 The World Health Organization's COVID-19 treatment guidelines are based on a network meta-analysis of published randomized controlled trials, RCTs. However, the severity classifications used in these trials varied greatly, making it difficult to accurately compare results between trials. Additionally, many trials did not provide information about the duration of illness or oxygen saturation levels, which further complicates the analysis. As a result, the guidelines may not reflect the true effects of treatments in all cases. To address this issue, researchers need to develop more consistent definitions of severity and conduct individual patient data, IPD, meta-analysis to better understand the effects of treatments in different populations. Furthermore, open science practices such as prompt data sharing should be adopted to facilitate IPD meta-analysis. This article was authored by Philippe J. Guerin, Alastair R. D. McLean, Samaya Rishan, and others.