Early this month, yet another story surfaced about allegations of undisclosed conflicts of interest affecting the author of a highly publicized clinical trial. This story has now taken such an odd twist that it seems worth discussing in some detail. Let me first try to present some relevant facts, derived from published articles, mainly peer-reviewed, which do not seem to be in dispute, in chronological order.2005
- In a multi-author review article on mood disorders in the medically ill, Dr Robert G Robinson, of the University of Iowa, disclosed that he served on the speakers bureau of Forest Laboratories.(1) In an article in Stroke: Clinical Updates, he made a similar disclosure.(2)28 May, 2008
- Dr Robinson was the first author of an article published in JAMA that described a randomized controlled trial comparing placebo, problem-solving (cognitive talking) therapy, and escitalopram (Lexapro, Forest Laboratories) in the prevention of depression in patients who have had strokes.(3) Patients who received placebo had a higher rate of depression (11 major, 2 minor cases, 22.4%) than patients who received escitalopram (3, 2, 8.5%) or who received problem-solving therapy (5,2, 11.9%). At the end of the article, Dr Robinson disclosed, "over the past 5 years, Dr Robinson reports serving as a consultant to the former Hamilton Pharmaceutical Company and Avanir Pharmaceutical Company," but made no disclosure about any financial relationship with Forest. In a news article published the day before, Dr Robinson was quoted as saying, "I think every stroke patient who can tolerate an antidepressant should be given one to prevent depression," but did not advocate the use of problem solving therapy.(4)15 October, 2008
- In a letter to JAMA, Lacasse and Leo asked whether Robinson et al had done an analysis directly comparing problem-solving therapy and escitalopram, and noted that the reported incidence data for stroke in the two treatment groups suggest that the difference in treatment results "does not appear to be either clinically or statistically significant."(5) Robinson et al responded that there was no statistically significant differences between the groups.(6)5 March, 2009
- In a rapid response section in the British Medical Journal, Leo and Lacasse raised their concerns about the interpretation of the results of the 2008 trial, and also stated that "during a subsequent internet search we were surprised to learn that four years previously the lead author had been listed on the speaker's bureau for Forest. The omission, however innocent or mistaken, is disturbing; neither the JAMA article nor subsequent media accounts noted that the lead author had served on the speaker's bureau for the manufacturer of Lexapro."(7) Their citation was to our reference 2.11 March, 2009
- In a letter to JAMA, Robinson and Arndt reported that the financial disclosure for their 2008 article was incomplete, "resulting from erroneous recollection of the appropriate dates for speaking presentations sponsored by pharmaceutical companies...."(8) Dr Robinson disclosed receiving honoraria for two presentations in 2004, and serving on the Forest Laboratories speakers bureau "in 2004 and perhaps 2005." No editorial comment accompanied this letter.
So far, this seems to be a familiar story about an author who seemed to be excessively enthusiastic about a product of the company with whom he had had a financial relationship, but reluctant to disclose this relationship. But wait,20 March, 2009
- A rather extraordinary editorial was published electronically in JAMA on 20 March, 2009.(9) Let me review its main points, section by section.
After acknowledging the attention conflicts of interest now receive, and briefly describing the 2008 study and the subsequent letter by Lacasse and Leo, the editorial stated that JAMA editors had received a communication from Leo on or after 16 October, 2008 which "indicated he had evidence that Robinson had not reported in his article that he had served on the speakers bureaus for pharmaceutical companies."
The editorial then devoted several paragraphs explaining the "due diligence" JAMA editors used to investigate this "allegation." They noted the "sensitive nature of these investigations," which required them to conduct them "confidentially," and again insisted "these investigations into undisclosed conflicts of interest are time-intensive and require careful attention." Thus, from the time Leo sent his "allegations," (apparently 16 October, 2008), it took until 31 January, 2009 to get a letter from Robinson et al acknowledging his undisclosed conflicts, and until 11 March, 2009 to publish it, a total of five months.
However, the concerns with confidentiality, and the repeated emphasis on the need for unusually painstaking investigation seemed disconnected to the particular case. It is clear that it may take quite a bit of time and effort to investigate some allegations of undisclosed conflicts of interest, especially when the allegations are vague, but the alleged conflicts are severe. However, in this case, it should have taken trivial effort to find Dr Robinson's previous, published disclosures. (It took me about 5 minutes of internet searching to find the two 2005 articles.) Once (easily) discovered, the disclosures in the 2005 articles starkly contrast with the disclosures, or lack thereof in the 2008 article. Furthermore, since these disclosures were already in the public domain, there should have been no concerns with confidentiality.
Nonetheless, the editors then asserted:
While the confidential investigation of unreported conflicts of interest is under way, we consider involvement of third parties— such as Leo had done by his posting on the BMJ site and by contacting the media—to be a serious ethical breach of confidentiality that not only potentially damages our ability to complete a fair and thorough investigation (of the specific issue that Leo had brought to our attention), but also potentially damages JAMA’s reputation by the insinuation that we would fail to do so.
