Hydroxychloroquine or chloroquine with or without a macrolide for treatment of COVID-19: a multinational registry analysis
|Date||May 22, 2020|
A multinational registry analysis assessing the effectiveness of hydroxychloroquine or chloroquine with or without macrolide therapy in treatment of 96,032 hospitalized COVID-19 patients showed significantly increased risk of both in-hospital mortality and de-novo ventricular arrhythmia during hospitalization.
This article was retracted by its authors on June 4, 2020, after mounting concerns regarding the veracity of an unvalidated dataset.
Most notably, the data was procured by the corporation Surgisphere, which claimed to have partnerships with 671 hospitals across six continents that captured detailed patient information on nearly every hospitalized COVID-19 patient in Australia and North America.   To date, no hospitals have verified shared data agreements with Surgisphere. When asked for an external audit of the database, Surgisphere declined and, subsequently, every author with the exception of the Surgisphere founder, formally asked The Lancet to retract.
Surgisphere officially closed operations and deleted its website in mid-June of 2020.
Within one week after the study was published, independent researchers worldwide discovered inconsistencies between the study’s dataset and total reported COVID-19 hospitalizations. On May 28, 2020, The Guardian reported that the study included 73 COVID-19 deaths in Australia, which was greater than the 67 COVID-19 deaths reported by multiple established COVID tracking databases. In interviews with The Guardian, major hospital systems in Australia denied any relationship or data sharing agreements with the corporation (Surgisphere) responsible for procuring the database. The study authors responded with a “Department of Error” published in The Lancet declaring that single new hospital was accidentally designated to Australia, but belonged in the Asian continental designation.
North American Data
Study data designated to North America also fell under scrutiny when compared against total COVID-19 hospitalizations. The study reported that 63,315 hospitalized patients with COVID-19 in North America met inclusion criteria prior to April 14, 2020. A review of data from the COVID Tracking Project by The Atlantic showed only 63,276 patients hospitalized with COVID-19 by this date. While theoretically possible that Surgisphere also collected patient data from Canada and Mexico, this would have added only an additional 2,019 and ~1,000 patients from Canada and Mexico, respectively. Researchers found it increasingly unlikely that the unknown corporation Surgisphere was capable of obtaining detailed patient records for nearly every COVID-19 patient in North America.
The study reported patient data from Africa that required sophisticated patient monitoring technology and electronic medical record systems. An open letter to The Lancet signed by nearly 200 physicians and medical researchers argued this quality of data from Africa to be unlikely. For the dataset to be valid, nearly 25% of all COVID-19 cases and 40% of all deaths in the continent would have needed to occur in Surgisphere-affiliated hospitals with sophisticated electronic patient data recording and monitoring capable of detecting and recording “nonsustained [at least 6 secs] or sustained ventricular tachycardia or ventricular fibrillation.”
First raised by researchers including infectious disease specialist Dr. Didier Raoult, the reported baseline characteristics included unusually small variances among continents. 
Surgisphere (est. 2007) was the sole provider of the dataset for the study, and is a self-proclaimed real-time global research network that "performs cloud-based healthcare data analytics" using machine learning and artificial intelligence.
Confronted with the aforementioned concerns on the dataset, the Surgisphere founder stood by the integrity of the data and refused to reveal additional information on data sources.
Independent researchers then focused attention on Surgisphere’s history to discover only a single prior publication utilizing its database, a digital history excluded from the internet archives, five employees that joined the corporation only months earlier (according to LinkedIn) including a science fiction writer and adult model as well as a founder with a checkered past of malpractice lawsuits.
Facing mounting pressure to verify the database, the study authors responded that the database was to undergo an independent audit. Under the guise of client agreements and confidentiality requirements, Surgisphere declined the audit and the remaining three study authors formally requested a retraction by The Lancet, which occurred on June 4, 2020, less than two weeks after the study was published.
No minor issues identified at this time.
The Lancet study was published on Friday, May 22, 2020. By the following Monday, May 25, the World Health Organization (WHO) announced it was suspending all COVID-19 clinical trials on hydroxychloroquine in 17 different countries.  This announcement disrupted clinical trials and caused an uproar among active physician researchers who did not detect any safety concerns with hydroxychloroquine that replicated the findings of the Lancet study.
