Impact of the COVID-19 Pandemic on Violence-Related Emergency Visits: Trend, Seasonality, and Interrupted Time-Series Analysis in Peru, 2015–2024

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This study quantifies and documents an increase in violent injuries seen at a hospital in Lima, Peru. The problem of rising violence during this time is important to study to understand what happened, and its causes and consequences. Research on this topic from Latin American countries is sparse. The paper is well written, the statistical analysis is appropriately specified, and the seasonal decompositions are interesting. I have the following questions and concerns:

 

• What is the population study base for this analysis? Who is served by the hospital in this study and who is not served by this hospital? What section of the population (defined geographically, socioeconomically, or otherwise) is likely to be the catchment area? I ask to better understand to who these results refer?
• Were there any major changes in service regions or referral patterns over the study period that could explain any of the observed time trends? Or has the catchment population been relatively stable? This information would aid in interpreting the results.
• Data collection: did the study include all violence-related injuries presenting to the study hospital over the study period? How certain is complete case ascertainment?
• WHO typology of violence needs a citation
• Table 1: It is difficult to compare the counts of injuries across rows because they are for different amounts of time. Would it be possible to report these in a standardized way for comparison (e.g. average cases per month or average cases per year) instead of just total cases for the time period? The percentages allow us to see how the composition of cases changed over time but not the volume.
• Table 1: the self inflicted violence, interpersonal community violence and family violence rows are missing percentages
• Figures 2 and 3: labels and legends are too small to read

Author Response

Comments 1: What is the population study base for this analysis? Who is served by the hospital in this study and who is not served by this hospital? What section of the population (defined geographically, socioeconomically, or otherwise) is likely to be the catchment area? I ask to better understand to who these results refer?

Response 1: Regarding to this observation, we added in the section Study design and settings the following explanation: “The hospital serves the general public and primarily attends low-income and uninsured populations from Lima and surrounding provinces, as it is a public emergency facility open to all individuals requiring urgent care, regardless of socioeconomic status or geographic origin.”

Comments 2:  Were there any major changes in service regions or referral patterns over the study period that could explain any of the observed time trends? Or has the catchment population been relatively stable? This information would aid in interpreting the results.

Response 2: Reporting on service region modifications or referral pattern changes was not a primary focus of this study. However, an indirect indicator of such variations can be observed in Table 1, which shows the distribution of violence cases according to the place of assault. Specifically, cases referred from provinces reached a maximum of eight in the post-pandemic period and six in the pre-pandemic period, whereas only one case was recorded during the pandemic. This fluctuation may reflect temporary disruptions in interprovincial mobility and referral flows due to pandemic restrictions rather than structural modifications in the hospital’s service network. The catchment area of the José Casimiro Ulloa Emergency Hospital has remained stable, as its jurisdiction and referral scope have not changed over the past 20 years.

Comments 3: Data collection: did the study include all violence-related injuries presenting to the study hospital over the study period? How certain is complete case ascertainment?

Response 3: The dataset includes all records of violence-related visits registered in the institutional database, based on the documented reason for admission to the emergency department. However, as with most hospital-based data in Latin America, there is a potential for underreporting of the reason for admission, which may arise from incomplete information provided by the patient, the attending physician, or the data entry personnel. To minimize selection bias and ensure consistent case identification, the definition of “violence-related visit” was independently reviewed and validated by three of the investigators, all experts in emergency medicine and epidemiological data analysis.

To clarify this point in the “Data collection and variables” section we added: “The dataset consisted of routinely collected emergency department records, including the date of admission, sex, age, type of document presented at admission, and reason for consultation (motive of violence-related care). To minimize selection bias and en-sure consistent case identification, the definition of “violence-related visit” was inde-pendently reviewed and validated by three of the investigators, all experts in emer-gency medicine and epidemiological data analysis.”

Comments 4: WHO typology of violence needs a citation.

Response 4: We added the citation as follows: The number 2 according to our reference list corresponding to:  Violence Prevention Alliance Approach Available online: https://www.who.int/groups/violence-prevention-alliance/approach (accessed on 12 September 2025).

