Predictive factors of clinical assays on hydroxychloroquine for COVID-19 mortality during the first year of the pandemic: A meta-synthesis

Abstract

Background: The COVID-19 pandemic led to a violent debate about the efficacy of a repurposed drug hydroxychloroquine (HCQ) and a new broad-spectrum antiviral (remdesivir) and about randomized controlled trials (RCTs) and observational studies. To understand conflicting results in the literature, we performed a metasynthesis to determine whether intrinsic qualitative criteria within studies may predict apparent efficacy or ineffectiveness of HCQ and remdesivir.
Methodology: Predictive criteria were identified through critical review of studies assessing HCQ and remdesivir for COVID-19 mortality from March to November 2020. Multiple correspondence analysis, comparative metaanalysis, and predictive value were used to explore and identify criteria associated with study outcomes.
Results: Among the 61 included studies, potential conflict of interest, detailed therapeutic protocol, toxic treatment (overdose or use in contraindicated patients), known centers and doctors, and private data computing company were the most predictive criteria of the direction of effect of the studies. All 18 observational studies evaluating HCQ and reporting detailed therapeutic protocol without conflict of interest were Pro. Potential conflict of interest was a perfect predictor for remdesivir efficacy. RCTs were associated with HCQ inefficacy and potential conflict of interest. The most predictive criteria were validated and allowed perfect classification of 10 additional studies.
Conclusion: In therapeutic trials on COVID-19, the major biases predicting the conclusions are not methodology nor data analysis, but conflict of interest and absence of medical expertise. The thorough search for declared or undeclared and direct or indirect conflict of interest, and medical expertise should be included in the quality criteria for the evaluation of future therapeutic studies in COVID-19 and beyond. A new checklist evaluating not only methodology but also conflict of interest and medical expertise is proposed.