Here is a thought experiment: Imagine what would happen if your local weatherman were financially liable for his predictions — e.g., he promises a bright and sunny day without a cloud in the sky, you leave the top down on your convertible, it rains mackerel and trout, and he has to pay for the damage to your upholstery.
What do you think would happen to the quality of meteorology in that situation?
We might end up with better weather forecasting — incentives matter, and under the current system there is only a marginal penalty for providing poor weather-prediction services: People start consulting other weathermen, and your audience share suffers. That’s not much of a threat compared to putting new upholstery in somebody’s Targa (or many somebodies’ Targas).
But we might end up with worse forecasting instead: The desire to avoid liability might drive some meteorologists out of the market or cause them to offer only vague forecasts. That’s the opposite of what you want if you are looking to improve forecasting, because predicting the weather with a high degree of accuracy is complex (it may even be impossible), and your best bet for understanding such recondite phenomena is having more brains examining the data and the models rather than fewer — the desirable thing is a large aggregation of focused expertise.
Incentives matter, brainpower matters, and incentives and brainpower interact in unpredictable ways. It follows that the worst way to organize weather forecasting — the way to ensure inferior forecasts — would be to have all the forecasting done by a single agency with no real incentive other than institutional self-interest. There is a popular myth that this is in fact how forecasting works: that there is one master forecast produced by a government agency, and all that changes is the color of the weatherman’s blazer. But that isn’t true. That would be nuts.
And yet that’s how we handle pharmaceuticals and medical devices.
A bill that has been introduced in Congress, the 21st Century Cures Act, would liberalize some of the Food and Drug Administration’s processes for allowing therapies to come onto the market. Naturally, this has the worrywarts worried. “Don’t weaken the FDA’s drug approval process,” pleads the New York Times, “placing patients at unnecessary risk of injury or death.” The 21st Century Cures Act is far short of the needed medicine for the FDA — which is abolition and replacement — but it does contain some desirable provisions: for instance, allowing for approval decisions to incorporate clinical experience where data from large-scale trials are not available, giving third-party evaluators a larger role in the approval of medical devices, etc. It would potentially bring more brainpower and better incentives to the evaluation of new therapies. As such, it would represent one of the most important reforms at the FDA since the creation of the accelerated-approval program in 1992, an innovation spurred by the AIDS crisis, which saw thousands die while federal regulators dawdled over life-saving drugs.
For those of you who do not recall the controversy over AIDS drugs, it was an interesting libertarian moment, albeit a strange one.
For those of you who do not recall the controversy over AIDS drugs, it was an interesting libertarian moment, albeit a strange one: In the late 1980s, radical gay groups and AIDS activists were demanding federal regulatory reform, and Vice President George Bush, who was Ronald Reagan’s point man on the President’s Task Force for Regulatory Relief, was working diligently to give them what they wanted. A host of conservative institutions — the Heritage Foundation, the Wall Street Journal editorial page, National Review — encouraged broad reform of the FDA’s approval process, which put the hard-left elements of organized homosexuality in an awkward position. Activists from ACT-UP (AIDS Coalition to Unleash Power) maneuvered to keep their project — FDA liberalization when it suited their interests — separate from the conservative project — FDA liberalization when it suited everybody else’s interests, too. At issue was loosening up the 1962 Kefauver amendment to the Federal Food, Drug, and Cosmetic Act, which added an efficacy criterion to the FDA’s existing consumer-safety criteria, and strengthened those, too.
The Kefauver amendment was adopted in response to the Thalidomide scandal, which was odd: The heart-rending birth defects produced by Thalidomide happened in other countries, because the FDA had disallowed the use of Thalidomide in the United States, the headline there being that a federal agency actually did its job for once. Twenty-odd years later, the agency was denying experimental drugs to men and women dying horrible deaths from HIV on the grounds that these late-stage AIDS patients might get sick. People with Lou Gehrig’s disease have been pleading with the FDA to expedite the approval of possible new treatments, and the result has been an unedifying political battle.
