Compassionate Use of Remdesivir in Covid-19
Jonathan Grein, Robert P Myers, Diana Brainard
Abstract
Compassionate Use of Remdesivir in Covid-19To the Editor: In their article, Grein et al.(published online on April 10 at NEJM.org) 1 report outcomes in a cohort of patients who were hospitalized for severe Covid-19 and who received remdesivir on a compassionate-use basis.Over a median follow-up of 18 days, the authors observed improvement in oxygen-support class in 36 of 53 patients (68%), and the condition of 8 patients (15%) worsened.Grein and colleagues reported that Kaplan-Meier analysis showed a cumulative incidence of clinical improvement of 84% (95% confidence interval [CI], 70 to 99).We have found that this result is not correct.A reanalysis of the patient-level data reported in Figure 2 of the article revealed a much lower estimate (approximately 70%).Apparently, the authors erroneously censored data on the deceased patients at the time of death.This is a serious methodologic error.A basic assumption in survival analysis is that censoring should be "noninformative" (i.e., patients for whom data have been censored have the same survival prospects as those who continue to be followed). 2Data on patients who died during the follow-up period should not have been censored; their status should have been considered "not improved" through day 28.We ask the authors to reanalyze and report the remdesivir data, taking into account the above methodologic aspects.