FDA To Mine Big Databases For Safety Problems
3 CommentsBy Ed Silverman // May 23rd, 2008 // 8:25 am
The effort, called Sentinel Initiative, will be the first time the FDA will have an opportunity to monitor almost immediately how drugs are affecting the public. To do so, the agency will mine databases of more than 20 million patients who receive their drugs through Medicare. The idea, of course, is to catch side effects that might otherwise go undetected for months or years.
“This initiative will tremendously increase the FDA’s capacity to monitor the use of medical products on the market,” Mike Leavitt, the HHS Secretary, says in a statement. “We are moving from reactive dependence on voluntary reporting of safety concerns - to proactive surveillance of medical products on the market.” Last month, by the way, the FDA and Wellpoint announced plans to do the same thing.
The concept is in stark contrast to the current system, which relies on sporadic reports from docs, patients and drug and device makers, whose judgements are inconsistent. Moreover, the FDA has readily acknowledged that only an undetermined fraction of side effects are ever reported and the system isn’t set up to detect specific side effect problems with a drug.
However, there may be issued with this latest approach as well. As The New York Times notes, Medicare collects data only claims data for billing purposes, which is generally regarded as less accurate than patient health records. And the project may take awhile to coalesce, because Medicare drug benefits are available through so many different private plans.
Another problem is that sometimes patients suffer problems after receiving drugs because they are sick, not because the drug is to blame, the Times writes. Also, Medicare beneficiaries use an average of 28 scrips a year, compared with an average of 13 among all Americans. So sorting out which medicine caused any single problem can be difficult in the elderly, the paper points out.
david
FDA insists the spontaneous system accounts for 1% of all episodes of serious adverse events. That estimate is from a study more than two decades old. The problems with the spontaneous system (GIGO) are so well known to pharmacoepidemiologists and other drug safety specialists that few are willing to invest the energy needed to examine methods to use the system. There has been a a collaboration between industry and FDA to bring data mining to the spontaneous system, but it is hardly an unqualified success, and not everyone agrees with its use. Hence this new approach (which has been used by pharmacoepidemiologists for more than two decades). Hopefully, we’ll finally start to get good data upon which to make policy.
Laurie
Using Medicare data, which is better than nothing, would only track those on Medicare….so what about the “younger” population?
CMC guy
“Another problem is that sometimes patients suffer problems after receiving drugs because they are sick, not because the drug is to blame, the Times writes.”
This statement may be news to some of the posters here who blame pharma for development of “bad” drugs. Even with well designed and conducted clinical studies it is hard to sort out causal relationships with complex or individual variations. At the same time does not mean companies should weaken vigilance or be dismissive in determining correlations.