100 Chapter 5 (LOS), after correction for age, sex, CCI and immunocompromised status. To avoid informative censoring, patients who were known to have survived and still had any activity (e.g. hospital visit) registered after hospital discharge were assumed to have survived until the end of follow-up at day 60. If patients did not die during hospital admission but had no activities registered after admission either, their outcome was set to missing. Finally, a linear regression analysis was done to analyze the effect of the different treatment groups on the ICU LOS and hospital LOS, after correction for age, sex and CCI. An extreme cases analysis was performed for the people with missing outcomes. In the all-died setting, all patients with missing outcomes were assumed to have died on the day of hospital discharge. In the allalive setting, all patients with missing outcomes were assumed to have survived until the end of follow-up. Information on patients who had two separate COVID-19 hospital admission can be found in the supplementary material. Given the competing outcomes for ICU and hospital LOS - patients who die may have a shorter stay than patients who survive- a sensitivity analysis was done only including patients who survived. To account for missing outcome data, a second sensitivity analysis for the Cox proportional hazards analysis was performed in which missing outcome and survival time data were imputed using multivariate imputation by chained equations (mice). Outcome was imputed using logistic regression and survival time was imputed using predictive mean matching. A total of 10 multiple imputation rounds were performed, and the pooled results were used to repeat the Cox proportional hazard analysis using the same predictor variables as in the main analysis. Hazard ratios from the Cox proportional hazard model and unadjusted survival probabilities were used to calculate the time specific number needed to treat (NNT), which is the number of patients that need to be treated to have one additional patient survive at a specific time point16. The mg/kg group was used as the reference group for these calculations. All statistical analyses were performed using R Statistical Software (version 3.6.1). Ethics This study was approved by the Institutional Review Board of the Leiden University Medical Center for observational COVID-19 studies and performed according to Dutch legislation on studies with clinical data. Results During the study period, a total of 5485 patients met the inclusion criteria. 168 (3.0%) of patients received a dose that did not fit within our prespecified groups and were excluded. Data were complete for most variables. No outcome was registered in 10.1% of patients. 19.2% had an incomplete registration of hospital and/or ICU LOS. Out of hospital (after discharge) all-cause mortality was 1.4%.
RkJQdWJsaXNoZXIy MTk4NDMw