Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Risk assessment of workplace violence: how data collection methods influence results. A Swedish and Italian cross-sectional study.

 The study aimed to demonstrate that the three samples would have maintained the same characteristics (association of WV with stress and disability) but would have shown different prevalence.

The article addresses a relevant and original methodological topic but presents significant limitations that compromise the validity and generalisability of the results.

 

1. The comparison between the self-selected (Brescia/Trollhättan) and census (Latium) samples is central, but the age difference is a strong potential confounder. Is Impossible to distinguish whether differences are due to the method or age.  The group also differ in 
   1. Sampling method (self-selection vs. census).
   2. Context (online vs. medical examination)
   3. Country

 

Impossible to isolate the effect of the reporting method.

Conclusion

The data are good presented and the analysis is appropriate. The study should be accepted after the authors address the major revisions regarding the age difference and the precise framing of the selection bias.

 

 

 

Author Response

REVIEWER #1

 The study aimed to demonstrate that the three samples would have maintained the same characteristics (association of WV with stress and disability) but would have shown different prevalence.

The article addresses a relevant and original methodological topic but presents significant limitations that compromise the validity and generalisability of the results.

Response: We sincerely thank the reviewer who praised our work as "relevant and original". As he/she correctly noted, the three samples have maintained the same characteristics (association of WV with stress and disability), but two of them (those collected online with an explicit request on WV and those collected through a census with an indirect request) have differences in age of participants and prevalence of violence. We absolutely agree that the difference between methods limits the generalizability of the results of this and all research on the subject but demonstrating this difference was exactly the aim of our study. We explained this choice in the abstract and in the introduction and then discussed it in the final part of the article.

 

1. The comparison between the self-selected (Brescia/Trollhättan) and census (Latium) samples is central, but the age difference is a strong potential confounder. Is Impossible to distinguish whether differences are due to the method or age.  The group also differ in 
1. Sampling method (self-selection vs. census).
2. Context (online vs. medical examination)
3. Country

 Impossible to isolate the effect of the reporting method.

Conclusion. The data are good presented and the analysis is appropriate. The study should be accepted after the authors address the major revisions regarding the age difference and the precise framing of the selection bias.

Response: We thank the reviewer for highlighting this point. Naturally, we took age and gender into account in multivariate analyses. In the new version, we have attempted to clarify this point. The method of our study was constructed so that the differences correctly listed by the reviewer are present only in pairs of samples, never in all three. Sampling: Two samples (Brescia and Trollhättan) were self-selected, one was voluntary but census-based (Latium). Online: All three samples were collected online using the same platform. Country: Two samples were Italian, one was Swedish. We can add that in two samples the survey focused on violence (Brescia and Trollhättan), while in one it focused on sleep (Latium). In two samples, the participants did not know the examiner, while in one, the examiner was the occupational physician who would subsequently examine them. This construction makes it possible to verify, through binary comparisons, the weight of the three factors indicated by the reviewer and the two we added. The only factor significantly correlated with the prevalence of violence is the data collection method. Age, gender, stress, and work capacity correlate with violence. We have detailed these results in the text at lines. All these results are consistent with what we know from literature and are confirmed in our study, but they add nothing to what was already known. The original finding is that the method chosen influences prevalence. We emphasized this concept in the discussion.

Reviewer 2 Report

Comments and Suggestions for Authors

Peer Review Report – Manuscript nursrep-3935897

I would like to thank the Authors for submitting this interesting and well-conceived manuscript. The study explores how data collection methods and sampling strategies influence the estimation of workplace violence (WV) prevalence among nurses in Italy and Sweden, while also examining its association with occupational stress (Effort/Reward Imbalance, ERI) and work ability (Work Ability Scale, WAS). The manuscript is overall clear, coherent, and methodologically sound. It contributes significantly to understanding how methodological choices affect the validity of WV assessment in healthcare. The cross-national comparison adds further depth and relevance.

Below I provide section-by-section comments including both positive feedback and specific suggestions for improvement.

Abstract

The abstract is concise, well-structured, and clearly communicates the study’s purpose, methodology, key findings, and conclusions. It effectively balances scientific precision with readability, making the main outcomes accessible to both researchers and practitioners. The inclusion of cross-national data and attention to methodological bias demonstrate originality and analytical depth.

Suggestions for improvement:
- Specify more clearly the study design (cross-sectional observational comparison) and total sample size (n = 236, three nurse groups).
- Briefly indicate the main statistical approach (e.g., regression analysis) to highlight analytical rigor.
- Consider mentioning the consistency of correlations between WV, stress, and work ability across samples.
- The last sentence could be phrased more cautiously to avoid prescriptive wording.

Introduction

The introduction provides a comprehensive and critical overview of WV in healthcare. It demonstrates strong command of the literature and clearly identifies methodological gaps justifying the study. The four hypotheses are well formulated and coherent with the study aims.

Suggestions for improvement:
- The extensive description of national prevalence studies (Tables 1–2) is informative but could be partially condensed to improve readability.
- Strengthen the explicit link between the discussion of definitional and measurement issues and the methodological innovation of this study.
- Clarify briefly the rationale for including Sweden to underline the cross-national significance.
- Add a short paragraph summarizing how the study’s methodological insights may inform better surveillance and prevention policies.

