Statistical review of Effect of Dexamethasone in Hospitalized Patients with COVID-19 – Preliminary Report
- 1. MRC Clinical Trials Unit at University College London
- 2. HRB Clinical Research Facility Cork. Senior Lecturer in Patient-Focused Research Methods, School of Public Health, University College Cork
- 3. Centre for Trials Research, Cardiff University
- 4. Cancer Research UK Clinical Trials Unit, Institute of Cancer and Genomic Sciences, University of Birmingham, UK
The following review has been prepared in collaboration with members of the MRC-NIHR Trials Methodology Research Partnership. The reviewers named above, and other, unnamed discussants of the paper, are all qualified statisticians with experience in clinical trials. Our objective is to provide a rapid review of publications, preprints and protocols from clinical trials of COVID-19 treatments, independent of journal specific review processes. We aim to provide timely, constructive, focused, clear advice aimed at improving both the research outputs under review, as well as future studies. Given our collective expertise (clinical trial statistics) our reviews focus on the designs of the trials and other statistical content (methods, presentation and accuracy of results, inferences). This review reflects the expert opinions of the named authors, and does not imply endorsement by the MRC-NIHR Trials Methodology Research Partnership, its wider membership, or any other organization. Here we review Effect of Dexamethasone in Hospitalized Patients with COVID-19 – Preliminary Report by Horby et al, published as a preprint on medRxiv on the 22nd of June, 2020.
Overall, this was a well-conducted, well-designed, practice-changing clinical trial providing the most convincing evidence to date of an efficacious treatment in hospitalised COVID-19 patients. However, there are a number of issues, primarily around the reporting of the trial, that should ideally be resolved prior to final publication. We also urge caution against over-interpreting the estimated subgroup-specific effects based on level of respiratory support, especially the suggestion of harm in the group not on oxygen or mechanical ventilation.