When thousands of psychiatrists attend their field’s largest annual meeting each year, the presentations they hear about research into drug treatments report overwhelmingly on positive results.
That’s the finding of a new study published in the Journal of Clinical Psychopharmacology by two young psychiatrists from the University of Michigan and Yale University, who analyzed the presentations given at two recent meetings of the American Psychiatric Association.
Of 278 studies presented at the 2009 and 2010 APA meetings that compared at least two medicines against each other for any psychiatric illness, they found that 195 had been supported by industry, and 83 funded by other means. The authors then evaluated the studies without knowing which kind of support each one had.
Of the industry-supported studies, 97.4 percent reported results that were positive toward the medicine that the study was designed to test, and 2.6 percent reported mixed results. No industry-sponsored studies with negative results were found.
In contrast, when industry was not the source of funding, 68.7 percent of the presentations were positive, and 24.1 percent contained mixed results, while 7.2 percent contained negative results.
This ‘presentation bias’, in which mostly good news about medicines gets reported at meetings, echoes the ‘publication bias’ that has been seen in research published in major journals, says Srijan Sen, M.D., Ph.D., an assistant professor of psychiatry at the U-M Medical School who led the study while in his residency at Yale.
While attending the APA’s massive annual meeting, he noted the large industry presence and emphasis on research involving medicines that were still “on patent” and being actively marketed to both psychiatrists attending the conference. He teamed with Yale psychiatry resident Maya Prabhu, M.D., M.Sc., now a consulting forensic psychiatrist at Yale, to do a formal review.
“This analysis suggests that the APA meeting might be being used as an opportunity to make drugs seem more effective than they are,” he says.
Meanwhile, research on “talk therapy” treatments such as cognitive behavioral therapy – which doesn’t have industry backing – gets less attention, even though there is growing evidence that such non-drug therapies can have as much effect as medicines in illnesses such as depression.
Since the APA meeting is a major source of continuing medical education credit for psychiatrists, and a hub for psychiatry residents just starting out in the field, the bias has the potential to affect their clinical care when they return home.
Sen notes that the research journals and funding agencies have tried to confront research bias in journal articles by requiring pharmaceutical companies to register the clinical trials they are conducting and include the registration number when publishing the study. This allows interested individuals to see which trials are being reported in the medical literature, and whether the trial results are interpreted according to the original study design.
For instance, if a drug trial is designed to test long-term results from a particular treatment, but a paper is published showing positive results over a much shorter term, that could be an indication of bias. Research meetings could require a similar practice, Sen says. And, APA could be more selective in accepting poster presentation submissions.
If nothing else, Sen says, attendees at the APA’s meeting – and perhaps other large gatherings of psychiatrists – should be aware of the positive bias of the research they will hear about at the meeting.
And, non-industry funding for research – especially research to compare older “off patent” drugs that aren’t being marketed by industry – is also important, he says. The federal government has funded large comparison studies in depression, schizophrenia and bipolar disorder, he notes – and these mostly showed that the generic medications that have lost their patent protection are just as effective as newer, patent-protected ones. But even still, psychiatrists prescribe the generic ones far less often than the brand-name patent-protected ones.
