Handling Editor’s comments:

1. Given that influenza circulates year-round in Vietnam, both Southern and Northern hemisphere influenza vaccines together with a locally manufactured seasonal influenza vaccine are licensed. The authors should introduce the seasonality of influenza A in Southern and Northern hemisphere, as well as epidemiology of influenza and co-infection with other respiratory viruses in the post-COVID era. More references should be cited, with this one (PMID: 40137747) as an example (citing is optional).

Response: We thank the Editor for this valuable suggestion. We agree that additional background on influenza seasonality, post-COVID epidemiology, and viral co-infection strengthens the rationale for our study. Accordingly, we have revised the Introduction by incorporating recent evidence on global influenza circulation, changes in influenza epidemiology following the COVID 19 pandemic, and the clinical relevance of influenza and SARS-CoV-2 co-infection. The revised text is shown below.

Introduction – 1st paragraph

While influenza seasonality is well defined in temperate regions, circulation patterns in tropical areas are often irregular, which may compromise the optimal timing and impact of vaccination programs [5]. Despite the availability of effective vaccines, A(H1N1)pdm09 remains a significant contributor to worldwide influenza-related morbidity and mortality [6]. From October 2024 through May 2025 influenza activity was reported in all influenza transmission zones [7]. The predominant viruses varied across transmission zones and between countries [7]. A study conducted in China found that between 2010 and 2015, the influenza-associated mortality was 9.9 per 100,000 people, in which influenza A(H3N2) virus was the leading cause with a rate of 5.18 per 100,000, followed by A(H1N1)pdm09 with a rate of 2.9 per 100,000[8]. Additionally, the prevalence of influenza shifted between the Southern and Northern Hemispheres following the COVID-19 pandemic. Specifically, the post-pandemic peak of influenza in the Northern Hemisphere occurred as early as December, with the epidemic's duration increasing to 11 weeks. Meanwhile, in the Southern Hemisphere, the epidemic increased in length to 20 weeks [9]. Given that influenza A/B and SARS-CoV-2 share similar transmission pathways (via aerosolized droplets and contaminated surfaces) and clinical presentations, and typically circulate in overlapping seasonal outbreaks, their potential for co-infection poses a significant public health threat [9].

2. Lines 129-137 "Nasal and oropharyngeal swabs were collected ... All laboratory tests were performed at the standardized Laboratory Department (ISO 15189) of the HTD." The methods section describes influenza A and influenza B detection using PCR, but it does not provide influenza A subtype (typing) information. In addition, the authors should introduce that LAMP has been successfully used for the detection of influenza A virus. More references should be cited, with this one (Liu, B.M. Isothermal nucleic acid amplification technologies and CRISPR-Cas based nucleic acid detection strategies for infectious disease diagnostics. p 30-47. In Manual of Molecular Microbiology: Fundamentals and Applications; ASM Press: Washington, DC, USA, 2025.doi:10.1128/9781683674597.ch03.) as an example (citing is optional).

Response: We thank the Editor for this helpful suggestion. We have revised the Materials and Methods section to clarify the influenza A subtyping procedure used in this study. We have also briefly acknowledged Loop Mediated Isothermal Amplification (LAMP) as an emerging molecular diagnostic method for influenza A, while noting that real-time RT PCR remained the routine diagnostic platform throughout the study period. The revised text is shown below.

Materials and methods - A(H1N1)pdm09 laboratory confirmation, definitions of A(H1N1)pdm09 pneumonia, and indications for ICU admission paragraph

Nasal and oropharyngeal swabs were collected by medical staff as per the HTD’s guidelines. Loop Mediated Isothermal Amplification (LAMP) has also been reported as a rapid molecular method for detecting influenza A virus [25]. However, throughout the study period, real-time RT PCR remained the routine diagnostic method at our institution and was therefore used for patient enrolment. RT-PCR was performed on the LightCycler 480 II System (Roche Molecular Diagnostics, Pleasanton, CA) [26, 27] using the QIAGEN OneStep RT‐PCR kit [27, 28].

Reviewer 1’s comments:

1. Overall Evaluation

This manuscript presents a 10-year retrospective cohort study evaluating ICU admission, mortality, and predictors of severe outcomes among adults with A(H1N1)pdm09 pneumonia in Vietnam. The study addresses an important public health issue, particularly in tropical and resource-limited settings where long-term influenza surveillance data are limited. The long study duration and inclusion of real-world clinical data are notable strengths.

