How our national obsession with drug safety is killing people -- and what we can do about it.
(FORTUNE Magazine) - For six hours on Nov. 8, 2005, time moved so slowly for Frank Burroughs and Steve Walker that it seemed to stop altogether. The two had come to a dreary ballroom at a Holiday Inn in Gaithersburg, Md., for a meeting of the FDA's Oncologic Drugs Advisory Committee, or ODAC. Here, at three tables set up in a horseshoe and draped in burgundy cloth ruffles, sat some of the most important people in the cancer establishment. Their role was to review presentations about new-drug applications and make recommendations regarding their approval.
For Burroughs and Walker, fidgeting as other presenters went through their PowerPoint slides, this meeting was not about dispassionate science. The panel had the power to offer hope to cancer victims, or withhold it. Both men had lost loved ones to the disease--Burroughs had buried his college-age daughter, Abigail, 4 1/2 years earlier; Walker's wife had died after a two-year struggle with the disease. Both had tried and failed to get access to last-ditch experimental medicine. Both had become activists. And both felt an agonizing frustration with the FDA that, if anything, was growing.
Some in the room guessed what was coming when Burroughs stepped up to the mike. But even they were surprised when, after a few words of introduction, his cheerful face took on a look of fury. His voice began shaking, and soon he was invoking Thomas Jefferson's admonition on tyranny to those on the dais, reminding them that people of good conscience can't remain silent. He was particularly angry with the committee's lone patient advocate. Who did she represent? Burroughs wanted to know. Certainly not dying patients with nowhere to turn!
Moments later, Walker took the podium. He, too, tried to keep calm, but his voice began to crack. "We need to remember who this is all for," he said. "It's for the patients. It's not about P values; it's not about endpoints; it's not about policies; it's not about your careers. It's about the patients out there--the patients we have lost waiting for the process to run its laborious, tortoise-like course."
There was a hush as the awkwardness of the moment filtered through the room. And then the moment was gone. The chairwoman thanked Walker for his remarks, said they were "food for thought," and, after a five-minute bathroom break, moved to the next agenda item.
Burroughs and Walker are familiar names to people involved in the drug-approval process--Walker is the chief spokesman of the Abigail Alliance for Better Access to Developmental Drugs, an advocacy group Burroughs set up and named for his daughter. And judging from the scattered smirks around the room, there are plenty of people who think these guys are wackos.
It's easy to see why. The Abigail Alliance and other advocacy groups are pushing to get potentially toxic drugs released to patients in the earliest stages of testing and development, long before they've been proved to work. They want faster access right at a time when regulators are forcing the drug industry to slow down--often with good reason.
Ten medicines, most notoriously Merck's painkiller Vioxx--which has been linked to as many as 55,000 deaths by FDA whistleblower David Graham--have been removed from pharmacy shelves for safety reasons since the start of 2000; in the 25 years before that, a total of just 16 were withdrawn, according to a study in the Journal of the American Medical Association. There have been twice as many "black box" warnings, which the FDA uses to alert doctors and patients to a particularly worrisome side effect, issued in the past two years (91) as there were in the quarter-century covered by the JAMA study (45).
With so many warnings and withdrawals, you might conclude that the drug-approval process is broken at the 100-year-old agency. That's certainly the view of advocates like Sidney Wolfe, a gravel-voiced medical doctor and founder of Public Citizen's Health Research Group. Wolfe, who cited disturbing data about Vioxx, Bextra, Baycol, and Rezulin in his newsletter Worst Pills, Best Pills long before their makers withdrew them, sits at the opposite extreme from the Abigail Alliance. He is calling for more oversight, more safety studies, and more caution. He sees dangers in an additional 180 prescription drugs that remain on the market.
