly and sometimes difficult to on September 28, 2021 by guest. Protected copyright. understand (e.e Page 3 line 8; page 4-lines 4-5, 11-12 and 14-19).

Answer: Thank you very much for your advice. We improved the quality of language in our manuscript with the assistance of an English-speaking colleague.

Page 3 line 8: This study is impossible to draw causal relationship due to cross-sectional design. The revision was as followed:

Due to the cross-sectional design, this survey was not able to determine causality among variables. (Page 3, Line 10-11) page 4-lines 4-5: Doctors in direct contact with patients and patients’ families may be exposed to high levels of workplace violence. The revision was as followed:

It was well documented that doctors were vulnerable to suffer workplace violence. (Page 4, Line 4-5) page 4- lines 11-12: In China the ratio of doctor to general population (1:735) is lower than that in western countries (1:280-1:640), so human resources in hospital are insufficient. The revision was as followed:

BMJ Open: first published as 10.1136/bmjopen-2018-024186 on 24 May 2019. Downloaded from According to China Statistical Yearbook, the number of doctors was 2.31 per 1000 persons. A study reported that human resources in Chinese hospitals were insufficient, which may make doctors under strain. (Page 4, Line 10- 12) page 4- lines 14-19: These may lead to unharmonious doctor-patient relationship. In addition, in China patients undertake huge medical costs so that they often complain about it. At the same time, doctors get low benefits from medical technology due to strictly government-controlled conditions, and some Chinese doctors may provide unreasonable treatments for patients in order to make more profit. Though government takes some measures to deal with this issue, this still causes adverse effects. The revision was as followed:

And with the development of medicine, most doctors are busy with technical problems, and thus ignore patients’ inner feeling in clinical process[7]. In addition, in China patients often complain about huge medical costs[8]. Because the price of medical services is under government-controlled conditions, Chinese doctors get low benefits from medical services[9]. As a result, in order to make more profits, some doctors may provide unreasonable treatments for patients[9]. Though government takes some measures to deal with this issue, this still causes adverse effects[8]. (Page 4, Line 12-18)

If you do not feel satisfied with the revision, we will improve the quality of language in our manuscript with a professional copy-editing agency.

Q2: Page 5 why should only Chinese doctors be vulnerable - line 15 well-being spelling needs correction, I suggest authors should give a concise and clear definition of psychological capital early in the paper and explain how they protected respondents' privacy and anonymity especially when they had such high response rates.

Answer:

① Page 5 why should only Chinese doctors be vulnerable

Thank you very much for your question. We are very sorry that our description made you misunderstand. For the sentence “It has been reported that Chinese doctors are vulnerable to depressive symptoms and burnout”, we didn’t demonstrate that only Chinese doctors were vulnerable. We replaced “It has been reported that Chinese doctors are vulnerable to depressive symptoms and burnout” with “It has been reported that doctors are vulnerable to depressive symptoms and burnout”. We also added some references. http://bmjopen.bmj.com/ ② line 15 well-being spelling needs correction

Thank you very much for your advice. We are very sorry that we were careless.

We corrected it.

③ I suggest authors should give a concise and clear definition of psychological capital early in the paper on September 28, 2021 by guest. Protected copyright. Thank you very much for your advice. We gave a concise and clear definition of psychological capital in the abstract and introduction. The revision was as followed:

In the abstract:

Setting: Though workplace violence may lead to many adverse outcomes, there is not enough evidence to test the relationships between workplace violence, psychological capital which refers to an individual’s positive psychological state of development, depressive symptoms and burnout in Chinese doctors. (Page 2, Line 10-11)

In the introduction:

Psychological capital refers to an individual’s positive psychological state ofdevelopment [26]. (Page 5, Line 17)

④ explain how they protected respondents' privacy and anonymity especially when they had such high response rates.

Thank you very much for your question. All questionnaires were anonymous, which contributed to protecting respondents' privacy. The following factors may contribute to enhancing response rates.

Before the questionnaire survey, we first contacted the hospital to explain the purpose, significance, methods, and contents of the survey. On the premise of obtaining permission and active cooperation from hospitals, we carried BMJ Open: first published as 10.1136/bmjopen-2018-024186 on 24 May 2019. Downloaded from out this investigation. Investigators were trained in how to fill out the questionnaire, related techniques, etc. Investigators who passed the evaluation were hired. During the field investigation, the investigators shall explain the purpose, significance, content, and so on. Anonymous questionnaires were directly distributed to doctors after obtaining written informed consent. At the same time, the investigators described specific requirements for filling in the questionnaires.

If you do not feel satisfied with the revision and interpretation, we are pleased to looking forward to your further advice.

Q3: I would also like to read a bit more about the impact of deleting two major items of physical violence and vicarious violence and its impact on the overall quality of the tool.

Answer: Thank you very much for your question. For workplace violence scale designed by Schat et al., we deleted two items (“Has your personal property or workplace property been damaged by someone at work” ; “Has anyone threatened to damage any of your personal or workplace property while you’ve been at work? ”) in “physical violence at work”; and all items (“Have you seen any of your co-workers/managers experiencing violent events at work? ”;“Have you heard about any of your co-workers/managers experiencing violent events at work? ”;“Have you seen any co-workers/managers being threatened with physical violence at work? ”; “Have you heard about any co-workers/managers being threatened with physical violence at work? ”) in “vacarious violence at work”. Based on the definition of workplace violence from WHO and workplace violence scale from WHO and work violence scale adapted by Peixi Wang, we consulted with relevant experts and deleted them. These didn’t impact the overall quality of the tool. At the same time, we found that workplace violence scale in our study had good reliability and validity.

If you do not feel satisfied with the interpretation, we are pleased to looking forward to your further advice.

Reviewer: 2

Reviewer Name: Oriol Yuguero

Institution and Country: IRBLLeida

Q1: Perhaps a shorter introduction would be fine. http://bmjopen.bmj.com/

Answer: Thank you very much for your advice. We revised the introduction to make it shorter.

If you do not feel satisfied with the revision, we are pleased to looking forward to your further advice.

Q2: As a non-Chinese reader, I would like to know the situation of the professionals that may induce aggression towards them.

on September 28, 2021 by guest. Protected copyright. Answer: Thank you very much for your question. There are many reasons that may induce aggression towards professionals. We stated it in our manuscript.

According to China Statistical Yearbook, the number of doctors was 2.31 per 1000 persons[5]. A study reported that human resources in Chinese hospitals were insufficient, which may make doctors under strain[6]. And with the development of medicine, most doctors are busy with technical problems, and thus ignore patients’ inner feeling in clinical process[7]. In addition, in China patients often complain about huge medical costs[8]. Because the price of medical services is under government-controlled conditions, Chinese doctors get low benefits from medical services[9]. As a result, in order to make more profits, some doctors may provide unreasonable treatments for patients[9]. Though government takes some measures to deal with this issue, this still causes adverse effects[8]. All above facts may lead to unharmonious doctor-patient relationship, which may make Chinese doctors at high risk of workplace violence[10].(page 4,line 10-19)

If you do not feel satisfied with the interpretation, we are pleased to looking forward to your further advice.

Q3: In the Methodology section I would reduce the explanation of the sample size calculation. We already deduce that the calculation has been made correctly.

Answer: Thank you very much for your advice. We reduced the explanation of the sample size calculation in our manuscript as you asked. The revision was as followed:

Based on previous studies, the prevalence of depressive symptoms was 30%-60%; for burnout, standard deviations of emotional exhaustion, depersonalization and personal accomplishment were 11.76, 4.76 and 10.96, BMJ Open: first published as 10.1136/bmjopen-2018-024186 on 24 May 2019. Downloaded from respectively[29-31].Tolerance errors of emotional exhaustion and personal accomplishment were set to 1; tolerance error of personal accomplishment was set to 0.50; tolerance error of depressive symptoms was set to 0.03[32-33]. Finally according to the prevalence and tolerance error of depressive symptoms (p=30%, d=0.03, Z ɑ =1.96), sample size was 897. If design effect was set to 1.50 and effective response rate was set to 75%[34], 1793 individuals needed to be included. Therefore finally 1800 doctors were surveyed. (Page 6, Line 16-21; Page 7, Line 1-2)

If you do not feel satisfied with the revision, we are pleased to looking forward to your further advice.

Reviewer: 3

Reviewer Name: Lynn Monrouxe

Institution and Country: Chang Gung Medical Education Research Centre (CG-MERC) Chang Gung Memorial Hospital, Taiwan.

