Obstacles to honesty in medical research
Dr Peter Wilmshurst, a consultant cardiologist, has spent the last two decades trying to expose research misconduct and has reported more than twenty doctors to the General Medical Council. In recognition of his dogged and selfless pursuit of the truth, Dr Wilmshurst was presented with the HealthWatch Award 2003.
I feel greatly honoured to receive the Health Watch Annual Award and I am grateful for the opportunity to speak to you about obstacles to honesty in medical research.
I have been interested in this subject for 20 years, since I first experienced research misconduct when I was a research registrar. I hope that a personal account of my experiences may explain why I believe this is a serious problem.
In 1986 I went to the Guardian Newspaper with the story after the medical and pharmaceutical regulators refused to take any action. I supplied the Guardian's lawyers with over 200 pages of documents and statements, which convinced them that they could successfully defend any legal action if sued. We were not.
My research was on heart failure. This is a common condition and it has a worse 5-year survival than many forms of cancer. Twenty years ago there were few treatment options to improve symptoms and none was proven to improve survival. I was offered the opportunity to do research on a promising new drug, named amrinone. It was patented by Sterling-Winthrop. Preliminary research looked promising. Research, mainly from the company, showed that the drug increased the strength of contraction of the heart in animals. But the most influential article and the one that persuaded me that the drug was worthy of research was on patients and was published in the New England Journal of Medicine in 1978.
The New England Journal is the most influential medical journal in the world. The article came from the Cardiology Department at Harvard and one of its five authors was the most well known cardiologist in the world and head of medicine at Harvard, Professor Eugene Braunwald. The paper was given extra prominence by being the first article in that issue of the Journal and it was accompanied by an editorial.
In a large series of experiments we showed that, although amrinone increased the strength of contraction of normal heart muscle, it did not affect contractility in patients with heart failure. We also found that amrinone frequently caused life threatening side effects.
With hindsight there were two things that should have raised my concerns when we started our research. The first were anomalies in the study from Braunwald's group. It was a small study, which made claims that were not substantiated by the observations reported.
Later I discovered that though the article stated that the 5 authors were employed in the Cardiology Department at Harvard Medical School, 2 were full-time employees of Sterling-Winthrop and had never worked at Harvard. Two of the three that worked at Harvard were paid consultants to the company. These conflicts of interest were not declared. In fact the New England Journal of Medicine had no policy on declaration of conflicts of interest at the time. The first statement on conflicts of interest was published in the New England Journal one month after I wrote to the Massachusetts Medical Society, which owns the Journal, complaining about the undeclared conflicts of interest in this case.[3,4]
The second thing that should have alerted me was a letter published in the New England Journal of Medicine from cardiologists in Los Angeles. The letter reported fatal side effects from amrinone. The first author, Dr Stanley Rubin, had a patient with severe heart failure. The patient's wife was a stock-broker. She saw the dramatic increase in the price of Sterling-Winthrop shares after the paper from Braunwald's group was published. She reasoned that this proved that amrinone was an important advance. She asked Dr Rubin to get amrinone for her husband. Rubin was able to persuade the company to let him have amrinone on a named-patient basis and the amrinone swiftly killed his patient. Rubin and colleagues sent the New England Journal the first report of side effects with amrinone. They did not tell Sterling-Winthrop that they had submitted the report. Within 48 hours Rubin was under pressure by the company to retract the report. The Journal admitted that it had sent Sterling-Winthrop a copy of Rubin's report. The Journal initially refused to publish the report but was forced to do so when Rubin said that if they did not he would go to the press.[3,6,7]
However the conflicts of interest involving the New England Journal, the Cardiologists at Harvard and Sterling-Winthrop did not end there. The company later produced a congener of amrinone, named milrinone. The initial human research on milrinone was also performed in Braunwald's department. Unusually it was agreed before the research had been completed that it would be published in the New England Journal. When the first 2 referees chosen by the journal to review the paper recommended rejection, the editor, Dr Arnold Relman sent the article to 2 more referees. They also recommended rejection, but the Journal published the paper on milrinone as previously agreed.[3,6,7] This says much about peer review in the World's most prestigious medical journal.
