Deaths induced by compassionate use of hydroxychloroquine during the first COVID-19 wave: the devil is in the details.

Several trials of different designs were conducted to investigate the efficacy and safety of hydroxychloroquine (HCQ) for the prevention and/or the treatment of COVID-19 patients. Recently, it has been reported that HCQ might have been associated with an excess of 16,990 deaths during the first wave of the COVID-19 pandemic in 6 countries. Such attributable risk analysis is associated with many limitations. These previous findings did not adequately address dose-subgroup and sensitivity analyses which precludes any overall firm conclusions on in-hospital mortality attributable to HCQ. We performed a meta-analysis and proposed a stratification by doses of HCQ. By pooling studies employing HCQ doses < or = 2400mg/5 days (i.e., k=12, n patients treated with HCQ=947, n controls=745), an OR of 0.94 (95%CI 0.56; 1.59) was found, indicating no increase in the mortality rate anymore. Importantly, there was no significant reduction in mortality rate with HCQ at < or = 2400mg/5 days neither. The same observation held true when pooling studies employing HCQ doses < or = 4800mg/5 days (i.e., k=25, n patients treated with HCQ=1672, n controls=1479) with an OR of 0.97 (95% CI 0.73; 1.29). Only high dose regimens of HCQ are associated with a significant increase in mortality. Applying an excess of mortality in the population treated with doses where no increase of mortality is found creates a misleading overestimation of deaths associated with the use of HCQ in hospitalized patients with COVID-19. On the other hand, even at low doses HCQ regimen, no reduction in mortality with HCQ was observed suggesting that, when it comes to mortality as the outcome, HCQ did not show a benefit in hospitalized patients suffering from COVID-19. This mainly justifies the past and still up-to-date recommendations and guidelines to not use HCQ in this indication.