>>9311682
WHO CONCLUSIONS AND THE STUDY ARE SUSPECT
translated from French
ABOUT THE CLINICAL EFFICACY OF CHLOROQUINE DERIVATIVES IN COVID-19 INFECTION: VIRTUAL ANALYSIS AGAINST THE MEDICAL WORLD
Home About the clinical efficacy of chloroquine derivatives in COVID-19 infection: Virtual analysis against the medical world
May 25, 2020
Matthieu Million, Yanis Roussel, Didier Raoult
Regarding the efficacy of chloroquine derivatives in the management of COVID19 infections, we were challenged by the major discrepancies between the results of the various published studies and our experience at the IHU where 7800 ECGs were performed in 4000 patients. In order to understand which elements could lead to contradictory results, we compared the results of studies carried out by clinicians (real world) and those carried out by database analysts (virtual world of "Big data" - Figure 1, Supp Table). The clinical studies used a standardized treatment protocol mentioning in the methods the evaluation of contraindications, the daily dosage, the adjuvant measures and the duration of treatment with at least 48 h of treatment before being able to evaluate the objective.For example, the evaluation of serum potassium and the electrocardiogram is essential before starting treatment, especially when the chloroquine derivative is combined with azithromycin (Million, 2020). At the same time, we observed that virtual “big data” studies did not mention these elements and binaryly considered the presence in the electronic files of the prescription of chloroquine. Obviously, the number of patients included in the database analyzes was very much higher than that included in the clinical studies, since these databases are made up of thousands of electronic medical files (EMF). As we have mentioned in the past (Million, 2019),this type of study has colossal statistical power but they are limited by a clinical inaccuracy which makes their conclusions difficult to believe. In the attached figure, we were surprised to find that 5/5 big data studies concluded that chloroquine and its derivatives were ineffective. Using big data was significantly associated with results concluding that the chloroquine derivatives were ineffective (p <.05).
One of these studies, published in the Lancet,even identified a toxicity of chloroquine derivatives with a mortality multiplied by 3, and a cardiac toxicity leading to ventricular tachycardia in 8% of patients (Mehra, 2020). Here at the IHU, we have treated and followed more than 3,500 patients ourselves by dual therapy (hydroxychloroquine + azithromycin) with a very detailed protocol (Million, 2020), we have never observed mortality and cardiac toxicity compatible with the data reported in this study based on patient record registers (Million, 2020). On more than 1000 unselected infected patients of all ages treated with hydroxychloroquine + azithromycin according to a detailed standardized therapeutic protocol (Million, 2020),8 deaths (74 to 95 years) were identified (8/1061) with 0% mortality before 70 years (0/995 - 95% confidence interval according to Wilson 0 - 3.73%) and a mortality of 12% (8 / 66 - CI95% 6.3 - 22.1) in those aged 70 and over. No ventricular tachycardiade novo was observed. These results are not compatible (p <10 -7 ) with those reported by analysis of a recent database (mortality of all ages, 1479/6221 - 23.8% - 95% CI 22.7 to 24.8) (Mehra, 2020).
We are communicating today this very preliminary analysis to allow everyone to judge with the elements that seem relevant since, in real life, we are not dealing with databases but humans. Without neglecting the theoretical potential of big data, it is time to put clinical expertise back at the center of medical research to treat real people.