Q1: The Introduction does not quite set the study up adequately. I believe that some of this is to do with the English language that needs to be addressed, some to do with lack of references and some to do with the argument not building sufficiently.

Answer: Thank you very much for your advice. We improved the quality of language in our manuscript with the assistance of an English-speaking colleague. And we added some references and descriptions in the introduction. The revision was as followed:

① According to China Statistical Yearbook, the number of doctors was 2.31 per

1000 persons[5]. A study reported that human resources in Chinese hospitals were insufficient, which may make doctors under strain[6]. ( Page 4, Line 10-12) [5]http://www.stats.gov.cn/tjsj/ndsj/2017/indexch.htm (accessed 13 Oct 2018) [6]Anand S, Fan VY, Zhang J, et al. China's human resources for health: quantity, quality, and distribution. Lancet 2008;372(9651):1774-1781.

② And with the development of medicine, most doctors are busy with technical http://bmjopen.bmj.com/ problems, and thus ignore patients’ inner feeling in clinical process[7]. In addition, in China patients often complain about huge medical costs[8]. Because the price of medical services is under government-controlled conditions, Chinese doctors get low benefits from medical services[9]. As a result, in order to make more profits, some doctors may provide unreasonable treatments for patients[9]. Though government takes some measures to deal with this issue, this still causes adverse effects[8]. All above facts may lead to unharmonious doctor-patient relationship, which may make Chinese doctors at high risk of workplace violence[10]. ( Page 4, Line 12-19)

[7]Liu GZ, Zhang ZJ, Ma CH, et al. Investigation and analysis of clinical doctors’ humanity quality. Chin Hospi on September 28, 2021 by guest. Protected copyright. Manage 2012;32:17–19.

[8]Sun T, Gao L, Li F, et al. Workplace violence, psychological stress, sleep quality and subjective health in Chinese doctors: a large cross-sectional study. BMJ Open 2017;7(12):e017182.

[9]Ran L, Shuang X, Lu W. Analysis of 1645 cases irrational outpatient Chinese medicine prescriptions intervention. China Med Her 2014;11:154–158.

[10]Shi J, Jiang Y, Hu P, et al. A surveying study on social satisfaction to current doctor-patient relationship in China. JSSM 2015;08:695–702.

③ In most studies of workplace violence targeting Chinese doctors, sample size was relative small, and results were varied. A meta-analysis showed that the prevalence of workplace violence targeting doctors was 61.1% (95% CI [56.1%,

66.1%])[12]. ( Page 5, Line 2-3)

[12]Lu L, Dong M, Wang SB, et al. Prevalence of Workplace Violence Against Health-Care Professionals in China: A Comprehensive Meta-Analysis of Observational Surveys. Trauma Violence Abuse 2018. doi: 10.1177/1524838018774429.

④ Though the impacts of hospital violence on depressive symptoms and burnout

BMJ Open: first published as 10.1136/bmjopen-2018-024186 on 24 May 2019. Downloaded from have been explored, most studies involved medical staff or nurses[22-24]. ( Page 5, Line 10-11) [22]Zhao S, Xie F, Wang J, et al. Prevalence of Workplace Violence Against Chinese Nurses and Its Association with Mental Health: A Cross-sectional Survey. Arch Psychiatr Nurs 2018;32(2):242-247.

[24]Chen S, Lin S, Ruan Q, et al. Workplace violence and its effect on burnout and turnover attempt among Chinese medical staff. Arch Environ Occup Health 2016;71(6):330-337.

⑤ And in-depth research on both direct and indirect effects of workplace violence on depressive symptoms and burnout should be conducted to take more effective measures for improving mental health[ 13,14,15,25]. ( Page 5, Line 14-16) [13]Schat AC, Kelloway EK. Reducing the adverse consequences of workplace aggression and violence: the buffering effects of organizational support. J Occup Health Psychol 2003; 8(2): 110-122.

[14]Rogers KA, Kelloway EK. Violence at work: personal and organizational outcomes. J Occup Health Psychol 1997; 2(1): 63-71.

[15]Gong Y, Han T, Yin X, et al. Prevalence of depressive symptoms and work-related risk factors among nurses in public hospitals in southern china: a cross-sectional study. Sci Rep 2014; 4: 7109.

[25]Schat ACH, Kelloway EK. Effects of perceived control on the outcomes of workplace aggression and violence. J Occup Health Psychol 2000; 5(3): 386-402.

⑥ Based on above theory, individuals with high levels of psychological capital respond flexibly to changing demands and demonstrate emotional stability when faced with adversity[26-27]. ( Page 6, Line 2-3)

[26]Luthans F, Youssef CM. Human, social, and now positive psychological capital management: investing in http://bmjopen.bmj.com/ people for competitive advantage. ORGAN DYN 2004; 33(2):143-160.

[27]Luthans F, Youssef CM, Avolio BJ. Psychological capital: Developing the human competitive edge. Oxford University Press, Oxford, 2007.

If you do not feel satisfied with the revision, we are pleased to looking forward to your further advice.

Q2: Workplace violence was measured via a categorical scale but other measurements were (possibly) on a on September 28, 2021 by guest. Protected copyright. continuous scale – I confess that I am not a high-level statistician but mixing nominal data with ordinal/continuous data isn’t possible is it? I suggest someone who understands statistics gives a view here – I found the analysis and results section hard to follow.

Answer: Thank you very much for your advice. We consulted with an experienced statistical teacher and read some relevant statistical books. According to the “Medical Statistics”, ordinal variable and unordered polytomous variable may be entered into multiple linear regression model when they become multiple binary variables[1]. And Categorical variables and continuous variables may be entered into a model at the same time[1]. Workplace violence was binary variable. Therefore, workplace violence and other continuous variables may be entered into a model at the same time.

We made some revisions for analysis and results section. The revision was as followed:

2.5 Statistical analysis (page 10, line 17-21;page 11, line 1-21)

For the categorical variables, groups for which the response rate was less than 5% were merged. In our study, only 37(2.7%) participants belonged to the “widow/divorced/separated” group, and thus this group was combined with “single” group. Univariate analysis of depressive symptoms and burnout in relation to categorical variables was examined by t-tests and one-way ANOVAs. Univariate analysis of depressive symptoms and burnout in relation to continuous variables was tested by Pearson’s correlation analysis .

We performed hierarchical linear regression analysis to examine the effects of workplace violence and psychological capital on depressive symptoms and burnout. Except for gender and age, variables related to BMJ Open: first published as 10.1136/bmjopen-2018-024186 on 24 May 2019. Downloaded from depressive symptoms and burnout (P <0.05) in univariate analysis were entered in the model. In step 1 of the hierarchical linear regression analysis, demographic characteristics were used as predictors; in step 2, workplace violence was added; in step 3, psychological capital was added.

Based on the results of the hierarchical linear regression analyses, we further used the asymptotic and resampling strategies developed by Preacher and Hayes to examine the mediating role of psychological capital in the relationships between workplace violence and depressive symptoms and burnout. In the regression equation, workplace violence was used as a predictor, and depressive symptoms and burnout as outcomes ( Figure 1). The first step in the analysis was to explore the associations of workplace violence with depressive symptoms and burnout (the c path) and the second was to explore the mediating role of psychological capital (the a*b products). And c’ path coefficient represents the effects of workplace violence on depressive symptoms and burnout after psychological capital was added in the second step. When the c’ path coefficient was smaller compared to c path coefficient, or became statistically insignificant, mediating effect may exist. Based on 5000 bootstrap samples, a bias-corrected and accelerated 95% confidence interval (BCa 95% CI) of each a*b product was explored, and a BCa 95% CI without 0 meant a significant mediation. All of the above analyses were conducted using SPSS for 13.0. Two-tailed probability value of <0.05 was considered to be statistically significant.

3.Results (page 12, line 3-21; page 13, line 1-21; page 14, line 1 -5)

3.1 Characteristics of workplace violence among doctors

Characteristics of workplace violence among doctors were shown in Table 1. The prevalence of workplace violence was 77.5%. About 72.7% of participants encountered psychological aggression. Other prevalence of workplace violence from highest to lowest was: physical violence (33.7%), threat (31.1%), verbal sexual harassment (19.5%) and sexual assault(12.6%). And 34.7% of participants experienced one type of violence; 14.5% of participants encountered two types of violence; 14.5% of participants suffered three types of violence.

3.2 Univariate analysis of depressive symptoms and burnout in relation to categorical variables

(emotional exhaustion and depersonalization) (p<0.05). Educational level, weekly work time and division were significantly associated with depressive symptoms and burnout (p<0.05). The above results were shown in Table 2.

