The Lancet and The New England Journal of Medicine (NEJM), two of the most prestigious and widely read medical journals in the world, each retracted papers in June 2020 regarding medication use in patients with COVID-19. The papers, “Hydroxychloroquine or Chloroquine With or Without a Macrolide for Treatment of COVID-19: A Multinational Registry Analysis” and “Cardiovascular Disease, Drug Therapy, and Mortality in Covid-19,” share three of the same authors: Drs. Mandeep Mehra, Amit Patel, and Sapan Desai. Dr. Desai is the founder of Surgisphere, the private data analytics company that provided the data used in both studies.
Since the two papers were retracted, a lot of questions have been raised about peer review and why the problems with these papers were not discovered during the review phase.
The reason for both retractions was the refusal of Surgisphere to provide the raw data on which the analyses were based. The authors of the two papers relied on data compiled and analyzed by Surgisphere, but after both articles were published, readers throughout the world raised concerns about inconsistencies in the data. When Surgisphere refused to provide the raw data underlying the analyses, the authors not associated with the data company agreed to retract the papers, noting that an independent review could not be done and the results, therefore, could not be confirmed.
The idea that peer review could have uncovered this kind of data problem highlights a fundamental misunderstanding of what peer review is. Peer review is not a mechanism for detecting fraud or data mining errors. Peer reviewers are charged with evaluating the logic and scientific quality of papers, not with determining if the data used to create the tables and graphs are faulty or fabricated. And even with this limited charge, the peer review system is far from perfect.
But compounding the existing deficiencies in peer review is the rush-to-publish atmosphere that has emerged during the COVID-19 pandemic. Submissions to journals—particularly the large and influential journals—have dramatically increased since the virus was identified. The overwhelming lack of understanding about the novel coronavirus is a strong impetus for publishing information as quickly as possible to inform prevention measures and treatment plans. Not surprisingly, the confluence of these factors has led to a number of papers being published that have since been retracted. As of July 17, 2020, Retraction Watch had identified 25 retracted COVID-19 papers, 3 temporarily retracted papers, and 1 paper with expressions of concern.
In our view, the responsibility for these retractions lies squarely with the authors. Deflecting blame to well-known deficiencies in peer review and/or editorial review obfuscates the fundamental failure of the authors of these papers to verify their data before submitting for publication.
So to answer the question we began with, that’s how it happened—the two retractions from The Lancet and the NEJM. The authors failed to verify their data before submission. And that’s not OK, no matter which way you look at it.