Materials and Methods

This section is detailed, transparent, and adheres to sound ethical standards. The description of populations, instruments (VIF, ERI, WAS), and procedures is exemplary.

Suggestions for improvement:
- Clarify whether the Swedish version of the VIF was translated or adapted and describe any back-translation or pilot testing undertaken.
- The rationale for the sample size could note that calculations were based only on physical violence prevalence.
- Briefly describe data quality control (handling of incomplete online responses, if any).
- Report reliability coefficients separately for the Italian and Swedish samples.
- Specify criteria for parametric vs. nonparametric analyses, and whether multicollinearity checks were performed before regression.
- Explicitly note that causal relationships cannot be inferred.

Results

The results are well presented, with clear tables and logical progression from descriptive to inferential statistics.

Suggestions for improvement:
- Emphasize explicitly that there were no significant differences between the Italian and Swedish online samples.
- Include effect sizes or odds ratios for main group comparisons.
- Add R² values for regression analyses to show explained variance.
- Clarify whether age was controlled for in the regression model.
- Comment briefly on power to detect gender effects.
- Confirm handling of missing data.
- Consider adding one figure to visualize relationships between WV, ERI, and WAS.

Discussion

The discussion is insightful and well argued. It contextualizes the results within the broader literature and articulates how methodological biases shape WV prevalence data.

Suggestions for improvement:
- Expand on practical implications for occupational health monitoring.
- Condense the long literature summary to enhance readability.
- Acknowledge that underreporting might still occur even during health surveillance.
- Discuss briefly why younger nurses appear more exposed or more likely to report WV.
- Explicitly highlight how this study empirically confirms theoretical expectations from previous meta-analyses.
- Include a short note on cultural/systemic differences between Italy and Sweden.
- End the section with a paragraph summarizing policy implications.

Conclusions

The conclusion clearly and effectively synthesizes the study’s findings, offering a coherent argument for embedding WV monitoring in occupational health surveillance.

Suggestions for improvement:
- Moderate the tone of certainty by framing recommendations as 'suggestions' rather than definitive conclusions.
- Mention possible future research directions such as longitudinal studies.
- Highlight the potential relevance to other occupational settings.
- Reinforce the ethical and organizational value of including WV assessment in health promotion activities.

Tables and References

Tables are clear and informative, effectively complementing the narrative. The reference list is extensive, current, and highly relevant.

Suggestions for improvement:
- For Tables 1–2, consider adding brief summary statements highlighting prevalence contrasts.
- Ensure consistent use of decimals and notation across tables.
- Add confidence intervals directly within Table 5 and report R² in Table 7.
- Verify uniform DOI formatting and access dates for online resources.
- Self-citations are appropriate and justified given the author’s expertise.

Overall Evaluation

This manuscript makes a significant methodological and practical contribution to occupational health research. It is scientifically sound, logically structured, and of interest to both researchers and practitioners concerned with violence prevention and health surveillance. After minor revisions focused on clarity, data reporting, and synthesis, it will be suitable for publication.

Recommendation: Minor Revision

Author Response

REVIEWER #2

I would like to thank the Authors for submitting this interesting and well-conceived manuscript. The study explores how data collection methods and sampling strategies influence the estimation of workplace violence (WV) prevalence among nurses in Italy and Sweden, while also examining its association with occupational stress (Effort/Reward Imbalance, ERI) and work ability (Work Ability Scale, WAS). The manuscript is overall clear, coherent, and methodologically sound. It contributes significantly to understanding how methodological choices affect the validity of WV assessment in healthcare. The cross-national comparison adds further depth and relevance.
Below I provide section-by-section comments including both positive feedback and specific suggestions for improvement.

Response: We are indebted to the reviewer for careful review of our study, providing many helpful suggestions for its improvement.

 

Abstract

The abstract is concise, well-structured, and clearly communicates the study’s purpose, methodology, key findings, and conclusions. It effectively balances scientific precision with readability, making the main outcomes accessible to both researchers and practitioners. The inclusion of cross-national data and attention to methodological bias demonstrate originality and analytical depth.

Response: We gladly accepted the reviewer's suggestions, striving to maintain the characteristics he/she praised. Specifically, we accepted all the points listed below.

 

Suggestions for improvement:
- Specify more clearly the study design (cross-sectional observational comparison) and total sample size (n = 236, three nurse groups)

R.: We have reformulated the description of the epidemiological design

- Briefly indicate the main statistical approach (e.g., regression analysis) to highlight analytical rigor.

R.: We have indicated the statistical technique and the results achieved

- Consider mentioning the consistency of correlations between WV, stress, and work ability across samples.

R.: We added the significance level of the bivariate correlations

- The last sentence could be phrased more cautiously to avoid prescriptive wording.

R.: We have reworded the last sentence to make it clear that using health surveillance to assess WV is a win-win opportunity, not a prescription.

 

Introduction

The introduction provides a comprehensive and critical overview of WV in healthcare. It demonstrates strong command of the literature and clearly identifies methodological gaps justifying the study. The four hypotheses are well formulated and coherent with the study aims.

Suggestions for improvement:
- The extensive description of national prevalence studies (Tables 1–2) is informative but could be partially condensed to improve readability.

