Scientific fraud or simple mistakes?
A recent article caught my attention in relation to what I used to do for a living before covid. Years ago, in my role as an expert in image authentication, I was involved in a number of cases relating to academic fraud. In those cases, one researcher would accuse another research of various types of academic fraud and their institution, in my case it was the US Veterans Affairs Dept., would conduct an investigation to sort things out.
It also caught my attention as I had previously called out the FBI’s FAVIAU for engaging in similar unsavoury practices years ago (see here and here).
The article reports that the Dana-Farber Cancer Institute, a Harvard University-affiliated teaching hospital, is retracting or correcting dozens of research papers authored by four of its top researchers, including the hospital's CEO Dr. Laurie Glimcher.
The retractions and corrections follow an investigation into allegations of data falsification made by a data sleuth named Sholto David, who published a blog post earlier in January identifying irregularities in 57 papers published by these researchers.
So far, the hospital has initiated 6 retractions and 31 corrections. The exact nature of the issues is not fully specified, but they relate to things like duplications of images, blots, bands, and plots. The hospital says that while discrepancies were identified, they do not necessarily indicate intentional deception or misconduct on the part of the authors.
The article situates this news in the context of a recent plagiarism scandal involving Harvard University’s president, Claudine Gay, who resigned after facing dozens of accusations of plagiarism and inadequate citation.
Overall, the article suggests that this is the latest case highlighting credibility and ethical issues around research coming out of Harvard-affiliated institutions and researchers. It remains to be seen what additional findings and actions may result as the hospital continues investigating the remainder of the flagged papers.
Sholto David’s points
Here is a summary of the key points from Solto David’s article:
Sholto analysed papers from researchers at the Dana-Farber Cancer Institute (DFCI), which is affiliated with Harvard University. He found evidence of problematic data manipulation and falsification in papers from several top DFCI researchers.
The researchers implicated include DFCI's president and CEO Dr. Laurie Glimcher; Executive VP and COO Dr. William C. Hahn; Senior VP Dr. Irene Ghobrial; and Harvard professor Dr. Kenneth Anderson. Examples provided include western blot manipulations, duplicated images, mislabeled data, and questionable animal experiments.
Sholto situates these issues in the wider context of the recent plagiarism scandal involving Harvard's president and suggests it reflects a broader lack of research integrity and accountability at Harvard-affiliated institutions. He argues obvious errors have gone uncorrected for years.
In total, Sholto discusses problematic papers from over 10 Harvard-connected researchers. He highlights how many are in positions of leadership at DFCI, edit major journals, or have links to industry.
The article concludes by noting DFCI has initiated 6 retractions and 31 corrections in response to the allegations of data manipulation. However, Sholto questions whether DFCI's research integrity officer leading this investigation may be too conflicted given his publishing history with authors implicated.
In summary, the article provides extensive evidence that high-profile cancer research from Harvard-affiliated DFCI suffers from pervasive and unaddressed problems with research integrity based on a review of the published literature. Sholto suggests reform is overdue.
So what?
The “or have links to industry” part caught my attention. What does it mean to have “links to industry?” Why might these links be important to sort out?
The article does mention connections to industry for a few researchers:
Dr. William C. Hahn has published research with Dr. David Sabatini from MIT, who has faced sexual misconduct allegations. Sabatini has been involved with founding multiple biotech companies.
Dr. Levi Garraway, former Global Head of Pharma Research and Early Development at Roche, has published concerning research collaborations with Dr. Hahn.
Dr. James E. Bradner, formerly an executive at Novartis and just appointed as Chief Scientific Officer at Amgen, has published problematic studies with both Dr. Hahn and Dr. Anderson.
Again, so what?
The potential influence of industry ties on research integrity has attracted growing attention and unease over the last 20 years. Financial relationships with drug or biotech firms have been flagged as risk factors that may unintentionally introduce bias or occasionally even deliberately distort research outcomes.
Several high-profile cases have catalysed wider debate around industry-sponsored trials and university-industry collaboration agreements. These cases revealed instances of compromised data, undeclared financial interests from authors, and restrictive publication terms set by corporate sponsors.
In response, medical journals and universities have updated their conflict of interest and disclosure policies. However, recent meta-research across fields like nutrition, clinical medicine, and engineering indicates industry-funded studies still preferentially report outcomes favourable to the sponsor's interests. Reviews have connected reliance on commercial research money to practices like selective publishing of positive results.
Why does this matter for reputations like Harvard and for cancer research progress? Industry provides opportunities to scale up research and develop new treatments. But preventable errors or overstated findings in early preclinical and clinical studies can steer whole fields down unproductive paths for years.
This risks squandering public money supporting translational research programmes, delaying real breakthroughs for patients. It may damage public trust in medical research, which relies on rigorous methods and data to advance.
As such, institutions training future researchers and partnering with industry carry particular responsibility to uphold research integrity. Where conflicts of interest around funding or commercial opportunities may unconsciously prejudice scientific judgement, open scrutiny, traineeship, and policies to strengthen objectivity remain vital safeguards.
Addressing perceived integrity issues, whether isolated mistakes or cultural norms enabling poor practices, should matter profoundly to Harvard’s cancer centre given its global status and strategic vision to accelerate progress against intractable diseases.
Why would they do it?
There are likely multiple potential motivations why researchers might compromise research integrity through practices like data manipulation or falsification:
Career incentives and pressures - In a hyper-competitive research climate, high-profile publications and groundbreaking results are valued for career advancement and notoriety. This may incentivise rushing studies or tweaking data to achieve an exciting story.
Financial incentives - Industry funding or commercialisation opportunities around research can introduce conflicts of interest that bias study design and reporting. The potential financial gains may lead some to disregard integrity.
Culture and role models - Working in research groups or institutions where questionable practices are normalised or tacitly tolerated can reinforce viewing this as acceptable scientific behavior to emulate. See the above links to FBI’s FAVIAU.
Ego and self-deception - A conviction that you “know” the underlying truths despite messy data coupled with a high degree of self-belief in your innate brilliance can motivate rationalisations for cutting corners.
Complacency and convenience - Established senior researchers may feel scrutiny is lower and they can get away with integrity lapses or that corrective actions are tedious. Data errors may be seen as expedient.
Many factors are likely intersecting in complex ways. High workloads, presentation pressures, confirmation bias, technical constraints, and simple mistakes may also play enabling roles even without malicious intent from researchers. But ultimately personal responsibility and leadership set the tone for integrity.
What does this mean?
It's a troubling possibility to confront, but based on the trends evidenced here and broader meta-research on research integrity issues, the scale of studies requiring correction or retraction across western academia is likely quite substantial.
Whilst most researchers undoubtedly act ethically and conduct careful work, the intense publish-or-perish environment today involves powerful pressures and incentives that can strain integrity.
Given the levels of problematic data recently exposed in a flagship institute like Harvard’s cancer center, there are likely deeply concerning norms being transmitted in other elite labs and high-stakes research areas too. Review processes seem to be frequently failing to detect even basic data inconsistencies before publication.
Estimates vary, but multiple meta-analyses gauging the scale of issues like data fabrication, falsification, plagiarism, and questionable research practices point to anywhere from 1-3% of scientists admitting to clear misconduct and perhaps 15-20% observing those behaviors from colleagues. Other studies examining specific integrity issues report higher rates still in some fields. (source) (source) (source) (source) (source)
Factoring in the unknown scope of unintentional errors and corrections needed, a conservative lower bound based on the frequencies above would suggest no less than 40,000 biomedical papers likely warranting retraction across western research institutions today. The true figure could be 200,000 or more depending on one’s thresholds for retracting versus correcting.
In aggregate, this constitutes an emergency for the research system … and for “the evidence.” The downstream wasting of resources, health risks from distorted findings, disillusioning of trainees, and loss of public trust require urgent collective action to undertake reforms. Research leaders can no longer pretend integrity lapses are rare bad apples instead of prompting cultural change.