3.3 Univariate analysis of depressive symptoms and burnout in relation to continuous variables on September 28, 2021 by guest. Protected copyright.

3.4 Hierarchical linear regression analysis results, with depressive symptoms and job burnout as the criterion variable, respectively Workplace violence was positively associated with depressive symptoms (β=0.11, p<0.05) and two dimensions of burnout (emotional exhaustion : β=0.18, p<0.05; depersonalization: β=0.17, p<0.05). Psychological capital was negatively associated with depressive symptoms (β=-0.32, p<0.05) and two dimensions of burnout (emotional exhaustion: β=-0.23, p<0.05 ; depersonalization: β=-0.23, p<0.05). The above results were shown in Table 3.

3.5 Regression analysis results, with depressive symptoms and two dimensions of burnout (emotional exhaustion and depersonalization) as outcomes and psychological capital as a mediator

First, the associations of workplace violence with depressive symptoms and two dimensions of burnout (emotional exhaustion and depersonalization)(the c path) were determined. Positive associations of workplace violence with depressive symptoms and two dimensions of burnout were observed (p<0.05). The mediating role of psychological capital in the relationships of workplace violence with depressive symptoms and two dimensions of burnout was then estimated. Workplace violence was significantly associated with psychological capital (the a path, p<0.05). Psychological capital was significantly negatively associated with depressive symptoms and two dimensions of burnout after controlling for workplace violence(the BMJ Open: first published as 10.1136/bmjopen-2018-024186 on 24 May 2019. Downloaded from b path, p<0.05). The c’ path coefficient was still significant (p<0.05), but smaller compared to c path coefficient. Psychological capital significantly mediated the relationships of workplace violence with depressive symptoms(a*b = 1.61, BCa 95% CI: 1.08-2.25; p < 0.05) , emotional exhaustion (a*b = 1.29, BCa 95% CI: 0.86- 1.83; p < 0.05) and depersonalization(a*b = 0.70, BCa 95% CI: 0.45-1.00; p < 0.05). [1]Sun ZQ. Medical Statistics.the third[M]. People's Medical Publishing House, 2010

If you do not feel satisfied with the interpretation and revision, we are pleased to looking forward to your further advice.

Q3: In terms of the results section – some of my difficulty was with the reporting of the findings and the lack of any detail at all in the main body of the report – while I understand the authors wished to summarise their findings, it’s too brief and the reader is left wondering what was really found to be statistically significant.

Answer: Thank you very much for your advice. We are very sorry that we didn’t state the results very well. We added some contents in order to understand our result better. The revision was as followed:

3.Results (page 12, line 3-21; page 13, line 1-21; page 14, line 1 -5)

3.1 Characteristics of workplace violence among doctors

3.2 Univariate analysis of depressive symptoms and burnout in relation to categorical variables

The average age was (38.90 ± 8.81) years. Gender and rank were significantly associated with personal accomplishment (p<0.05). Marriage was significantly related to two dimensions of burnout (emotional exhaustion and personal accomplishment) (p<0.05). Income was significantly associated with depressive symptoms and personal accomplishment (p<0.05). Shift work, night shift and workplace violence were significantly associated with depressive symptoms and two dimensions of burnout (emotional exhaustion and depersonalization) (p<0.05). Educational level, weekly work time and division were significantly associated with depressive symptoms and http://bmjopen.bmj.com/ burnout (p<0.05). The above results were shown in Table 2.

3.3 Univariate analysis of depressive symptoms and burnout in relation to continuous variables

The correlation analysis showed that age was significantly correlated with personal accomplishment(r=0.07,p<0.05). Psychological capital was significantly associated with depressive symptoms(r=- 0.35,p<0.05) and burnout(emotional exhaustion : r=-0.28,p<0.05; depersonalization : r=-0.27,p<0.05; personal accomplishment: r=0.26,p<0.05). on September 28, 2021 by guest. Protected copyright.

3.5 Regression analysis results, with depressive symptoms and two dimensions of burnout (emotional exhaustion and depersonalization) as outcomes and psychological capital as a mediator

Psychological capital was significantly negatively associated with depressive symptoms and two dimensions of burnout after controlling for workplace violence(the b path, p<0.05). The c’ path coefficient was still significant (p<0.05), but smaller compared to c path coefficient. Psychological capital significantly mediated the relationships of workplace violence with depressive symptoms(a*b = 1.61, BCa 95% CI: 1.08-2.25; p < 0.05) , emotional BMJ Open: first published as 10.1136/bmjopen-2018-024186 on 24 May 2019. Downloaded from exhaustion (a*b = 1.29, BCa 95% CI: 0.86-1.83; p < 0.05) and depersonalization(a*b = 0.70, BCa 95% CI: 0.45- 1.00; p < 0.05). If you do not feel satisfied with the revision, we are pleased to looking forward to your further advice. Q4: In the discussion section - that “psychological capital mediated the effects of workplace violence on depressive symptoms and two dimensions of burnout” is, of course, important. But then the authors go on to say that Doctors who “suffered workplace violence would be more likely to experience lower levels of psychological capital, which in turn increased the levels of depressive symptoms and burnout.” doesn’t actually follow logically. Answer: Thank you very much for your question. The mediator function of a third variable represents the generative mechanism through which the focal independent variable is able to influence the dependent variable of interest[1].

Psychological capital

Workplace violence Depressive symptoms, burnout

Figure 1: Theoretical model of the mediating role of psychological capital on the relationships between workplace violence and depressive symptoms, burnout.

Psychological capital mediated the effects of workplace violence on depressive symptoms and burnout. In other words, workplace violence increased the level of depressive symptoms and burnout of doctors through damaging psychological capital.

The expression of the sentence (“doctors who suffered workplace violence would be more likely to experience lower levels of psychological capital, which in turn increased the levels of depressive symptoms and burnout”) may be inappropriate. Therefore we replaced “doctors who suffered workplace violence would be more

likely to experience lower levels of psychological capital, which in turn increased the levels of depressive symptoms and burnout.” with “workplace violence could increase the level of depressive symptoms and burnout of http://bmjopen.bmj.com/ doctors through damaging their psychological capital”.(page 15,line 21; page 16,line 1-2)

[1]Baron RM, Kenny DA. The moderator-mediator variable distinction in social psychological research: conceptual, strategic, and statistical considerations. J Pers Soc Psychol 1986;51(6):1173-82.

If you do not feel satisfied with the interpretation and revision, we are pleased to looking forward to your further advice.

on September 28, 2021 by guest. Protected copyright. Q5: Following this the authors say “Therefore it may be a effective strategy to develop some programs to increase psychological capital, thus decreasing doctors’ depressive symptoms and burnout in the long run” – again – this is too much of a claim based on the authors’ data and too simplistic to have ‘some programs to increase psychological capital’ (so many constructs involved) and then depressive symptoms and burnout are decreased – I can’t see this being the case as these are really complex mental illness phenomena.

Answer: Thank you very much for your question. The mediator function of a third variable represents the generative mechanism through which the focal independent variable is able to influence the dependent variable of interest[1].

Psychological capital

Workplace violence Depressive symptoms, burnout

Figure 1: Theoretical model of the mediating role of psychological capital on the relationships between workplace violence and depressive symptoms, burnout.

BMJ Open: first published as 10.1136/bmjopen-2018-024186 on 24 May 2019. Downloaded from We are very sorry that our expression was not appropriate. We replaced “Therefore it may be a effective strategy to develop some programs to increase psychological capital, thus decreasing doctors’ depressive symptoms and burnout in the long run” with “Efforts should be made to develop interventions of improving psychological capital to reduce doctor’s depressive symptoms and burnout”. (page16, line2-3) We added the description of psychological capital intervention. “In 2005, Luthans et al. put forward psychological capital intervention model which improved the level of psychological capital in four aspects including self-efficacy, hope, resilience and optimism.” (page 16,line 7-8)

If you do not feel satisfied with the interpretation and revision, we are pleased to looking forward to your further advice.

Q6: Despite the authors admitting that their study is such that it is impossible to draw causal relationships – they continually attempt to do such a thing. Caution should be taken here.

Answer: Thank you very much for your advice. Cross-sectional study design was not able to determine causality among variables. We firstly did a cross-sectional study to explore their relationship. Next we plan to do a longitudinal study to further draw causality.

If you do not feel satisfied with the interpretation, we are pleased to looking forward to your further advice.

Q7: The authors have filled in the STROBE-Vet which is an extension of the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) statement. However, their study is not an observational study in epidemiology (hence they responded ‘no’ to many items). I suggest they switch to using: Kelley K, Clark B, Brown V, Sitzia J. Good practice in the conduct and reporting of survey research. Int J Qual Health Care. 2003;15(3):261-266.

