Study design and population
We conducted a monocentric prospective observational study in the medical intensive care unit of Rennes University Hospital, a tertiary teaching hospital.
Data were recorded prospectively and retrospectively analyzed. We included all consecutive patients over 18 years with laboratory-confirmed Covid-19 infection and mechanically ventilated (MV) who were admitted to the ICU between March 10th, 2020 to April 16th, 2021 for patients with Covid-19, and between April 10th, 2006 to February 8th, 2020 for patients with influenza. Notably, patients were tested for both influenza and SARS-Cov-2 during the Covid-19 pandemic period.
Only laboratory-confirmed cases were included. A confirmed case of Covid-19 or influenza was defined by a positive result on a reverse-transcriptase–polymerase-chain-reaction (RT-PCR) using the Influenza A/B r-geneTM (Argene®, bioMérieux, Marcy-l’Etoile, France) and TaqPath™ Covid-19 (Thermo Fisher Scientific, Illkirch-Grafenstaden, France) assay of a specimen collected on an endotracheal aspiration. ARDS definition was based on Berlin criteria . This study was conducted in accordance with the Declaration of Helsinki and approved by the hospital’s ethical committee (No. 20.52) and we followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) recommendations for cohort studies. Due to its observational nature, patient signed informed consent was waived by the ethical committee in compliance with French legislation on observational studies of anonymized data.
Sedation and pain management are standardized in our ICU according to recommended guidelines to decrease the risk of oversedation . Thus, sedation and analgesia were managed through a nurse-driven protocol with assessment of the Richmond Agitation Sedation Scale  (RASS) and the Behavioral Pain Scale  (BPS) or Numeric Rating Scale (NRS) in those able to communicate. Sedation was primarily targeted at a light level (RASS between 0 and − 1), except for ARDS patients, for whom sedation was targeted at a deep level (RASS − 5) before introducing neuromuscular blocking agents, according to the ACURASYS sedation protocol . After improvement of the ARDS, sedation was reduced to target a light level (RASS between 0 and − 1). Analgesia was targeted to obtain a BPS < 5 or NRS < 3. There was no daily sedation interruption. Sedatives used were midazolam or propofol, and opioid used was morphine.
All ARDS patients in both groups received protective ventilation according to published guidelines , especially: assist-control mode, initial tidal volume targeted at 6 mL per kilogram of predicted body weight, titration of positive end-expiratory pressure (PEEP) level and end-inspiratory pressure measured at least every 2 h to be kept below 28 cm of water.
We collected demographic data, clinical symptoms at presentation, and comorbidities (hypertension, obesity, diabetes mellitus, and neurologic history). Diabetes mellitus was defined by a history of diabetes requiring chronic therapy with insulin or an oral hypoglycemic agent. Obesity was defined by a body mass index greater than or equal to 30 kg per square of height in meters . The severity of illness was assessed through severity scores at admission to ICU, including Simplified Acute Physiology Score  (SAPS) II within 24 h after admission and Sequential Organ Failure Assessment  (SOFA) score calculated on the first day after admission. As previously described, we defined agitation as a day with RASS score greater than 0 during the ICU stay  not explained by pain (i.e. BPS < 3) and other causes of delirium (alcoholic, iatrogenic, or metabolic). We also recorded parameters that can contribute to encephalopathy or agitation in critically ill patients: the cumulative dose of midazolam, propofol, and opioids during the ICU stay, length of sedation, and time to defecation defined as the delay between admission in the ICU and the first defecation . Moreover, we collected events that may result from agitation, such as the need for anti-agitation drugs, and the self-extubation, defined as a deliberate action taken by the patient to remove the endotracheal tube. Finally, patients were followed from admission to day 28, and we recorded the 28 days of survival.
Normally distributed continuous variables are presented as the means ± standard deviations (SDs), whereas non-normally distributed data are presented as medians (interquartile ranges (IQRs)). Categorical variables are presented as numbers (percentages). Continuous variables were compared by using the Mann–Whitney U test. Proportions for categorical variables were compared using the χ2 test or Fisher’s exact test when more appropriate.
To improve the balance of baseline characteristics and reduce the effects of selection bias and potential confounding factors in this observational study, a propensity score analysis was performed. The propensity score was calculated by a multivariable logistic regression model using a priori defined variables with a clinical relevance between the two groups (age, BMI, MV and sedation duration, ileus duration, cumulative doses of midazolam, propofol, and morphine, day-28 survival, Glasgow coma scale at admission). These variables were chosen based on the results of previous studies [9, 10, 18]. The Covid-19 and influenza patients were then matched 1:1 on these propensity scores.
Second, we used a Cox-proportional hazard model to determine whether agitation during the ICU stay was independently associated with mortality on day 28. For this analysis, susceptibilities known to produce agitation achieving a p value of 0.10 were used for adjustments. Results were expressed as Hazard Ratios (HR) with their 95% confident interval (CI). Survival curves were constructed until day 28 using the Kaplan–Meier method and compared with the log-rank test. Patients alive on day 28 were censored. We defined agitation-free days as the number of days in the first 28 days after admission during which the patient was alive without agitation and not in a coma for any cause. Patients who died within the 28-day study period were recorded as having zero days free of agitation. As agitation is associated with mortality, death was deemed as a competing risk for agitation in a multivariate analysis using a Fine-gray model to fit cumulative incidence. All probability values reported were 2-sided. Statistical analyses were performed using R 4.1.2 (R Foundation for Statistical Computing, Vienna, Austria), and p values of less than 0.05 were considered significant.