According to French law, the VIRUS-ATTAC study falls within the scope of a low-interventional clinical study (RIPH2: Recherche Impliquant la Personne Humaine de type 2) that is prospective and multicenter. The study is being conducted in 5 French university centers. The sponsor is the University Hospital of Rennes. Recruitment began in February 2021. The protocol is presented using the SPIRIT (Standard Protocol Items: Recommendations for Interventional Trials) 2013 checklist (Checklist 1). VIRUS-ATTAC is a nonrandomized, low-interventional study (RIPH2); its study-specific procedure is a research nasopharyngeal swab with centralized, masked respiratory virus testing, and SPIRIT items specific to randomized allocation are marked as not applicable. The completed study will be reported following the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) statement for cohort studies [17].

Trial site investigators will identify consecutive eligible patients according to the predefined inclusion criteria. Eligible patients will receive written and oral information about the study and will be enrolled after providing written informed consent (Multimedia Appendix 1). The participant eligibility criteria are shown in Textbox 1.

For elective cases, patients are typically admitted the day before surgery for preoperative evaluation. Surgery is typically postponed if the patient has a fever or respiratory symptoms; these patients will be excluded. For eligible patients who consent to the study, nasopharyngeal flocked swabs will be collected on the day of surgery, after anesthesia induction and before surgical incision, for viral identification. Results of the assay will be kept blinded to the physician in charge.

In the event of postoperative pneumonia or ARDS, testing for respiratory pathogens, including viruses, will be systematically performed (nasopharyngeal swab or bronchoalveolar lavage) as part of routine care. Results of any postoperative virological assays will be communicated to the physician in charge, and antiviral therapy will be prescribed at the discretion of the attending physician.

Nasopharyngeal flocked swabs are collected in 3 mL of universal transport medium (Copan Diagnostics Inc) and stored at −80 °C until testing. All samples will be sent to Rennes University Hospital for viral detection. A multiplex molecular assay (Allplex Respiratory Panel Assays; Seegene) will be performed according to the manufacturer’s instructions [18]. This assay tests for 16 respiratory viruses: influenza virus types A and B (with influenza A subtyping), adenovirus, coronaviruses HKU1, NL63, 229E and OC43, human metapneumovirus, human rhinovirus or enterovirus, parainfluenza viruses types 1‐4, and respiratory syncytial viruses A and B. The respiratory panel was selected at study inception, before the emergence of SARS-CoV-2. Throughout the study period, SARS-CoV-2 was screened separately as part of routine perioperative care at each center, and a positive preoperative test led to postponement of surgery according to standard institutional practice. SARS-CoV-2 was therefore not included in the study panel.

Demographic data; data required to calculate the European System for Cardiac Operative Risk Evaluation II (EuroSCORE II) [19], Simplified Acute Physiology Score II [20], and Sequential Organ Failure Assessment scores [21]; CPB duration; perioperative ventilator settings; perioperative blood product infusion; perioperative fluid infusion; vasopressor use and duration; renal replacement therapy use and duration; postoperative mechanical ventilation duration; reintubation (including for surgical revision); postoperative pulmonary complications according to Hulzebos et al [22]; pneumonia defined according to the 2018 French Society of Anesthesia and Intensive Care Medicine (SFAR)–French-speaking Intensive Care Society (SRLF) French guidelines [23]; ICU and hospital length of stay; ICU and hospital mortality; and 28-day mortality will be collected.

