A collaboration of researchers from Emory University, the University of Washington, Cornell University, Johns Hopkins University, and the University of Utah have shown potential for speeding up randomized COVID-19 treatment trials, through relatively simple adjustments in how the trial data are analyzed. The results of the study were recently published in the journal Biometrics.

Among the University of Washington faculty involved, Assistant Professor Alex Luedtke from the Department of Statistics has played an important role in identifying statistical methods for the analysis of COVID-19 treatment trials. By utilizing data of over 500 COVID-19 positive and hospitalized patients collected from Weill Cornell Medicine New York Presbyterian Hospital, the researchers simulated clinical trials of a hypothetical treatment for COVID-19, and compared the performance of different approaches to analyzing the data. The paper demonstrates that covariate adjustment, a statistical approach that evaluates the effect of treatments on COVID-19 outcomes while accounting for important factors like age and pre-existing conditions, provides up to 22% more information about COVID-19 treatments compared to the typical approach employed in many randomized trials that ignore risk factors associated with COVID-19. 

“This means that we can find effective treatments and abandon ineffective ones more quickly,” said co-author David Benkeser, an Assistant Professor in Biostatistics and Bioinformatics at Emory’s Rollins School of Public Health. 
“This work exemplifies the important role that statisticians can play in helping to reduce the burden of the COVID-19 pandemic,” said Alex Luedtke.
Other co-authors on the work include Michael Rosenblum from Johns Hopkins Bloomberg School of Public Health, Jodi Segal from Johns Hopkins University School of Medicine, Daniel Scharfstein from the Department of Population Health Sciences at University of Utah School of Medicine, and Iván Díaz from the Department of Population Health Sciences at Weill Cornell Medicine.