Association Between the Introduction of Pediatric Influenza Vaccination and Influenza Diagnoses in Primary Care and Hospitalizations: An Interrupted Time Series Study

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Estimated Authors,

I've read with interest the present paper entitled "ASSOCIATION BETWEEN THE INTRODUCTION OF PEDIATRIC INFLUENZA VACCINATION AND INFLUENZA DIAGNOSES IN PRIMARY CARE AND HOSPITALIZATIONS: AN INTERRUPTED TIME SERIES STUDY".

In this study, a total of 6804 influenza diagnoses in primary care settings, and 3252 hospitalizations associated with diagnosis of influenza, were included into the analyses. Authors dichotomized their data by pre-universal vaccination strategy vs. post-universal vaccination strategy. In the end, a substantial decrease was observed in the percentage of influenza diagnoses in primary care settings in the age groups 2 to 4 years (13.5% vs. 10.6%) and 5 to 14 years (26.1% vs. 16.3%). Focusing on the interrupted time series analysis, a reduction in the expected influenza cases in the 15–64 age group in primary care was noticed (RR 0.87, 95%CI 0.81 to 0.96). In the hospital setting, a certain reduction in admissions was observed in the 5–14 age group (RR=0.25, 95%CI 0.06 to 0.92). These results are substantially in line with expected estimates from international experiences. However, in my opinion, some improvements are in need before the full acceptance of the paper.

Please take a note of following remarks:

1) Table 1 should be moved to appendix material

2) The same for Table 2

3) in my opinion, reporting strategy may be somehow confusing and misleading. According to your aims, the present paper is designed to provide an assessment of "the association between the introduction of systematic influenza vaccination in children aged 6 to 59 months and changes in the number of influenza diagnoses in primary care (PC) and hospitalizations for influenza for all age groups in the Central Catalonia health region, Spain".  Table 3 should be therefore simplified by including a summary column ("TOTAL") and only retaining the columns for "period". Data on vaccination are interesting, but not consistently addressed in results, and therefore could be moved to appendix material alongside tables 1 to 2.

4) Chapter 3.3 should be refined being more consistent with the results from supplementary material. As supplementary material could be not so easy spotted as appendix material, Authors should report in full details estimates with incertitude index (i.e. 95%CI) the risk estimates. Please note that Appendix material will be included in the main file, attached at the end of the paper, while supplementary material will be included as a separate file.

5) Figure 3 is misleading as it includes only some of the age groups, and the included age groups are not associated with significant differences previously addressed. Please either remove or expand with the whole of addressed age groups.

Author Response

We thank the reviewer for their constructive suggestions, which have helped improve the clarity and structure of the manuscript.

 

• Table 1 should be moved to appendix material

 

We agree and have moved Table 1 to the supplementary material. 

 

• The same for Table 2

Table 2 has also been moved to the supplementary material.

• In my opinion, reporting strategy may be somehow confusing and misleading. According to your aims, the present paper is designed to provide an assessment of "the association between the introduction of systematic influenza vaccination in children aged 6 to 59 months and changes in the number of influenza diagnoses in primary care (PC) and hospitalizations for influenza for all age groups in the Central Catalonia health region, Spain".  Table 3 should be therefore simplified by including a summary column ("TOTAL") and only retaining the columns for "period". Data on vaccination are interesting, but not consistently addressed in results, and therefore could be moved to appendix material alongside tables 1 to 2.

 

We have simplified Table 3 to include only the summary column (“Total”) and the columns for “period,” as suggested. Data related to vaccination have been moved to the supplementary material as the reviewer has suggested. 

 

• Chapter 3.3 should be refined being more consistent with the results from supplementary material. As supplementary material could be not so easy spotted as appendix material, Authors should report in full details estimates with incertitude index (i.e. 95%CI) the risk estimates. Please note that Appendix material will be included in the main file, attached at the end of the paper, while supplementary material will be included as a separate file.

 

Thank you for your comment. Chapter 3.3 has been refined. It has been reported full details estimates with incertitude index the risk estimates in the text. 

 

• Figure 3 is misleading as it includes only some of the age groups, and the included age groups are not associated with significant differences previously addressed. Please either remove or expand with the whole of addressed age groups.

 

Thank you for your comment. Figure 3 has been removed as the reviewer has suggested. 

 

All comments have been addressed, and we believe the manuscript is now clearer and more consistent.

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

This manuscript presents a quasi-experimental interrupted time series (ITS) analysis evaluating the population-level impact of the 2023 introduction of universal influenza vaccination for children aged 6–59 months in Central Catalonia, Spain. The work addresses a critical evidence gap in Southern European influenza immunization policy research, aligns directly with the scope of Vaccines, and has notable strengths: it uses a methodologically appropriate design for public health intervention evaluation, leverages comprehensive 7-year real-world data from a well-defined health region (including 6804 primary care influenza diagnoses and 3252 influenza-related hospitalizations), maintains full ethical compliance, and appropriately acknowledges core limitations without overstating preliminary findings. However, the manuscript contains critical methodological, statistical, and narrative flaws that must be rigorously addressed to meet the journal’s publication standards.

Major Revisions

1. The authors performed stratified ITS analyses across 5 age groups for both primary care and hospitalization outcomes without adjusting for multiple comparisons, which substantially increases the risk of false-positive results—especially for the borderline significant finding of reduced hospitalizations in the 5–14 years age group (RR=0.25, 95%CI 0.06–0.92, p=0.043). Apply a validated multiple testing correction (e.g., Bonferroni or Holm-Bonferroni method) to all subgroup analyses, update all p-values and 95% confidence intervals with corrected values in the Abstract, Results section, and supplementary tables, and revise all narrative content to only present corrected statistically significant findings as core positive results. Any results that lose significance after correction must be described with neutral, non-causal language.
2. The authors note that rapid influenza diagnostic tests were introduced in primary care in December 2020, which markedly improved case detection sensitivity. This change acts as a strong time-varying confounder: increased diagnostic capacity would artificially elevate recorded influenza cases post-2020, directly counteracting and diluting the true protective effect of the pediatric vaccination program, yet this critical variable was not included in the segmented negative binomial regression models. Incorporate a dummy variable for the introduction of rapid diagnostic tests into all ITS regression models, re-calculate all relative risk (RR) estimates, 95% CIs, and p-values associated with the vaccination intervention, and fully update the Methods, Results, and Discussion sections to reflect the adjusted findings and their impact on the study’s conclusions.
3. The authors classified influenza cases complicated by otitis media as "severe influenza" in the primary care analysis, which is inconsistent with the global consensus for severe influenza in vaccine epidemiology (otitis media is universally categorized as a mild, non-severe complication; severe influenza should be restricted to life-threatening or invasive complications such as pneumonia, encephalopathy, and myocarditis). Revise the severe influenza case definition to exclude otitis media cases, re-analyze trends in severe influenza using the corrected definition, and update all relevant content in the Methods, Results, and Discussion sections with a clear rationale for the definition revision.
4. Throughout the manuscript, the authors use causal language (e.g., "reduction", "decrease") to describe non-statistically significant downward trends in RR estimates, which is methodologically inappropriate. Non-significant results cannot be interpreted as evidence of a true protective effect, and the ecological study design precludes robust causal inference. Revise all narrative related to non-statistically significant results to use only neutral descriptive language (e.g., "non-significant downward trend"), remove all definitive causal inference language from the Abstract and Conclusion sections, and explicitly state that the study identifies an association, not a causal effect, between the vaccination program and influenza outcome trends, with clear emphasis on the limitations that restrict causal interpretation.
5. While the authors acknowledge some study limitations, the Discussion section lacks a detailed, quantitative analysis of how key limitations affect result interpretation. Substantially expand the section to include: a detailed analysis of how the introduction of rapid diagnostic tests biased effect estimates, including a quantitative description of the potential dilution of the true vaccine effect; an in-depth discussion of the impact of low vaccination coverage (19% in 2023/24, 27% in 2024/25), which is far below the 50–60% threshold required for robust herd protection, with contextualization against high-coverage published studies; and an explicit description of the "ecological fallacy" of this aggregate-level analysis, clarifying that population-level associations cannot confirm individual-level vaccine effectiveness, with a proposal for subsequent individual-level studies to validate findings.
6. Perform a sensitivity analysis by excluding the 2020–2021 influenza season (during which nearly no influenza cases were recorded due to COVID-19 restrictions) and re-running the ITS models, with results included in the supplementary materials to validate the robustness of core findings.
7. Add data on the matching rate between seasonal influenza vaccine strains and circulating strains for each study season, and discuss its potential impact on observed vaccine effect estimates in the Discussion section.
8. Fully reformat supplementary Table S1, which is currently poorly organized with incomplete data, typographical errors, and low readability, to present complete, standardized results of the adjusted negative binomial regressions including full RR estimates, 95% CIs, and corrected p-values for all variables across all age groups.
9. Revise the Introduction section to more specifically focus on the research gap of universal pediatric influenza vaccination policy in Southern Europe/Spain, rather than overly generalized background on influenza, to strengthen the study’s rationale.
10. Complete the truncated concluding sentence in the Abstract, and ensure the abstract adheres to Vaccines formatting guidelines with a full, concise conclusion aligned with the study’s final findings.
11. Standardize abbreviation use throughout the manuscript, ensuring all abbreviations (e.g., PC, ICS, ITS) are defined with their full terms at first mention in the Abstract and main text.
12. Clarify the specific influenza diagnostic criteria (e.g., PCR confirmation, rapid antigen test, clinical diagnosis) for primary care cases in the Methods section to improve study transparency and reproducibility.