Answer: thank you very much for your advice. We are very sorry that we misused “STROBE-VET checklist”. We replaced it with “STROBE Statement—Checklist of items that should be included in reports of cross-sectional studies” (https://strobe-statement.org/index.php?id=strobe-home).The revision was as followed:

STROBE Statement—Checklist of items that should be included in reports of cross-sectional studies http://bmjopen.bmj.com/

Item page No Recommendation

Title and abstract 1 (a) Indicate the study’s design with a commonly used term in the Page 2,line 8

title or the abstract on September 28, 2021 by guest. Protected copyright. (b) Provide in the abstract an informative and balanced summary of Page 2,line what was done and what was found 4-21

Introduction

Background/rationale 2 Explain the scientific background and rationale for the investigation Page 4,line being reported 1-21; Page 5,line 1-21; Page 6,line 1-9 Objectives 3 State specific objectives, including any prespecified hypotheses Page 6,line 10-13

Methods

Study design 4 Present key elements of study design early in the paper Page 7,line 9 Setting 5 Describe the setting, locations, and relevant dates, including periods Page 7,line of recruitment, exposure, follow-up, and data collection 9-18 BMJ Open: first published as 10.1136/bmjopen-2018-024186 on 24 May 2019. Downloaded from

Participants 6 (a) Give the eligibility criteria, and the sources and methods of Page 7,line selection of participants 9-18 Variables 7 Clearly define all outcomes, exposures, predictors, potential Page confounders, and effect modifiers. Give diagnostic criteria, if 7,line19-21; applicable Page 8,line 1-21; Page 9,line 1-21; Page 10,line 1-15 Data sources/ 8* For each variable of interest, give sources of data and details of Page measurement methods of assessment (measurement). Describe comparability of 7,line19-21; assessment methods if there is more than one group Page 8,line 1-21; Page 9,line 1-21; Page 10,line 1-15 Bias 9 Describe any efforts to address potential sources of bias Page 3,line 12-14 Study size 10 Explain how the study size was arrived at Page 6,line 15-21;page 7,line 1-2 Quantitative variables 11 Explain how quantitative variables were handled in the analyses. If Page http://bmjopen.bmj.com/ on September 28, 2021 by guest. Protected copyright. BMJ Open: first published as 10.1136/bmjopen-2018-024186 on 24 May 2019. Downloaded from applicable, describe which groupings were chosen and why 7,line19-21; Page 8,line 1-18; Page 10,line 1-7 Statistical methods 12 (a) Describe all statistical methods, including those used to control Page 10,line 17- for confounding 21; Page 11,line 1-21 (b) Describe any methods used to examine subgroups and Page 10,line 17- interactions 21; Page 11,line 1-21 (c) Explain how missing data were addressed no (d) If applicable, describe analytical methods taking account of Page 11,line sampling strategy 8-21 (e ) Describe any sensitivity analyses no Results

Participants 13* (a) Report numbers of individuals at each stage of study—eg Page 7,line 18- numbers potentially eligible, examined for eligibility, confirmed 19; eligible, included in the study, completing follow-up, and analysed

http://bmjopen.bmj.com/ (b) Give reasons for non-participation at each stage no (c) Consider use of a flow diagram no Descriptive data 14* (a) Give characteristics of study participants (eg demographic, Page 24;Page clinical, social) and information on exposures and potential 25 confounders (b) Indicate number of participants with missing data for each no

variable of interest on September 28, 2021 by guest. Protected copyright. Outcome data 15* Report numbers of outcome events or summary measures Page 24;Page 25 Main results 16 (a) Give unadjusted estimates and, if applicable, Page 27;Page confounder-adjusted estimates and their precision (eg, 95% 28 confidence interval). Make clear which confounders were adjusted for and why they were included (b) Report category boundaries when continuous variables were no categorized (c) If relevant, consider translating estimates of relative risk into no absolute risk for a meaningful time period Other analyses 17 Report other analyses done—eg analyses of subgroups and no interactions, and sensitivity analyses

Discussion

Key results 18 Summarise key results with reference to study objectives Page 14,line 13;page 14,line BMJ Open: first published as 10.1136/bmjopen-2018-024186 on 24 May 2019. Downloaded from

http://bmjopen.bmj.com/ on September 28, 2021 by guest. Protected copyright. BMJ Open: first published as 10.1136/bmjopen-2018-024186 on 24 May 2019. Downloaded from

18-20;page 15,line 13-15,19-21 Discuss limitations of the study, taking into account Limitations 19 sources of Page16,line potential bias or imprecision. Discuss both direction and magnitude 13-18 of any potential bias Give a cautious overall interpretation of results Page 14; Interpretation 20 considering Page objectives, limitations, multiplicity of analyses, results from similar 15; Page 16 studies, and other relevant evidence Discuss the generalisability (external validity) of the study Generalisability 21 results no Other information

Give the source of funding and the role of the Funding 22 funders for the no present study and, if applicable, for the original study on which the present article is based

*Give information separately for exposed and unexposed groups.

Note: An Explanation and Elaboration article discusses each checklist item and gives methodological background and published examples of transparent reporting. The STROBE checklist is best used in conjunction with this article (freely available on the Web sites of PLoS Medicine at http://www.plosmedicine.org/, Annals of Internal Medicine at http://www.annals.org/, and http://bmjopen.bmj.com/ Epidemiology at http://www.epidem.com/). Information on the STROBE Initiative is available at www.strobe-statement.org.

If you do not feel satisfied with the revision, we are pleased to looking forward to your further advice.

Q8: I suggest the authors fond someone to English Edit their work.

Answer: Thank you very much for your advice. We improved the quality of language in our manuscript on September 28, 2021 by guest. Protected copyright. with the assistance of an English-speaking colleague. If you do not feel satisfied with the revision, we will improve the quality of language in our manuscript with a professional copy-editing agency.

Q9: Abstract

Please briefly define psychological capital as it is not clear. The design just says it’s a “cross sectional study” but of what kind? Many different qualitative and quantitative studies can be cross sectional – can the authors please be more specific. Also, the authors state “Psychological capital mediated the effects of workplace violence on depressive symptoms and burnout” – but in what way? For the conclusion section it is going beyond the data to say that in China “most doctors encountered various types of workplace violence”: the authors are extrapolating their findings too much.

Answer:

① Please briefly define psychological capital as it is not clear.

Thank you very much for your advice. We gave a concise and clear definition of psychological capital in the abstract. The revision was as followed:

BMJ Open: first published as 10.1136/bmjopen-2018-024186 on 24 May 2019. Downloaded from Setting: Though workplace violence may lead to many adverse outcomes, there is not enough evidence to test the relationships between workplace violence, psychological capital which refers to an individual’s positive psychological state of development, depressive symptoms and burnout in Chinese doctors. (Page 2, Line 9-12)

② The design just says it’s a “cross sectional study” but of what kind? Many different qualitative and quantitative studies can be cross sectional – can the authors please be more specific.

Thank you very much for your advice. We revised it as you asked. The revision was “a quantitative, cross-sectional study”. (Page 2, Line 8)

③ Also, the authors state “Psychological capital mediated the effects of workplace violence on depressive symptoms and burnout” – but in what way?

Thank you very much for your question. The mediator function of a third variable represents the generative mechanism through which the focal independent variable is able to influence the dependent variable of interest[1].

Psychological capital

Workplace violence Depressive symptoms, burnout

Figure 1: Theoretical model of the mediating role of psychological capital on the relationships between workplace violence and depressive symptoms, burnout.

Figure 1: Theoretical model of the mediating role of psychological capital on the relationships between workplace violence and depressive symptoms, burnout. http://bmjopen.bmj.com/ Psychological capital mediated the effects of workplace violence on depressive symptoms and burnout. In other words, workplace violence increased the level of depressive symptoms and burnout of doctors through damaging psychological capital.

We replaced “Psychological capital mediated the effects of workplace violence on depressive symptoms and burnout” with “workplace violence increased the level of depressive symptoms and burnout through damaging psychological capital”. ( Page 2, Line 17-19)

on September 28, 2021 by guest. Protected copyright. [1]Baron RM, Kenny DA. The moderator-mediator variable distinction in social psychological research: conceptual, strategic, and statistical considerations. J Pers Soc Psychol 1986;51(6):1173- 82.

④ For the conclusion section it is going beyond the data to say that in China “most doctors encountered various types of workplace violence”: the authors are extrapolating their findings too much.

