March 20, 2011
FOLLOWING A number of high-profile drug safety scandals in recent years — such as Vioxx and Avandia — the pharmaceutical industry and the Food and Drug Administration are spending a great deal of energy trying to uncover, as early as possible, rare side effects that make a drug too risky. Sometimes, the agency asks companies to go back to the drawing board to redesign their studies or expand clinical trials to answer specific concerns about safety. But in some cases, the expanded data already exists in the FDA bureaucracy.
There is a vast amount of clinical information sitting in the agency’s archives that could be analyzed and mined to flag worrisome side effects. Multiple years’ worth of clinical trial data from studies of diabetes, high blood pressure, heart disease, Alzheimer’s disease, multiple sclerosis, cancer, and many other conditions could be used to answer important questions about drug safety and efficacy. But most of that data is unsearchable, existing in the form of paper submissions or unwieldy electronic files that can’t be downloaded and analyzed.
As recently as 2007, only 37 percent of drug applications were submitted to the FDA in electronic format. The figure has improved — it was around 70 percent last year — but that means 30 percent of data still comes in as binders full of paper. In the era of the iPad, this is unfathomable.
Even among electronic submissions, a lot of the data arrives in forms that are not easily searchable because companies submit the information in custom formats. Patients’ gender, for example, might be categorized as “male’’ or “female,’’ “M’’ or “F,’’ or “1’’ or “0,’’ depending on the study. The result, as described by the FDA itself, is an “informatics tower of Babel,” in which an extremely valuable resource for public health — millions of data points from a multitude of clinical studies — is virtually impossible to use.