Retracted COVID-19 study remains ‘widely cited’ in medical literature

A retracted New England Journal of Medicine study on the link between CVD, renin-angiotensin–aldosterone system inhibitor therapy and COVID-19 outcomes continues to be “widely cited” in medical literature, researchers wrote.

The study was retracted because of concerns that the data, which were pulled from the “little-known” company Surgisphere, were “fraudulent,” Todd C. Lee, MD, MPH, FIDSA, an associate professor in the department of medicine at McGill University in Montreal, and colleagues wrote in JAMA Internal Medicine. Another study in The Lancet that used the same database to examine the safety and efficacy of hydroxychloroquine for hospitalized patients with COVID-19 was also retracted, Lee and colleagues noted.

The NEJM article and Lancet article had Altmetric attention scores of 3,727 and 23,084, respectively, as of June 11, according to the researchers. In general, the higher an article’s Altmetric score is, the greater the public’s interest in that article, according to the Altmetric website.

Lee and colleagues analyzed data from Google Scholar to determine how many works cited the retracted NEJM article. The researchers said they chose this article “because it was retracted soon after publication, received prominent media attention, and was not subject to controversies, as has been the case for some studies of hydroxychloroquine and COVID-19, the topic of the other prominent article.”

According to Lee and colleagues, 934 articles cited the retracted NEJM article. Of that total, 15.6% were either preprints or not from peer-reviewed journals, 6.7% were replica citations and 0.2% were false links. The researchers could not identify the full text for 1.1% of the citations.

A citation was confirmed in 91.4% of the remaining articles. Of those, 10.7% were published the month before the retraction, 34.8% were published within the first 2 months, 54.4% were published 3 months or more after the retraction and 27.8% were published 6 months or later. In addition, 11 months after the retraction, the article was cited 21 times. Only 17.6% of the 652 articles with a confirmed citation noted that the article had been retracted.

Lee and colleagues also noted that although most of the confirmed citations used the NEJM article data to support a statement, 2.6% used data from the retracted article as part of a new analysis. Among those articles, 64.7% were published 3 or more months after the article was retracted and 41.2% were published 6 months or later.

Lee and colleagues acknowledged that they may have missed a few citations.

“Nevertheless, there is no reason for a retracted study to continue to be widely cited in the medical literature months after it was retracted, and, in some instances, for the retracted data to be incorporated into new analyses,” Lee and colleagues wrote.

They encouraged medical journal stakeholders to do a “better job of addressing the broader issues of ongoing citations of retracted scientific studies and protecting the integrity of the medical literature.”

References:

Altmetric.com. The Altmetric score is now the Altmetric Attention Score. https://www.altmetric.com/blog/the-altmetric-score-is-now-the-altmetric-attention-score/. Accessed July 30, 2021.

Lee TC, et al. JAMA Intern Med. 2021; doi:10.1001/jamainternmed.2021.4112.