Author Response

REVIEWER 2

 

We are grateful for the reviewer’s detailed and methodologically rigorous comments, which have significantly strengthened the manuscript.

 

1)The authors performed stratified ITS analyses across 5 age groups for both primary care and hospitalization outcomes without adjusting for multiple comparisons, which substantially increases the risk of false-positive results—especially for the borderline significant finding of reduced hospitalizations in the 5–14 years age group (RR=0.25, 95%CI 0.06–0.92, p=0.043). Apply a validated multiple testing correction (e.g., Bonferroni or Holm-Bonferroni method) to all subgroup analyses, update all p-values and 95% confidence intervals with corrected values in the Abstract, Results section, and supplementary tables, and revise all narrative content to only present corrected statistically significant findings as core positive results. Any results that lose significance after correction must be described with neutral, non-causal language.

 

We have applied a Bonferroni correction to all subgroup analyses. All p-values and confidence intervals have been updated accordingly in the Abstract, Results section, and supplementary tables. The narrative has been revised to reflect only corrected statistically significant findings.

2)The authors note that rapid influenza diagnostic tests were introduced in primary care in December 2020, which markedly improved case detection sensitivity. This change acts as a strong time-varying confounder: increased diagnostic capacity would artificially elevate recorded influenza cases post-2020, directly counteracting and diluting the true protective effect of the pediatric vaccination program, yet this critical variable was not included in the segmented negative binomial regression models. Incorporate a dummy variable for the introduction of rapid diagnostic tests into all ITS regression models, re-calculate all relative risk (RR) estimates, 95% CIs, and p-values associated with the vaccination intervention, and fully update the Methods, Results, and Discussion sections to reflect the adjusted findings and their impact on the study’s conclusions.

 

We acknowledge the importance of this point. However, the change points corresponding to the onset of the pandemic and the introduction of testing occur in a period of low incidence and limited variability, which prevents estimating their effects independently due to issues of collinearity and model identifiability. For this reason, it was decided to include only a single interruption point (the onset of the pandemic), which captures the main structural change observed in the series.We have clarified this rationale in the Methods and Discussion sections and expanded the discussion on how changes in testing may have biased incidence estimates.       

 

3) The authors classified influenza cases complicated by otitis media as "severe influenza" in the primary care analysis, which is inconsistent with the global consensus for severe influenza in vaccine epidemiology (otitis media is universally categorized as a mild, non-severe complication; severe influenza should be restricted to life-threatening or invasive complications such as pneumonia, encephalopathy, and myocarditis). Revise the severe influenza case definition to exclude otitis media cases, re-analyze trends in severe influenza using the corrected definition, and update all relevant content in the Methods, Results, and Discussion sections with a clear rationale for the definition revision.

 

We have revised the definition of severe influenza and excluded otitis media cases accordingly. The Methods and Results sections have been updated to reflect these changes. Upon review of the database, we confirmed that no cases of otitis media were identified; therefore, no reanalysis was required.

 

4) Throughout the manuscript, the authors use causal language (e.g., "reduction", "decrease") to describe non-statistically significant downward trends in RR estimates, which is methodologically inappropriate. Non-significant results cannot be interpreted as evidence of a true protective effect, and the ecological study design precludes robust causal inference. Revise all narrative related to non-statistically significant results to use only neutral descriptive language (e.g., "non-significant downward trend"), remove all definitive causal inference language from the Abstract and Conclusion sections, and explicitly state that the study identifies an association, not a causal effect, between the vaccination program and influenza outcome trends, with clear emphasis on the limitations that restrict causal interpretation.

 

We have thoroughly revised the manuscript to remove causal language. All findings are now described as associations, and non-significant results are presented using neutral terminology.