But dig deeper, and you might come to another conclusion: The approval process is broken--but not in the way most people think. It is in thrall to a well-intentioned but ultimately misguided national obsession: the quest for certainty about drug safety and efficacy. And in fact, scores of interviews with scientists, regulators, industry officials, patients, and policymakers over the past four months suggest that's the case--that our regulators have fallen prey to a deadly caution. Simply put, the need for certainty in drug approval is killing people. Excessive caution is delaying the availability of potentially helpful treatments for cancer, multiple sclerosis, Parkinson's, and a host of other ailments; it's slowing the absorption of new knowledge and diagnostic tools into medical practice; and it's discouraging the pursuit of vaccines and next-generation antibiotics that could save tens of millions of lives. It's sacrilege to say so, but Hippocrates' prime directive--the narrow regulatory interpretation of it, anyway--is having a devastating effect. The four most dangerous words in medicine may be "First, do no harm."
Consider the predicament that FDA reviewers find themselves in. Regulators can't be sure that drugs are safe until long after they are out in the marketplace. Clinical drug trials simply cannot be large enough or long enough to yield absolute certainty. That's why withdrawals and warnings are inevitable. In the case of drugs intended for people with terminal illnesses like cancer, safety is clearly less of a factor. But reviewers face an equally daunting dilemma: whether to greenlight compounds that have not definitively been shown to work or to insist on more testing and let people die waiting. And in the post-Vioxx age, they are erring on the side of extreme caution.
This preoccupation with drug safety contributed to the shortages of flu vaccine last winter. It has contributed to the fact that we have few weapons left to combat some scary, mutated, antibiotic-resistant bacteria showing up in U.S. hospitals. It has been a factor in both the blistering rise of drug costs and in what the FDA itself has dubbed the "pipeline problem" at the world's pharmaceutical companies. Today a drug entering early clinical trials has just a one-in-13 chance of getting to market--odds that are worse, not better, than they were decades ago. In fact, the number of medicines withdrawn or mandated to carry major warnings over the past six years actually exceeds the number of innovative drugs (so-called new molecular entities) approved for the market.
The caution has been building for two generations, says Jim Greenwood, a former Republican Congressman from Bucks County, Pa., who now runs BIO, the biotech industry's trade group. That is, ever since it became known that thalidomide, an antinausea drug approved in Europe but not in the U.S., was the cause of horrific birth defects in thousands of babies. "The FDA, post-thalidomide, was given a mandate," says Greenwood. "And the mandate was, 'Don't you ever let any product get to the public unless you're absolutely sure that it's safe and effective.' No one ever told them--Congress certainly never told them--that another objective should be to do that as quickly as possible, because lives are hanging in the balance."
It's no surprise that the nation's chief biotech lobbyist would want to speed up the approval process. But Greenwood is part of an eclectic and growing group of people agitating for a revolutionary change in the way the FDA evaluates drugs. In their ranks are everyone from patient advocates to infectious-disease specialists, from health economists to statisticians, and even some top staffers at the FDA. Some are pushing the agency to think more flexibly about how it tests drugs and to give patients greater say. Others are agitating to change the incentives for developing new treatments and the ground rules by which drugmakers disclose their clinical research. And all want a better system for keeping track of adverse reactions to treatments and drugs.
Their arguments are compelling. But change won't come easily.
Rick Pazdur is aware that some people see him as the devil incarnate. He was at that meeting in Gaithersburg--in fact, he was the main target of Walker's anger. As the nation's cancer-drug czar (his formal title is director of the Office of Oncology Drug Products at the FDA's Center for Drug Evaluation and Research), he has been called a murderer and received death threats. The Wall Street Journal editorial page has vilified him three times for delaying the approval of drugs for dying patients. Such criticism is unfair--Pazdur isn't some wild-eyed apparatchik on a power trip; he's a deeply conscientious cancer doctor who took a six-figure pay cut to come to the FDA. The problems lie in the culture and legally mandated mechanisms of the whole drug- approval process. And the ad hominem attacks hurt, he says, although not as much as actually hearing the stories of those who feel as though their options are used up. But Pazdur also hears from people who are bitter that loved ones endured harrowing end-of-life treatments with toxic drugs that had little chance of working. "We ourselves have had family members with cancer," he says. "And all of them are close to our hearts too."