Thank you very much for your advice. We are very sorry that we didn’t state it very well. We revised it. The revision was as followed: In China, most doctors might be exposed to workplace violence, especially psychological aggression.( Page 2, Line 20-21)

If you do not feel satisfied with the revision and interpretation, we are pleased to looking forward to your further advice.

Q10: Strengths and limitations

BMJ Open: first published as 10.1136/bmjopen-2018-024186 on 24 May 2019. Downloaded from “ This is the first study exploring mediating role of psychological capital in the associations of workplace violence with depressive symptoms and burnout.” Is not focused on the study methods/methodology so should be revised.

Answer: thank you very much for your advice. We are very sorry that we didn’t state it very well. We had revised it as you asked. The revision was as followed:

3.We used the asymptotic and resampling strategies developed by Preacher and Hayes to examine the mediating role of psychological capital in the relationships between workplace violence and depressive symptoms and burnout. ( Page 3, Line 7-9)

If you do not feel satisfied with the revision, we are pleased to looking forward to your further advice.

Q11: Introduction

IBBN? Please spell out acronyms first.

Answer: Thank you very much for your question. We are very sorry that we made a mistake. It is not IBBN. We replaced it with “International Labour Office/International Council of Nurses/World Health Organization/Public Services International”. ( Page 4, Line 5-6)

Q12: “In China the ratio of doctor to general population (1:735) is lower than that in western countries (1:280-1:640)” – this range is really very large and maybe not significantly higher than the top end of 1:640. I think it is sufficient to give us the figures for China and emphasise that doctors are under strain due to insufficient human resources.

Answer：Thank you very much for your advice. We are very sorry that we didn’t state it very well. We got some data from China Statistical Yearbook. And we also cited the reference to demonstrate that human resources in hospitals were insufficient, which may make doctors under strain. The revision was as followed:

According to China Statistical Yearbook, the number of doctors was 2.31 per 1000 persons[5]. A

study reported that human resources in Chinese hospitals were insufficient, which may make doctors http://bmjopen.bmj.com/ under strain[6]. ( Page 4, Line 10-12)

[5]http://www.stats.gov.cn/tjsj/ndsj/2017/indexch.htm (accessed 13 Oct 2018) [6]Anand S, Fan VY, Zhang J, et al. China's human resources for health: quantity, quality, and distribution. Lancet 2008;372(9651):1774-1781.

If you do not feel satisfied with the revision, we are pleased to looking forward to your further advice.

Q14: Lines 13 – 19 on Page 4 really needs some references. on September 28, 2021 by guest. Protected copyright.

Answer: thank you very much for your advice. We added some references as you asked. The revision was as followed:

[7]Liu GZ, Zhang ZJ, Ma CH, et al. Investigation and analysis of clinical doctors’ humanity quality. Chin Hospi Manage 2012;32:17–19.

[8]Sun T, Gao L, Li F, et al. Workplace violence, psychological stress, sleep quality and subjective health in Chinese doctors: a large cross-sectional study. BMJ Open 2017;7(12):e017182. BMJ Open: first published as 10.1136/bmjopen-2018-024186 on 24 May 2019. Downloaded from

[9]Ran L, Shuang X, Lu W. Analysis of 1645 cases irrational outpatient Chinese medicine prescriptions intervention. China Med Her 2014;11:154–158.

[10]Shi J, Jiang Y, Hu P, et al. A surveying study on social satisfaction to current doctor-patient relationship in China. JSSM 2015;08:695–702.

If you do not feel satisfied with the revision, we are pleased to looking forward to your further advice.

Q5: Page 5: “sample size was relative small, and results were varied” – more information on this required.

Answer: Thank you very much for your advice. We added some information. The revision was as followed:

In most studies of workplace violence targeting Chinese doctors, sample size was relative small, and results were varied. A meta-analysis showed that the prevalence of workplace violence targeting doctors was 61.1% (95% CI [56.1%, 66.1%])[12]. ( Page 5, Line 2-3)

[12]Lu L, Dong M, Wang SB, et al. Prevalence of Workplace Violence Against Health-Care Professionals in China: A Comprehensive Meta-Analysis of Observational Surveys. Trauma Violence Abuse 2018. doi:

10.1177/1524838018774429.

If you do not feel satisfied with the revision, we are pleased to looking forward to your further advice.

Q6: Line 13; and lines 16-18 on page 5 – references please?

Answer: Thank you very much for your advice. We added the references.

Line 13 on page 5

http://bmjopen.bmj.com/ Though the impacts of hospital violence on depressive symptoms and burnout have been explored, most studies involved medical staff or nurses[22-24]. ( Page 5, Line 10-12) [22]Zhao S, Xie F, Wang J, et al. Prevalence of Workplace Violence Against Chinese Nurses and Its Association with Mental Health: A Cross-sectional Survey. Arch Psychiatr Nurs 2018;32(2):242-247.

[23]Liu W, Zhao S, Shi L, et al. Workplace violence, job satisfaction, burnout, perceived organisational support and their effects on turnover intention among Chinese nurses in tertiary hospitals: a cross- sectional study. BMJ Open 2018;8(6):e019525. on September 28, 2021 by guest. Protected copyright. [24]Chen S, Lin S, Ruan Q, et al. Workplace violence and its effect on burnout and turnover attempt among Chinese medical staff. Arch Environ Occup Health 2016;71(6):330-337.

Lines 16-18 on page 5

And in-depth research on both direct and indirect effects of workplace violence on depressive symptoms and burnout should be conducted to take more effective measures for improving mental health[ 13,14,15,25]. ( Page 5, Line 14-16)

[13]Schat AC, Kelloway EK. Reducing the adverse consequences of workplace aggression and violence: the buffering effects of organizational support. J Occup Health Psychol 2003; 8(2): 110-122.

[14]Rogers KA, Kelloway EK. Violence at work: personal and organizational outcomes. J Occup Health Psychol 1997; 2(1): 63-71.

[15]Gong Y, Han T, Yin X, et al. Prevalence of depressive symptoms and work-related risk factors among nurses in public hospitals in southern china: a cross-sectional study. Sci Rep 2014; 4: 7109.

BMJ Open: first published as 10.1136/bmjopen-2018-024186 on 24 May 2019. Downloaded from [25]Schat ACH, Kelloway EK. Effects of perceived control on the outcomes of workplace aggression and violence. J Occup Health Psychol 2000; 5(3): 386-402.

If you do not feel satisfied with the revision, we are pleased to looking forward to your further advice.

Q7: Line 4 on Page 6 – reference please?

Answer: Thank you very much for your advice. We added the references.

Based on above theory, individuals with high levels of psychological capital respond flexibly to changing demands and demonstrate emotional stability when faced with adversity[26-27]. ( Page 6, Line 1-3)

[26]Luthans F, Youssef CM. Human, social, and now positive psychological capital management: investing in people for competitive advantage. ORGAN DYN 2004; 33(2):143-160.

[27]Luthans F, Youssef CM, Avolio BJ. Psychological capital: Developing the human competitive edge. Oxford University Press, Oxford, 2007.

If you do not feel satisfied with the revision, we are pleased to looking forward to your further advice.

We must thank you and all reviewers and editor for these critical comments, which help us with the improvement of our manuscript. Based on these comments, careful modifications have been made to our manuscript. It would be highly appreciated if you are kind to give a favorable consideration on it. We look forward to hearing from you.

If there are other errors or further requests, please contact us by e-mail.

VERSION 2 – REVIEW

REVIEWER Shailesh Kumar

Private bag 3200 Hamilton 3210 New Zealand http://bmjopen.bmj.com/ REVIEW RETURNED 30-Oct-2018

GENERAL COMMENTS changes are acceptable and have improved overall quality of the paper

REVIEWER Oriol Yuguero on September 28, 2021 by guest. Protected copyright. Institut de Recerca Biomedica de Lleida REVIEW RETURNED 03-Nov-2018

GENERAL COMMENTS I am pleased with all the changes made in the manuscript. Great job.

REVIEWER Lynn Monrouxe CG-MERC, Chang Gung Memorial Hospital, Taiwan REVIEW RETURNED 20-Dec-2018

GENERAL COMMENTS ABSTRACT The abstract needs further work. The content of the sections do not match the headings. For example, the 'Setting' section needs BMJ Open: first published as 10.1136/bmjopen-2018-024186 on 24 May 2019. Downloaded from to talk about setting - rather than giving the reader background information. The 'Objectives' set out three research questions - but I think the third question is part of the second.