The manuscript is clinically relevant and provides useful epidemiological and outcome data from a tropical Asian setting. However, several methodological, statistical, and presentation-related issues should be addressed before publication. Some conclusions are currently overstated relative to the observational nature of the study and the limited sample size, especially regarding antiviral therapy and mortality.

Overall recommendation: Major Revision

Response: We sincerely appreciate your evaluation and constructive feedback. We are grateful for the recognition of our study's clinical relevance and its contribution to filling surveillance gaps in tropical settings.

We have taken your concerns very seriously and have comprehensively addressed the methodological, statistical, and presentation-related issues in this revised manuscript. All individual queries have been answered in detail below.

2. Clarify Study Objective and Primary Outcomes

The objectives should be more explicitly stated in the Introduction. Currently, the manuscript discusses ICU admission, mortality, and treatment outcomes, but the primary and secondary outcomes are not clearly defined.

Suggested improvement:

At the end of the Introduction, clearly state:

• Primary outcomes: ICU admission and mortality

• Secondary outcomes: ventilation requirement, ARDS, hospital-acquired infection, and length of stay

Response: We thank the Reviewer for this excellent suggestion. While these endpoints were initially detailed in the Materials and Methods section, we agree that explicitly defining them at the end of the Introduction enhances the clarity of our study objectives. We have revised the final paragraph of the Introduction as follows:

Introduction – 3rd paragraph

To strengthen influenza prevention and control in Vietnam and similar settings, continual assessment of disease severity is needed [22]. This study aimed to examine the clinical burden and identify predictors of severe disease among adult patients with A(H1N1)pdm09-related pneumonia. The primary outcomes were ICU admission and all-cause mortality. The secondary outcomes included respiratory support requirements (such as invasive ventilation), development of moderate-to-severe ARDS, occurrence of hospital-acquired infections, and overall hospital and ICU lengths of stay.

3. Concern Regarding Small Number of Mortality Events

The mortality analysis appears statistically underpowered. Only 13 deaths occurred, yet many variables were entered into the multivariable logistic regression model. This raises concern for:

• Model overfitting

• Unstable estimates

• Extremely wide confidence intervals

• Reduced reliability of adjusted odds ratios

Examples:

• Invasive ventilation AOR 55.355 (95% CI 1.486–2062.375)

• Double-dose oseltamivir/combination therapy AOR 32.625 (95% CI 1.594–667.661)

These wide confidence intervals indicate substantial statistical instability.

Suggestions:

• Reduce the number of variables included in the mortality regression model.

• Consider using penalized logistic regression or Firth logistic regression.

• Include an events-per-variable justification.

• Acknowledge possible model instability more explicitly in limitations.

Response: We thank the Reviewer for raising this important statistical concern. We acknowledge that the small number of deaths (n = 13) limits the precision of the mortality model and increases the risk of model overfitting. Variable selection followed a pre-defined purposeful selection strategy based on both statistical criteria (P < 0.25 in the univariable analysis) and clinical relevance, as recommended in the methodological literature. We retained clinically important variables to reduce potential confounding, but we also recognised that the limited number of events resulted in wide confidence intervals for several estimates. Based on your suggestion, we used Firth logistic regression to strengthen the power of mortality. Additionally, we have included model diagnostics appropriate for each multivariable model. For the ICU admission model, calibration was assessed using the Hosmer–Lemeshow goodness-of-fit test, and discrimination was evaluated using the area under the receiver operating characteristic curve (AUC). For the mortality model, which was analysed using Firth's penalized logistic regression, model performance was evaluated using the penalized likelihood ratio test together with the AUC. Multicollinearity was assessed using variance inflation factors (VIFs), and no evidence of problematic multicollinearity was identified among the variables included in the mortality model. Despite these strategies, we acknowledge that the number of covariates remains high relative to the number of events and therefore regard the mortality model as exploratory. In response to this, we further strengthened the Limitations section to explicitly acknowledge the exploratory nature of the mortality model and the possibility of model instability. The revised text is shown below:

