Publication ethics: an embarrassing amount of room for improvement

1. Publication ethics: an embarrassing amount of room for improvement Richard Smith Editor, BMJ www.bmj.com/talks

2. What I want to talk about • Why research misconduct matters • Britain’s most dramatic case of misconduct • Other cases • What is research misconduct? • How common is it? • Conflict of interest as a case study • Why does misconduct happen? • What does a country need to do to respond? • A comment on COPE (Committee on Publication Ethics) • An editor’s intray

3. Why research misconduct matters • It’s like child abuse: we didn’t recognise it, now we see a lot • It undermines public trust in medical research and health workers • It corrupts the scientific record and leads to false conclusions • Most countries do not have good systems of either treatment or prevention

4. Britain’s most dramatic case of fraud

5. August 1996: a major breakthrough • Worldwide media coverage of doctors in London reimplanting an ectopic pregnancy and a baby being born • Doctors had been trying to do this for a century. It was a huge achievement

6. August 1996: a major breakthrough • Achieved by Malcolm Pearce, a senior lecturer in at St George’s Hospital Medical School in London • A world famous expert on ultrasonography in obstetrics • A story from a paper in the British Journal of Obstetrics and Gyneacology. Pearce was an assistant editor. • A second author on the case report was Geoffrey Chamberlain, editor of the journal, president of the Royal College of Obstetricians and Gynaecologists, and professor and head of department at St George’s. • The same issue contained a randomised controlled trial also by Malcolm Pearce & others

7. Autumn 1996: both papers are fraudulent • A front page story in the Daily Mail exposed the two papers as fraudulent. • It had a full length picture of Geoffrey Chamberlain saying that he hadn’t known that the work was fraudulent despite his name being on the paper. • Chamberlain said it was common within medicine for people to have their name on papers when they hadn’t done much.

8. What had happened? • A young doctor at St George’s Hospital Medical School had raised questions about the two papers • An investigation was promptly started and showed: • The patient did not exist • The patients supposedly in the randomised trial could not be found • Among studies investigated back to 1989 - three others fraudulent, two of them in the BMJ.

9. The case of Peter Nixon, cardiologist • A television programme accused him of fraud. He sued for libel. His case collapsed. • Rigged the results of a breathing test • Had not written or even read papers published in the Journal of the Royal Society of Medicine over his name

10. The case of Peter Nixon, cardiologist • Admitted errors could not be due to honest error • In case control studies applied different tests to cases and controls • Found “effort syndrome” in virtually all patients who consulted him with chronic fatigue syndrome--but failed to report that 55% of controls were also positive

11. John Anderton, former registrar of the Royal College of Physicians of Edinburgh • In a drug trial forged consent of 17 patients who were never given the drug • Invented echocardiographs and magnetic resonance images for patients

12. Britain’s slowest case?

13. Britain’s slowest case? • Anjan Banerjee and Tim Peters: paper in Gut 1990 on drug induced enteropathy in the and inflammatory bowel disease (Gut 1990--contained falsified data • The same issue contained an abstract due to be presented at the British Society of Gastroenterology. Withdrawn but still published in Gut • Both papers retracted in March 2001

14. Britain’s slowest case? • Banerjee was awarded a Master of Surgery degree by the University of London for work that included the fraudulent work--still not retracted • December 2000. Banerjee found guilty of serious professional misconduct for falsifying data and suspended • September 2002. Banerjee found guilty of serious professional misconduct for financial fraud and struck off

15. Britain’s slowest case? • March 2001. Tim Peters, the professor who supervised Banerjee, was found guilty of serious professional misconduct for failing to take action over the falsified research • The GMC hearings were hampered by notebooks being “selectively shredded” by Kings,the medical school • Authorities at Kings conducted an inquiry in 1991 but did not inform the GMC or Gut

16. Does medicine have a culture that turns a blind eye to research misconduct?

17. What is research misconduct? • The Americans have argued for years over a definition • The Europeans have tended to take a broad view and not attempt a specific, operational definition

18. US Commission on Research Integrity (1996) • Research misconduct is significant misbehaviour that improperly appropriates the intellectual property or contributions of others, that intentionally impedes the progress of research, or that risks corrupting the scientific record or compromising the integrity of scientific practices. Such behaviours are unethical and unacceptable in proposing, conducting, or reporting research, or in reviewing the proposals or research reports of others.

19. Definition of research misconduct proposed by a British consensus panel (1999) • "Behaviour by a researcher, intentional or not, that falls short of good ethical and scientific standards."

20. A preliminary taxonomy of research misconduct (ranked by seriousness) I • Fabrication: invention of data or cases • Falsification: wilful distortion of data • Plagiarism: copying of ideas, data or words without attribution • Failing to get consent from an ethics committee for research

21. A preliminary taxonomy of research misconduct (ranked by seriousness) II • Not admitting that some data are missing • Ignoring outliers without declaring it • Not including data on side effects in a clinical trial • Conducting research in humans without informed consent or without justifying why consent was not obtained from an ethics committee

22. A preliminary taxonomy of research misconduct (ranked by seriousness) III • Publication of post hoc analyses without declaration that they were post hoc • Gift authorship • Not attributing other authors • Redundant publication • Not disclosing a conflict of interest

23. A preliminary taxonomy of research misconduct (ranked by seriousness) IV • Not attempting to publish completed research • Failure to do an adequate search of existing research before beginning new research

24. What is fraud? • We need a full taxonomy • Better we need codes of good research practice--and we now have several

25. How common is fraud? • Obviously depends on how fraud is defined? • How does serious fraud relate to minor fraud? • Are they quite separate? • Does minor progress to serious?