I discovered this much later. In the early days of our research my colleagues and I were more concerned that we could not confirm in our large number of experiments claims made in the small study from Braunwald's department.
We reported to Sterling-Winthrop that we were unable to find evidence that amrinone injections increased contractility in patients with heart failure and we reported our experience of serious adverse effects with the oral preparation of the drug. Company employees asked us to exclude some patients from the analysis. These were ones where there was a downward trend in contractility. The effect of excluding them would have been to produce an apparent but spurious increase in contractility in the remainder. We refused. My supervisor and I were then threatened with litigation. We published.
Our on-going research studies on amrinone ended when company employees removed the drug stocks from the pharmacy in the hospital and research institute. As a result, 2 of our publications contain statements pointing out that the studies were smaller than planned because Sterling-Winthrop had prematurely discontinued our trials without our agreement.[9,10]
A number of tactics were used to try to prevent my colleagues and I presenting our findings at meetings and to discredit us when we did present. One strange incident involved one of my colleagues, Alex Crowther, who was due to present some of our work on amrinone on the second day of a meeting in Luxembourg. He just managed to get on the last flight of the day that would permit him to attend the first session of the meeting. When he arrived he discovered that his talk had been rescheduled for the previous day. The organisers had received a forged letter that appeared to be from him asking for his talk to be brought forward a day. Those responsible were never identified.
When I presented our findings on side effects a company employee stood up and said that I had made up the findings. I had to point out that I was an independent investigator, but that my accuser was a company employee. I had nothing to gain by claiming that the drug was unsafe. I asked the chairman to appoint people to review our data. A few days after the meeting I received an apology from the company, but the hundreds who heard the allegations at the meeting would not be aware of the company's retraction.
At a number of other meetings at which I presented our findings, three eminent professors of cardiology, each of who was a paid consultant to Sterling-Winthrop, made public statements that they had tried to replicate our findings and failed. None of them acknowledged their affiliation to the company. Twenty years later none of those failures to replicate has been published. This tactic came to an end at a European Congress of Cardiology, in front of several hundred doctors. I pointed out that a professor who made these claims was a paid consultant to the company and that he had been making the claims for two years. I suggested that if he continued to make the claims without publishing his data people might think that he was lying. My findings were not challenged again.
At one point, my supervisor and I were asked to meet with the company and a different American professor of cardiology who is an opinion leader in the treatment of heart failure and who was a consultant to the company. The American professor told us that we were mistaken about the drug. He said that he was aware of finding by other investigators and that these entirely refuted ours. He advised us that we should not publish any more of our findings. He said that we would be found to be wrong and our reputations would be adversely affected. We went on to present 14 abstracts, and 15 publications.
One of the presentations was at the American Heart Association meeting in November 1982. I presented data, which showed that amrinone did not have the cardiac effects claimed. After my presentation, 3 professors of cardiology at separate American university hospitals told me that they had also obtained results similar to ours. They were unaware of each other's research or of our research. They informed Sterling-Winthrop. The company arranged meetings between each of them individually and the same professor of cardiology, who had told us that our findings were aberrant. He also told each of them the same thing. He persuaded two of them not to publish. The third did publish, after much soul searching because he was afraid that he would lose research contracts with Sterling-Winthrop and other pharmaceutical companies. After he published he received threats, including a threatening phone call at 2am.
The Netherlands Committee for the Evaluation of Medicines spotted our paper on the side effects of amrinone. There were major discrepancies when compared with the clinical record cards submitted by the company on our patients. We showed that the company had sent the Netherlands Committee forged clinical records for our patients with the information on adverse events deleted.
Because of this I contacted the UK Committee on Safety of Medicines and discovered that Sterling-Winthrop had also failed to notify the CSM of side effects in our patients. During discussions I discovered that contrary to statements made to us at the outset of our research, Sterling-Winthrop had not obtained a Clinical Trials Certificate for oral amrinone, though they had got a CTC for amrinone injection. This meant that the research with oral amrinone conducted by us as well as by doctors in the National Heart Hospital in London, in Newcastle-upon-Tyne and in Birmingham had been illegal.