Response: We completely agree with the reviewer. In fact, the data presented in Tables 1 and 2 are part of an unfinished scoping review, which motivated us to make this observation. In the Introduction, we limited the commentary on the two tables to about 150 words. The text referring to this argument in the Discussion is too long, so we've shortened it. Overall, we reduced the text by about 400 words.

 

- Strengthen the explicit link between the discussion of definitional and measurement issues and the methodological innovation of this study.

R.: We thank the reviewer because, upon rereading the manuscript, we realized that the relevant reasons for our study had remained unwritten. We have therefore dedicated a few brief sentences to explaining the reasons that had led us to construct a design with three observation groups.

 

- Clarify briefly the rationale for including Sweden to underline the cross-national significance.

R.: This suggestion is also essential, and we have accepted it.

 

- Add a short paragraph summarizing how the study’s methodological insights may inform better surveillance and prevention policies.

R.: This recommendation, too, is absolutely necessary. Knowing the actual rate of workers suffering from WV is essential for developing appropriate counseling for victims, developing measures to combat violence tailored to their needs, and monitoring the effectiveness of these interventions over time.

 

Materials and Methods

This section is detailed, transparent, and adheres to sound ethical standards. The description of populations, instruments (VIF, ERI, WAS), and procedures is exemplary.

Suggestions for improvement:
- Clarify whether the Swedish version of the VIF was translated or adapted and describe any back-translation or pilot testing undertaken.

R.: The three questions in the VIF and the eleven in the other two questionnaires were originally written in English and were administered in the original language, taking into account that Swedish nurses are all fluent in English. We had indicated this in the original manuscript but have now emphasized it.

 

- The rationale for the sample size could note that calculations were based only on physical violence prevalence.

R.: We based the sample size calculation on physical violence (an attack, which may or may not involve weapons, and has the potential to result in physical harm) because this type of violence is more universally recognizable than other forms of violence (threats, harassment) which depend on the victim's perception and sensitivity. We have added this explanation to the text.

 

- Briefly describe data quality control (handling of incomplete online responses, if any).

R.: Given the brevity of the questionnaire, we decided to eliminate cases with incomplete responses. After collecting responses online, only three Italian nurses and one Swedish nurse had limited themselves to personal data, without responding to the questionnaires, and were therefore not included in the study.

 

- Report reliability coefficients separately for the Italian and Swedish samples.

R.: We calculated Cronbach's alpha scores in the three groups and reported them.

 

- Specify criteria for parametric vs. nonparametric analyses, and whether multicollinearity checks were performed before regression.

Response: The Kolmogorov-Smirnov and Shapiro-Wilk tests were employed to assess adherence to normal distribution. The sample size indicated that parametric methods were applicable despite the ordinal nature of the variables, as noted by Lumley; however, we employed robust methods when this assumption was violated. Consequently, we examined the relationship between the variables by computing both Pearson's r and Spearman's rho. The prevalence uncertainty was assessed using the Clopper–Pearson exact binomial test, yielding a 95% confidence interval (CI95%). Prior to examining the interactions among the variables of interest through multiple linear regression, we ensured that these variables were not excessively correlated, as such relationships could compromise the integrity of the regression models. We calculated the variance inflation factor (VIF) to assess the interrelationship among predictor variables in the regression model. We have specified these criteria in the statistical section.

 

- Explicitly note that causal relationships cannot be inferred.

Response: We expressed this concept and reiterated it in the Discussion. In the article, we were careful to avoid sentences that implied causality, even in cases where this was demonstrated by the literature.

Results

The results are well presented, with clear tables and logical progression from descriptive to inferential statistics.

Suggestions for improvement:
- Emphasize explicitly that there were no significant differences between the Italian and Swedish online samples.

Response: The reviewer's advice is very valuable. We had already compared the violence rates of the three groups. We thought it appropriate to include a new table comparing stress and workability levels, highlighting the similarities between the two self-selected groups and the differences compared to the nurses interviewed online during the health surveillance.

The statistical analyses were performed using SPSS version 28.0. During the review, we became aware of version 30.0, and we therefore rechecked the calculations using this version.

 

- Include effect sizes or odds ratios for main group comparisons.

Response: Thanks for this observation. We evaluated the odds ratio of being distressed. Experiencing violence increases the risk of experiencing distress ninefold. We added a table and a short comment on this finding.

 

- Add R² values for regression analyses to show explained variance.

Response: We added the adjusted R2 values to the linear regression equation and the Nagelkerke R2 to the logistic regression equation, and briefly commented on their significance.

 

- Clarify whether age was controlled for in the regression model.

Response: Of course, we controlled for age and gender in the regression equations. We have explained the concept both in the statistics section and in the results section.

 

- Comment briefly on power to detect gender effects.

Response: The topic of gender in workplace violence is highly controversial. Some studies have observed no differences in risk between genders, while others have observed a greater prevalence of certain forms of violence among males or females. Some of the authors of this paper have studied this aspect of violence in previous studies involving very large and homogeneous case studies. This study was not designed to evaluate this aspect, but we have taken it into account both in the presentation of the results and in the multivariate analyses.

 

- Confirm handling of missing data.

Response: we excluded cases with missing data. The electronic data collection system required all responses to proceed; therefore, participants could complete the questionnaire or exit before completing it. We decided to exclude anyone who did not answer at least 18 of the 19 questions, that is, those who exited before the last question. Given the brevity of the questionnaire, only a few exited at the first few questions and were excluded.