Materials and methods – statistical analysis paragraph

“…A standard multivariable logistic regression for ICU admission prediction model was constructed. Model calibration and goodness-of-fit were formally evaluated using the Hosmer–Lemeshow test, while discrimination capacity was assessed via the Area Under the ROC Curve (AUC / C-statistic). Multicollinearity among predictors was monitored using Variance Inflation Factors (VIF). For mortality model, due to low-event constraints (sparse mortality data), Firth’s penalized logistic regression was employed to minimize small-sample bias and avoid issues of separation. Complete-case analysis was applied to this model. Goodness-of-fit was evaluated using the Penalized Likelihood Ratio Test, and discrimination power was determined via the AUC / C-statistic. Multicollinearity was similarly assessed using VIF. Given the potential for confounding among covariates, the purposeful selection process was used to identify covariates for the regression models. Following the methodology described elsewhere [34, 35], a more generous P-value cutoff of <0.25 in the univariate analysis was used. This approach was chosen because traditional cutoffs (such as 0.05) often fail to identify variables that, while not independently significant, act as important confounders or become significant when adjusted for other covariates. Furthermore, variables considered clinically important based on the authors' expertise were retained irrespective of their univariable statistical significance to ensure that potential confounding was appropriately addressed.”

Results

Table 7. Firth's penalized logistic regression analysis of predictors of mortality among 158 patients with laboratory-confirmed A(H1N1)pdm09 pneumonia

To ensure complete transparency and help readers better interpret our findings, we have expanded the Limitations section to explicitly address how the small number of events affects model precision and stability:

Discussion – Limitations paragraph

“Our study has some notable limitations. Firstly, this study was based on an analysis of medical records and thus, data such as patients’ BMI [63] that may confound the study findings have not been documented. Additionally, some predictor variables contained missing data. Variables with substantial missingness were excluded from the multivariable analyses, which may have limited our ability to evaluate the effects of some potentially relevant predictors. Importantly, individual influenza vaccination status was not recorded in the historical medical records. Since vaccination can significantly reduce clinical severity, its absence may have confounded the observed associations with ICU admission, disease severity, and mortality. Secondly, although our study included all eligible patients treated during the study period, only 13 deaths occurred. Consequently, the multivariable mortality model should be interpreted cautiously, as the limited number of outcome events may have resulted in model overfitting and imprecise effect estimates, reflected by the wide confidence intervals for several adjusted odds ratios. Systematic microbiological investigations for bacterial co-infection at hospital admission were not performed for all patients. Therefore, the potential contribution of community acquired bacterial co-infection to disease severity and mortality could not be evaluated in this study. In addition, although the institutional criteria for ICU admission remained unchanged throughout the study period, we cannot completely exclude temporal changes in clinical practice, supportive care, or diagnostic approaches that may have influenced patient management over the 10-year study period. Finally, because the Hospital for Tropical Diseases (HTD) serves as a specialized tertiary referral center, our cohort is inherently prone to selection bias toward more severe clinical cases. Therefore, the reported ICU admission rates and mortality estimates may be higher than those in community hospital settings and may not fully represent the broader population with general influenza infections…”

4. Interpretation of Antiviral Therapy Findings is Overstated

The manuscript repeatedly implies that double-dose oseltamivir or combination therapy may increase mortality. However, this conclusion is not sufficiently supported because:

• This is an observational retrospective study.

• Sicker patients are more likely to receive aggressive therapy (confounding by indication).

• Disease severity itself likely influenced treatment decisions.

Therefore, the association should not be interpreted causally.

Current problematic statements:

• “Neither double-dose oseltamivir nor combination therapy is recommended…”

• “Alternative antiviral regimens should not be recommended.”

These conclusions exceed the evidence generated by this study.

Suggested revision:

Replace with more cautious wording such as:

“The observed association between intensified antiviral therapy and mortality likely reflects treatment allocation to more severely ill patients rather than a direct harmful effect of therapy.”

And:

“Further prospective studies are required to evaluate the efficacy of intensified antiviral regimens in severe influenza.”

Response:

We thank the Reviewer for this important comment. We agree that the observed association should not be interpreted as causal because this retrospective observational study is susceptible to confounding by indication. Therefore, we have revised the Discussion and Conclusion to avoid causal language and to emphasise that patients receiving intensified antiviral therapy were likely those with more severe disease. The revised text is shown below.

Discussion – 6th paragraph

“…The observed association between intensified antiviral therapy and mortality most likely reflects confounding by indication. Patients with more severe disease were preferentially treated with intensified antiviral regimens rather than a direct harmful effect of the therapy itself. Current evidence does not demonstrate a clear survival benefit from intensified antiviral regimens in severe influenza, and prospective studies are needed before definitive conclusions regarding their clinical effectiveness can be drawn…”

5. ICU Admission Criteria Need Be

Attachments

Attachment

Submitted filename: Response to reviewers and editor 14072026 .pdf