When I raised this with the company, the senior vice president bragged that they were telling the government that if the company was prosecuted it would close down its large manufacturing plant near Newcastle upon Tyne. The company was not prosecuted for breaches of the Medicines Act.
I tried unsuccessfully to get sanctions against the company or its employees, but the Association of the British Pharmaceutical Industry, the Faculty of Pharmaceutical Medicine of the Royal College of Physicians and the General Medical Council were not interested. I spoke to editors of medical journals, including BMJ, Lancet and Nature. None disputed the facts but all were afraid to take on a multinational pharmaceutical company with unlimited financial and legal resources. One editor mentioned the loss of advertising revenue from the company.
The process of being rejected by all the official bodies that I believe should have dealt with the issues took nearly 5 years. While this was going on, in 1984, the company told a hearing of the Food and Drugs Administration in the USA that there had been over 1400 serious adverse events in 1200 patients given amrinone in trials and the company announced that they would cease trials and applications for product licences worldwide. Officially the drug was unsafe to take even on a doctor's prescription. Two years later, in 1986, I discovered that the company was still marketing amrinone in parts of Africa and Asia. In those countries it was being sold as an over the counter treatment for heart failure. I approached Oxfam, which had workers in the developing countries where this was happening. They collected evidence, which was presented at a meeting of the World Health Association in Geneva. Sterling-Winthrop was finally embarrassed into withdrawing the drug world wide in 1986.
It was my contact at Oxfam who put me in touch with James Erlichman, a Guardian reporter. He and the deputy editor, Peter Preston, were convinced by the evidence I had and so were the Guardian's lawyers. The paper covered the story on the front, back and the whole of an inside page of one issue and in follow-up stories in other issues.
I had seen how corporate greed and personal ambition had tended to distort scientific evidence. Sterling-Winthrop believed that my supervisor and I could be bribed or threatened into suppressing our data. Others, such a Drummond Rennie, Deputy Editor of the Journal of the American Medical Association, have documented this occurrence. Some professors preferred to suppress their findings rather than run the risk of losing prestige by appearing mistaken or losing lucrative contracts for future pharmaceutical research. Financial conflicts of interest caused some opinion leaders to behave dishonestly. Conflicts of interest, affected publication decisions at the New England Journal of Medicine. The institutions including government, which one might expect to help preserve research integrity, were not prepared to take on a multinational pharmaceutical company.
However these are not the only obstacles to honesty in medical research I have come across. In one case an eminent clinician, who was the president of his specialist society, and who had a large private practice doing a particular interventional procedure wished to publish a series of 400 cases. It was then the largest series in the United Kingdom. When the data was analysed it was found that his mortality rate for the procedure was unacceptably high compared with rates in other countries. If this became known it would have a disastrous impact on his private practice. So the mortality rate was falsified. However, they had already published an abstract at an obscure meeting at which amongst other things they reported the deaths in the first 254 patients. The number of deaths reported in the abstract was greater than in the 400 reported in the paper. This discrepancy became common knowledge in the specialty. I was present during a meal at which a junior doctor that was a co-author of the paper admitted that the falsification had occurred. He implied that he and other junior doctors had little option but to go along with their boss. Five other junior doctors heard the admission. I contacted the editor of the journal. It was part owned by the specialist society of which the senior author of the paper was the president. The editor knew of the rumours. He said that if I could get one of those who heard the incriminating admission to confirm it, he would act. I went back to those who had heard the admission. Now, years after those events, some have provided me with written statements confirming that they heard the admission, but at the time all said that they would not support my efforts to get the paper retracted. Some said that it would be bad for their careers. Some said that it would be bad for medicine or the specialty. One said that he thought that it was the sort of thing that any of us would do. Those 5 junior doctors went on to get consultant posts and one went on to be a president of the society himself.
