How well does a prescription drug work? It can be hard for even doctors to know. Pharmaceutical companies frequently withhold the results of negative or inconclusive trials. Without a full accounting, a physician who wants to counsel a patient about whether a drug works better than a sugar pill is frequently at a loss. Drug companies share only airbrushed versions of data on safety and usefulness.

As a consequence, regulators can approve drugs that have hidden health hazards. Clinical trials of GlaxoSmithKline's diabetes drug Avandia (rosiglitazone) and Merck's anti-inflammatory Vioxx (rofecoxib) revealed an elevated cardiac risk from the drugs, but relevant findings were held back from regulators or never published.* Far more drugs have gone to market with critical safety data kept secret. These scandals have tarnished the reputation of the pharmaceutical industry.

Such revelations have made the industry come to realize that greater transparency is inevitable. “The question is not whether but how these data should be broadly shared,” noted an article in the New England Journal of Medicine last fall. The article had a co-author from the leading U.S. drug industry trade group.

Yet the challenge of how to share the data is not simple. A few manufacturers, including GlaxoSmithKline, Roche and Pfizer, have set up Web portals to open their files to outsiders with few, if any, restraints on access. Other companies worry that opening their doors too wide will compromise trade secrets, as well as the confidentiality of patient records.

Europe has recently taken the lead in adopting measures to ensure openness of data collected throughout drug trials. In early April the European Parliament voted to require that clinical trial results be published within a year of completion, whether or not the data are positive—a regulation that mirrors a similar effort being developed by the European Medicines Agency, an organization roughly equivalent to the U.S. Food and Drug Administration. (Physician and transparency advocate Ben Goldacre has pointed out that the vote is only a first step because it does not make public the data for already approved drugs.)

The FDA, meanwhile, has followed these proceedings intently as it contemplates requiring new levels of openness from pharmaceutical companies. As the agency deliberates, it should consider that the companies' poor track record is mirrored by its own. Since 2007 the FDA has required drugmakers to post some trial results in the government registry ( within a year of a drug's approval, but the agency has failed to enforce this edict. A 2012 study showed that fewer than one in four approved drugs had results that were filed in time.

Fortunately, there are other ways to ensure that drug data get shared. The FDA should carefully consider a collaboration with the kind of independent institution that is already up and running at Yale University. In 2011 Yale's Open Data Access (YODA) Project reached an agreement with medical device maker Medtronic to act as an intermediary for releasing all data on clinical trials of a controversial bone-growth protein whose safety had been questioned. In an effort to defend its reputation, the company gave up any right to decide who would get the information. YODA then commissioned two systematic reviews of the protein, which conveyed mixed results that were then published. Following Medtronic's example, Johnson & Johnson pledged in January to make all its clinical trial data available for perusal by outsiders through YODA. Such early signs of successes might serve as the basis for devising a national system that replicates a YODA-like model for all U.S. drug trials, perhaps backed up by FDA-enforced penalties for companies that refuse to comply.

The benefits of this approach will assist not just independent evaluators trying to determine whether a pharmaceutical actually works. It will help drug companies do their job better. Large open data sets will improve the design of future clinical trials. Such an approach will also let pharmaceutical makers avoid committing tens of millions of dollars for late-stage studies that others have already found to be money sinks.

The most important reason for moving ahead has nothing to do with costs. An open data system—perhaps one like Yale's, backed with some regulatory clout—is the only way that physicians can weigh available evidence to make informed, timely decisions about what to tell their patients.

*Correction (6/9/14): This sentence was edited after posting. The original erroneously stated that celecoxib is the generic name for Vioxx.