 

- Consider adding one figure to visualize relationships between WV, ERI, and WAS.

Response: In accordance with the suggestion, we added Figure 1 to represent the bivariate relationship between WV, stress, and work ability.

 

Discussion

The discussion is insightful and well argued. It contextualizes the results within the broader literature and articulates how methodological biases shape WV prevalence data.

Suggestions for improvement:
- Expand on practical implications for occupational health monitoring.

R.: The authors, who are half occupational physicians and half nurses, welcomed this suggestion very much because it corresponds to their shared aim of providing elements for improving working conditions. We have added a paragraph to section 4.6 on the prospects of this study.

 

- Condense the long literature summary to enhance readability.

Response: We gladly heeded this advice and reduced the description of the results of the studies on violence conducted in Italy by over 400 words, in order to streamline the text and focus the reader on the original findings of our research.

 

- Acknowledge that underreporting might still occur even during health surveillance.

Response. We completely agree. We added a statement at the end of paragraph 4.5 stating that underreporting may also occur during health surveillance, either because the worker believes that assault by a patient who is not in full possession of his or her mental faculties should not be reported, or because they are afraid of revealing assaults suffered by colleagues and superiors. However, the occupational physician's attention to the topic of WV will educate workers to report and prevent these phenomena as well.

 

- Discuss briefly why younger nurses appear more exposed or more likely to report WV

Response: The reviewer urged us to address a very complex topic, to which some of the authors have devoted particular attention in the past; we have briefly touched on one possible cause, but cannot elaborate further to avoid weighing down this article.

- Explicitly highlight how this study empirically confirms theoretical expectations from previous meta-analyses.

Response: This study empirically confirms theoretical expectations that emerged from the analysis of the marked difference between retrospective and spot studies and from the related synthetic analyses. We have added this statement to section 4.6

 

- Include a short note on cultural/systemic differences between Italy and Sweden.

Response: To accommodate this fair observation from the reviewer, we have added a paragraph to the first part of the discussion.

 

- End the section with a paragraph summarizing policy implications.

Response: As mentioned above, we have added a paragraph on policy implications.

 

Conclusions

The conclusion clearly and effectively synthesizes the study’s findings, offering a coherent argument for embedding WV monitoring in occupational health surveillance.

Suggestions for improvement:
- Moderate the tone of certainty by framing recommendations as 'suggestions' rather than definitive conclusions.

Response: We agree. What we have formulated are suggestions, not recommendations. We have therefore reworded the first sentence of the conclusions

 

- Mention possible future research directions such as longitudinal studies.

Response. It will thus be possible to verify whether the relationship between WV and stress and between WV and work ability is reciprocal, as observed in previous longitudinal studies.

 

- Highlight the potential relevance to other occupational settings.

R.: We are confident that other workplace health and safety services can apply the method used in this study, which requires minimal time commitment for workers but provides reliable data for occupational medicine and prevention services. We have added this statement to the conclusions.

 

- Reinforce the ethical and organizational value of including WV assessment in health promotion activities.

R.: This suggestion is also appreciated and we have applied it. Collecting workers' experiences of violence directly allows us not only to provide them with adequate assistance, but also to educate them on how to combat violence and limit its effects. The problem of violence in healthcare settings is too serious to base prevention efforts on literature rather than effectively assessing the circumstances and discussing possible solutions with those involved.

 

Tables and References

Tables are clear and informative, effectively complementing the narrative. The reference list is extensive, current, and highly relevant.

Suggestions for improvement:
- For Tables 1–2, consider adding brief summary statements highlighting prevalence contrasts

Response: We thank the reviewer for this recommendation. In a forthcoming synthesis study, which is currently underway, we will conduct a meta-analysis of the data reported in the two tables. For now, we have limited ourselves to indicating the range of variations in the text.

- Ensure consistent use of decimals and notation across tables.

Response: We verified the consistency of the tables. In Table 1, we used two decimal places because the results are close to one; in Table 2, we considered one decimal place sufficient because the values are an order of magnitude higher.

- Add confidence intervals directly within Table 5 and report R² in Table 7.

Response: We have added the determination coefficients to Tables 8 and 9 (the latter being new). We chose not to include confidence intervals for each item in Table 5 to avoid making the table more readable, and instead simply report the overall values in the text.

 

- Verify uniform DOI formatting and access dates for online resources.

R.: We've prefixed the doi: particle where it was missing. We checked that all the cited sites were working before submitting the article, and we've double-checked it now..

 

- Self-citations are appropriate and justified given the author’s expertise.

Response: We thank the reviewer for this clarification. The high number of self-citations is actually due to the fact that one of the authors has translated and validated many of the questionnaires currently in use in occupational health into Italian and has conducted numerous studies on the topics covered in this article. Where possible, self-citations have been replaced by similar work by other authors.

 

Overall Evaluation

This manuscript makes a significant methodological and practical contribution to occupational health research. It is scientifically sound, logically structured, and of interest to both researchers and practitioners concerned with violence prevention and health surveillance. After minor revisions focused on clarity, data reporting, and synthesis, it will be suitable for publication.

Recommendation: Minor Revision

Response: We thank the reviewer for his numerous suggestions. We believe that reviewing is an essential part of the publication process and we believe that the reviewer's contribution was particularly useful.

