Guest blogger: Ida Sim, Associate Professor of Medicine and Director, Center for Clinical and Translational Informatics, University of California, San Francisco, USA.
I’ve just started reading One Thousand and One Nights with my young son. He likes the Seven Voyages of Sinbad the Sailor most, but I’d like to talk about Ali Baba and the Forty Thieves. Ali Baba, of course, is the poor Arabian woodcutter who overhears the Forty Thieves as they intone “Open Sesame!” to gain access to their cave of ill-gotten treasure. Using this privileged information, Ali Baba avails himself of treasure, and soon after, I’m hesitating at the part about hot oil being poured on the thieves in their jars.
So is open access the modern-day equivalent of open sesame, prying open the rock that keeps scientific treasures in darkness? No, not if open access means only open access to articles about research rather than the real treasure, the research data itself.
Let’s look at one research area, clinical trials, where our ultimate quest should resoundingly be for open data, not just open access to trial reports. In a dismaying parade of front-page exposes – from Paxil to Vioxx to Zyprexa – we have learned how several Big Pharma companies have withheld clinical trial results to protect their bottom lines. With each scandal, patients are injured, doctors are misled, or company stock prices drop, and the march of clinical science is a few steps slower than it should be.
In response, the World Health Organization is working to ensure that all clinical trials worldwide are publicly registered before the first patient is enrolled. (I led this International Clinical Trial Registry Platform from its founding in 2005 until last month). Such universal trial registration would provide the public with open access to a standard set of 20 data items that describe a trial’s basic design. Critically, trial registration allows trial results to be tracked down later regardless of whether the results are “good” or “bad.”
This is all very well — and very, very good — as a first step. The second, no less challenging step is to link trial registration with mandatory reporting of the basic results of each registered trial, a Minimum Trial Report. This would provide a systematic bulwark against trial investigators or sponsors cherry picking results for career or commercial gain, and would probably prevent another “vioxx” from happening again. And if the whole motivation for trial registration and reporting is to meet the scientific, ethical, and moral duties of experimenting on humans (see Declaration of Helsinki and the Nuremburg Code), then surely trial reports should be open access for the public good. Keeping trial reports behind subscription or other access barriers just provides other ways for skewing the public record of clinical trial research.
But what about the data behind the reports? Don’t all the arguments for open access to the trial report also apply to the actual data? Add to this that having trial results data in computable form (along with the study designs that gave rise to them) would enable much more powerful data mining and decision support, which should speed improvements to human health and increase society’s return from the billions spent on clinical trials every year.
So we shouldn’t just aim for open access of trial reports, and then ask beseechingly for scraps of open data as icing on the cake. No, we should unabashedly set open clinical trial data as the long-term goal and require compelling reasons for backing off that goal. Granted, there are many, many complicated issues to resolve (e.g., here and here), but there do exist examples of clinical trial data sharing. The treasure is in the data. Open sesame.