Guidelines for reporting this retrospective study were from the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement [13]. This was a retrospective analysis, conducted over 1 month, from March 4, 2020, to April 6, 2020, in a medical intensive care unit (32 beds) before that any randomized controlled trial on SARS-CoV-2 had started. The Research Ethics Committee of Sorbonne University approved the project (CER 2020-36). Oral information about this retrospective analysis was given to patients or relatives.
Study population
Since our hospital is a national reference center for emerging biological risks, only patients with suspected or confirmed SARS-CoV-2 infection were admitted in our medical ICU during this period as long as they presented clinical signs of severity. The data from all prospective patients admitted to our medical ICU and who fulfilled the following criteria were studied: (1) acute respiratory failure as defined by severe hypoxemia requiring either a high level of oxygen via facemask (> 6 L/min to achieve SpO2 > 90%), high flow oxygen therapy (with a minimum of 30 L/min and 50% FiO2 to achieve SpO2 > 90%), or invasive mechanical ventilation and (2) proven infection by SARS-CoV-2 defined by positive reverse transcriptase polymerase chain reaction (RT-PCR) assay targeting the E (envelope) gene of SARS-CoV-2, obtained from nasopharyngeal swab or lower respiratory tracts [14]. Patients who received anti-viral treatments other than lopinavir/ritonavir or hydroxychloroquine or who received both of these treatments were not analyzed.
Treatments
Somewhat by accident, three therapeutic approaches were consecutively implemented. During the first period, each new patient admitted in our ICU was receiving, in addition to standard of care, lopinavir/ritonavir (400 mg twice daily, oral route down the nasogastric tube in syrup form) for 5 days as per our local disease control policies. During the second period, due to a lopinavir/ritonavir shortage, the local policy was changed to hydroxychloroquine (200 mg, twice a day, oral route) for each new patient admitted in our ICU. During the last period, no specific treatment was given besides the standard of care. Soon after, patients started to be included in randomized controlled trials. Across each period, patients received standard of care that consisted of ventilatory support, antibiotic agents whenever needed, vasopressors, renal replacement therapy, and ECMO.
Data collection
The following data were extracted from each patient’s electronic medical chart: age, gender, clinical and biological variables upon admission, and time between symptoms’ onset and ICU admission. Simplified Acute Physiology Score (SAPS) 2 and Sequential Organ Failure Assessment (SOFA) were calculated upon ICU admission. Advanced life support measures taken during the ICU stay such as invasive mechanical ventilation, ECMO, vasopressor support, and renal replacement therapy were also collected. Finally, we recorded the length of ICU stay, time spent under invasive mechanical ventilation, and biological variables. Mortality was assessed at day 14 and day 28 after ICU admission. SARS-CoV-2 load in respiratory tracts (nasopharyngeal swab or tracheal aspiration) and plasma SARS-CoV-2 load were collected within 24 h of ICU admission and at day 7. The cycle threshold (CT) value of RT-PCR was used as an indicator of the viral load in clinical samples, the lower the CT value, and the higher the viral load. PCR was considered negative when CT was > 45.
Outcomes analyzed
The primary outcome was treatment escalation occurring after day 1 after ICU admission until day 28. Treatment escalation was defined by the initiation of at least one life support intervention among intubation, ECMO, or renal replacement therapy. Secondary clinical outcomes included ventilator-free days at day 28 (zero ventilator-free day was attributed to a patient who died) and mortality on day 14 and day 28. Last, secondary virological outcomes were viral load changes in respiratory tracts and plasma between admission and day 7.
Statistical analysis
Continuous variables are expressed as median (25–75, interquartile range, IQR) and categorical variables are expressed as number and relative frequencies (%). Patients were categorized a posteriori into the three groups according to the treatment that they received. Continuous variables were tested for normality using the Kolmogorov-Smirnov normality test. Gaussian variables were compared using an ordinary ANOVA test and non-normally distributed variables using a Kruskal-Wallis test. Categorical variables were compared with chi-square test or Fisher’s exact test. Kaplan-Meier curves were computed for the proportion of patients who needed treatment escalation and were compared using log-rank test. Due to the retrospective nature of the study, no sample size calculation was performed. A convenience sample of patients corresponding to the number of patients admitted in the ICU during the first 4 weeks of the outbreak was deemed appropriate. For final comparisons, a two-tailed p value less than or equal to 0.05 was considered statistically significant. The statistical analysis was performed by using Prism 8.3.0 software (GraphPad Software, USA).