Abstract
Purpose A international registry analysis led by Mehra et al. to investigate the use of hydroxychloroquine (HCQ) and chloroquine (CQ) with or without a macrolide in 96,032 hospitalised COVID-19 patients were published on Lancet, which has raised considerable discussions in the public health community. This study aimed to critically review the quality and limitations of the Mehra et al. publication and discuss the potential influences on the use of HCQ/CQ worldwide. Method A critical review of this publication was conducted to examine the potential study bias in the study objectives, methodology, confounding factors and outcomes and summarise the external reviews. Results The very high homogeneity of the patients' characteristics at baseline was inconsistent with region specific epidemiology and several critical confounding factors. The results indicated that angiotensin converting enzyme inhibitors were associated with a hazard ratio of 0.5, which suggested a technical problem in the estimation of the propensity scores. Several major risk factors for mortality identified in the analysis were treated as a minor risk or neutral or even protective factors. Antiviral treatments were recognised as an effective method to reduce mortality and were neither further studied nor integrated in the multivariate Cox model. Conclusion This research appeared to carry multiple biases. An extensive audit of the study, conditions of review and acceptance for publication in the Lancet of that study are requested to avoid damage to the publics' trust on the scientific community at this critical time of COVID-19 pandemic.