Comments 5: Table 1: It is difficult to compare the counts of injuries across row s because they are for different amounts of time. Would it be possible to report these in a standardized way for comparison (e.g. average cases per month or average cases per year) instead of just total cases for the time period? The percentages allow us to see how the composition of cases changed over time but not the volume.

Response 5: This is a very appropriate observation. However, due to the ethical committee’s data access regulations, the José Casimiro Ulloa Emergency Hospital provided the research team only with anonymized datasets corresponding to visits specifically classified as violence-related. The investigators do not have access to the complete institutional database for all emergency visits across the study years; therefore, it is not possible to calculate standardized proportions (e.g., average cases per month or per year) relative to the total emergency caseload. The analysis was thus limited to the available subset of violence-related records.

Comments 6: Table 1: the self inflicted violence, interpersonal community violence and family violence rows are missing percentages

Response 6: We have added the missing percentages. Additionally, we changed the color of the subcategories within interpersonal community violence to gray in the table, as shown below, to clarify that the proportions of these subcategories correspond to the overall interpersonal community violence category.

Comments 7:  Figures 2 and 3: labels and legends are too small to read

Response 7: We modified the legends to make them clearer as shown in the adjunt document

 

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

Dear authors

In this manuscript, the status of alcohol and drug use is important. It would have been better to explain this because it can lead to bias.

It seems necessary to discuss alcohol consumption in these cases, yet it is not mentioned in your research.

Author Response

COMMENT 1:

In this manuscript, the status of alcohol and drug use is important. It would have been better to explain this because it can lead to bias.

It seems necessary to discuss alcohol consumption in these cases, yet it is not mentioned in your research.

RESPONSE:

Unfortunately, the database available from the José Casimiro Ulloa Emergency Hospital does not include information on alcohol or drug use at the time of admission. However, this is a valid and relevant observation. We have therefore acknowledged this limitation in the revised manuscript, noting that the absence of such variables may restrict the interpretation of potential behavioral or contextual risk factors associated with violent events as ollows:

• “Taken together, our results situate Peru within a broader global narrative: while community violence was transiently suppressed by lockdown measures, domestic violence and self-harm followed more complex patterns, influenced by both access to care and underlying social determinants. These findings emphasize the need to interpret health-service data in light of complementary sources and to consider both abrupt shocks and cyclical processes when analyzing violence dynamics. It is also important to note that the available database did not include information on alcohol or drug use at the time of admission, which represents a limitation that may restrict the interpretation of behavioral or contextual risk factors underlying these patterns.”

 

 

Reviewer 3 Report

Comments and Suggestions for Authors

The present manuscript addresses a topic of great relevance in the field of public health and social epidemiology, by analyzing the impact of the COVID-19 pandemic on violence-related emergency visits at a referral hospital in Peru. It is a methodologically solid piece of work, with a clear structure and a rigorous analytical approach that employs advanced time-series techniques and the identification of structural breakpoints. The study covers an extensive period—from 2015 to 2024—providing a valuable longitudinal perspective to understand trends, seasonality, and the effects of the pandemic on different types of violence. Overall, the manuscript demonstrates internal coherence, appropriate use of statistical methods, and a sound interpretation of the results.

 

Regarding the Introduction, this section provides a precise contextualization of the problem of violence and the effects of confinement on various forms of aggression. The text shows a strong command of international literature and adequately justifies the relevance of the study. However, it would be advisable to incorporate a more explicit reference to the Peruvian context, including information on surveillance systems or national violence prevention policies, which would reinforce the local relevance of the work. It might also be useful to make the research hypothesis or questions explicit at the end of the section, allowing the reader to clearly identify the analytical purpose. Finally, a brief mention of how this study differs from previous work in the region could be added, emphasizing its original contribution.

With respect to the Materials and Methods, this section stands out for its transparency and level of detail. The description of the analytical process is comprehensive, the use of segmented regression is well justified, and the statistical criteria employed are clearly stated. This level of rigor represents one of the main strengths of the manuscript. Nevertheless, it would be advisable to justify more explicitly the choice of the hospital as the data source, indicating whether its records can be considered representative of the urban or national population. Clarification of how missing data were handled—especially the random imputation of cases with no information on migratory status—would also be helpful, along with a brief note on the validation or verification of diagnostic codes.