This makes no sense. There was no need for confidentiality, because the allegations were not based on whistle-blowing or confidential information. They were based on published articles, articles which clearly disclosed Dr Robinson's financial ties to Forest Laboratories. Anyone who bothered to do a simple Google search could have found that Dr Robinson's disclosure in his 2008 JAMA article did not agree with his disclosure in several 2005 articles. Lacasse and Leo used no confidential information whatsoever in making their allegations. Actually, they were not making "allegations," but simply pointing out that an author had published statements which were inconsistent. Since Lacasse and Leo were simply pointing discrepancies among published articles, how could they have breached "confidentiality," ethically or not?
Second, there was no need for a complex or prolonged investigation. The discrepancies were apparent as soon as one viewed the published articles side by side.
The third point was a non sequitur, and a classic example of blaming the messenger for the message. JAMA had published an article. Other journals published other articles. Publishing these articles put them in the public domain. If comparison of these articles in the public domain were to reflect badly on on of their publishers, it would not be the fault of making the comparison.
Regardless of the illogical nature of their concerns with non-existent confidentiality, assertions that it takes a long, complex investigation to demonstrate the obvious, and their aspersions on the messenger for the message, the JAMA editors then revealed that:
A telephone conversation intended to inform Leo that his actions were inappropriate transformed into an argumentative discussion, as Leo continued to refuse to acknowledge any problems with his actions, even after he was informed that the investigation was completed and was advised to read the upcoming March 11 issue of JAMA (where the letter of explanation and apology from Robinson and the formal correction were in press).
Furthermore, the editors proposed punishing Leo, first by effectively banning him from publishing in the pages of JAMA,
Leo also was informed that, if his actions represented his apparent lack of confidence in and regard for JAMA, he certainly should not plan to submit future
manuscripts or letters for publication.
Then, by complaining to his supervisor
However, since Leo apparently did not appreciate the serious implications of his actions, despite our attempts to explain, we felt an obligation to notify the dean of his institution about our concerns of how Leo’s actions were potentially damaging to JAMA’s reputation. We sought the dean’s assistance in resolving this issue involving a member of the faculty of his institution, to assure there would be no need to publicly identify that faculty member. No dean wants his or her institution implicated in a publication reflecting improper behavior by a faculty member.
Finally, they seemed bitterly offended that the British Medical Journal would publish anything that disagreed with anything printed in JAMA:
In addition, we were dismayed that BMJ would post the article by Leo with the allegations against Robinson and the negative insinuations about JAMA, without at least contacting JAMA to verify the veracity of the report.
Again, rather than raising unsubstantiated allegations, Leo and Lacasse's letter to the BMJ simply showed the discrepancy between the disclosures in Robinson and colleagues' 2008 JAMA article and those in one of his 2005 articles. Both articles are publicly available. What useful information about the letter above and beyond that provided by viewing the discrepant articles could the JAMA editors have provided?
While the JAMA editors castigated Leo for unethical behavior, and sought to punish him for it, they said nothing negative about Robinson. Yet it was Dr Robinson who was supposed to disclose his conflicts to JAMA, but who managed to forget conflicts that he had disclosed before. Why would would it be more unethical to point out publicly available evidence about failure to disclose conflicts of interest than to conceal these conflicts in the first place?
This story is saddening. JAMA has published many important and useful articles on conflicts of interest, and other matters relevant to Health Care Renewal. We have probably favorably cited more editorials by JAMA on these topics than those in any other journal. Yet now the journal's leadership seems to have somehow lost their way. Instead of trying to constructively respond to criticism, they now seem intent on punishing the critics. I hope they find their compass soon, before an important medical institution really is irreparably damaged.
Note: another detailed analysis of this case is here
in the Hooked: Ethics, Medicine and Pharma blog. Here
is Professor Leo's response to the JAMA ediorial.
ADDENDUM (24 March, 2009) - see also the extensive analysis
of this case on the Respectful Insolence blog, and the comments here
on the Effect Measure blog, here
on Gooznews, and here
1. Evans DL, Charney DS, Lewis L et al. Mood disorders in the medically ill: scientific review and recommendations. Biol Psychiatry 2005; 58: 175-189. [Link here
2. Robinson RG, Zorowitz RD. Pseudobulbar affect and stroke. Stroke: Clinical Updates 2005; XV (Jan-Feb). [Link here
3. Robinson RG, Jorge RE, Moser DJ et al. Escitalopram and problem-solving therapy for prevention of poststroke depression: a randomized controlled trial. JAMA 2008; 299: 2391-2400. [Link here
4. Elias M. Study: antidpressants help stroke victims. USA Today, May 27, 2008.
5. Lacasse J, Leo J. Escitalopram, problem-solving therapy, and poststroke depression. JAMA 2008; 300: 1757-1758.
6. Robinson RG, Jorge RE, Arndt S. Escitalopram, problem-solving therapy, and poststroke depression. JAMA 2008; 300: 1758-1759.
7. Leo J, Lacasse J. Clinical trials of therapy versus medication: even in a tie, medication wins. Brit Med J, 2009: 338: b464. [Link here
8. Robinson RG, Arnd S. Incomplete financial disclosure in a study of escitalopram and problem-solving therapy for prevention of poststroke depression. JAMA 2009; 301: 1023-1024. [Link here
9. DeAngelis CD, Fontanarosa PB. Conflicts over conflicts of interest. JAMA 2009; 301: [Link here