Cross-posted at PLoS Blog's "Speaking Medicine
A recent case of publication bias fuels the debate over if results of clinical studies sponsored by pharmaceutical companies must be published in European countries. I have become interested in this issue since researching for an editorial in Value in Health about a similar problem in studies assessing the health-related quality of life.
Researchers at IQWIG, Germany's NICE equivalent, did not cave in when they could not get their hands on the unpublished results of clinical trials of the selective norepinephrine reuptake inhibitor reboxetine (an anti-depressant, branded as Edronax(R), Norebox(R), Davedax(R), or Vestra(R)) for major unipolar depression. Instead, they asked the manufacturer Pfizer time and again for the unpublished data (the drug was developed in the 1990s by Pharmacia which merged with Pfizer in 2003).
The drug was approved by European drug regulatory agency EMEA in 1997, but only conditionally approved by the U.S. equivalent FDA in 1999. In 2001, the FDA declined a final approval on the grounds of insufficient evidence for its efficacy.
A new systematic review with meta-analysis published by BMJ displays the striking difference between the published and the unpublished trials: the Forrest plot in figure 6 shows that taking reboxetine did not result in a significantly better response or remission in the unpublished or the total combined results (they were significantly different in the published ones). Withdrawal was also more likely than placebo only in the combined results.
In the spirit of comparative effectiveness research, the performance of reboxetine was compared to SSRIs, a major class of antidepressants. Response, remission, and withdrawal all were worse. Results were not compared to tricyclics or other drug classes. However, it should be noted that even within the class of SSRIs, they are differences in efficacy and acceptance. To this day, it is still not clear how psychiatrist can best find the most suitable drug for an individual. Interpersonal differences, pharmacogenetics and proteomics might play a role in different responses. In any event, this is a beautiful execution of a meta-analysis and highlights the value of this study type well.
It is shocking that this useless and possibly harmful medication was on the German (as well as other European markets) for well over a decade while it was withdrawn in the U.S. about nine years ago. EMEA seems not to have made the best decision back in 1997. Re-evaluations based on efficacy (other than safety) concerns remain difficult.
IQWIG said in a press release, that Pfizer only provided the data under "massive public pressure". Along with their original research, they describe in BMJ the back and forth with Pfizer in a "Tale of Hide and Seek". The press release goes a bit further than the companion article. It mentions a draft for a German "Law on the Reorganization of the Pharmaceutical Market". This bill would stipulate mandatory publication of studies by the pharmaceutical companies. However, you have to go to the actual text to understand that posting raw data on the internet would suffice. Why not a European requirement? In the US, there is an analogous rule; all clinical trial data of studies after September 27, 2007 must be published on ClinicalTrials.gov. The value of such transparency for patients, payors, and society at large could not be more obvious.