Hydroxychloroquine or chloroquine with or without a macrolide for treatment of COVID-19: a multinational registry analysis

Therapeutic Transversal
Mehra MR et al

Main result

  •    Inclusion of 96,032 patients hospitalized for SARS-CoV-2 infection from 671 hospitals, a majority of whom were North Americans (67.9%) with an average age of 53.8 years, 53.7% male, an average BMI of 27.6 and numerous co-morbidities: 31% hypercholesterolemia, 27% hypertension and 14% diabetes. The average length of hospitalization was 9.1 days with an 11% hospital mortality rate. Other antiviral drugs were used in 41% of cases, with lopinavir-ritonavir first, followed by ribavirin.
  •     The treatment groups included 1868 patients with chloroquine alone, 3016 with hydroxychloroquine alone, 3783 with chloroquine + macrolide, and 6221 with hydroxychloroquine + macrolide. There were no significant differences in measured variables between the treatment groups.
  •     Non-survivors were on average older, more often obese black or Hispanic men with several co-morbidities including diabetes, coronary artery disease, high cholesterol, heart failure and arrhythmia mainly.
  •     Being female, being of Asian origin, being treated with converting enzyme inhibitors or statins seems on the contrary to be protected from the risk of hospital death.
  •     In terms of mortality, treatment with hydroxychloroquine or chloroquine appears to increase the risk of death during hospitalization compared to other antiviral treatment via an increase in heart rhythm disorders (p<0. 001) with an increased risk for all treatment groups compared to the control group (9.3%): hydroxychloroquine alone (18%, HR=1.34), hydroxychloroquine + macrolide (24%, HR=1.45), chloroquine alone (16%, HR=1.37) and chloroquine + macrolide (22%, HR=1.37).
  •     Subjects most at risk of developing arrhythmia are those treated with hydroxychloroquine + macrolide (HR=5.1), then chloroquine + macrolide (HR=4.0), chloroquine alone (HR=3.6) and hydroxychloroquine alone (HR=2.4). Cardiac damage prior to treatment exposure and COPD are also predisposing factors.


  • Following the retraction of 3 of the 4 authors of the study, its conclusions must be taken with caution. The quality and accuracy of the source data used is not checked. However, on the volume of data, the trends measured are undoubtedly valid, being consistent with the trends of the majority of studies on hydroxychloroquine   
  • This multinational, observational, retrospective study of registry data of patients with COVID-19 requiring hospitalization found that the use of hydroxychloroquine or chloroquine therapy (with or without macrolide) was not associated with any evidence of benefit, but rather was associated with an increased risk of ventricular arrhythmias and an increased risk of in-hospital death with COVID-19
  •     The risk of death during hospitalization with the tested treatments increases from 9% to a minimum of 16% and even 24% if hydroxychloroquine is combined with azythromycin.
  •     These results suggest that these treatment regimens should not be used outside of clinical trials and urgent confirmation of randomized clinical trials is needed.

Strength of evidence Weak

- Representation of pragmatic use of treatment (real life)
- Despite the power of the study and the well-conducted statistical analysis with propensity score, it does not replace a double-blind randomized controlled trial.
- One may question the reliability of data extracted automatically by computer from manually completed medical records (even though "The Surgical Outcomes Collaborative" is a regularly used registry, "The Surgical Outcomes Collaborative" is a registry that has been used for many years).
- Control group not strictly comparable
- No QT interval measurement or recording of peak torsades e.g. to distinguish the most severe arrhythmias from others


Assess the benefit and safety of the use of hydroxychloroquine, often used in combination with azythromycin, from multinational registries.


  •     Multinational retrospective observational cohort study using registries: medical records of patients treated with hydroxychloroquine or chloroquine, whether or not in combination with azythromycin
  •     Registry containing data from 671 hospitals on all continents, with data extracted automatically from digital medical records.
  •     Selection of subjects on medical record data, with external validation based on the following criteria: hospitalization between December 20 and April 14, 2020, for PCR-confirmed COVID-19 infection and known hospital discharge or death status
  •     Patients had to be treated with hydroxychloroquine or chloroquine, with or without second-generation macrolides, within 48 hours after hospitalization, excluding patients treated with remdesivir or mechanical ventilation at the start of treatment.
  •     5 patient groups consisting of 4 treatment groups with hydroxychloroquine or chloroquine alone or in combination with a macrolide + a control group for other patients not receiving these treatments
  •     Primary endpoint = hospital mortality
  •     Secondary endpoints: occurrence of significant ventricular arrhythmia during hospitalization + need for mechanical ventilation + total hospitalization time + length of intensive care hospitalization
  •     Statistical analysis taking into account many confounding factors (gender, age, co-morbidities, severity, co-treatments). No imputation for missing data. Use of a Cox model to assess the risk of arrhythmia. Use of a propensity score with all the potentially confounding variables mentioned above for the main assessment criterion.

bibliovid.org and its content are bibliovid property.

Legal Notice