FDA Should Not Mandate Comparative Effectiveness Trials

AEI resident fellow Scott Gottlieb has a new paper out explaining why the FDA should not force pharmaceutical companies to prove their new drugs are superior to existing treatments before they may be approved. The idea has been kicked around by pharmaceutical industry critics for years, particularly as a solution to their claim that the industry produces too many “me too” drugs that are all essentially identical. CEI has debunked that argument before, but it nevertheless gets a lot of traction any time the issue of rising drug prices comes up.

Still, the FDA itself has held fast to the idea that randomized placebo-controlled trials (i.e. comparing a new drug against nothing at all, rather than another active compound) is the best way to generate clean, statistically reliable data on safety and efficacy. Gottlieb’s new paper does a fantastic job of explaining why that’s so. “Superiority trials require an enormous number of study subjects to discern clinically meaningful differences between two drugs. These trials can require tens of thousands of patients when two closely matched drugs are being compared.”

Moving from 2,000 or 3,000 patients in a typical Phase III trial to tens of thousands of patients is guaranteed to cause and explosion in R&D expenses. Note that each additional patient enrolled in a clinical trial can add an average of $30,000 to trial costs.  Many products could justify the expense — though critics would complain even more about rising drug costs. But the bigger question is just where are all these patients going to come from? It’s already difficult to enroll the smaller placebo-controlled trials. Enrolling an order of magnitude more patients into a trial in a world of increasingly narrowly-targeted personalized medicines would be difficult, if not impossible. The new drug development process would grind to a halt due to the sheer impracticality of it — which just may be why drug industry critics love the idea so much.

An alternative, called a non-inferiority trial, permits a smaller patient enrollment, but it’s designed to show that two drugs are at least equally effective, not that one is necessarily better.  It also provides less reliable data. Gottliebe notes that drug firms are already doing lots of these trials for the very purpose of trying to convince insurance companies and other health care payers of the value of their newest offerings.  A study out earlier this year found that 70 percent of the new drugs approved by the FDA from 2000 to 2010 for which an alternative was already available, manufacturers had conducted at least one trial comparing the new drugs to active comparators.  But this must almost always accompany placebo-controlled trials too — except in cases where running placebo-controlled trials would be unethical. If FDA officials weren’t such sticklers for perfectly pristine data packages, some compromise middle position might be arranged. But they are. So, for now at least, it can’t.

Still, even if we did have more or better comparative effectiveness data at the time of approval, it’s not at all clear whether it would provide useful guidance to prescribing physicians. As I’ve pointed out before:

Patients vary substantially in their individual physiology, their response rates to drugs and surgical procedures, and their willingness to tolerate side effects. So, for many conditions, from cancer and cardiac care to asthma medicines and antidepressants, the choice of appropriate treatment requires doctors and patients to carefully balance the benefits and drawbacks of individual interventions. But, in order to produce statistically significant results, the clinical research conducted to compare different treatments must evaluate groups of patients who are highly similar and who therefore are not representative of the population at large.

Gottlieb further notes that, “In clinical practice, patients often try many similar drugs until they find one that works best for them.” There was a bit of a minor scandal a few years ago, when The Wall Street Journal reported that Novartis CEO Daniel Vasella was using Pfizer’s cholesterol lowering statin Lipitor, not Novartis’s Lescol. Like countless millions of patients, Vasella tried one drug (in his case Lescol) and didn’t respond to it very well, so his doctor prescribed another. But if the FDA forbids competing products if they’re not “superior” for the average clinical trial enrollee, it’ll leave lots of patients in the real world with fewer alternatives.