The first time you hear it, “evidence-based medicine” sounds like one of those goofily redundant phrases like “animated cartoon” or “past experience.” Aren’t doctors always carrying out studies of one sort or another? Isn’t medicine evidence-based already?
Well, no, not really. One of the biggest and least-understood problems in the U.S. healthcare system is that where many new drugs, medical devices and surgical techniques are concerned, there’s relatively little data as to which benefit patients most and under what circumstances. (See, for instance, today’s NYT column from David Leonhardt.) In part, that’s because there are relatively few incentives in today’s healthcare system for anyone to carry out such studies. For instance, it’s taken well over a decade to learn that unblocking partially clogged arteries fails to prevent heart attacks anywhere near as well as inexpensive drug treatment.
In principle, then, evidence-based medicine, which aims to scientifically measure and ultimately standardize medical practice systematically, sounds like something just about everyone could get behind. In practice, though, it all depends on whose ox — or ideology — is being gored by the evidence.
To see that, you don’t need to look any farther than the controversies that rage whenever the FDA rejects or postpones a drug approval for lack of sufficient evidence. Earlier this week, for instance, the NYT reported that two researchers who served on an FDA advisory panel received threatening e-mails and other messages after they opposed the approval of Provenge, a new type of “cancer vaccine” designed to trigger the body’s defenses against prostate tumors.
The researchers — Howard Scher of Memorial Sloan-Kettering Cancer Center and Maha Hussain of the University of Michigan — were concerned that the data offered by Provenge maker Dendreon didn’t actually prove that the vaccine worked. Although Scher and Hussain were in the minority when their panel voted to recommend approving Provenge, the FDA eventually concurred with the dissenters and said it would withhold approval until Dendreon produces more data.
To judge by the reaction, you’d think the FDA was having men with prostate cancer lined up and shot. My former WSJ colleague Marilyn Chase reported that “stunned” patient advocates vowed to deluge the agency with protest e-mails and to continue pushing for early access to the drug. One typical reaction came from Steve Fleischmann, a prostate-cancer survivor and advocate. “I feel very, very let down,” he told the WSJ. “What does this say to men who have prostate cancer and want to stay alive?”
Those were the measured reactions. A Sloan-Kettering spokeswoman told the NYT that Scher received phone calls and e-mails, including one titled “your murder.” Another letter writer told Scher, “You should get cancer.” Commenters on the WSJ Health Blog, whose three items on the FDA decision drew an outpouring of vitriolic attacks on Scher, Hussain and the FDA, accused the researchers of unethical behavior and FDA officials of shorting Dendreon’s stock or colluding with hedge funds. One commenter called the FDA decision “MASS MURDER” because it denied patients a promising drug.
It’s impossible not to empathize with people who consider Provenge the last hope for their loved ones or themselves. On the other hand, the evidence that Provenge actually works is slim indeed, while Dendreon seemed determined to spin it as hard as possible. (For a recap of the company’s incredible sloppiness with those clinical trials, see my earlier piece on Provenge.) In other words, Dendreon opportunistically took whichever path seemed to offer the quickest path to approval, even if it meant heaving a Hail Mary pass at the end. If patients really want to get angry at someone, they might want to take a closer look at Dendreon management — particularly CEO Mitchell Gold, who coincidentally sold off a big chunk of stock virtually the moment Dendreon shares bounced up on the positive advisory-panel vote.
Still, the patients consistently ask one question that deserves answering: Why not approve a drug that seems relatively safe, as Provenge does, and let it stand or fall in the market? While there’s a practical answer — the FDA is legally bound to ensure both the safety and effectiveness of drugs — a better response is both moral and economic.
No one wants prostate-cancer patients to die agonizing deaths. Allowing an unproven and undoubtedly expensive treatment like Provenge onto the market, however, means that its costs will be picked up by taxpayers and anyone who pays health-insurance premiums — again, in the absence of proof that it will help anyone. That seems like a bad gamble at a time when healthcare costs are already rising far faster than inflation. And since those skyrocketing costs are also boosting the ranks of the uninsured — partly because employers are scaling back coverage — you have to ask if providing an unproven treatment to a small group of cancer patients is worth the additional misery in store for ordinary people who lose or can’t get health insurance as a result.
In addition, once a treatment is available, it’s far more difficult to figure out whether it actually works or not. Patients aren’t likely to volunteer for controlled clinical trials in which they might end up receiving a dummy shot if they can be assured of getting the real thing from their doctor. While it’s still possible to compare marketed treatments to other marketed treatments, the logistics and expense can be hairy.
Patients, of course, aren’t the only ones grousing about the costs and delays that result from waiting for evidence that a given treatment works. Right around the time the FDA postponed approval of Provenge, for instance, the WSJ editorial page featured two pieces in which biotech-industry executives effectively argued for setting aside hard statistical evidence in favor of wishful thinking.
The first piece was by Richard Miller, the CEO of Pharmacyclics, a company that has been notoriously unsuccessful in proving that a drug called Xcytrin works against tumors in the brain the way it is supposed to. Miller’s complaint is that while clinical trials suggested that Xcytrin works, the data failed tests of “statistical significance” designed to distinguish actual results from random chance. Similarly, in a May 14 piece, a former FDA official named Mark Thornton lamented the day the agency postponed approval of Provenge and another cancer immunotherapy, calling it “Black Wednesday” and accusing the FDA of “kneeling before the altar of statistics.” (Oddly, the WSJ didn’t see fit to note that Thornton is currently a vice president at GenVec, a gene-therapy biotech that might also benefit should the FDA relax its statistical standards.)
The two men do have a point: The significance level required by the FDA is indeed arbitrary. The problem is that you have to set that bar somewhere, and Miller’s argument that it should depend on vague notions like “context” and “very real advances in science and medicine” is simply incoherent. (Thornton’s notion that “solid immunology science” should trump statistical significance isn’t any easier to understand.) My favorite part, though, is when Miller argues that “[s]tatistical standards, of course, should not be set aside” — this right in the middle of an article pleading with the agency to grant a statistical exemption for Xcytrin.
Such “up with patients, down with pointy-headed statisticians” sentiments have a politically conservative underpinning that isn’t hard to find. Check out, for instance, DrugWonks, a blog run by the Center for Medicine in the Public Interest. There, conservative pundit and former Manhattan Institute official Robert Goldberg criticizes the FDA’s “deadly decision to delay Provenge” and its advisors’ dedication to the “probabalistic priesthood” in language virtually identical to that used by Miller and Thornton — almost as if they’d all been reading from the same script. Ultimately, it’s hard to escape the impression that these folks are less concerned about patients than they are about the freedom to market drugs without all that pesky FDA interference.
On the other hand, every drug-safety scandal that erupts over treatments like Vioxx and Avandia helps make the case for the importance of good medical evidence. Among other encouraging signs that evidence-based medicine is slowly getting some traction are these articles on a House bill that would provide $3 billion for evidence-based studies by the federal Agency for Healthcare Research and Quality. AHRQ has fallen on hard times over the past decade, a trend that might be about to reverse itself. More on that in a future post.
VB's research team is studying web-personalization... Chime in here, and we’ll share the results.