In the Results section, the manuscript presents information in a clear and well-structured manner, maintaining a proper balance between descriptive analysis, regression modeling, and graphical interpretation. A significant decrease in violence-related visits is evident during the early phase of the pandemic, followed by subsequent recovery and stabilization—findings consistent with international literature.

Regarding the Discussion, this is one of the most notable sections of the work. The text coherently integrates the results with previous studies, and the interpretation is cautious, well-supported, and consistent with available evidence. However, the discussion could be further strengthened by exploring in greater depth the social and structural factors that may explain the observed differences among types of violence. It is also suggested to expand the comparative analysis with other Latin American countries—such as Chile, Brazil, or Colombia—to situate the findings within a broader regional perspective. As for the limitations, although the author addresses them appropriately, a more critical reflection could be added on potential underreporting during lockdown, changes in registration criteria, or the lack of information regarding the severity of cases. Finally, it would be useful to include a closing paragraph discussing policy implications, emphasizing the need to strengthen hospital surveillance systems and to link them with intersectoral violence prevention strategies.

The Conclusions are concise and coherent with the study objectives. However, they could be reinforced with a final sentence that more explicitly summarizes the practical relevance of the work, highlighting the importance of strengthening information and surveillance systems in anticipation of future health crises. It is also suggested to avoid repeating limitations already mentioned in the discussion, in order to keep this section succinct and effective.

Author Response

COMMENTS 1: 

Regarding the Introduction, this section provides a precise contextualization of the problem of violence and the effects of confinement on various forms of aggression. The text shows a strong command of international literature and adequately justifies the relevance of the study. However, it would be advisable to incorporate a more explicit reference to the Peruvian context, including information on surveillance systems or national violence prevention policies, which would reinforce the local relevance of the work. It might also be useful to make the research hypothesis or questions explicit at the end of the section, allowing the reader to clearly identify the analytical purpose. Finally, a brief mention of how this study differs from previous work in the region could be added, emphasizing its original contribution.

RESPONSE 1

We sincerely thank the reviewer for the constructive observations. In response, the Introduction was revised to reinforce the Peruvian and regional context and to clarify the analytical purpose of the study. Specifically, new content was added to describe the national situation of violence in Peru, referencing data from the Instituto Nacional de Estadística e Informática (INEI) and recent reports indicating a marked increase in homicide rates during 2024–2025, which highlight the urgency of improving surveillance and prevention efforts. Additionally, the revised text now situates the problem within the broader Latin American context, incorporating studies from Brazil, Colombia, and Chile and acknowledging that the regional literature remains scarce and mainly focused on gender-based violence.

Furthermore, a clear research hypothesis was introduced at the end of the section, specifying that the COVID-19 pandemic likely produced a structural break in the temporal distribution of violence-related emergency visits, with heterogeneous effects depending on the type of violence. Finally, the original contribution of this study was emphasized, noting that it represents one of the few decade-long, hospital-based time-series analyses of violence in Latin America, providing valuable evidence to strengthen surveillance systems and inform public health policies in the region.

The changes can be found in the fourth and fifth parragraph of the introduction - lines 62 to 85- as follows:

Similar dynamics have been reported across Latin America, however the scientific literature on violence during this period is scarce and predominantly focused on gender-based and intimate partner violence, with limited evidence addressing community or self-inflicted forms. Evidence from the region revealed significant under-reporting of domestic violence in Brazil during lockdown[12], and transient increases in domestic and declines in community violence in Colombia and Brazil [13,14]. In Peru, by 2023, the National Institute of Statistics and Informatics (INEI) reported that more than 50% of women aged 15–49 had experienced some form of violence perpetrated by their partner [15]. Projections for Brazil suggest a concerning scenario, with estimates indicating a potential 95% increase in violence against women over the next decade [16] . Meanwhile, violence in Peru has shown a worrying upward trend, with around 500 homicides recorded every quarter during 2024 and 2025, reflecting the intensification of public insecurity nationwide [17]. In this context, examining how violence-related cases evolve within hospital emergency services becomes increasingly relevant to better understand the dynamics and health impact of this growing social problem in the region.