 

Reviewer 3 Report

Comments and Suggestions for Authors

Introduction:
The authors address an important gap in workplace violence (WV) research by explicitly comparing outcomes across different data-collection approaches. The cross-national sample—Italian and Swedish participants—adds potential value to the study. The authors also hypothesize associations between WV, occupational stress, and work ability.

Methods:
The authors state that “the survey, which was conducted in June 2025, was terminated when, on the basis of a previously calculated sample size, enough responses had been received.” This statement requires further clarification, and the authors should provide a supporting citation for the sample-size calculation.

Additionally, although the authors mention using English versions of the instruments for the Swedish nurses, they should clarify whether cultural or linguistic adaptation was performed and whether validation of these instruments for Swedish participants was necessary.

Results and Discussion:
The results section is clearly presented. However, in the discussion, the authors should consider addressing whether cross-cultural differences between Italy and Sweden—such as organizational climate, reporting norms, or national policies—could influence the reporting of workplace violence.

Good luck.

Author Response

REVIEWER #3

Introduction:
The authors address an important gap in workplace violence (WV) research by explicitly comparing outcomes across different data-collection approaches. The cross-national sample—Italian and Swedish participants—adds potential value to the study. The authors also hypothesize associations between WV, occupational stress, and work ability.

Response: We thank the reviewer for acknowledging the usefulness of our work. We constructed an experimental design with three groups to demonstrate, through comparisons across countries and different methods, how the latter are particularly influential on the estimated prevalence of WV.

 

Methods:
The authors state that “the survey, which was conducted in June 2025, was terminated when, on the basis of a previously calculated sample size, enough responses had been received.” This statement requires further clarification, and the authors should provide a supporting citation for the sample-size calculation.

Response: Following the reviewer's recommendation, we detailed the method used to calculate the sample size. Since the purpose of the article was to highlight the difference between self-selection and census, and the speed of response to the request for participation is important, we kept the numbers small so that the difference was more evident.

Response. Following the reviewer's recommendation, we detailed the method used to calculate the sample size. Since the purpose of the article was to highlight the difference between self-selection and census, and the speed of response to the request for participation is important, we kept the numbers small so that the difference was more evident.

Additionally, although the authors mention using English versions of the instruments for the Swedish nurses, they should clarify whether cultural or linguistic adaptation was performed and whether validation of these instruments for Swedish participants was necessary.

Response: This topic is certainly of interest. We explained that the Swedish nurses used the original English version of the questionnaires, given the excellent level of English proficiency in Sweden.

 

Results and Discussion:
The results section is clearly presented. However, in the discussion, the authors should consider addressing whether cross-cultural differences between Italy and Sweden—such as organizational climate, reporting norms, or national policies—could influence the reporting of workplace violence.

Good luck.

Response: We fully accepted the reviewer's suggestion, as the purpose of this study was precisely to clarify how cultural and healthcare organizational differences between Italy and Sweden were less important than the prevalence differences caused by the different survey design. We attempted to further explain this concept in the Introduction, Results, and Discussion.

 

Reviewer 4 Report

Comments and Suggestions for Authors

Dear authors

The article presents an observational and cross-sectional study focused on assessing the risk of workplace violence among healthcare professionals.

The topic is quite relevant and current, however, the study presents significant methodological weaknesses.

The authors should have taken into account the recommendations of STROBE , in order to ensure alignment with international criteria for transparency and quality in this type of study.

1. Title and Abstract

a) The title indicates the phenomenon and the type of study, but does not present the population.

b) The abstract, despite having an adequate structure, does not include: the justification for the relevance of the study; the aims of the study; the type of study and type of sample.

c) Keywords: It is unclear why the keywords "Systematic review" and "Meta-analysis" are used, since neither of these studies was conducted. Some keywords, such as "workplace violence" and "negative behaviors," should have been used.

2. Introduction

a) The definition of violence in the workplace is not clearly discussed. It can refer to various situations: violence between peers, violence within a multidisciplinary team, violence between clients and healthcare professionals. What type of violence does this study refer to?

b) To justify this study, the authors compare several studies, presenting two tables. What methodology is used to select these studies? What are the criteria for selecting these studies? How did the authors ensure that these studies are studying the same type of workplace violence? Is it the same type of violence that this study intends to analyze? The basis for the authors' claims based on this comparison is very weak from a methodological point of view, not allowing for the inferences drawn.

c) Lines 116-117: What is the criteria for selecting the questions? How do the authors ensure that these questions correspond to the phenomenon under study?

d) The hypotheses are not aligned with the defined objective.

3. Materials and Methods

a) The methodological design needs clarification. What type of sample was used? Although the authors included Systematic review and Meta-analysis in the keywords, neither of these studies was conducted, making it impossible to meet the overall objective of this study.

b) The data collection instrument needs clarification. Why weren't validated instruments used in their entirety? How were the questions selected?

Due to all these concerns and methodological limitations that call into question the validity of this study, I believe the results may not be reliable. In this sense, I believe that, regrettably, this article should be rejected.

Author Response

REVIEWER #4

Dear authors

The article presents an observational and cross-sectional study focused on assessing the risk of workplace violence among healthcare professionals.

The topic is quite relevant and current, however, the study presents significant methodological weaknesses.

