OBJECTIVES:

Mehra et al. conducted an international registry analysis in 96032 patients with COVID-19, indicating that hydroxychloroquine (HCQ) or chloroquine (CQ) was associated with no improvement in outcomes, but increased risk of ventricular arrhythmias and mortality. These results were published in Lancet. We aim to critically the study results.

METHODS:

Results were systematically scrutinized for veracity.

RESULTS:

The study report data on patients from the 5 continents collected in 600 hospitals. The baseline disposition of included patients showed an unexpected very high homogeneity, which was not consistent with the specific epidemiology and socio-demographic in different regions. Arhythmia had to last 6seconds to be confirmed while none of the involved developing countries have the infrastructure to perform such monitoring. In Australia the number of deaths in a sample of hospitals was larger than the total reported deaths. Race was reported in France while by law these data could not be collected. The angiotensin converting enzyme inhibitors were reported to reduce mortality by 50% while hypertension was a risk factor for mortality. Age and obesity were not a risk factor while diabetes was a protective factor in several geography. Several other major risk factors recognised in many previous studies were identified as minor risk factors or even protective factors. Antiviral treatments were proven to be effective in reducing mortality but were not integrated in the multivariate Cox model or studied further.

This study as well as a similar one published by the same authors in NEJM were withdrawn and recognized as fake data.

CONCLUSIONS:

Publication of database analysis in top level medical journals questioned the credibility of real-world evidence. Despite incredible results the study accepted for publication with limited comments. There is a need for guidelines for reviewing and assessing veracity of real-world studies.