INTRODUCTION This require further work. Firstly, I suggest the authors find an English editor to look at the grammar and the style of writing to make it more commensurate with the journal. Also, it comprises a list of features (definitions/claims/studies) that are not well integrated so hard to get the real picture. It doesn't yet tall the story - what is the problem and why is it problematic (yes - any workplace violence is problematic - but given that we already know "It was well documented that doctors were vulnerable to suffer workplace violence" - why is this study needed? So the authors need to work more at hooking the reader in early on. Yes, the second paragraph provides a list of key issues - shortage of doctors, high medical prices and patient complaints, lack of empathy and poor doctor-patient relationships, doctors' mental health concerns and so on - but it feels disjointed.

A quick google scholar search of hospital workplace violence produces many interesting studies that are omitted from the introduction but that are quite key articles: https://scholar.google.com.tw/scholar?as_ylo=2014&q=hospital+w orkplace+violence+statistics&hl=en&as_sdt=0,5&as_vis=1

At the end of introduction, the purposes of the study are specified as follows as follows (with my questions after each one): (1) identify the prevalence of workplace violence among doctors in Liaoning, China; - But why? Why is this important? (2) examine the relationships between workplace violence, psychological capital and depressive symptoms and burnout of Chinese doctors; - yes - this is interesting - and I feel that this http://bmjopen.bmj.com/ should be the focus of the introduction - and any other studies internationally that have looked at associations of factors with workplace violence should be the 'story' of the introduction. (3) explore the mediating role of psychological capital. - but surely - this is embedded in the RQ2?

SAMPLE SIZE This section is confusing - new research is alluded to - and the reader is left bemused. What the authors are doing is trying to on September 28, 2021 by guest. Protected copyright. justify their sample size but it seems laboured and confusing and one wonders if this was really how the justification came about or whether this is a post-hoc explanation. When we get to the section on STUDY POPULATION I think we are getting to the real rationale for the sample.

QUESTIONNAIRES This section contains some results and it quite hard to follow. Also - it's hard to understand why these particular inventories are used - it's likely that they have been used in previous research so they are good comparable measures but the reader cannot tell.

STATISTICAL ANALYSES This section is also confusing. I kept with it but then when we reached the regression part it wold have been good to just tell the reader that a regression was done and the details then would be in the results section.

BMJ Open: first published as 10.1136/bmjopen-2018-024186 on 24 May 2019. Downloaded from How were missing data addressed?

Did the authors undertake any sub-group analyses? So what about any effect of specialty? Given the plethora of research that suggests different specialties having different cultural profiles in terms of violence - this would be an interesting analysis. There are some data reported in the Tables but it's hard to make sense of - so something in the results section would be ideal.

RESULTS There are a lot of results and sometimes it gets quite confusing. I wonder whether they can be described in a 'lay' manner after each set - so the reader can work out exactly what this issues are - with so many constructs it's hard to remember what each means.

DISCUSSION

the sentence "The sample size (1392 doctors) was large and the effective responsive rate in this study was 77.3% greater than 70% that was considered to be acceptable for questionnaire survey[40]. These may increase the generalization of our study conclusion." - please explain further - generalisation to whom? Across the whole of the Liaoning Province? To other provinces? Across China? To other country settings?

For many findings - there was the statement that the findings echoed other research - and I was beginning to wonder "what does this study add"? But then the authors state "This is the first study to examine the mediating role of psychological capital in the relationships of workplace violence with depressive symptoms and burnout". If this is the case, then this needs to be up-front and take president in the discussion section. The discussion section needs to begin by describing the novel findings - or at least flagging them http://bmjopen.bmj.com/ up at the start so the reader is primed.

The limitation section is rather brief - more can be said about issues such as generalisability, study engagement and so on.

Review 3

1.The abstract needs further work. The content of the sections do not match the headings. For example, the 'Setting' section needs to talk about setting - rather than giving the reader background information. The 'Objectives' set out three research questions - but I think the third question is part of the second.

Answer: Thank you very much for your advice. We had revised it as you asked.

'Setting' section: Eight hospitals in Liaoning Province were surveyed using a self-reported questionnaire. (page 2, line 9-10)

'Objectives' section: The purpose of this study was to (1) identify the prevalence of workplace violence among doctors in Liaoning, China; and (2) examine the relationships between workplace violence and psychological capital with depressive symptoms and burnout in Chinese doctors. (page 2, line 4-7) BMJ Open: first published as 10.1136/bmjopen-2018-024186 on 24 May 2019. Downloaded from If you do not feel satisfied with the revision, we are pleased to looking forward to your further advice.

2. INTRODUCTION

This require further work. Firstly, I suggest the authors find an English editor to look at the grammar and the style of writing to make it more commensurate with the journal.

Answer: Thank you very much for your advice. We had improved the quality of language in our manuscript with a professional copy-editing agency .

3. INTRODUCTION

Also, it comprises a list of features (definitions/claims/studies) that are not well integrated so hard to get the real picture. It doesn't yet tall the story - what is the problem and why is it problematic (yes - any workplace violence is problematic - but given that we already know "It was well documented that doctors were vulnerable to suffer workplace violence" - why is this study needed? So the authors need to work more at hooking the reader in early on. Yes, the second paragraph provides a list of key issues - shortage of doctors, high medical prices and patient complaints, lack of empathy and poor doctor-patient relationships, doctors' mental health concerns and so on - but it feels disjointed.

A quick google scholar search of hospital workplace violence produces many interesting studies that are omitted from the introduction but that are quite key articles: https://scholar.google.com.tw/scholar?as_ylo=2014&q=hospital+workplace+violence+statistics&hl=en &as_sdt=0,5&as_vis=1

Answer: Thank you very much for your advice. Your advice is very helpful. The sentence "It was well documented that doctors were vulnerable to suffer workplace violence" may be not appropriate and was deleted. We tried our best to search more literatures. We added some sentences and refernces in the introduction. (page 4, line1-22；page 5, line1-9)

what is the problem: hospital violence http://bmjopen.bmj.com/ why is it problematic: Workplace violence was defined as “incidents where staff were abused, threatened or assaulted in circumstances related to their work, including commuting to and from work, involving an explicit or implicit threat to their safety, well-being or health”[1]. International Labour Office,International Council of Nurses, World Health Organization (WHO), and Public Services International have reported that approximately one-quarter of the total number of violent incidents in all workplaces occurs in a hospital[2]. The Occupational Safety and Health Administration reported on September 28, 2021 by guest. Protected copyright. that from 2011 to 2013, the number of workplace violence incidents averaged nearly 24000 annually, with approximately 75% occurring in the health care setting[3]. In Turkey, 78.1% of doctors suffered workplace violence[4]. In Iran, almost 90% of residents experienced violence[5]. Hospital violence has become an increasingly serious public health problem worldwide and is also common in China, which has attracted considerable attention.

According to the China Statistical Yearbook, the number of doctors is 2.31 per 1000 persons[6]. A study reported that human resources in Chinese hospitals were insufficient, which may cause strain on doctors[7]. With the development of medicine, most doctors are busy with technical chanllenges and thus ignore patients’ inner feelings in clinical settings[8]. In addition, in China patients often complain about extensive medical costs[9]. Because the price of medical services is under governmental control, Chinese doctors receive low benefits from medical services[10]. As a result, to make more profits, some doctors may provide unnecessary treatments for patients[10]. Although the government has taken some measures to address this issue, it still causes adverse effects[9]. All of the above factors may lead to an unharmonious doctor-patient relationship, which may make Chinese doctors at high risk of workplace violence[11]. Sun et al. found that 83.4% of doctors experienced workplace violence from 30 provinces of China [9]. Yao et al. reported that 63.2% of doctors suffered BMJ Open: first published as 10.1136/bmjopen-2018-024186 on 24 May 2019. Downloaded from workplace violence in Henan Province, China[12]. The results of previous studies varied considerably[13]. Inconsistency in the existing data may make interpretation of the results quite difficult[14]. A meta-analysis showed that the prevalence of workplace violence targeting doctors was 61.1% (95% CI [56.1%, 66.1%])[13]. This study further estimated the prevalence of workplace violence among doctors in China.

If you do not feel satisfied with the revision and interpretation, we are pleased to looking forward to your further advice.

4. INTRODUCTION

At the end of introduction, the purposes of the study are specified as follows as follows (with my questions after each one):

(1) identify the prevalence of workplace violence among doctors in Liaoning, China; - But why? Why is this important?

(2) examine the relationships between workplace violence, psychological capital and depressive symptoms and burnout of Chinese doctors; - yes - this is interesting - and I feel that this should be the focus of the introduction - and any other studies internationally that have looked at associations of factors with workplace violence should be the 'story' of the introduction.