After the study was retracted, on June 4, 2020, the WHO resumed hydroxychloroquine trials for COVID-19.
Soumya Swaminathan, Chief Scientist for the WHO, explained that the WHO did not check data quality themselves and instead trusted the authors had adhered to basic standards.
Department of Error
On May 30, 2020, The Lancet published a Department of Error where authors changed the designation of a single hospital from Australia to Asia.
On June 4, 2020, The Lancet published a Retraction after Surgisphere refused to transfer the full dataset for independent audit.
- Mehra, MR; Desai, SS; Ruschitzka, F; Patel, AN (May 2020). Hydroxychloroquine or chloroquine with or without a macrolide for treatment of COVID-19: a multinational registry analysis. Lancet. doi:10.1016/S0140-6736(20)31180-6. PMC 7255293. PMID 32450107.
- Retraction: "Hydroxychloroquine or chloroquine with or without a macrolide for treatment of COVID-19: a multinational registry analysis". The Lancet. 4 June 2020. Retrieved 25 June 2020.
- Todaro, James (29 May 2020). "A Study Out of Thin Air". MedicineUncensored. Retrieved 25 June 2020. “North American data from the study is highly suspicious. The study reports that 63,315 hospitalized patients with COVID-19 met inclusion criteria prior to April 14, 2020. A review of the well-curated data from the COVID Tracking Project by The Atlantic shows that there were only 63,276 patients hospitalized with COVID-19 by April 14. It is theoretically possible that Surgisphere also collected patient data from Canada and Mexico. However, both of these countries had a tiny number of COVID-19 hospitalizations in comparison to the USA.”
- Offord, Catherine (30 May 2020). "Disputed Hydroxychloroquine Study Brings Scrutiny to Surgisphere". The Scientist. Archived from the original on 3 June 2020. Retrieved 25 June 2020.
- Davey, Melissa (28 May 2020). "Questions raised over hydroxychloroquine study which caused WHO to halt trials for Covid-19". Guardian. Retrieved 25 June 2020.
- Davey, Melissa (12 June 2020). “Covid-19 studies based on flawed Surgisphere data force medical journals to review processes”. Guardian. Retrieved 25 June 2020.
- “Department of Error: “Hydroxychloroquine or chloroquine with or without a macrolide for treatment of COVID-19: a multinational registry analysis”. Lancet. 30 May 2020. Retrieved 25 June 2020.
- Watson, James (28 May 2020). “An open letter to Mehra et al and The Lancet”. Retrieved 25 June 2020. “Data from Africa indicate that nearly 25% of all COVID-19 cases and 40% of all deaths in the continent occurred in Surgisphere-associated hospitals which had sophisticated electronic patient data recording, and patient monitoring able to detect and record “nonsustained [at least 6 secs] or sustained ventricular tachycardia or ventricular fibrillation”. Both the numbers of cases and deaths, and the details provided, seem unlikely.”
- Didier Raoult [@raoult_didier] (26 May 2020). “About the paper published in The Lancet: data coming from five different continents cannot be so homogeneous. There is either data manipulation (not mentionned in Material and Methodes), or incorporation of faked data.” (Tweet) via Twitter.
- Todaro, James (29 May 2020). "A Study Out of Thin Air". MedicineUncensored. Retrieved 25 June 2020. “Surgisphere appears to be the sole provider of the data for the Lancet study, and boasts itself to be a real-time global research network that "performs cloud-based healthcare data analytics" using machine learning and artificial intelligence.”
- Piller, Charles (8 June 2020). “Who’s to blame? These three scientists are at the heart of the Surgisphere COVID-19 scandal”. Science. Retrieved 25 June 2020.
- Beaubien, Jason (25 May 2020). “WHO Halts Hydroxychloroquine Trial Over Safety Concerns.” NPR. Retrieved 25 June 2020.
- Davey, Melissa (4 June 2020). “WHO to resume hydroxychloroquine trial after earlier halt over safety concerns”. Guardian. Retrieved 25 June 2020.
- Soumya Swaminathan [@doctorsoumya] (4 June 2020). “We acted in the safety interests of patients in the trial, relying on the advice of our steering comm. It is v difficult for us to check data quality of each published paper & we trust authors to adhere to basic stds. HCQ restarted today after data safety committee approval” (Tweet) via Twitter.