November 15, 2005
What the FDA Needs for Xmas
One of the major points University of Chicago economist Steven D. Levitt makes in “Freakonomics” is that information is power. I was reminded of that truism yesterday as I sat listening to presentations at the Food and Drug Administration’s hearing on the Prescription Drug User Fee Act. The law expires in 2007 and the FDA wanted public input as it begins the process of seeking renewal from Congress.
That law, enacted in 1992 after extensive pharmaceutical industry lobbying, was designed to reduce the time it takes the FDA to approve new drugs. And by that sole metric, it has been successful. The amount of time from a company’s new drug application – when it’s completed clinical trials and wants the FDA to approve marketing a drug – to the final FDA decision has been reduced from three years in the mid-1980s to slightly over a year today.
Over the same period, however, the amount of time that the average new drug remains in clinical development, which is measured from the time a company files an investigative new drug application (which takes place at the beginning of human safety studies) to its new drug application (after all the efficacy trials have been completed) has risen sharply, from about six years to over eight years. In other words, the total time for the average new drug to move through the pipeline hasn’t budged in a quarter century.
During a break in the hearing, I asked Janet Woodcock, the deputy director for operations at the FDA, for the agency’s analysis of the slowdown in the clinical trial phase, which is largely under industry control. Since agency staff interact with company officials throughout the process, they would know why one drug got dropped from development or why another had to go back to do a second safety trial before moving onto the next phase of testing for efficacy. Properly captured and analyzed, this data would be crucial information for companies hoping to understand where they had gone wrong. It also inform other companies about unproductive pathways, help them avoid meaningless clinical trials (and the heartbreak that causes the test subjects) and reduce overall drug development costs.
Woodcock told me neither she nor anybody else at the agency had ever studied the reasons why most drugs in the FDA system fail. They have no data on which experimental classes of medicines may have caused severe safety problems in early human testing. There is no databank that aggregates all clinical trial data for particular classes of drugs or that can be tapped to compare different classes of drugs aimed at the same condition.
I asked her why. She began describing the FDA’s computer system, which as near as I can tell from her description is still in the pre-Windows mainframe era.
Bringing the FDA’s information systems into the 21st century would do a lot for safety, too. Right now, the agency’s self-reporting MedWatch program, where physicians are supposed to report adverse events among their patients on drugs, has no capacity to provide those physicians with feedback about what other doctors are experiencing with those drugs. The result is that not many physicians use the system and safety problems remain locked away in physicians’ files.
During a presentation at the outset of the hearing, deputy commissioner Scott Gottlieb flashed a picture on the overhead screen of a white-coated technician screwing in vacuum tubes in the original Univac computer. He said it was a snapshot of the FDA’s Information Technology department. One item that should be high on Congress’ agenda when they get around to reauthorizing PDUFA is buying the agency a new computer system.