Despite the pressure--and maybe because of it--Pazdur and others defend their mission the way Davy Crockett did the Alamo. They would love to approve drugs more quickly, but they believe that the only way to really know whether a therapy works and is safe is through rigorous clinical trials. As acting FDA commissioner Andrew von Eschenbach puts it, "We have to go rapidly, but not recklessly."
Drug trials typically involve three phases that can take a decade or longer to complete. The concept of comparing new drugs and treatment combinations with known and accepted therapies got its start in the 1950s and 1960s in the early days of experimentation with cancer medications. The drugs were inordinately toxic, and so it was crucial that investigators establish that they worked at least a little before allowing them to be used widely. In time, clinical trials grew more rigorous, and they soon became the gold standard for conducting medical research.
The process generally comprises three separate, sequential tests in human patients: Phase I tests toxicity; Phase II attempts to see if the compound is effective at all in a small number of patients. And Phase III is a much bigger and longer experiment to determine whether the new drug is truly better than existing treatments or than doing nothing at all. This final hurdle is tough enough--but then, to make sure there's no cheating or unintentional bias, patients and investigators are typically "blinded." In theory, they don't know who's been randomly assigned to get the experimental drug and who's getting the control. In trials for life-threatening diseases, it generally isn't until a preset number of patients have died that anybody peeks at the results.
If this process seems unduly long or excessively cold, it is only in the name of good science. Medical orthodoxy says there is no other way to know with certainty whether a new drug is safe and effective. Indeed, even after a therapy is long accepted, clinical trials can reveal problems. It was through follow-up studies, points out Ward Casscells, a leading authority in cardiovascular medicine, that physicians learned that Flecainide, a drug designed to correct irregular heartbeats, was killing more patients than it was saving. Clinical trials revealed that high-dose chemotherapy followed by a bone-marrow transplant, a once-common, brutal, and often deadly therapy for breast cancer, wasn't necessary. "More than 20,000 American women had endured this treatment for no compelling reason," says breast cancer advocate Musa Mayer. Even Vioxx's nasty side effects came to light in clinical trials.
No wonder Pazdur sees any attempt at rushing drugs through the process as an assault on science and public health. Lowering the testing bar might help desperately ill people get a drug more quickly today, but it could hurt thousands in the future. "We're not just talking about the immediate patient," he says. "We're talking about the long-term goal here, and you have generations of patients who want to have evidence that something works."
But does the process work? For all the time and expense these tests entail, do they tell us whether a drug is safe and effective? Amazingly, in many cases the answer appears to be no. The problem is that virtually identical experiments often yield contradictory conclusions. Witness what happened when researchers tried to determine if early-stage colon cancer patients, post-surgery, actually fared better when given a standard chemotherapy regimen. Five rigorously designed trials came up with very different answers. One, for example, showed a 67% reduction in death rates for those given the therapy; another said deaths in the treated group rose by 28%. In 1999 a much-anticipated analysis of the five trials arrived at its own conclusion: There was no statistically significant benefit to the treatment. End of story? Not quite: Yet another so-called meta-analysis of four more trials said patients do benefit from the treatment. In all, 2,581 cancer patients were studied from 1977 to 1990 to come up with a big academic shrug.
This isn't the exception; it's the rule. Flip through any medical journal, and you will find a similar argument in progress. Enormous trials have offered conflicting data to justify or reject mammograms in younger women and PSA tests for prostate cancer; battles rage over whether synthesized GDNF (a nerve growth factor) works in Parkinson's disease and how low cholesterol levels should be in healthy people. Does flu vaccine make sense for children under 2? After 51 studies involving 263,987 children around the world, we still don't know. (The seven authors of a recent meta-analysis of the trials concluded, "Large-scale studies ... are urgently required.")
Such uncertainty has kept a number of potentially very helpful therapies in a kind of regulatory limbo. Even the FDA's "accelerated approval" mechanism can be slow: The cancer drug Erbitux (made infamous in the ImClone scandal) hung for months in statistical purgatory, as did Nexavar and Revlimid, two recently approved drugs for fatal cancers. The promising multiple sclerosis drug Tysabri was withdrawn by its manufacturer, Biogen, over its connection with a rare side effect. And the FDA delayed (perhaps interminably) Merck's and Bristol-Myers Squibb's Pargluva, a treatment for diabetes. In all those cases, clinical trials seemed to raise more questions and controversies than they answered.
