Online Program

Improving Precision and Power in COVID-19 Treatment Trials (301258)

View Presentation

David Benkeser, Department of Biostatistics and Bioinformatics, Rollins School of Public Health, Emory University 
Ivan Diaz, Division of Biostatistics, Department of Population Health Sciences, Weill Cornell Medicine 
Alex Luedtke, Department of Statistics, University of Washington Vaccine and Infectious Disease Division, Fred Hut 
*Michael Rosenblum, Johns Hopkins Bloomberg School of Public Health 
Daniel Scharfstein, Department of Biostatistics, Johns Hopkins Bloomberg School of Public Health, Johns Hopkins Universi 
Jodi Segal, Department of Medicine, School of Medicine, Johns Hopkins University 

Keywords: covid-19, corona, randomized trial

Background: There are currently over 400 clinical trials (phase 2 and 3) of treatments for COVID-19 registered on clinicaltrials.gov. A statistical analysis method called covariate adjustment has potential to improve precision and reduce the required sample size for a substantial number of these trials. Though covariate adjustment is recommended by the U.S. Food and Drug Administration and the European Medicines Agency, it is underutilized, especially for ordinal and time to event outcomes, which are common in COVID-19 trials.

Objective: To demonstrate the benefits and limitations of covariate adjustment for COVID-19 treatment trials and to give practical recommendations on implementation.

Design: We simulated randomized trials comparing a hypothetical COVID-19 treatment versus standard of care, where the primary outcome is ordinal or time-to-event.

Setting: Clinical trials of hospitalized, COVID-19 positive patients.

Patients: Data sampled from an observational study of COVID-19 positive patients at Weill Cornell Medicine New York Presbyterian Hospital, and a CDC summary report of COVID-19 positive cases.

Measurements: Outcomes include intubation, ventilator use, and death.

Results: In our simulated trials with at least 200 participants, the required sample size to achieve a desired power was reduced by 10-20% by using covariate adjustment.

Limitations: At sample size 100, covariate adjustment led to no improvement in some cases.

Conclusion: In randomized trials of COVID-19 treatments with at least 200 participants, precision may be improved by using covariate adjustment. Precision gains can be translated into smaller sample sizes (and faster trials) or larger power.

Joint work with: Benkeser, D., Diaz, I., Luedtke, A., Segal, J., Scharfstein, D. Working paper: https://medrxiv.org/cgi/content/short/2020.04.19.20069922v1