(3) explore the mediating role of psychological capital. - but surely - this is embedded in the RQ2?

Answer:Thank you very much for your question.

(1) identify the prevalence of workplace violence among doctors in Liaoning, China; - But why? Why is this important? (page 4, line1-22；page 5, line1-9)

Workplace violence was defined as “incidents where staff were abused, threatened or assaulted in

circumstances related to their work, including commuting to and from work, involving an explicit or http://bmjopen.bmj.com/ implicit threat to their safety, well-being or health”[1]. International Labour Office,International Council of Nurses, World Health Organization (WHO), and Public Services International have reported that approximately one-quarter of the total number of violent incidents in all workplaces occurs in a hospital[2]. The Occupational Safety and Health Administration reported that from 2011 to 2013, the number of workplace violence incidents averaged nearly 24000 annually, with approximately 75% occurring in the health care setting[3]. In Turkey, 78.1% of doctors suffered workplace violence[4]. In Iran, almost 90% of residents experienced violence[5]. Hospital violence has become an increasingly on September 28, 2021 by guest. Protected copyright. serious public health problem worldwide and is also common in China, which has attracted considerable attention.

According to the China Statistical Yearbook, the number of doctors is 2.31 per 1000 persons[6]. A study reported that human resources in Chinese hospitals were insufficient, which may cause strain on doctors[7]. With the development of medicine, most doctors are busy with technical chanllenges and thus ignore patients’ inner feelings in clinical settings[8]. In addition, in China patients often complain about extensive medical costs[9]. Because the price of medical services is under governmental control, Chinese doctors receive low benefits from medical services[10]. As a result, to make more profits, some doctors may provide unnecessary treatments for patients[10]. Although the government has taken some measures to address this issue, it still causes adverse effects[9]. All of the above factors may lead to an unharmonious doctor-patient relationship, which may make Chinese doctors at high risk of workplace violence[11]. Sun et al. found that 83.4% of doctors experienced workplace violence from 30 provinces of China [9]. Yao et al. reported that 63.2% of doctors suffered workplace violence in Henan Province, China[12]. The results of previous studies varied considerably[13]. Inconsistency in the existing data may make interpretation of the results quite difficult[14]. A meta-analysis showed that the prevalence of workplace violence targeting doctors was BMJ Open: first published as 10.1136/bmjopen-2018-024186 on 24 May 2019. Downloaded from 61.1% (95% CI [56.1%, 66.1%])[13]. This study further estimated the prevalence of workplace violence among doctors in China.

Yes, “the relationships between workplace violence, psychological capital and depressive symptoms and burnout in Chinese doctors” was the focus of the introduction.

(3) explore the mediating role of psychological capital. - but surely - this is embedded in the RQ2?

The mediating role of psychological capital had been embedded in the RQ2. If you do not feel satisfied with the revision and interpretation, we are pleased to looking forward to your further advice.

5.SAMPLE SIZE

This section is confusing - new research is alluded to - and the reader is left bemused. What the authors are doing is trying to justify their sample size but it seems laboured and confusing and one wonders if this was really how the justification came about or whether this is a post-hoc explanation. When we get to the section on STUDY POPULATION I think we are getting to the real rationale for the sample.

Answer: Thank you very much for your question. We are very sorry that the description was confusing. We transferred the “sample size” to “study population”. Before the survey, we estimated the sample size based on previous studies.Thus based on estimated sample size, we started the investigation.The revision was as followed: (page 6, line 17-22；page 7, line 1-16)

2.1Study population http://bmjopen.bmj.com/ Based on previous studies, the prevalence of depressive symptoms was 30%-60%; for burnout, the standard deviations of emotional exhaustion, depersonalization and personal accomplishment were 11.76, 4.76 and 10.96, respectively[29-31]. Tolerance errors of emotional exhaustion and personal accomplishment were set to 1; the tolerance error of personal accomplishment was set to 0.50, and the tolerance error of depressive symptoms was set to 0.03[32-33]. Finally according to the prevalence and tolerance error of depressive symptoms (p=30%, d=0.03, Zɑ=1.96), the sample size was 897. If the design effect was set to 1.50 and the effective response rate was set to 75%[34], 1793 on September 28, 2021 by guest. Protected copyright. individuals needed to be included. Therefore, 1800 doctors were surveyed.

During June-July 2015, a cross-sectional survey was conducted in Liaoning Province. According to the geographic division of Liaoning Province, the entire province was divided into five regions. One city in each region was randomly selected (Dandong, Fuxin, Anshan, Fushun and Shenyang). Based on the number of “grade two and above” general hospitals (>100 beds) and population size, one “grade two and above” general hospital was randomly selected in Dandong, Fuxin, Anshan and Fushun, respectively, and four “grade two and above” general hospitals were randomly selected in Shenyang (a total of 8 hospitals). Based on the estimated sample size, 225 (1800 / 8) individuals who worked for 12 months or longer were selected in every hospital. Self-administered questionnaires were directly distributed to 1800 doctors after obtaining written informed consent. The questionnaire in which missing values exceeded 20% was regarded as invalid. We received an effective response from 1392 doctors with an effective response rate of 77.3%.

If you do not feel satisfied with the revision and interpretation, we are pleased to looking forward to your further advice. BMJ Open: first published as 10.1136/bmjopen-2018-024186 on 24 May 2019. Downloaded from 6.QUESTIONNAIRES

This section contains some results and it quite hard to follow. Also - it's hard to understand why these particular inventories are used - it's likely that they have been used in previous research so they are good comparable measures but the reader cannot tell.

Answer:The scales in our study are international common scales, such as The Center for Epidemiology Studies Depression Scale (CES-D) , Maslach Burnout Inventory Human Services Survey (MBI-HSS), 24-item Psychological Capital Questionnaire (PCQ-24) and so on. At the same time, these scales had good reliability and validity, and have been widely used in Chinese populations.Therefore, we choosed these scales. We added some descripiton in the article. The CES-D is an international common scale and has been translated, back-translated, and revised to make its items culturally and linguistically applicable in China. The Chinese version of the CES-D has shown good reliability and validity and has been widely used in Chinese populations[30]. (page 8, line6-9 )

The MBI-HSS is an international common scale, and the Chinese version of the MBI-HSS has good reliability and validity, and has been widely used in Chinese populations[37]. (page 8, line17-19 )

The PCQ-24 is an international common scale, and the Chinese version of the PCQ-24 has good reliability and validity and has been widely used in Chinese populations[39]. (page 10 , line 8-10)

If you do not feel satisfied with the revision and interpretation, we are pleased to looking forward to your further advice.

7.STATISTICAL ANALYSES

This section is also confusing. I kept with it but then when we reached the regression part it wold have been good to just tell the reader that a regression was done and the details then would be in the results section. http://bmjopen.bmj.com/ Answer: Thank you very much for your advice. We revised it as you asked. (page 10, line 20-22； page 11 line 1-14)

For the categorical variables, groups for which the response rate was less than 5% were merged. In our study, only 37 (2.7%) participants belonged to the “widow/divorced/separated” group, and thus this group was combined with the “single” group. Univariate analysis of depressive symptoms and burnout in relation to categorical variables was examined by t-tests and one-way ANOVAs. Univariate on September 28, 2021 by guest. Protected copyright. analysis of depressive symptoms and burnout in relation to continuous variables was tested by Pearson’s correlation analysis .

We performed hierarchical linear regression analyses to examine the effects of workplace violence and psychological capital on depressive symptoms and burnout. Except for gender and age, variables related to depressive symptoms and burnout (P <0.05) in t-tests or one-way ANOVAs were entered in the model.

If you do not feel satisfied with the revision, we are pleased to looking forward to your further advice. BMJ Open: first published as 10.1136/bmjopen-2018-024186 on 24 May 2019. Downloaded from 8. STATISTICAL ANALYSES

How were missing data addressed?

Answer: Thank you very much for your question. The revision was as followed:

Missing values were addressed with mean substitution[40]. (page 11, line 13-14)

If you do not feel satisfied with the revision, we are pleased to looking forward to your further advice.

9.Did the authors undertake any sub-group analyses? So what about any effect of specialty? Given the plethora of research that suggests different specialties having different cultural profiles in terms of violence - this would be an interesting analysis. There are some data reported in the Tables but it's hard to make sense of - so something in the results section would be ideal.

Answer: Thank you very much for your advice. We added some data to explore the effect of the division on workplace violence in Table 1. We found that the prevalence of workplace violence in the surgical department was highest. Due to insufficent sample size in sub-group, we will examine the relationships between workplace violence, psychological capital and depressive symptoms and burnout in different divisions in further study.