Response: We thank the reviewer for his/her cooperation. Our study was not intended to assess the risk of violence among nurses, but rather to demonstrate that the data collection method can significantly influence the prevalence of reported violence without altering the characteristics of the phenomenon, namely, the association of violence with stress and work ability.

 

The authors should have taken into account the recommendations of STROBE , in order to ensure alignment with international criteria for transparency and quality in this type of study.

Response: We followed the STROBE checklist for cross-sectional observational studies. We have reported this specification at the end of the article.

 

1. Title and Abstract
2. a) The title indicates the phenomenon and the type of study, but does not present the population.

Response: We welcomed the reviewer's suggestion, indicating the professional category in the title

 

1. b) The abstract, despite having an adequate structure, does not include: the justification for the relevance of the study; the aims of the study; the type of study and type of sample.

Response: Following the reviewer's suggestion and recognizing that the methods were not clearly reported in the abstract, we have slightly modified the text so that it is clear that it includes all the required elements.

 

1. c) Keywords: It is unclear why the keywords "Systematic review" and "Meta-analysis" are used, since neither of these studies was conducted. Some keywords, such as "workplace violence" and "negative behaviors," should have been used.

R.: We accepted the reviewer's suggestion, eliminating the keywords he/she indicated. We had extensively discussed synthesis studies, systematic reviews, and meta-analyses in the manuscript, also noting in the abstract that they can be significantly biased if they fail to distinguish between different data collection methods. We included the terms "spot survey" and "secondary research" among the keywords to clarify the topic of the article. We didn't use the term "workplace violence" because it was already in the title. For the same reason, we didn't add "nursing" to the keywords, as suggested by the editorial team, because we added the term "nurses" to the title.

 

2. Introduction
3. a) The definition of violence in the workplace is not clearly discussed. It can refer to various situations: violence between peers, violence within a multidisciplinary team, violence between clients and healthcare professionals. What type of violence does this study refer to?

Response: In the Introduction, we noted not only the existence of different types of violence, but above all the fact that, regardless of the definitions proposed by international organizations, each worker has their own opinion of what acts constitute a form of violence. Consequently, any survey must be conducted with a precise operational definition of the terms used. We therefore used a widely used standardized questionnaire, the VIF, which employs a operational definition of violence. This choice is explained in the various sections of the article.

 

1. b) To justify this study, the authors compare several studies, presenting two tables. What methodology is used to select these studies? What are the criteria for selecting these studies? How did the authors ensure that these studies are studying the same type of workplace violence? Is it the same type of violence that this study intends to analyze? The basis for the authors' claims based on this comparison is very weak from a methodological point of view, not allowing for the inferences drawn.

Response: The data reported in Tables 1 and 2 refer to 28 studies on violence against healthcare workers published in Italy over the past 10 years. The collection of these studies is part of a future review, which is not yet ready. The authors decided to anticipate some data, which the reviewer can consider a simple narrative review, when they realized that the eight retrospective studies and the 20 cross-sectional studies reported highly discordant prevalence rates. The former published rates between 0.18% and 0.25% for physical violence, and between 0.93% and 1.92% for all forms of WV; the latter reported rates between 6.1% and 46.6% for physical violence, and between 29.6% and 96.0% for all forms of violence. The authors believe that the differences are primarily due to the method used and designed this experimental study to test this hypothesis. Systematic analyses and meta-analyses, which are discussed extensively in this article, generally do not differentiate between the different methods used to collect data. The authors believe this could lead to bias. This is why we believe this research should be published.

 

1. c) Lines 116-117: What is the criteria for selecting the questions? How do the authors ensure that these questions correspond to the phenomenon under study?

Response: As we said above, the operational definition of violence is that of the Arnetz VIF questionnaire, an instrument used since the 1990s in many European countries to report episodes of violence.

 

1. d) The hypotheses are not aligned with the defined objective.

Response: While respecting the reviewer's opinion, we strongly disagree. On the contrary, we are convinced that only by demonstrating that violence in the three samples (self-selected Italy, self-selected Sweden, and Italy part of a census) has the same association with stress and workability, can we conclude that the observed marked differences in prevalence are not due to different individual experiences, but rather to seemingly minimal differences in methodology. The workers in the three samples responded to the same online questions in the same month of June 2025, all had the same profession, and were partly Italian and partly Swedish; however, the spot surveys with an explicit focus on violence resulted in significantly higher reporting than the survey of a fraction of the entire population of nurses in Latium, visited in June 2025 and consulted on a general health topic (sleep).

 

3. Materials and Methods
4. a) The methodological design needs clarification. What type of sample was used? Although the authors included Systematic review and Meta-analysis in the keywords, neither of these studies was conducted, making it impossible to meet the overall objective of this study.

Response: In the current version of the manuscript, we have repeatedly reiterated, starting with the abstract, the characteristics of the study: a rapid observational cross-sectional study on three samples. Although the idea for the study stems, as always, from an in-depth review of literature, it is not a synthetic study or a meta-analysis.

 

1. b) The data collection instrument needs clarification. Why weren't validated instruments used in their entirety? How were the questions selected?

Response: in the methods section we had already reported the characteristics of the questionnaires used, all validated in the original languages and in the version used here.