 

5) While the authors acknowledge some study limitations, the Discussion section lacks a detailed, quantitative analysis of how key limitations affect result interpretation. Substantially expand the section to include: a detailed analysis of how the introduction of rapid diagnostic tests biased effect estimates, including a quantitative description of the potential dilution of the true vaccine effect; an in-depth discussion of the impact of low vaccination coverage (19% in 2023/24, 27% in 2024/25), which is far below the 50–60% threshold required for robust herd protection, with contextualization against high-coverage published studies; and an explicit description of the "ecological fallacy" of this aggregate-level analysis, clarifying that population-level associations cannot confirm individual-level vaccine effectiveness, with a proposal for subsequent individual-level studies to validate findings.

 

A detailed analysis of how the introduction of rapid diagnostic tests biased effect estimates and a quantitative description of the potential dilution of the true vaccine effect has been added. 

 

The inserted paragraph is as follows:  “Before the availability of these tests, the diagnosis of influenza was based mainly on clinical criteria, with limited sensitivity, which contributed to underdiagnosis and the classification of many cases as nonspecific respiratory infections [20]. Rapid test, although they have moderate sensitivity (approximately 50-70%), show high specificity (>95%), which allows confirmation of a greater number of true cases [21]. In population terms, their implementation may result in an apparent increase in recorded incidence due to the reduction of  prior underdiagnosis. “

 

An in-depth discussion of the impact of low vaccination coverage with contextualization against high-coverage published studies has been added. 

 

The inserted paragraph is as follows: “The United Kingdom's experience during the 2014–15 season, with an average coverage of 56% (children aged 4 to 11), resulted in a 94% reduction in primary care consultations for influenza-like illness, a 74% decrease in emergency visits for respiratory disease, and a 93% reduction in hospitalizations for laboratory-confirmed influenza among schoolchildren. A community benefit was also observed, corresponding to a 59% decrease in consultations for influenza-like illness in adults [4].”

 

Another paragraph has been added to explain the “ecological fallacy”: “Furthermore, the ecological nature of this analysis introduces the possibility of an ecological fallacy, as associations observed at the population level cannot be directly extrapolated to infer individual-level vaccine effectiveness. Therefore, the findings should be interpreted with caution, and future studies based on individual-level data will be necessary to validate these results and better estimate the direct and indirect effects of vaccination”.

 

6)Perform a sensitivity analysis by excluding the 2020–2021 influenza season (during which nearly no influenza cases were recorded due to COVID-19 restrictions) and re-running the ITS models, with results included in the supplementary materials to validate the robustness of core findings.

 

A sensitivity analysis by excluding the 2020-2021 influenza season (during which nearly no influenza cases were recorded due to COVID-19 restrictions) has been performed.  Results are presented in supplementary Tables S4–S6, and the main text has been updated accordingly.

 

7)Add data on the matching rate between seasonal influenza vaccine strains and circulating strains for each study season, and discuss its potential impact on observed vaccine effect estimates in the Discussion section.

 

Data on matching rate between seasons influenza vaccine strains and circulating strains for each study season has been added. 

 

The inserted paragraph is as follows: “The annual variability in the effectiveness of the influenza vaccine may also have influenced the observed results, given its dependence on the degree of antigenic match between the vaccine strains and circulating viruses. In Spain, during the 2023-2024 season, influenza A (H1N1)pdm09 predominated, with most circulating viruses belonging to clade 5a.2a, whereas the vaccine strain corresponded to subclade 5a.2a.11, suggesting a close but not complete antigenic match. This partial concordance may have contributed to the relatively favorable vaccine effectiveness observed, while still allowing for some attenuation of its impact. In contrast, the 2024-2025 season was characterized by the co-circulation of influenza A and B viruses, indicating greater viral heterogeneity. Although this pattern is broadly compatible with the composition of quadrivalent vaccines, intra-subtype variability-particularly for A (H3N2)- may still have limited the overall vaccine effectiveness. These seasonal differences in strain match may therefore have contributed to variability in the estimated impact of the vaccination program across seasons [23,24].”

 

8) Fully reformat supplementary Table S1, which is currently poorly organized with incomplete data, typographical errors, and low readability, to present complete, standardized results of the adjusted negative binomial regressions including full RR estimates, 95% CIs, and corrected p-values for all variables across all age groups.

 

Supplementary Table S1 has been reorganized into three separate tables (S1–S3) to improve readability.