The solution to the problem, though, is not to throw out the clinical-trials process: Don Berry has a better way. The head of biostatistics at M.D. Anderson Cancer Center in Houston, Berry has been preaching the virtue of a technique called Bayesian analysis for 30 years. This centuries-old statistical approach is well established in genomics, astrophysics, computer science, and business--and is used for everything from spam-filtering software to oil exploration to locating lost submarines. (Harvard Business School has integrated the approach into some of its seminars.) Already around 10% of FDA approvals for medical and radiological devices are based on Bayesian analysis. They include spinal implants and cardiovascular stents that, if required to go through traditionally designed Phase III trials, would have taken perhaps a year longer to become available.
Rather than lock you into a single testable hypothesis for the duration of an experiment, the Bayesian method lets you adjust the questions you're asking to take into consideration new information or even prior knowledge, says Berry. You can pause or stop a clinical trial at any time, for example, and evaluate which treatment is working. Unlike with traditional Phase IIIs, if a drug appears to be effective early in a Bayesian-designed trial, future patients have a better chance of getting that new agent instead of standard therapy. That allows for much smaller sample sizes and shorter trials. It's also much closer to the spirit of both science and clinical practice: Just as doctors pick up vitally important knowledge in the course of taking care of patients, Bayesian rules add a crucial "learning as you go" element to experimentation.
What's more, Bayesian methods can enable researchers to learn from how treatments are used in the real world--information that traditional clinical trials can't take into account. Pharmacists misread handwritten scrips; patients pop too many pills or suddenly stop taking their meds. Drugs interact in unpredictable ways, and most important, they may simply cause different biochemical reactions from one individual to the next. (Care to guess what the No. 1 cause of acute liver failure is in this country? Acetaminophen, the pain reliever in Tylenol.)
The FDA says it hopes to encourage more Bayesian analysis. "Even when we agonize and refine and add another 10,000 people to the clinical trial, we still don't have the answers," says von Eschenbach. "We have to get out of the past and embrace a vision of the future." That includes not only new statistical tools but also the use of high-tech diagnostics--including tests for biological markers in the blood that can demonstrate quickly whether a new drug works or is making the patient sicker. It's all part of an ambitious campaign called Critical Path, which has been shepherded by Janet Woodcock, the FDA's deputy commissioner for operations. The only problem? Nobody has bothered to fund it. The project is getting a mere $6 million this year--which is barely enough to register in the multibillion-dollar world of drug development. Pfizer sells more than that in Lipitor every six hours.
Jonathan Baron worries about an even deeper flaw in the system. The bearded 61-year-old psychology professor at the University of Pennsylvania has been studying judgment and decision-making for many years. "We have a natural tendency to dismiss our sins of omission," he says. "We take harms of action to be worse than harms of inaction." For example, most people think that lying is worse than withholding the truth, and that killing someone is worse than letting them die. This tendency, which Baron calls "omission bias," can be catastrophic in public health.
Omission bias in medicine goes all the way back to Hippocrates, who exhorted young physicians to focus first on the effect of their actions. Doctors are taught to act only if the benefits of their ministrations outweigh the risks to the patient. The FDA is even more susceptible than doctors. Many of its bureaucrats would sooner do nothing than risk being blamed for a mistake. Baron would like to see the drug agency switch to what Dan Troy, a former chief counsel for the FDA, calls a "risk-risk" mentality: Reviewers should weigh the risk of approving a drug against the consequences of keeping it off the market--and not helping thousands of desperate people.
A tragic example of omission bias is the case of RotaShield, a vaccine made by American Home Products (now Wyeth). Launched in 1998, RotaShield was considered nothing less than a godsend by infectious-disease specialists. It was virtually 100% effective in preventing rotavirus--an omnipresent diarrhea-causing virus that directly leads to 600,000 deaths a year, nearly all of them children in the developing world. Selling for a little over $100 for a complete three-dose regimen, it was quickly taken up in the U.S. schedule for childhood vaccinations. Within a year some 1.5 million American children had been inoculated.