26. What is the relation of minor to serious research misconduct?

27. What is the relation of minor to serious research misconduct?

28. How common is fraud? • How many of you know of a case? • In how many of those cases was there a proper investigation, punishment if necessary, and a correction of the scientific record?

29. Study by Stephen Lock • Asked 80 researchers who were friends, mostly British and mostly professors of medicine. Not a random sample. • 100% response rate. • Over half knew of cases: • Over half the dubious results had been published - only 6 “retractions” - all vague and not using that term

30. How common is fraud? • US congressional inquiry heard of over 700 cases • The British General Medical Council has dealt with over 30 cases • Committee on Publication Ethics has discussed over a 100 cases

31. How common is research misconduct? • Redundant publication occurs in around a fifth of published papers • About a fifth of authors of studies in medical journals have done libtle or nothing • Most authors of studies in medical journals have conflicts of interest, yet they are declared in less than 5% of cases

32. Conflict of interest: a case study in poor performance within biomedicine

33. How common are competing interests? • 75 articles • 89 authors • 69 (80%) responded • 45 (63%) had financial conflicts of interest • Only 2 of 70 articles disclosed the conflicts of interest • Stelfox HT, Chua G, O'Rourke K, Detsky AS. Conflict of interest in the debate over calcium channel antagonists. N Engl J Med 1998; 338: 101-105

34. Why don’t authors declare conflicts of interest? • Some journals don’t require disclosure • The culture is one of not disclosing • Authors think that it’s somehow “naughty” • Authors are confident that they are not affected by conflicts of interest

35. Does conflict of interest matter? • Financial benefit makes doctors more likely to refer patients for tests, operations, or hospital admission, or to ask that drugs be stocked by a hospital pharmacy. • Original papers published in journal supplements sponsored by pharmaceutical companies are inferior to those published in the parent journal. • Reviews that acknowledge sponsorship by the pharmaceutical or tobacco industry are more likely to draw conclusions that are favourable to the industry.

36. Does conflict of interest matter? • Is there a relationship between whether authors are supportive of the use of calcium channel antagonists and whether they have a financial relationship with the manufacturers of the drugs? • Stelfox HT, Chua G, O'Rourke K, Detsky AS. Conflict of interest in the debate over calcium channel antagonists. N Engl J Med 1998; 338: 101-105

38. Does conflict of interest matter? • 106 reviews, with 37% concluding that passive smoking was not harmful and the rest that it was. • Multiple regression analysis controlling for article quality, peer review status, article topic, and year of publication found that the only factor associated with the review's conclusion was whether the author was affiliated with the tobacco industry. • Only 23% of reviews disclosed the sources of funding for research. • Barnes DE, Bero LA. Why review articles on the health effects of passive smoking reach different conclusions. JAMA 1998; 279: 1566-1570

39. Does conflict of interest matter?: third generation contraceptive pills • At the end of 1998 three major studies without sponsoring from the industry found a higher risk of venous thrombosis for third generation contraceptives; three sponsored studies did not. • To date, of nine studies without sponsoring, one study found no difference and the other eight found relative risks from 1.5 to 4.0 (summary relative risk 2.4); four sponsored studies found relative risks between 0.8 and 1.5 (summary relative risk 1.1) • The sponsored study with a relative risk of 1.5 has been reanalysed several times, yielding lower relative risks; after this failed to convince, a new reanalysis was sponsored by another company.

40. Why does scientific fraud happen? • Why wouldn’t it happen? It happens in all other human activities. • Pressure to publish. • Inadequate training. Not taught good practice. Indeed, sometimes taught the opposite. • Does sloppy behaviour spill over to fraud? • You can get away with it. The system works on trust.

41. What does a country need to respond to research misconduct? • A recognition of the problem by the medical community and its leaders • An independent body to lead with investigations, prevention, teaching and research • An agreement on what fraud is • Protection for whistleblowers • A body to investigate allegations • A fair system for reaching judgements • A code of good practice • Systems for teaching good practice

42. Does your country have these characteristics

43. Committee on Publication Ethics (COPE) • Founded in 1997 as a response to growing anxiety about the integrity of authors submitting studies to medical journals. • Founded by British medical editors--including those of the BMJ, Gut, and Lancet

44. COPE’s five aims • Advise on cases brought by editors • Publish an annual report describing those cases. Three published (www.publicationethics.org.uk) • Produce guidance on good practice • Encourage research • Offer teaching and training • (Shame the British establishment into mounting a proper response)

45. COPE’s first 103 cases • In 80 cases there was evidence of misconduct. • Several cases have been referred to employers and to regulatory bodies • Problems were • undeclared redundant publication or submission (29), • disputes over authorship (18) • falsification (15) • failure to obtain informed consent (11) • performing unethical research (11) • failure to gain approval from an ethics committee (10)

46. Three cases from an editor’s intray

47. Case 1 • A major RCT showing that an nutritional intervention can improve cognitive function in the elderly • An editor, a statistician, and a reviewer all raise doubts about the data and the tests used • Approach the university • It “investigates”: no problem • BMJ queries investigation • Researcher flees: university seems to see this as an admission of guilt • Should we notify the editors of a journal that published another paper from the same trial?

48. Case 2 • Two people write to me to suggest (with evidence) that many papers by an author, including several that we have published and two we are considering are fraudulent • Statistician agrees • We ask for raw data • Raw data arrives in a box, written out by hand after six months • Statistician takes years to analyse--says “I would go to court to say that these data are fraudulent” • Researcher owns his own hospital

49. Case 3 • I’ve been on this case for 12 years • Should we publish something in the BMJ? • We are about to publish a paper on chronic fatigue syndrome • An anonymous letter says the work is fraudulent • What should we do?

50. Conclusion • Research misconduct is a problem • Most countries have not developed a coherent response to the problem • They need to in order to avoid a collapse in public trust in medical research