 

9) Revise the Introduction section to more specifically focus on the research gap of universal pediatric influenza vaccination policy in Southern Europe/Spain, rather than overly generalized background on influenza, to strengthen the study’s rationale.

 

The Introduction has been revised to better emphasize the research gap regarding universal pediatric influenza vaccination in Southern Europe and Spain.

 

10) Complete the truncated concluding sentence in the Abstract, and ensure the abstract adheres to Vaccines formatting guidelines with a full, concise conclusion aligned with the study’s final findings.

 

The abstract has been revised and now includes a complete and concise conclusion aligned with the study findings. The conclusion in the abstract has been changed to: “Systematic influenza vaccination in children aged 6 to 59 months has not been shown to be associated with a reduction in influenza cases in primary care or in hospitals during the early stages of implementation of the new vaccination program”.

 

11) Standardize abbreviation use throughout the manuscript, ensuring all abbreviations (e.g., PC, ICS, ITS) are defined with their full terms at first mention in the Abstract and main text.

All abbreviations have been standardized and defined at first mention.

 

12) Clarify the specific influenza diagnostic criteria (e.g., PCR confirmation, rapid antigen test, clinical diagnosis) for primary care cases in the Methods section to improve study transparency and reproducibility.

 

We have clarified diagnostic criteria in the Methods section, specifying the use of clinical diagnosis, rapid tests, and PCR depending on the study period and setting.This paragraph has been added in the text: “Influenza records in primary care (PC) refer to children with a clinical diagnosis of influenza before December 2020, as rapid influenza tests were not available in PC prior to this date. After December 2020, patients included in the study were those with a diagnosis of influenza, whether clinical or confirmed by a rapid test. Patients registered with influenza in the hospital had either a rapid diagnostic test or a PCR test for the influenza virus.”

 

We believe these revisions have substantially strengthened the methodological rigor and transparency of the study.

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

The manuscript addresses an important public health question: the population-level impact of introducing pediatric influenza vaccination. The use of real-world data and an interrupted time series design is appropriate and potentially valuable. However, methodological ambiguities limit analytical depth, and the overinterpretation of weak signals currently undermines the robustness of the conclusions.

Major points

1. The ITS design is appropriate, but causal conclusions are limited by the short post-intervention period, the impact of COVID-19, and concurrent changes (such as rapid testing). Although this is acknowledged, the manuscript still suggests causality. It would be better to frame the results as associations and, if possible, include a control outcome or group.
2. The 2020–21 season had almost no cases, creating a break in the time series. It is unclear whether the model properly accounted for post-COVID recovery or seasonality. This may bias the results. The authors should clarify how this was handled and whether sensitivity analyses were performed.
3. The model is appropriate, but important details are missing (dispersion, offsets, autocorrelation, choice of interruption points, and lags). The model description should be expanded to include diagnostics and sensitivity analyses.
4. Vaccination coverage is low (19–27%) and not well explored. There is no stratified analysis or use as a continuous variable. Ideally, coverage should be incorporated into the model.
5. Primary care diagnoses may have been influenced by changes in testing and post-COVID healthcare-seeking behavior. Hospitalizations are more robust but may still vary due to coding and testing practices. Case definitions and changes in testing rates should be clarified.
6. Some results are highlighted despite lacking statistical significance and having wide confidence intervals. RRs <1 are interpreted as trends, which may introduce bias. The focus should be on uncertainty rather than direction.
7. The main significant effect appears in a non-vaccinated group (5–14 years). This is interpreted as an indirect effect, but there is no supporting data. It should be presented as a hypothesis rather than evidence of herd immunity.

Minor points

1. Standardizing metrics across sections.

Author Response

REVIEWER 3

 

We thank the reviewer for their insightful and constructive comments.

 

• The ITS design is appropriate, but causal conclusions are limited by the short post-intervention period, the impact of COVID-19, and concurrent changes (such as rapid testing). Although this is acknowledged, the manuscript still suggests causality. It would be better to frame the results as associations and, if possible, include a control outcome or group.

We have revised the manuscript to consistently present findings as associations rather than causal effects.

 

• The 2020–21 season had almost no cases, creating a break in the time series. It is unclear whether the model properly accounted for post-COVID recovery or seasonality. This may bias the results. The authors should clarify how this was handled and whether sensitivity analyses were performed.

 

We have conducted a sensitivity analysis excluding the 2020–2021 season and updated the Abstract, Methods,Results, and Discussion sections accordingly.The time series (TS) models have been rerun and are presented in the Supplementary Materials (Tables S4, S5, and S6). 