My efforts to get the paper retracted were common knowledge in the specialty. I was asked to see the post-graduate dean who advised me to stop upsetting influential people. Until that point things had gone well in my career. As an undergraduate, I had obtained honours or distinction in 10 out of 11 subjects. I had been awarded an Honours degree overall, plus six undergraduate prizes and an Intercalated B.Sc. My house jobs were in my teaching hospital, and included the professorial medical job. Then I was senior house officer at the Hammersmith and in Oxford, medical registrar at Northwick Park, and cardiac registrar and senior registrar at St Thomas'. After these events, for the first time in my career, I had difficulty getting a job. I stopped counting the rejections after the 42nd. In many cases individuals with much less clinical and research experience were appointed. It was clear to me that loyalty, no matter how misplaced, was valued more highly in medicine than honesty.
I believe that obstacles to honesty in medical research generally fall into a few categories. One is personal ambition for promotion, advancement, money, kudos and power.
A second obstacle is that those who achieve success by becoming heads of departments or institutions can only maintain their position if their institution continues to succeed. Success is judged in many ways, but the most common measure of success is the balance sheet. Department heads are expected to pull in research grants. So money is another obstacle to honesty in research. This does not apply purely to pharmaceutical companies. I do not imagine that executives of Elsevier, which owns the Lancet, asks the editors much about the research published. I imagine that Elsevier asks how much was earned from drug advertising, how much was earned from sales to pharmaceutical companies of reprints of trials showing their drugs in a positive light and how the current citation rating will affect circulation profits. Of course academic institutions are the most mercenary of all.
However the greatest obstacle to honesty in medical research is the code of silence that pervades the medical profession and the research establishment. There is still considerable reluctance to shop another doctor, no matter how dishonest he is. In this setting of tolerance is there any wonder that ambitious young doctors, aware that to progress they need lots of publications with exciting findings, will embellish their findings and some will falsify the lot? Should we be surprised that a search for funding for their department and personal gain, from drug company consultancies, result in dishonest behaviour by senior academics and opinion leaders? Who will blow the whistle on them? Institutions seeking high rating in the research assessment exercise will try to suppress knowledge of dishonesty in their establishments, even to the extent of letting the guilty escape punishment. Those institutions demand success from their department heads and do not look too carefully at whether that success was achieved honourably or honestly. In this setting it is almost invariable that whistle blowers are damaged more than the guilty they expose. Academic institutions and journals do not want to be associated with dishonest research and treat harshly anybody that brings it to attention.
I have, with difficulty, persuaded a few journals to publish a small number of articles describing research misconduct.[3,13,14] Each article has been reviewed sentence by sentence by lawyers wanting evidence to support individual statements. This was because the editors of the journals were concerned that they might by sued if individuals or institutions were libelled. In a libel case it is no defence to say I am only the publisher not the author. This is in stark contrast to scientific publications. I have submitted many scientific articles for publication and many had implications for survival of patients, but no journal has ever asked me to prove that I got the results claimed. This might suggest that medical journal editors are more concerned with the reputations of academics and their institutions than the lives of patients. The simple truth is that editors are most concerned with money. Journals are never sued for publishing false results no matter how many patients died. In scientific research they can have the best of both worlds. They are absolved from blame if a study is wrong and gain an improved impact rating if the research is an important advance. A higher impact rating increases revenue from sales and advertising. Editors know that research can bring major reward to individuals and organisations, which may act as a temptation for dishonesty, but journals accept submissions on trust without checking their accuracy. Journals almost never retract work shown to be false. When they do, they make it clear that publication of the false research was entirely the fault of the authors. I would like to see whether the policies at journals changed if some were sued by patients harmed by implementation of treatments based on their publications.
There are few objective medical scientists, because they all know that success in their career is dependent on the results they obtain. Every one has a conflict of interest, everyone is human and some are venal.