The regulatory mind is, in its resting state, a binary mind, and though the FDA recognizes some finer distinctions, it operates in a world in which “approved” and “prohibited” work like day and night, with the occasional twilight of accelerated action. The agency shone in the Thalidomide matter, but it does not have an especially distinguished record as a regulator, having approved any number of drugs that later turned out to be dangerous and had to be withdrawn. Astemizole may not be as shocking to the eye in its effects as Thalidomide, but fatal cardiac arrhythmia is a kind of excitement that you do not want. Those who fear that dangerous drugs will make it onto the market if the FDA is reformed should appreciate that dangerous drugs make it onto the market now.
And with the FDA, it pays to keep Frédéric Bastiat’s “seen and unseen” in mind: The FDA causes harm by approving dangerous drugs, but it almost certainly causes dramatically more harm by failing to approve useful therapies, or by simply delaying them and making them unnecessarily expensive. Like the liability-averse weathermen in our opening thought experiment, the FDA is strongly biased toward inaction, because the costs of action are potentially very high — for the agency and for its personnel. There is rarely a scandal about something a regulator didn’t approve, though there should be. And as with any political agency, the safety that most concerns the FDA is its own.
Like safety, the criterion of effectiveness has proven unamenable to an either/or approach: Some treatments are very effective for a small share of patients but otherwise generally ineffective. As we learn more about the human body and especially about the genome, we have more of an ability to predict which treatments are likely to be effective for which patients — which is to say, the question isn’t just whether the drug is effective, but for whom it is effective. The general question of effectiveness is no more straightforward: In its accelerated-approval process, the FDA relies on “surrogate endpoints,” clinical outcomes that are not beneficial in themselves but are suggestive of benefit — e.g., a cancer therapy might be thought effective if it reduces the size of tumors, and the FDA might act on that result before there is clinical data about whether the treatment in fact extends life. Is a drug truly effective if all it does is allow you to die with smaller tumors?
The general posture of the FDA suggests that the agency thinks that its job is to block drugs and medical devices; its industry critics believe that its job should be to approve drugs and medical devices. What it should really do is try to describe drugs and medical devices.
The FDA should act more like the National Weather Service, and function as a clearinghouse for research, clinical data, and scholarship.
Which is to say, the FDA should act more like the National Weather Service, and function as a clearinghouse for research, clinical data, and scholarship. It should be more in the rating and ranking business, and less in the prohibition/stamp-of-approval business. Just as there are many meteorologists producing different forecasts from the same baseline data — a process that leads to more accurate forecasting — we should encourage the emergence of a competitive market of agencies that certify the safety and efficacy of drugs and medical devices, with certifications that are more nuanced than “approved” and “prohibited.” The reality of safety and efficacy is more complex than that.
Understanding exactly how we would want to structure the incentives of such agencies is the purpose of the opening thought experiment. We have seen some bad examples of agencies of this sort, notably the NRSROs — the nationally recognized statistical rating organizations, Moody’s and Standard & Poor’s and the other acts in the pinstriped clown show that failed to cover themselves in glory in the matter of all those securitized subprime mortgages a few years back. That is not how you want third-party certification to work: The government creates a cartel (you’re not an NRSRO unless Uncle Stupid says you are) and invests it with extraordinary power (those AAA securities you need on your insurance company’s balance sheet aren’t AAA unless an NRSRO says they are), while also making it partially subservient to the institutions it is supposed to be scrutinizing. That’s not entirely unlike what we have at the FDA now, with the important distinction that the FDA is a regulatory monopolist instead of a cartel member.
What would be preferable is a regulatory structure in which the presumption is that doctors, patients, and researchers can do a great deal among themselves to evaluate the safety and efficacy of drugs and medical devices; in which strict corporate-liability rules provide drug developers and device manufacturers with very strong incentives to be cautious and to disclose the inevitable risks associated with such products; in which private certifying agencies earn their profits by providing reliable information about pharmaceutical safety and efficacy to consumers; and, if we must, in which an agency something like the FDA retains the power to declare certain therapies unsafe and to restrict their use in certain circumstances.
By all means, pass the 21st Century Cures Act — piecemeal reform often is the best course of action. But the direction of reform should be toward a post-FDA order in which prohibition is not treated as the sole instrument of the public good. It is time for a change in the regulatory weather.
— Kevin D. Williamson is a roving correspondent for National Review and the author of The End Is Near and It’s Going to Be Awesome.