To address these gaps, we analyzed ten years (2015–2024) of violence-related visits at a major emergency hospital in Lima, Peru. Using the WHO typology, cases were classified into self-inflicted, interpersonal community, and interpersonal family violence. We combined descriptive statistics, stratified time-series analysis, segmented regression to identify structural breaks during and after the COVID-19 pandemic, and seasonal-trend decomposition to evaluate recurrent patterns. Our findings provide new insights into how violence-related emergencies evolved before, during, and after the pandemic, with implications for surveillance and prevention strategies in similar urban contexts.

 

COMMENTS 2

With respect to the Materials and Methods, this section stands out for its transparency and level of detail. The description of the analytical process is comprehensive, the use of segmented regression is well justified, and the statistical criteria employed are clearly stated. This level of rigor represents one of the main strengths of the manuscript. Nevertheless, it would be advisable to justify more explicitly the choice of the hospital as the data source, indicating whether its records can be considered representative of the urban or national population. Clarification of how missing data were handled—especially the random imputation of cases with no information on migratory status—would also be helpful, along with a brief note on the validation or verification of diagnostic codes.

RESPONSE 2

We greatly appreciate this thoughtful comment. Indeed, the initial intention of the study was to design a multicenter analysis to obtain a population representative of the different areas of Metropolitan Lima. However, after screening at least five major hospitals, we identified that none of them maintained an institutional record as extensive or detailed as that of the José Casimiro Ulloa Emergency Hospital (HEJCU). Most hospitals only register visits using ICD-10 codes, which leads to significant underreporting of violence-related cases compared to the “reason for visit” variable available at HEJCU. Expanding the study to other hospitals under these conditions would not have been methodologically coherent.

HEJCU is one of the oldest emergency hospitals in Lima and, although it has no assigned jurisdiction and is open to the public from any region of Peru, it predominantly serves the low-income population of the central–southern area of Lima. Approximately 97% of its patients are covered by the Seguro Integral de Salud (SIS), a national insurance program provided by the Peruvian government for individuals of limited financial resources.

Regarding data handling, missing information was excluded from the final dataset, but these cases represented less than 5% of the total records and did not affect the analysis. As for case identification, we relied on the “reason for visit” registry rather than ICD-10 codes. The hospital’s Office of Statistics and Informatics provided the research team with the complete list of “reason for visit” categories, from which three investigators of this study—each with expertise in emergency medicine and epidemiological research—independently reviewed and selected the categories related to violence. Their selections were subsequently discussed in joint sessions until a consensus was reached. Once the final set of categories was approved, the corresponding dataset including all visits under those selected categories was released to the research team.

We added the changes in Study design and setting section (lines 94 to 99): The hospital serves the general public and primarily attends low-income and uninsured populations from Lima and surrounding provinces, as it is a public emergency facility open to all individuals requiring urgent care, regardless of socioeconomic status or geographic origin. The analysis aimed to describe the epidemiological characteristics of these visits and to examine temporal patterns, including trends, seasonality, and structural breaks associated with the COVID-19 

Data collection and variables section lines 104-109

The hospital’s Office of Statistics and Informatics provided the research team with the complete list of “reason for visit” categories, from which three investigators of this study—each with expertise in emergency medicine and epidemiological research—independently reviewed and selected those corresponding to violence-related care. Their selections were subsequently discussed in joint sessions until a consensus was reached, ensuring consistent and unbiased case identification.

 

COMMENTS 3

In the Results section, the manuscript presents information in a clear and well-structured manner, maintaining a proper balance between descriptive analysis, regression modeling, and graphical interpretation. A significant decrease in violence-related visits is evident during the early phase of the pandemic, followed by subsequent recovery and stabilization—findings consistent with international literature.

RESPONSE 3

 We clarified this points on the “Data collection and variables” section as follows:

The dataset consisted of routinely collected emergency department records, including the date of admission, sex, age, type of document presented at admission, and reason for consultation (motive of violence-related care). The hospital’s Office of Statistics and Informatics provided the research team with the complete list of “reason for visit” categories, from which three investigators of this study—each with expertise in emer-gency medicine and epidemiological research—independently reviewed and selected those corresponding to violence-related care. Their selections were subsequently dis-cussed in joint sessions until a consensus was reached, ensuring consistent and unbi-ased case identification.

