H. Gilbert Welch, MD, was professor at the Dartmouth Institute of Health Policy and Clinical Practice in New Hampshire and also held positions at several sister colleges. He is well known as a critic of routine cancer screening, emphasizing that potential harms may outweigh potential benefits, and charging that such screening leads to “overdiagnosis,” that is, a finding of cancer that would never lead to clinical symptoms.
Welch resigned on September 14, saying that he could not agree with the “punishment” that Dartmouth had outlined for him.
The Institute had found him guilty of plagiarism on a paper about mammography screening published in 2016 in the New England Journal of Medicine (NEJM). Two researchers, one of whom was colleague at the same institution, claimed ownership of the idea explored in the paper: that mammography finds small tumors that are not clinically significant and thus leads to overdiagnosis.
Welch disagreed with the finding of plagiarism and said it was an “authorship issue.”
The NEJM also ruled that this was an “authorship dispute” and did not retract the paper.
STAT now reports that Welch has resigned as a consequence of the dispute.
The report reveals that in an email to colleagues, Welch wrote that the dean of Dartmouth’s medical school had proposed that, as punishment, Welch contact the journal to revise the authorship of his paper by listing the complainant as first author and that Welch could stay employed by the institution but not be allowed to teach.
“I cannot in good conscience accept the demand that I make the complainant an author — much less the demand that I make him the first author,” he wrote. “Doing so requires that I falsely attest that he meets the requirements of authorship: namely, that he materially participated in the work and is able to defend it. Much as I have enjoyed working at Dartmouth, I am not willing to falsely attest to anything simply to stay here.”
“Furthermore, the demand that I no longer teach subverts the very reason I came to work at Dartmouth,” he added.
Journal Does Not Retract Paper
In August, Dartmouth sent a summary of its investigation to the NEJM and requested that the study be retracted.
However, the NEJM declined to withdraw the article, saying the issue was an “authorship dispute” and that its opinion was in line with the US Office of Research Integrity.
The journal described the situation as a case of “alleged research misconduct.”
“The complainant party in this matter claims that he was not given adequate recognition for his contribution to the work we published. We do not deem this sufficient grounds for retraction of the article,” summarized Jeffrey Drazen, MD, editor in chief, and Dan Longo, MD, deputy editor in a letter posted online.
The NEJM editors also said they would like to amend the study as needed. “We are happy to work with you and the article authors to reach a solution whereby sufficient acknowledgment is given so that the contribution of the complainant is adequately recognized,” they wrote.
Dispute Over Who Had the Idea
The study at the center of the dispute, which was published in 2016 in NEJM, concluded the overdiagnosis of breast cancer via mammography screening is “larger than is generally recognized.” The study‘s findings and methods were controversial at the time of publishing, as reported by Medscape Medical News.
Welch was first author on the paper, with a coauthor from Dartmouth and two other coauthors from the National Cancer Institute.
However, two other researchers claimed that the idea behind the paper came from them, and Welch was accused of research misconduct by fellow Dartmouth faculty member Samir Soneji, PhD, and another academic, Hiram Beltran-Sanchez, PhD, from the University of California Los Angeles.
Internal documents from Dartmouth obtained in August 2018 by Retraction Watch showed that after an investigation, Dartmouth eventually concluded that Welch “engaged in research misconduct, namely, plagiarism, by knowingly, intentionally, or recklessly appropriating the ideas, processes, results or words of Complainants without giving them appropriate credit.”
At the time, however, Welch emphasized to Retraction Watch that the dispute is “about the origin of the idea — not about the validity of the work.”
That idea relates to tracking the size of breast tumors over time epidemiologically as a way of determining whether screening was effective.
Soneji says he came up with the idea. In an email posted in the Retraction Watch story, he wrote: “Professor Welch and I met in his office and he said that he had never thought of my research ideas to a) compute the share of incident breast cancers by tumor size over time and b) use the temporal patterns in tumor size-specific incidence rates to quantify the contribution of breast cancer screening.”
However, Welch argued that the 2016 NEJM paper is a “natural progression of my work.”
For example, he was the lead author of a 2012 epidemiologic study, also published in the NEJM and reported by Medscape Medical News, that tracked the incidence rates of breast cancer stage (early vs late) over time as a way of determining overdiagnosis.
Welch Tells His Side of the Story
The latest STAT article, published September 14, contains excerpts from an email that Welch sent to his Dartmouth colleagues about his resignation, in which he tells his side of the story.
Welch says that he had respected the confidential nature of the Dartmouth investigation for nearly 2 years “and never wanted it to become public. But that changed when the complainant chose to release confidential documents from the investigation to STAT in mid-August, and I now feel compelled to respond.”
He argues that Dartmouth’s verdict in the case was at odds with the conclusions of the NEJM and the US Office of Research Integrity. Both agreed that his actions did not rise to the level of “idea plagiarism” or research misconduct but rather were a dispute over credit.
Welch acknowledged that he was intrigued by his colleague Soneji’s data, which he first saw in 2015 at a seminar Soneji was giving. He asked for and received a slide and noticed what he says were several errors. Welch said that he communicated with Soneji about trying to improve the analysis, but Soneji didn’t seem interested. Meanwhile, Welch said he fixed the errors and made his own figure with the corrected data and then used this modified figure in the 2016 NEJM article.
In other words, Welch argued, Soneji’s data were an influence, but in reverse, according to the STAT report. “Ironically, it’s [sic] biggest influence was simply because I realized it contained errors. That is what drove me to look directly at the data myself,” he wrote.
Welch said Dartmouth focused on the similarity between the figure in the NEJM paper and that in Soneji’s slide rather than looking at the data themselves.
Welch also dismissed the notion that the information was novel.
“It was the product of ideas and methods I have been using for decades — ideas and methods that are the result of the influences that many others have had on my career,” he wrote.
Welch concluded the email by telling his colleagues, “For nearly 30 years, I have been blessed to be able to work with the many fine staff, faculty and clinicians at Dartmouth. It has been an honor to work with all of you.”
The STAT report also notes that Soneji has filed a grievance with Dartmouth in which he claims to have been the victim of retaliation for his complaint against Welch. The grievance alleges that he felt pressured to consider entering into mediation with Welch — presumably putting a halt to the investigation. The investigation went forward, however, and there was no mediation. Soneji also implies in the grievance that the 10% salary rise he received when he was promoted earlier this year was less than it should have been and may have been an effort to penalize him for speaking out against his renowned colleague.