Dr. Abbott argues that the new state and federal health information exchanges (HIEs) created by the HITECH Act and the Affordable Care Act will provide a rich source of post-market data that can be mined for safety information. HIEs raise a host of important privacy and information technology questions, but let's bracket these issues and assume that the growth of HIEs leads to large, anonymized datasets of health information that is accessible by diverse researchers. How can these researchers be incentivized to mine this data?
Some researchers will already have sufficient incentives to data mine for pharmacovigilance (i.e., drug safety): the FDA and NIH will likely use HIE data to improve existing research, pharmaceutical companies have incentives to monitor the safety of their own drugs (to avoid liability for failing to act after they should have been aware of a problem) or to discredit the safety of rivals' drugs, insurance companies will mine HIE data to avoid paying for ineffective therapies, and academics receive reputational benefits from such research. But Dr. Abbott argues that these existing incentives are inadequate, as illustrated by cases such as Vioxx (for example, some insurers had restricted access to Vioxx before its withdrawal, "but this information was not effectively disseminated").
If additional incentives are needed, which is the most effective of the various incentive policy tools for pharmacovigilance? The problem (at least as Dr. Abbott presents it) seems to satisfy all the conditions under which Daniel Hemel and I argue that prizes are optimal: the government can set a clear goal (production of information sufficient to show that an approved drug is unsafe) and will be better than the market at setting an appropriate reward (since the social benefit of this information is much greater than its market value), but depending on the complexity of mining, the government may be at a disadvantage in identifying the most promising projects. And if Dr. Abbott is correct that data mining likely has "modest costs," the incentive effect will not be dulled significantly by the ex post nature of a prize reward.
The solution Dr. Abbott proposes is indeed much like a prize: he suggests a bounty proceeding modeled after the False Claims Act qui tam provision, which would provide a reward to data miners who succeed in having a drug withdrawn from market through an adversarial proceeding. He reviews the various options for setting the reward size and determining who pays the reward, and suggests the following structure:
In the event a petitioner submission results in the FDA removing a product from the market or amending labeling, petitioners could be paid an award from the federal government based on the government’s estimated cost savings over a time period to be determined. If the product’s sponsor were found to have underlying negligence, the sponsor could be responsible for paying the award, at a level based on a percentage of a drug’s revenue. If negligence-plus were established, the product sponsor could be responsible for treble damages, half paid to the petitioner and half to the government.The article concludes by reviewing how such a proceeding might have played out in the Vioxx case. The details of implementation are complicated, and Dr. Abbott acknowledges that he does not have all the solutions—including for the significant challenge of "finding the political will" and overcoming likely opposition from the pharmaceutical industry. And, of course, the proposal depends on having large, accessible HIE datasets to mine in the first place, which requires overcoming significant privacy concerns. But if such data becomes available to independent researchers, and if existing incentives still appear insufficient, it may be worth giving more consideration to this creative proposal.