Sociodemographic variables included sex (female, male), life stage, classified into child (0–11 years), adolescent (12–17 years), youth (18–29 years), adult (30–59 years), and elderly (≥60 years), and migratory status. For the latter, document categories were col-lapsed into two groups: foreigners (foreign documentation, foreign identification card, passport, were assigned to this category) and natives (including national ID, national auto-generated temporary ID codes, newborn codes, and national military/police codes). Missing information was excluded from the final dataset; however, these cases represented less than 5% of the total and did not affect the overall analysis.

The type of violence was reorganized into three analytical categories following the World Health Organization (WHO) typology of violence [2]. (1) Self-inflicted violence included all cases coded as self-harm or auto-aggression. (2) Interpersonal community violence grouped physical assault, robbery with assault, gunshot injuries, stabbing, robbery with firearm, sexual assault, rape, kidnapping, and referred sexual or physical violence. (3) Interpersonal family violence included all cases recorded as domestic or intrafamilial violence. This WHO-based harmonization allowed for consistency and comparability across the study period. Although the HEJCU has no assigned jurisdic-tion and is open to patients from across Peru, it predominantly serves low-income populations from the central–southern area of Lima, with approximately 97% of pa-tients covered by the Seguro Integral de Salud (SIS), a national insurance program for individuals of limited financial resources

COMMENTS 4

Regarding the Discussion, this is one of the most notable sections of the work. The text coherently integrates the results with previous studies, and the interpretation is cautious, well-supported, and consistent with available evidence. However, the discussion could be further strengthened by exploring in greater depth the social and structural factors that may explain the observed differences among types of violence. It is also suggested to expand the comparative analysis with other Latin American countries—such as Chile, Brazil, or Colombia—to situate the findings within a broader regional perspective. As for the limitations, although the author addresses them appropriately, a more critical reflection could be added on potential underreporting during lockdown, changes in registration criteria, or the lack of information regarding the severity of cases. Finally, it would be useful to include a closing paragraph discussing policy implications, emphasizing the need to strengthen hospital surveillance systems and to link them with intersectoral violence prevention strategies.

RESPONSE 4

We sincerely appreciate the reviewer’s positive assessment and constructive suggestions regarding the Discussion section. In response, several improvements were incorporated to strengthen its analytical depth and contextual relevance. Specifically, we expanded the discussion of social and structural factors that may explain the observed differences among types of violence, highlighting the role of socioeconomic inequality, mobility restrictions, and reduced access to health and social protection services during the pandemic. We also added a comparative regional perspective, referencing studies from Latinameria, to situate our findings within the broader Latin American context.

Additionally, the limitations section was refined to include a more critical reflection on potential underreporting during lockdown periods, changes in registration practices, and the absence of information on case severity. Finally, we incorporated a new concluding paragraph on policy implications, emphasizing the urgent need to strengthen hospital-based surveillance systems, improve data quality through registries that include the reason for visit (rather than relying solely on ICD-10 codes), and link these systems with intersectoral strategies for violence prevention and response.

COMMENTS 5

The Conclusions are concise and coherent with the study objectives. However, they could be reinforced with a final sentence that more explicitly summarizes the practical relevance of the work, highlighting the importance of strengthening information and surveillance systems in anticipation of future health crises. It is also suggested to avoid repeating limitations already mentioned in the discussion, in order to keep this section succinct and effective.

RESPONSE

We appreciate the reviewer’s insightful feedback. In response, the Conclusions section was revised to include a final sentence emphasizing the practical relevance of the study, highlighting the importance of strengthening health information and surveillance systems to improve preparedness for future health crises. Additionally, previously mentioned limitations were removed from this section to maintain conciseness and ensure that the conclusions remain focused on the main findings and their implications for policy and public health practice.

Round 2

Reviewer 3 Report

Comments and Suggestions for Authors

I have reviewed the document once again and confirm that the changes and adjustments suggested in the previous version have been properly implemented. A significant improvement can be observed in the conceptual clarity, argumentative coherence, and theoretical foundation of the work. Likewise, the structure of the text reflects careful and consistent work, with a clear line of reasoning that enhances its readability and overall comprehension.

I would like to congratulate you on the effort, dedication, and notable improvement compared to the initial version. The final result demonstrates a strong commitment to scientific quality and precision in the presentation of information.