According to a new study published by PLoS Medicine (a new online journal that is likely to end up in my regular rotation), less than half of the registered clinical trials in America over the past decade went unpublished. The trials sponsored by industry were the least likely ever to end up on the shelves.
This new research trial was elegant in its simplicity. The authors surveyed the clinicaltrials.gov website for all registered clinical trials between 1999 and 2007. Of these, they took a randomized subset of 10% of the 7515 total research trials. After eliminating incomplete trials and CAM studies (!), they were left with 677 data points.
Of these 677 trials, only 311 (or 46%) had been published. Of the studies sponsored by industry, only 40% had been published.
Of course, it would be useful to know why so many of these results go unpublished, and how all of these missing data would impact current medical practice. The authors provided a few clues.
One more benign explanation is that many of the trials were not posted with an end date, and included other clues that they were meant more for preliminary information gathering than for publication. Presumably, these trials would be followed up with more formal research meant to educate the physician community.
The fact that 70% of the studies reviewed presented positive results does suggest that publication bias may be at work, as well. As anyone who has ever considered a career in medical research would probably attest, publishing positive results is pretty much imperative for career building. And, of course, the pharmaceutical and device industries may have financial incentive to not share inconclusive data.
It is very difficult to know exactly what to do about these findings, even though they are consistent with a number of previously published studies on the subject. But it seems likely that there are treatments used routinely in clinical practice that might not be indicated if clinicians had access to a more full research picture. In the worst case scenario, there are research trials tucked away in the vaults showing that routinely used treatments are potentially dangerous. If this sounds like conspiracy theory, note that Merck in-house documents demonstrate that the company was aware of potential cardiovascular complications from the medication years before this information was shared with physicians.
As an aside, it was not clear why the researchers chose to exclude CAM studies from their analysis. It was interesting to see that less than 1% of registered clinical trials in this country looked at CAM interventions. I’d be very interested to see if these trials were published at a rate similar to the conventional medicine studies.