This was a retrospective cohort study utilizing national data from a single influenza epidemiological surveillance database (SIVEP-Gripe). The data are derived from the universal surveillance of Severe Acute Respiratory Syndrome (SARS). SARS is a notifiable disease in Brazil, with immediate reporting required [9]. The study followed the guidelines of the REporting of studies Conducted using Observational Routinely-collected health Data (RECORD) Statement [10].

Cases were defined as individuals with Influenza-like Illness (ILI) defined as fever (even if self-reported) accompanied by cough or sore throat, with symptom onset within the last 7 days—presenting with: dyspnea/respiratory distress, OR persistent chest pressure, OR O2 saturation < 95% in room air, OR bluish coloration of the lips or face. For notification purposes in SIVEP-Gripe, hospitalized SARS cases or SARS-related deaths were considered, regardless of hospitalization status [11].

The study was conducted in Brazil, a country of continental dimensions in South America characterized by a universal public health system (Unified Health System - SUS) and a robust epidemiological surveillance system. As the world’s fifth-largest country, Brazil possesses a vast territorial extent harboring rich geographic and hydrographic diversity, including the Amazon Basin. It is divided into five macro-regions (North, Northeast, Central-West, Southeast, and South) with distinct environmental and socioeconomic characteristics, which directly influence the nation’s economic activities and development patterns [12].

Climatically, Brazil presents a wide range of climate types, predominantly tropical. These vary from the humid equatorial climate in the Amazon (high temperatures and abundant rainfall) to the semi-arid climate in the Northeast (characterized by long droughts) and the subtropical climate in the South (with well-defined seasons and the occurrence of frost) [13, 14].

Data were obtained from the Influenza Epidemiological Surveillance Information System (Sistema de Informação de Vigilância Epidemiológica da Gripe - SIVEP-Gripe), an official, open-access database managed by the Brazilian Ministry of Health that collects information on SARS cases in the country (https://opendatasus.saude.gov.br/dataset/?q=srag). Participants included individuals hospitalized for influenza in Brazil in 2024.

Inclusion criteria comprised only cases with laboratory confirmation for Influenza via RT-PCR and closed cases—i.e., those with complete information and a known outcome (death or survival), excluding ongoing cases or those with incomplete data. Exclusion criteria were cases attributed to other etiologies.

Missing data were handled using a complete-case approach, including only observations with available information for the variables included in each analysis. Records with missing outcome data were excluded from the study, as previously described. For covariates, categories labeled as “ignored” or undefined in the SIVEP-Gripe database were treated as missing and were not analyzed as separate categories to minimize misclassification bias. No imputation procedures were performed given the observational nature of the data and the large sample size.

The proportion of missing data varied substantially across variables. While demographic variables such as age and sex were complete, and COVID-19 vaccination status showed negligible missingness due to automatic integration with the National Health Data Network (RNDS), other variables particularly comorbidities and influenza vaccination status presented higher levels of missing data, reflecting manual data entry processes within the surveillance system. This pattern is consistent with known limitations of routinely collected administrative datasets. Although complete-case analysis may introduce bias if missingness is not completely at random, the consistency of the main findings and the large sample size support the robustness of the estimates [15, 16].

Regarding outcome classification, cases recorded as “death from other causes” were excluded to ensure a more specific assessment of influenza-related mortality.

The outcome variable (dependent) was death (binary: yes/no). The independent variables (predictors) were categorized as follows: Demographic: Age (continuous), Sex (Male/Female). Clinical/Symptoms: Cough, Fever, Dyspnea, Respiratory distress, O2 saturation < 95%, Fatigue, Sore throat, Vomiting, Diarrhea, Abdominal pain, Loss of smell, Loss of taste. Comorbidities: Chronic Cardiovascular Disease, Diabetes mellitus, Other Chronic Pneumopathy, Asthma, Chronic Neurological Disease, Immunodeficiency/Immunosuppression, Chronic Kidney Disease, Obesity, Chronic Hematologic Disease, Down Syndrome, Chronic Liver Disease. Interventions/History: Vaccinated (COVID-19 booster dose), Vaccinated (Influenza in the last campaign), Oseltamivir use, ICU admission, Invasive ventilation. Influenza Type: Influenza A, Influenza B. Subtypes/Lineages: Influenza A Subtype (A(H1N1)pdm09, A(H3N2), A not subtyped, A not subtypable, inconclusive, other); Influenza B Lineage (Victoria, Yamagata, not performed, inconclusive, other, blank).

Descriptive statistics were calculated for the study population, stratified by clinical outcome (death vs. survivor) and influenza type (A vs. B). A detailed comparative analysis of clinical characteristics, symptoms, and outcomes across influenza types, subtypes, and lineages was performed and is presented in the Supplementary Appendix to provide additional epidemiological context. For descriptive proportions, 95% confidence intervals were calculated using the Wilson binomial method.

The Kolmogorov-Smirnov test was used to assess the normality of the age distribution. A p-value < 0.001 indicated that age did not follow a normal distribution, justifying the use of the median [17]. Consequently, the Mann-Whitney U test was employed to compare age medians between independent groups [18] (death vs. survivor and Influenza A vs. B), with p-values < 0.001 indicating statistically significant differences.

The association between COVID-19 booster vaccination and mortality was assessed using multivariable logistic regression. To control for confounding, a propensity score was estimated based on demographic characteristics, clinical variables, and comorbidities, including age, sex, signs and symptoms, and pre-existing conditions. Inverse probability of treatment weighting (IPTW) was then applied, with weights truncated at 10 to reduce the influence of extreme values, as recommended in the literature. Covariate balance after weighting was assessed using standardized mean differences to evaluate the adequacy of the propensity score adjustment. Covariate balance was considered adequate when absolute standardized mean differences were below 0.10 after weighting. To provide a transparent assessment of the propensity score adjustment, pre- and post-IPTW balance metrics were summarized in the Supplementary Appendix. Model calibration was evaluated using the Hosmer–Lemeshow goodness-of-fit test and complemented by inspection of agreement between observed and predicted mortality across risk strata [19].

Weighted models were fitted excluding variables considered potential mediators of the exposure–outcome relationship (such as intensive care unit admission and ventilatory support), in order to minimize overadjustment. Model discrimination was evaluated using the area under the receiver operating characteristic curve (AUC) based on predicted probabilities [20], and calibration was assessed using the Hosmer–Lemeshow test [21].

Multicollinearity was examined using variance inflation factors (VIFs) and condition indices [22]. Additionally, E-values were calculated for statistically significant associations as a sensitivity analysis to assess the potential impact of unmeasured confounding [23].

Analyses were performed using the Statistical Package for the Social Sciences (SPSS) software, version 26.0. A p-value < 0.05 was considered statistically significant in all analyses.

This study used secondary, de-identified data from the publicly available SIVEP-Gripe database (Brazilian Ministry of Health). According to national regulations, studies using anonymized public data are exempt from institutional review board approval and informed consent requirements. All procedures were conducted in accordance with relevant ethical standards and the Declaration of Helsinki.