From the Journals

Despite retraction, study using fraudulent Surgisphere data still cited


A retracted study on the safety of blood pressure medications in patients with COVID-19 continues to be cited nearly a year later, new research shows.

Floaria Bicher/iStock/Getty Images Plus

The study in question, published on May 1, 2020, in the New England Journal of Medicine, showed no increased risk for in-hospital death with the use of ACE inhibitors or angiotensin-receptor blockers (ARBs) in hospitalized patients with COVID-19.

Concerns about the veracity of the Surgisphere database used for the study, however, led to a June 4 retraction and to the June 13 retraction of a second study, published in the Lancet, that focused on hydroxychloroquine as a COVID-19 treatment.

Although the Surgisphere scandal caused a global reckoning of COVID-19 scientific studies, the new analysis identified 652 citations of the NEJM article as of May 31.

More than a third of the citations occurred in the first 2 months after the retraction, 54% were at least 3 months later, and 2.8% at least 6 months later. In May, 11 months after the article was retracted, it was cited 21 times, senior author Emily G. McDonald, MD, MSc, McGill University, Montreal, and colleagues reported in a research letter in JAMA Internal Medicine.

“In early May and June there were already more than 200 citations in one of the world’s leading scientific journals, so I do believe it was a highly influential article early on and had an impact on different types of studies or research taking place,” she said in an interview.

Dr. McDonald said she’s also “certain that it impacted patient care,” observing that when there are no guidelines available on how to manage patients, physicians will turn to the most recent evidence in the most reputable journals.

“In the case of ACE [inhibitors] and ARBs, although the study was based on fraudulent data, we were lucky that the overall message was in the end probably correct, but that might not have been the case for another study or dataset,” she said.

Early in the pandemic, concerns existed that ACE inhibitors and ARBs could be harmful, increasing the expression of ACE2 receptors, which the SARS-CoV-2 virus uses to gain entry into cells. The first randomized trial to examine the issue, BRACE CORONA, showed no clinical benefit to interrupting use of the agents in hospitalized patients. An observational study suggested ACE inhibitors may even be protective.

Of two high-profile retractions, McDonald said they chose to bypass the hydroxychloroquine study, which had an eye-popping Altmetric attention score of 23,084, compared with 3,727 for the NEJM paper, because it may have been cited for “other” reasons. “We wanted to focus less on the politics and more on the problem of retracted work.”

The team found that researchers across the globe were citing the retracted ACE/ARB paper (18.7% in the United States, 8.1% in Italy, and 44% other countries). Most citations were used to support a statement in the main text of a study, but in nearly 3% of cases, the data were incorporated into new analyses.

Just 17.6% of the studies cited or noted the retraction. “For sure, that was surprising to us. We suspected it, but our study confirmed it,” Dr. McDonald said.

Although retracted articles can be identified by a watermark or line of text, in some cases that can be easily missed, she noted. What’s more, not all citation software points out when a study has been retracted, a fate shared by the copyediting process.

“There are a lot of mechanisms in place and, in general, what’s happening is rare but there isn’t a perfect automated system solution to absolutely prevent this from happening,” she said. “It’s still subject to human error.”

The findings also have to be taken in the context of a rapidly emerging pandemic and the unprecedented torrent of scientific papers released over the past year.

“That might have contributed to why this happened, but the takeaway message is that this can happen despite our best efforts, and we need to challenge ourselves to come up with a system solution to prevent this from happening in the future,” Dr. McDonald said. “Current mechanisms are probably capturing 95% of it, but we need to do better.”

Limitations of the present analysis are that it was limited to the single retracted study; used only a single search engine, Google Scholar, to identify the citing works; and that additional citations may have been missed, the authors noted.

McDonald and coauthor Todd C. Lee, MD, report being signatories on a public letter calling for the retraction of the Surgisphere papers. Dr. Lee also reported receiving research support from Fonds De Recherche du Quebec-Sante during the conduct of the study.

A version of this article first appeared on

Next Article: