By Geoff Webb
I have been writing about major errors in the biomedical sciences since the late 1980s. My area of interest widened to include research fraud when I discovered that two authors who I had frequently cited in my own books and papers had been accused of fabricating their data.
A wide range of methods is available to the biomedical researcher: observational epidemiological methods, in vitro and animal studies and experimental studies using human subjects. Meta-analysis has become a very popular method of aggregating similar studies with the same outcome measures; it effectively combines them into a larger study of greater statistical power. Evidence hierarchies or pyramids are used by groups like NICE to decide whether the accumulated evidence favouring a hypothesis is sufficient to authorize treatments or other health-related behaviours. Nevertheless, errors or false theories escape scrutiny and not only reach publication but persist in the scientific and public health consciousness. For example:
- The almost unanimous belief that there was a huge and increasing shortfall in world protein supply (the protein gap) and that primary protein deficiency was the most common nutritional disorder in the world;
- The promotion of front sleeping for infants that led to a worldwide epidemic of cot deaths in the 1970s and 1980s;
- The belief that antioxidant supplements would substantially increase life expectancy even in well-nourished people; and
- The assertion that a defect in the heat-generating capacity of brown adipose tissue was a cause of human obesity and that thermogenic drugs were a promising therapeutic strategy.
Some scholars have proposed that the conclusions of most published research are wrong and that up to 85% of research expenditure is wasted, that is, that the four error case studies above are extreme examples at the tip of a very large iceberg of false theories and conclusions. One reason for such a pessimistic estimate is the pressure on academic scientists to publish. Paper counting refers to the metric whereby scientists are assessed by the number of their publications (Editor’s note: see News in Brief for a link to a talk by Professor David Colquhoun on this). Responding to this demand are many predatory journals that will publish anything for a fee, even meaningless strings of computer-generated jargon. The result has been an avalanche of low value papers that will remain largely unread, sometimes even by peer reviewers who should be the gatekeepers for quality. This could give the impression that all of the advances in science and medicine stem from about 15% of research expenditure.
As well as low value research, there are also cases of research misconduct, primarily the deliberate fabrication of results. I have encountered examples spanning 150 years and spread across the biological and medical sciences, perpetrated by doctors, dentists, biochemists, palaeontologists, botanists, zoologists and psychologists.
one deceived co-author of a fraudulent paper hanged himself after its exposure
Detected cases of research fraud may be relatively few but some have caused serious and long-lasting harm. Many are committed by serial offenders, sometimes spread over a decade or more; one would be unwise to accept any data such researchers cite. These fraudsters squander and misuse scarce research resources, mislead and waste the time and efforts of other researchers and harm the careers of students and co-workers. One deceived co-author of a fraudulent paper hanged himself after its exposure.(1) Several have been responsible for inappropriate treatments or health advice being given, and fabricated or falsified research data has led to patient harm and deaths. How much of the current vaccine hesitancy can be blamed on the questionable research about the MMR vaccine causing autism?
Werner Bezwoda was an oncologist based at the University of Witwatersrand in South Africa until his dismissal in 2001. In 1995, he published clinical trial data which claimed to show that using high doses of chemotherapy (HDC) with haematopoeitic rescue increased survival time in some women with advanced breast cancer; it seemed to effectively cure almost a third of them. HDC uses high doses of multiple chemotherapeutic drugs in an effort to kill all of the cancer cells and so prevent recurrence. These cytotoxic drugs also destroy the bone marrow, so bone marrow (or peripheral stem cells) is harvested prior to treatment and then re-introduced after chemotherapy i.e., haematopoetic rescue. Tens of thousands of women underwent this very expensive, gruelling and potentially life-ending treatment for no discernible benefit. Some phase II trials had suggested that HDC increased average survival times but Bezwoda’s trial was the only positive controlled trial published. Auditors were sent to South Africa when he presented data from a second positive trial at a 1999 conference in Atlanta; his positive data contrasted sharply with negative data presented by four other groups. A group in Seattle replicated Bezwoda’s published treatment protocol in a small phase II trial. It was quickly abandoned as four of the first six patients suffered severe cardiac toxicity causing the death of two and severe, permanent cardiac damage in two others. He later admitted falsifying his results, and retracted the publication.(2)
What safeguards are there to prevent the publication of fraudulent and erroneous data? Peer review is traditionally seen as the key barrier to publication of flawed data but it was not designed to combat deliberate deception and is ineffective in preventing fraud. Co-authors must take responsibility for ensuring the integrity of data published under their names in which senior departmental figures uninvolved in the research are named as authors (so-called “gift authorship”). In some cultures this is deemed a courtesy, but is no longer acceptable. Some potential fraudsters might be dissuaded from misconduct if they believed that discovery was more probable and that the consequences were likely to be severe. Should this in fact include criminal prosecution and imprisonment?
should the consequences include criminal prosecution and imprisonment?
Expert scrutiny of already-published papers has occasionally raised suspicion of data fabrication that was missed during peer review, but only rarely have major fraudsters been unmasked by either peer review or failed reproducibility – the traditional quality-control of published research. Whistle-blowers, colleagues or students who have reported suspicions of misconduct, have been responsible for unmasking many offenders. But some whistle-blowers whose accusations have ultimately been proven have been harassed, intimidated or had their career prospects blighted. Those who, in good faith, report suspicions of misconduct should be protected. Everything possible should be done to encourage potential whistle-blowers to report genuine suspicions.
Who should investigate cases of alleged research misconduct? Employers are often best placed to carry out any investigation and gain access to key data and co-workers. Some employer investigations are a credit to the institutions and should be a model for others to follow, with thorough, well-conducted and transparent investigations carried out by employers, professional bodies, journals and research funders. Other employers have abrogated their responsibilities in this regard, perhaps in an attempt to avoid adverse publicity.
A dilemma faces journal editors: how to disinfect the literature once misconduct has been identified and how to reduce further publication of fraud. Ideally, once one fraudulent paper is identified, all the guilty party’s published work should be scrutinized for indications of misconduct and any suspect work retracted unless co-authors can demonstrate that the data was honestly generated. Retraction should become the default position for any suspect work of a fraudulent author. Research misconduct by an author is rarely a “one-off” aberration but often an indicator of a pattern of behaviour spread over many years.
it is common for retracted papers to continue to be cited positively after retraction
Journals must make it clear when a paper has been retracted - it is common for retracted papers to continue to be cited positively after retraction. Some journals print “RETRACTED” across every page of the electronic article whilst in other cases it is easy to miss the retraction notice. A few journals still charge readers to access retracted papers and commentaries relating to them. Ideally the reasons for a retraction should also be published.
Greater awareness of research fraud and some of its signs and symptoms would be an important step in reducing its prevalence. Credible accusations must always be seriously and thoroughly investigated. The aura of secrecy around accusations should be lessened and they should be handled as transparently as possible and the results of formal investigations made public. “Sunlight is the best disinfectant”.
Geoffrey P Webb
Senior Lecturer, University of East London
Geoff Webb is the author of “Error and Fraud. The Dark Side of Biomedical Research”, due to be published by Taylor and Francis at the end of May 2021. He blogs as Dr Geoff.
- Webb G. In "Error and Fraud. The Dark Side of Biomedical Research". Published by Taylor & Francis, May 2021.
- Grady D. Breast Cancer Researcher Admits Falsifying Data. New York Times, 5 Feb 2000. https://www.nytimes.com/2000/02/05/us/breast-cancer-researcher-admits-falsifying-data.html
22 April 2021