 

• The model is appropriate, but important details are missing (dispersion, offsets, autocorrelation, choice of interruption points, and lags). The model description should be expanded to include diagnostics and sensitivity analyses.

 

Thank you for your comment. A negative binomial regression was used to correct for the overdispersion present in the data; for this reason, a Poisson regression was not fitted. In all models, theta values close to 1 were observed, indicating clear overdispersion. No offset was included because the model was fitted using counts and the population remained stable during the study period.

Regarding the time points, the onset of COVID-19 and the start of vaccination were selected for two reasons: first, the emergence of COVID-19 represented a turning point in the number of influenza infections, leading to a reduction; and second, systematic influenza vaccination constitutes the intervention of primary interest to be evaluated.

The model description has been expanded to include diagnostics and sensitivity analyses. The table can be found in the supplementary materials. This table has been rearranged into three tables to be more readable. 

 

• Vaccination coverage is low (19–27%) and not well explored. There is no stratified analysis or use as a continuous variable. Ideally, coverage should be incorporated into the model.

We thank the reviewer for this insightful comment. We have clarified in the Discussion that the relatively low vaccination coverage limits the ability to detect population-level effects and discussed this in the context of existing literature.

We agree that vaccination coverage is an important factor when evaluating the population-level impact of a vaccination program. However, in our study, incorporating vaccination coverage into the interrupted time series (ITS) model was not feasible due to several methodological constraints.

First, our analysis is based on aggregated monthly data at the population level, and we did not have access to time-resolved or stratified vaccination coverage (e.g., by month, age group, or healthcare area) that would allow its inclusion as a continuous or stratified variable in the model. The available coverage estimates (approximately 19% and 27% for the two post-intervention seasons) are season-level summaries and therefore lack sufficient temporal variability to be meaningfully incorporated into the ITS framework.

Second, vaccination coverage is intrinsically linked to the intervention itself (i.e., the introduction of the vaccination program), which is already modeled as a structural change in the ITS design. Including coverage as a covariate under these conditions could introduce collinearity and complicate the interpretation of the intervention effect.

Third, given the ecological nature of the study, incorporating aggregate coverage data into the model could increase the risk of ecological bias, without necessarily improving the causal interpretation of the results.

To address this important point, we have clarified in the Discussion section that the relatively low vaccination coverage during the first seasons of implementation likely limits the detection of population-level effects, and we have emphasized the need for future studies with higher coverage and individual-level data to better assess dose–response relationships and indirect effects.

We hope this clarification adequately addresses the reviewer’s concern.

• Primary care diagnoses may have been influenced by changes in testing and post-COVID healthcare-seeking behavior. Hospitalizations are more robust but may still vary due to coding and testing practices. Case definitions and changes in testing rates should be clarified.

 

.We have clarified diagnostic criteria and changes in testing practices in the Methods section.. It has been added the next text in the “Material and methods sections”: “Influenza records in primary care (PC) refer to children with a clinical diagnosis of influenza before December 2020, as rapid influenza tests were not available in PC prior to this date. After December 2020, patients included in the study were those with a diagnosis of influenza, whether clinical or confirmed by a rapid test. Patients registered with influenza in the hospital had either a rapid diagnostic test or a PCR test for the influenza virus”.

• Some results are highlighted despite lacking statistical significance and having wide confidence intervals. RRs <1 are interpreted as trends, which may introduce bias. The focus should be on uncertainty rather than direction.

 

We have revised the manuscript to focus on uncertainty and removed interpretations based solely on the direction of non-significant estimates.

 

• The main significant effect appears in a non-vaccinated group (5–14 years). This is interpreted as an indirect effect, but there is no supporting data. It should be presented as a hypothesis rather than evidence of herd immunity.

 

Thank you for your comment. In response to the reviewers’ comments, the Bonferroni correction was applied and a sensitivity analysis was conducted. As no association was identified between the new influenza vaccination program and the 5–14-year age group, these results have been omitted from the revised manuscript.

 

Minor points

 

• Standardizing metrics across sections.

 

Thank you for your comment. Metrics across sections has been reviewed. 

We believe the manuscript now provides a more cautious and methodologically robust interpretation of the findings.

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

This manuscript has been revised accordingly and is now acceptable.

Reviewer 3 Report

Comments and Suggestions for Authors

The authors accepted my suggestions and made the necessary revisions satisfactory or provided clear and satisfactory explanations for most issues. These changes allowed greater clarity for the reader, improving the quality of the work. I recommend publishing the revised version in its current form.