 

Due to all these concerns and methodological limitations that call into question the validity of this study, I believe the results may not be reliable. In this sense, I believe that, regrettably, this article should be rejected.

Response: We can understand that the reviewer, noting that we cited 158 references, understood that this study is based on substantial theoretical work. He/she asks for clarification of the selection criteria for the references cited in the introduction, but fails to consider that this study is clearly, from the title onwards, cross-sectional. No comments were made regarding the epidemiological design, the statistical analyses, the results and their discussion, or the conclusions. The reviewer's suggestions regarding the title and keywords have been implemented. We will be happy to consider and implement any other contributions he/she wishes to make to improve the work. Please remember that rejection of a paper can only be proposed when methods or analyses are found to be irremediably flawed.

 

Reviewer 5 Report

Comments and Suggestions for Authors

 

Dear Authors,

Thank you for the opportunity to review the manuscript titled "Risk assessment of workplace violence: how data collection methods influence results. A Swedish and Italian cross-sectional study" submitted to the journal Nursing Reports.

The study addresses a critically important topic in occupational health: the methodology for assessing Workplace Violence (WV). Its primary contribution lies in demonstrating how sample selection and the manner in which the research is presented significantly influence the reported prevalence of WV. The confirmation of the hypotheses—that WV is positively correlated with occupational stress and inversely correlated with work ability across all samples—strengthens the validity of the responses.

However, for the article to achieve the scientific relevance necessary for publication in a high-impact periodical, certain methodological improvements and a more robust discussion of limitations are crucial.

Below are the suggested improvements, indicating the specific parts of the text that require revision and the respective justification:

1. Section: Materials and Methods (Sampling and Study Design) and Results

Required Improvement: Address limitations regarding sample stability and size.

Justification: Section 4.7, Limitations, acknowledges that the study utilized a limited observation period and "barely sufficient" sample dimensions, which reduced the stability of the results. Although the text mentions that the limited sample size was intentionally chosen to enhance the effect of self-selection, it is vital for the study to reflect rigorous scientific standards, or for this limitation to be more strongly mitigated or addressed statistically.

Required Improvement: Address age-related selection bias more explicitly.

Justification: The results show that the self-selected nurses from Brescia and Trollhättan were significantly younger (mean age 37.5±11.8 and 37.1±11.5, respectively) than the census group from Latium (47.9±10.2) (p<0.001). The discussion recognizes that younger respondents are more exposed to violence than their older colleagues. It is important to clarify in the methods and discussion whether this age difference was controlled for in the prevalence analyses or how it may have confounded the observed difference in WV rates, beyond just the selection method and question presentation.

1. Section: Methodology and Discussion (Design Limitations)

Required Improvement: Strengthen the discussion on causality.

Justification: The study design is cross-sectional. The authors already mention in the Limitations (Section 4.7) that the cross-sectional nature of the observations prevented the interpretation of causality of the observed associations (e.g., between WV, stress, and work ability). This conclusion had to be based on previous longitudinal studies. For a journal of relevance, this limitation needs to be highlighted prominently, ensuring that the language used in the Results and Discussion section (regarding associations) does not imply causation, but rather correlation.

Required Improvement: Clarify objective verification of incidents.

Justification: The manuscript indicates that only in the Latium sample (collected during health surveillance) was the occupational physician able to verify the occurrence of WV episodes. For data collected anonymously online (Brescia and Trollhättan), no objective verification could be made. In the discussion, the authors should emphasize that the lack of objective verification in the online samples makes them vulnerable to subjective biases (such as social desirability bias and chronological recall bias), thereby limiting confidence in the extremely high prevalence (e.g., 73% for all forms of WV) reported by these groups.

III. Section: Conclusions

Required Improvement: Align conclusions with internal evidence.

Justification: The current conclusion states that violence assessment should be conducted by surveying all workers during health surveillance. While the study strongly suggests this method eliminates the bias of cross-sectional studies and underreporting from official records, and appears to be the most accurate method, the text must ensure that this strong recommendation is balanced with the recognized limitations (sample size and cross-sectional nature). Scientific relevance will be enhanced if the conclusion explicitly reflects the need for further studies (perhaps longitudinal ones) using this health surveillance method to confirm its accuracy.

In summary, the main improvements focus on the need to mitigate the limitations of the cross-sectional design and sampling (size and age difference) so that the conclusions about prevalence, which are the central focus of the article, are considered more stable and applicable in a context of high scientific relevance.

Author Response

REVIEWER #5

Dear Authors,

Thank you for the opportunity to review the manuscript titled "Risk assessment of workplace violence: how data collection methods influence results. A Swedish and Italian cross-sectional study" submitted to the journal Nursing Reports.

The study addresses a critically important topic in occupational health: the methodology for assessing Workplace Violence (WV). Its primary contribution lies in demonstrating how sample selection and the manner in which the research is presented significantly influence the reported prevalence of WV. The confirmation of the hypotheses—that WV is positively correlated with occupational stress and inversely correlated with work ability across all samples—strengthens the validity of the responses.

However, for the article to achieve the scientific relevance necessary for publication in a high-impact periodical, certain methodological improvements and a more robust discussion of limitations are crucial.

Response: We thank the reviewer for carefully reading the paper and focusing on the characteristics of this study, which uses a unique short cross-sectional observational design to reveal the reasons for the marked discrepancies in assault rates reported by studies conducted with different methods.