Do academic institutions or journals recognise the humanity and venality of their staff? They do in some areas of activity. When paying wages, do any of these organisations leave out a bag of money and trust their staff to take the wages to which they are entitled? Of course they don't, because they realise that for some the temptation for dishonesty would be too great. The gains from dishonesty in research can be greater but institutions and journals trust researchers not to fall prey to these. We need to put in place robust checks on research. I believe that there should be random checks of raw data of work in progress and of submitted work. We know that use of performance enhancing drugs is common in competitive sports because of enforced drug checks without warning at sporting events and between events. If we did not have these checks we might mistakenly conclude that doping was not common in sport. I believe that the checks reduce the dishonesty in sport. We need a similar approach to research. The raw data could be demanded at a routine check during a visit to the research institution or when the research is submitted for publication. Failure to produce the raw data should be considered the equivalent of failing the inspection and should result in a ban on future research for a specified period and a review of previous research published. A finding that a department in an institution had falsified research should be a negative factor when assigning ratings in the research assessment exercise. In this setting justified whistle-blowing would be welcomed by institutions. Publication of dishonest research by a journal should affect its impact rating. The failure of a journal to publish a retraction of dishonest research should have a multiplied negative effect on the journals rating.
However the most important thing is that we must change the culture in medicine in which research success is viewed as the passport to success in ones career. For most clinicians only a limited experience of research is required to enable you to understand what you read in research articles and to participate in multicentre trials, organised by career medical scientists.
However there is a more fundamental problem, which is the issue of honesty. Most medical students start with high ideals. Research, which I hope is honest, has shown that as medical students go through medical school a progressively greater proportion believe that cheating in exams is acceptable. The institutions tolerate it. Three years ago Richard Smith wrote in the BMJ about a medical school that permitted a student caught cheating in the final exams to pass. I know of examples where Universities have refused to withdraw higher research medical degrees that are known to contain falsified research. I know of an academic institution in London in which senior officers know that one of their professors lied about his qualifications when he was appointed to that institution. Specifically he claimed to have a MD that he had not been awarded. The institution does not think he should be sacked and the GMC does not feel that he should appear before it. In that and other institutions there is tolerance of dishonesty at all levels. Only a sea change in opinion will produce the required improvement. I fear that it must be imposed from without because our leaders in medicine and academe lack the appetite to produce the required changes.
1. Erlichman J. Drug firm "made threats". Company tested heart drug with DHSS clearance. The Guardian 3rd November 1986; 1 and 6.
2. Benotti JR, Grossman W, Braunwald E, Davolos DD, Alousi AA. Hemodynamic assessment of amrinone. N Engl J Med 1978; 299: 1373-7.
3. Wilmshurst P. The politics of disclosure. Lancet 1997; 349: 510.
4. Relman AS. Dealing with conflicts of interest. N Engl J Med 1984; 310: 1182-3.
5. Rubin SA, Lee A, O'Connor L, Hubenette A, Tober J, Swann HJC. Thrombocytopenia and fever in a patient taking amrinone (letter). N Engl J Med 1979; 310: 1185.
6. Relman AS. The politics of disclosure. Lancet 1997; 349: 885.
7. Wilmshurst P. The politics of disclosure. Lancet 1997; 349: 1558.
8. Baim DS, McDowell AV, Cherniles J et al. Evaluation of a new bipyridine agent - milrinone - in patients with severe heart failure. N Engl J Med 1983; 309: 748-56.
9. Wilmshurst PT, Walker JM, Fry CH, et al. Inotropic and vasodilator effects of amrinone on isolated human tissue. Cardiovasc Res 1984, 18: 302-9.
10. Wilmshurst PT, Thompson DS, Juul SM, Dittrich HC, Dawson JR, Walker JM, Jenkins BS, Coltart DJ, Webb-Peploe MM. Effects of intracoronary and intravenous amrinone infusion in patients with cardiac failure and patients with near normal cardiac function. Br Heart J 1985; 53: 493-506.
11. Wilmshurst PT, Webb-Peploe MM. Side-effects of amrinone therapy. Br Heart J 1983; 49: 447-51.
12. Rennie D. Thyroid storm. JAMA 1997; 277: 1238-43.
13. Wilmshurst P. The code of silence. Lancet 1997; 349: 567-9.
14. Wilmshurst P. Institutional corruption in medicine. BMJ 2002; 325: 1232-5.
15. Smith R. Cheating at medical school. BMJ 2000; 321: 398.
16. Wilmshurst P. Doctors seem not to be punished for dishonesty in their cv. BMJ 2001; 323: 1309.