The primary analysis follows a causal inference framework. The target estimand is the average treatment effect of asymptomatic influenza virus carriage on the risk of ARDS within 7 postoperative days. The causal model is specified through a directed acyclic graph (DAG) reflecting the assumed causal structure relating preoperative viral carriage, postoperative ARDS, and potential confounders (Figure 1). The primary analysis will use a multivariable logistic regression model, fitted with ARDS at day 7 as the outcome; asymptomatic influenza carriage as the exposure; and age, chronic obstructive pulmonary disease, immunosuppression, and influenza vaccination status as confounders (identified from the DAG). G-computation [25] will be used to compute the predicted probability of ARDS for each patient under both counterfactual scenarios (carrier vs noncarrier). Center will be modeled as a random intercept to account for between-center variation in case mix and clinical practice, reflecting a design effect rather than a confounder. The same analytical approach will be applied for the secondary objectives concerning noninfluenza viral carriage and any-virus carriage. Covariate selection followed the back-door criterion applied to the DAG. Perioperative variables (surgical procedure type, EuroSCORE II, CPB duration, intraoperative transfusion, and baseline left ventricular ejection fraction) were excluded because they lie downstream of the outcome or are not common causes of both exposure and ARDS; conditioning on them would risk overadjustment or collider bias. Sex, smoking status, and the Charlson comorbidity index were excluded because we found no evidence that they are common causes of both carriage and ARDS.

Given the expected low mortality within the 7-day primary end point window in this elective cardiac surgery population, death before day 7 is unlikely to substantially bias the primary analysis. Patients who die before day 7 without a prior ARDS diagnosis will be classified as not having developed ARDS. A sensitivity analysis using a Fine and Gray subdistribution hazard model [26] treating death as a competing event will be conducted to assess robustness.

The feasibility of the influenza-specific primary analysis depends on the observed prevalence of asymptomatic influenza carriage, which may vary substantially due to seasonal variation and the potential impact of the COVID-19 pandemic on respiratory virus circulation during the recruitment period (2021‐2023) [27,28]. The following prespecified decision rule will be applied based on exposure prevalence, without knowledge of the outcome distribution:

Secondary outcomes will be reported descriptively by exposure group (carriers vs noncarriers). No multivariable modeling or hypothesis testing will be performed for secondary outcomes, as these analyses are exploratory and not supported by a formal sample size calculation. For binary outcomes (reintubation, postoperative pulmonary complications, pneumonia, and 28-day mortality), crude proportions with 95% CIs will be provided for each group, along with unadjusted risk differences with 95% CIs. For free-day outcomes (ventilator-free days, ICU-free days, and hospital-free days at day 28), medians with IQRs will be reported by group; patients who die before day 28 will be assigned zero free days.

Missing data on adjustment covariates will be handled using multiple imputation by chained equations [29]. The imputation model will include the outcome, exposure, all adjustment covariates, and auxiliary variables predictive of missingness. Rubin rules will be used to pool the estimates. Missing data are assumed to be missing at random conditional on the observed variables. The proportion and pattern of missing data will be reported for all variables.

Continuous variables will be described as means with SDs or medians with IQRs, as appropriate. Categorical variables will be described as counts and percentages. Baseline characteristics will be compared between patients with and without viral carriage using standardized mean differences [30].

A sample size of 1250 patients, including 250 asymptomatic influenza carriers, provides 80% power to detect an absolute risk increase in ARDS of 5% in asymptomatic carriers, using a significance level of .05 (2-sided). This assumes a 20% prevalence of asymptomatic influenza carriage in the general population and a 5% prevalence of ARDS after cardiac surgery [8,31]. Because recruitment took place between 2021 and 2023, a period during which nonpharmaceutical interventions against COVID-19 markedly reduced the circulation of influenza and other respiratory viruses [27], the observed prevalence of asymptomatic carriage may be lower than this assumption. The prespecified decision rule described in the Statistical Analysis section ensures that the study remains informative if carriage is lower than expected by redirecting the primary analysis to any respiratory virus carriage or, if necessary, to descriptive analyses.

The results of VIRUS-ATTAC will be disseminated through publications in peer-reviewed international journals and presentations at congresses in anesthesiology, intensive care, and cardiothoracic surgery and will be reported in accordance with the STROBE statement for cohort studies [17]. Summary results will be posted on ClinicalTrials.gov (NCT04562207) and communicated to the participating centers.