Table 1 Characteristics of workplace violence among doctors

Variables N(%)

Violence styles

physical violence 469(33.7)

psychological aggression 1012 (72.7)

threat 433(31.1) http://bmjopen.bmj.com/

verbal sexual harassment 272(19.5)

sexual assault. 175(12.6)

0 type 315(22.6)

One type 483(34.7)

Two types 202(14.5)

Three types 202(14.5)

Four types 82(5.9)

Five types 108(7.8)

Division Workplace violence (N(%))

Internal medicine 378(80.8) BMJ Open: first published as 10.1136/bmjopen-2018-024186 on 24 May 2019. Downloaded from Surgical department 245(83.3)

Department of obstetrics and gynecology 66(76.7)

Ancillary department 139(69.5)

Other department 251(73.0)

If you do not feel satisfied with the revision and interpretation, we are pleased to looking forward to your further advice.

10.RESULTS

There are a lot of results and sometimes it gets quite confusing. I wonder whether they can be described in a 'lay' manner after each set - so the reader can work out exactly what this issues are - with so many constructs it's hard to remember what each means.

Answer:Thank you very much for your advice. We revised the results in a “lay” manner. The revision was as followed: (page 11, line18-22；page 12, line1-22；page 13, line1-22；page 14, line1-11)

3.1 Characteristics of workplace violence among doctors

The characteristics of workplace violence among doctors are shown in Table 1. The prevalence of workplace violence was 77.5%. Approximately 72.7% of participants encountered psychological aggression. The prevalence of workplace violence from highest to lowest was physical violence (33.7%), threat (31.1%), verbal sexual harassment (19.5%) and sexual assault(12.6%). Additionally, 34.7% of participants experienced one type of violence; 14.5% of participants encountered two types of violence; and 14.5% of participants suffered three types of violence. The prevalence of workplace http://bmjopen.bmj.com/ violence was highest (83.3%) in the surgical department and lowest (69.5%) in the ancillary department.

3.2 T-tests or one-way ANOVAs of depressive symptoms and burnout in relation to categorical variables

The average age was (38.90 ± 8.81) years. Educational level, weekly work time, shift work, night shift, on September 28, 2021 by guest. Protected copyright. division, income, and workplace violence were significantly associated with depressive symptoms (p<0.05). Marriage, educational level, weekly work time, shift work, night shift, division, and workplace violence were significantly associated with emotional exhaustion (p<0.05). Educational level, weekly work time, shift work, night shift, division, and workplace violence were significantly associated with depersonalization (p<0.05). Gender, marriage, educational level, weekly work time, rank, division, income, and workplace violence were significantly associated with personal accomplishment (p<0.05). The above results are shown in Table 2.

3.3 Pearson’s correlation analysis of depressive symptoms and burnout in relation to continuous variables

The correlation analysis showed that age was significantly correlated with personal accomplishment(r=0.07, p<0.05). Psychological capital was significantly associated with depressive symptoms(r=-0.35, p<0.05) and burnout(emotional exhaustion : r=-0.28, p<0.05; depersonalization : r=-0.27, p<0.05; personal accomplishment: r=0.26, p<0.05). BMJ Open: first published as 10.1136/bmjopen-2018-024186 on 24 May 2019. Downloaded from 3.4 Hierarchical linear regression analysis results, with depressive symptoms and job burnout as the criterion variables

In step 1 of the hierarchical linear regression analysis, demographic characteristics were used as predictors. In step 2, workplace violence was added, and in step 3, psychological capital was added. Workplace violence was positively associated with depressive symptoms (β=0.11, p<0.05) and two dimensions of burnout (emotional exhaustion : β=0.18, p<0.05; depersonalization: β=0.17, p<0.05). Psychological capital was negatively associated with depressive symptoms (β=-0.32, p<0.05) and two dimensions of burnout (emotional exhaustion: β=-0.23, p<0.05 ; depersonalization: β=-0.23, p<0.05). The above results showed that workplace violence and psychological capital were associated with depressive symptoms and two dimensions of burnout (emotional exhaustion and depersonalization).(Table 3)

3.5 Regression analysis results, with depressive symptoms and two dimensions of burnout (emotional exhaustion and depersonalization) as outcomes and psychological capital as a mediator

In the regression equation, workplace violence was used as a predictor, and depressive symptoms and burnout were used as outcomes (Figure 1). The first step in the analysis was to explore the associations of workplace violence with depressive symptoms and burnout (the c path). The second step was to explore the mediating role of psychological capital (the a*b products), and c’ path coefficient represented the effects of workplace violence on depressive symptoms and burnout after psychological capital was added in the second step. When the c’ path coefficient was smaller than the c path coefficient or became statistically insignificant, a mediating effect may have existed. Based on 5000 bootstrap samples, a bias-corrected and accelerated 95% confidence interval (BCa 95% CI) of each a*b product was explored, and a BCa 95% CI without 0 meant a significant mediating effect.

Workplace violence was significantly associated with psychological capital (the a path, p<0.05). Psychological capital was significantly negatively associated with depressive symptoms and two dimensions of burnout after controlling for workplace violence(the b path, p<0.05). The c’ path

coefficient was still significant (p<0.05), but less than the c path coefficient. Psychological capital http://bmjopen.bmj.com/ mediated the relationships of workplace violence with depressive symptoms and two dimensions of burnout (emotional exhaustion and depersonalization) (a*b = 1.61, BCa 95% CI: 1.08-2.25; p < 0.05), emotional exhaustion (a*b = 1.29, BCa 95% CI: 0.86-1.83; p < 0.05) and depersonalization (a*b = 0.70, BCa 95% CI: 0.45-1.00; p < 0.05). The above results showed that workplace violence increased the level of depressive symptoms and burnout by damaging psychological capital. (Table 4)

If you do not feel satisfied with the revision and interpretation, we are pleased to looking forward to your further advice. on September 28, 2021 by guest. Protected copyright.

11.DISCUSSION the sentence "The sample size (1392 doctors) was large and the effective responsive rate in this study was 77.3% greater than 70% that was considered to be acceptable for questionnaire survey[40]. These may increase the generalization of our study conclusion." - please explain further - generalisation to whom? Across the whole of the Liaoning Province? To other provinces? Across China? To other country settings?

Answer: Thank you very much for your question. These findings may increase the generalization of our study conclusion across the whole of Liaoning Province. (page 14, line 18-19)

12.For many findings - there was the statement that the findings echoed other research - and I was beginning to wonder "what does this study add"? But then the authors state "This is the first study to examine the mediating role of psychological capital in the relationships of workplace violence with depressive symptoms and burnout". If this is the case, then this needs to be up-front and take BMJ Open: first published as 10.1136/bmjopen-2018-024186 on 24 May 2019. Downloaded from president in the discussion section. The discussion section needs to begin by describing the novel findings - or at least flagging them up at the start so the reader is primed.

Answer: Thank you very much for your advice.We flagged the important discovery at the first paragraph in the discussion. The revision was as followed: (page 14, line 13-21)

This study explored the prevalence of workplace violence among doctors in China, the associations of workplace violence and psychological capital with depressive symptoms and burnout, and the mediating roles of psychological capital. The sample size (1392 doctors) was large and the effective responsive rate in this study was 77.3%, which was greater than 70%, which considered acceptable for the questionnaire survey[41]. The workplace violence scale in this study had good reliability and validity. These findings may increase the generalization of our study conclusion across the whole of Liaoning Province. In this study, the most important discovery was the discovery that workplace violence increased the level of depressive symptoms and burnout by damaging psychological capital.

If you do not feel satisfied with the revision, we are pleased to looking forward to your further advice.

13. The limitation section is rather brief - more can be said about issues such as generalisability, study engagement and so on.

Answer: Thank you very much for your advice. We added the limitations in the generalisability. The revision was as followed: (page 17, line1-8)

Several limitations must be mentioned. First, due to the cross-sectional design, this survey was not able to determine causality among variables. Therefore, all of these results need to be confirmed in longitudinal studies. Second, workplace violence, psychological capital, depressive symptoms and burnout were measured by a self-reported questionnaire. Thus, common reporting variance may affect the results of effects of workplace violence and psychological capital on depressive symptoms and burnout. Third, we only investigated doctors in Liaoning Province, so the results may not be fully http://bmjopen.bmj.com/ representative of the entire situation in China. Future studies should conduct in-depth exploration in other provinces.

If you do not feel satisfied with the revision, we are pleased to looking forward to your further advice.

If there are other errors or further requests, please contact us by e-mail.