 

Below are the suggested improvements, indicating the specific parts of the text that require revision and the respective justification:

1. Section: Materials and Methods (Sampling and Study Design) and Results

Required Improvement: Address limitations regarding sample stability and size.

Justification: Section 4.7, Limitations, acknowledges that the study utilized a limited observation period and "barely sufficient" sample dimensions, which reduced the stability of the results. Although the text mentions that the limited sample size was intentionally chosen to enhance the effect of self-selection, it is vital for the study to reflect rigorous scientific standards, or for this limitation to be more strongly mitigated or addressed statistically.

Response: Accepting the reviewer's correct observation, we have integrated the Limitations section with the statement that the purpose of the study was not to describe in detail the violence suffered by nurses in different European countries, but to demonstrate, through a rapid cross-sectional comparison of three samples, the impact that the choice of method has on the final results. Across the three samples, WV maintained the same correlation with stress and workability, demonstrating that the observations were sufficiently stable. The differences between the self-selected Italian and Swedish samples, despite the many cultural and healthcare system differences, supported our belief that the way questions are asked and respondents are invited is crucial to the results.

 

Required Improvement: Address age-related selection bias more explicitly.

Justification: The results show that the self-selected nurses from Brescia and Trollhättan were significantly younger (mean age 37.5±11.8 and 37.1±11.5, respectively) than the census group from Latium (47.9±10.2) (p<0.001). The discussion recognizes that younger respondents are more exposed to violence than their older colleagues. It is important to clarify in the methods and discussion whether this age difference was controlled for in the prevalence analyses or how it may have confounded the observed difference in WV rates, beyond just the selection method and question presentation.

Response: We thank the reviewer for this observation, which we promptly addressed. We clarified that the regression equations (both the linear regression presented in the first version and a logistic regression included in this new version) have been adjusted for age and gender. We noted in the text that the age difference is typical of spot studies compared to census surveys and therefore affects not only this study but the entire literature.

 

1. Section: Methodology and Discussion (Design Limitations)

Required Improvement: Strengthen the discussion on causality.

Justification: The study design is cross-sectional. The authors already mention in the Limitations (Section 4.7) that the cross-sectional nature of the observations prevented the interpretation of causality of the observed associations (e.g., between WV, stress, and work ability). This conclusion had to be based on previous longitudinal studies. For a journal of relevance, this limitation needs to be highlighted prominently, ensuring that the language used in the Results and Discussion section (regarding associations) does not imply causation, but rather correlation.

Response: We believe this observation is also important and have attempted to address it to the best of our ability. In the manuscript, we repeatedly emphasized that the cross-sectional nature of the study does not allow us to clarify the direction of the associations. In this version, we have added a figure showing the bivariate relationships between violence, stress, and work ability, and we have reiterated in the commentary that longitudinal studies are necessary to correctly interpret the phenomena. When using regressions, we emphasized that we assigned violence as an independent variable only to assess the impact this experience may have on stress, without attributing a finalistic value to the observations.

 

Required Improvement: Clarify objective verification of incidents.

Justification: The manuscript indicates that only in the Latium sample (collected during health surveillance) was the occupational physician able to verify the occurrence of WV episodes. For data collected anonymously online (Brescia and Trollhättan), no objective verification could be made. In the discussion, the authors should emphasize that the lack of objective verification in the online samples makes them vulnerable to subjective biases (such as social desirability bias and chronological recall bias), thereby limiting confidence in the extremely high prevalence (e.g., 73% for all forms of WV) reported by these groups.

Response. We absolutely agree. We added the statement that the inability to verify the claims of online studies that are not followed up by medical visits can be another factor in inappropriate reporting.

 

III. Section: Conclusions

Required Improvement: Align conclusions with internal evidence.

Justification: The current conclusion states that violence assessment should be conducted by surveying all workers during health surveillance. While the study strongly suggests this method eliminates the bias of cross-sectional studies and underreporting from official records, and appears to be the most accurate method, the text must ensure that this strong recommendation is balanced with the recognized limitations (sample size and cross-sectional nature). Scientific relevance will be enhanced if the conclusion explicitly reflects the need for further studies (perhaps longitudinal ones) using this health surveillance method to confirm its accuracy.

Response: We accepted the reviewer's valid suggestions, first clarifying that the use of health surveillance to collect observations is a recommendation, not a requirement. This recommendation stems from experience in which this method has been applied for over 20 years, some of which was conducted by the authors of this article. To remain consistent with the findings of this study, we emphasized the need for further longitudinal studies conducted by other health surveillance services.

 

In summary, the main improvements focus on the need to mitigate the limitations of the cross-sectional design and sampling (size and age difference) so that the conclusions about prevalence, which are the central focus of the article, are considered more stable and applicable in a context of high scientific relevance.

Response: We agree with the reviewer. We have outlined the study's future directions. We, for our part, are committed to deepening and systematizing the literature on the various research approaches, giving it a meta-analytic character; and to continuing longitudinal observations of violence risk, to clarify the directionality of the relationship with stress and work ability.

Round 2

Reviewer 4 Report

Comments and Suggestions for Authors

Dear authors,

The clarification of the issues raised and the changes introduced have made the manuscript clearer and more robust, therefore, I consider that the new version of the article is suitable for publication.