We hypothesize that preoperative asymptomatic carriage of influenza or other seasonal respiratory viruses is independently associated with an increased risk of postoperative ARDS within 7 days of elective cardiac surgery under CPB. VIRUS-ATTAC is designed to test this hypothesis directly; this section outlines the anticipated contribution of the study rather than its results, which are not yet available. To our knowledge, VIRUS-ATTAC is the first investigator-initiated, prospective, multicenter cohort study to evaluate directly the impact of preoperative asymptomatic carriage of influenza and other respiratory viruses on the risk of postoperative ARDS following elective cardiac surgery under CPB.

To date, the most relevant evidence on the association between respiratory viral exposure and postoperative ARDS after cardiac surgery comes from a single-center observational study by Groeneveld et al [8], who found that performing elective cardiac surgery during influenza season was associated with increased ARDS incidence (OR 1.85, 95% CI 1.06‐3.23). However, that study relied on seasonal exposure as a surrogate for viral carriage without virological testing, could not distinguish between influenza and other circulating respiratory viruses, and did not adjust for potential confounders using a causal framework. Moreover, the high influenza vaccination rate in that population (77%) may have attenuated the true effect of viral carriage. VIRUS-ATTAC addresses these limitations by directly testing each patient with multiplex RT-PCR, covering 16 respiratory viruses, blinding the clinical team, and applying a formal causal inference approach to the analysis.

VIRUS-ATTAC has several methodological strengths: a prospective multicenter design, systematic virological testing of all enrolled patients irrespective of symptoms, masking of virological results to the clinical team, independent blinded outcome adjudication, centralized virological analysis, and a prespecified causal inference framework. Virological results are kept blinded to the clinical team throughout the perioperative period, thereby preventing differential management and outcome ascertainment bias. The primary outcome (ARDS) and pneumonia diagnosis are adjudicated by an independent blinded committee comprising an intensivist, a pulmonologist, and a radiologist, following the PROBE (Prospective Randomized Open Blinded End-point) methodology. Centralization of all virological analyses at Rennes University Hospital ensures standardized and harmonized processing across all study sites. The primary analysis uses a causal inference framework, with DAG-based confounder selection and g-computation, to estimate the average treatment effect of viral carriage on postoperative ARDS, which is more robust to confounding than standard multivariable adjustment. This study has several limitations. First, asymptomatic viral carriage is only investigated in elective cardiac surgery patients, which may not reflect the full cardiac surgery population, including emergency and urgent cases. Second, the prevalence of viral carriage is highly dependent on the respiratory virus season and may be affected by year-to-year variability in circulating viruses; in particular, the COVID-19 pandemic substantially disrupted respiratory virus circulation during the recruitment period (2021‐2023), potentially affecting the prevalence of influenza and other respiratory viruses and the representativeness of the sample. These factors may lead to uneven distribution of positive cases across recruitment periods and centers, which is addressed in the prespecified analytical decision rule. In addition, because the panel was defined before the COVID-19 pandemic, SARS-CoV-2 was not among the assayed viruses; it was screened separately as part of routine perioperative care, and patients testing positive had their surgery postponed and thus did not enter the operated cohort [32].

If VIRUS-ATTAC confirms an independent association between asymptomatic respiratory virus carriage and postoperative ARDS, several lines of research would follow. Interventional trials could test preventive strategies such as systematic preoperative virological screening with postponement of elective surgery in confirmed carriers or systematic preoperative influenza vaccination [33]. The question could be extended to urgent and emergency cardiac surgery and to other major surgical procedures, in which preoperative screening and postponement are not feasible. Mechanistic substudies could clarify how subclinical viral carriage interacts with the pulmonary insults of CPB to precipitate lung injury.

If an independent association between preoperative asymptomatic respiratory virus carriage and postoperative ARDS is confirmed, this would provide a rationale for evaluating preventive strategies, such as systematic preoperative influenza vaccination or postponement of elective surgery in virologically confirmed carriers, in future interventional trials aiming to reduce postoperative respiratory complications after cardiac surgery.