Increasing the transparency of clinical drug trials reduces number of positive results | Ars Technica UK
"There are many good reasons to think that requiring transparency in clinical trials will improve the quality of medical research. A recent paper in PLOS One has added new evidence to the pile, finding that requiring drug makers to file a full description of cardiovascular drug trials before starting them coincided with a dramatic drop in the number of positive findings. Starting in 2000, all large clinical trials supported by the National Heart, Lung and Blood Institute (NHLBI) in the US were required to pre-register with clinicaltrials.gov. This filing involved stating ahead of time what methods they planned to use, the health outcomes they would measure, and what they were expecting to find. This created a perfect natural experiment. Before 2000, it was easy to generate a positive finding. Experimenters could measure a number of different results, look through the data, and report the most successful findings, ignoring the rest. After 2000, they had to report back on the exact question they had initially set out to answer. When researchers report only on positive findings, it has a huge and important impact. Obviously, hiding actively negative findings—where a drug is actively harmful—is a huge concern. But keeping quiet about null results, where nothing really happened at all, is also damaging. It means that the positive results that get reported might just be false positives: fluke positive findings in amongst a mass of null results. Treatments that aren’t as effective as they’re cracked up to be, but that make it to market because boring results have been ignored, have a huge opportunity cost. They eat up funding that could be spent researching and buying more effective drugs, and they leave people sicker than they might otherwise have been. For these reasons, it’s important to shine just as much of a spotlight on null results as on positive ones ..."