But during this time, government researchers caught sight of what looked to be a serious problem. In children who were vaccinated, the occurrence of a rare bowel obstruction was higher than in the general population--by a rate of one to two cases per 10,000. That prompted the U.S. Centers for Disease Control to withdraw its endorsement for the vaccine and the FDA to encourage the company to remove it from the U.S. market. Wyeth soon abandoned the whole effort.
Now, six years later, two of Wyeth's rivals, Merck and GlaxoSmithKline, have launched new rotavirus vaccines. A pair of studies published in January in the New England Journal of Medicine, involving massive and phenomenally expensive trials of more than 60,000 patients each, showed the new vaccines to be safe and effective; there was no evidence of increased risk for bowel obstructions. Merck's vaccine was approved by the FDA on February 3; Glaxo's is likely to be soon. So you could argue that the system worked as well as it was supposed to: A dangerous vaccine was withdrawn, and drugmakers went back to the lab to cook up a better medicine.
But consider the cost of not having the vaccine during those six years: An estimated 3.6 million children have been lost worldwide to a preventable disease. (In the U.S., where rotavirus is rarely deadly, it still results in some 95,000 annual trips to the emergency room and 227,000 doctor visits for children under age 5.) And if 3.6 million deaths weren't tragic enough, further study has led some researchers to believe that the 1999 vaccine may not have caused the rare bowel obstruction after all: The complication sometimes occurs for no known reason.
In addition to its appalling human cost, the rotavirus saga illustrates the regulatory pressures vaccine makers face. In 1967, according to the Institute of Medicine in Washington, D.C., there were 26 U.S. producers of vaccines; in 2003 there were five. A clear reason for the decline is that the high cost of gigantic trials and impossible expectations of risklessness push drugmakers toward investing in something else.
The chilling effect isn't limited to the treatment of Third World plagues: We're maneuvering ourselves into an equally unfathomable risk even with ordinary germs. Carl Nathan, an authority on infectious disease at Cornell University's Weill Medical College in New York City, worries about the rapid rise of drug-resistant bacteria. "We're racing back to pre-penicillin days," he warns. For growing numbers of infections, he says, "there aren't going to be any more antibiotics to turn to very soon." In 1992, according to the NIH, 13,300 patients died of an infection they picked up in the hospital. In 2004 the figure was 90,000--up almost sevenfold in 12 years.
The gentle-voiced Nathan, it should be noted, is one of the least alarmist of experts. The science of making new classes of anti-infectives is difficult, certainly. But the biggest obstacles, he says, are economic and regulatory. Development can easily take five or ten years longer than for other drugs. And because bacteria mutate so quickly, any new antibiotic that is discovered is going to have a short life span of effectiveness. It's just a lousy investment for a company to make, and many manufacturers have dropped out. But the U.S. is making a big mistake by not doing all it can to pack the pipeline with new antibiotics and vaccines, says Nathan: "Prosperity and stability are built on the foundation of not having your workforce drop dead suddenly. It just seems obvious that policymakers should look a little further ahead."
The FDA can't solve all these problems by itself. A powerful move Congress could make would be to require that drug companies register in a database every single clinical trial they do--and post the results promptly. That way, when nasty side effects show up years later (and they will--drugs are rarely so specific that they affect only a precise biological target), no one can say that companies knew of the risks in earlier trials and hid them.
PhRMA, the industry's powerful trade group, continues to fight the idea of mandatory reporting, but promises that its member companies will offer more data voluntarily. Most drug companies, in fact, claim that they're already posting what they call "hypothesis-driven" trials--jargon for late-stage studies. But a look at the industry's online database shows that few companies have done even that. Only two trials are listed for all studies in patients with multiple sclerosis. (The website of the NIH, www.clinicaltrials.gov, lists 88--and that's just a fraction of human studies known to be going on, advocates say.) Complete disclosure will add a layer of safety without slowing the development process.
Congress should also mandate the reporting of adverse drug reactions (ADRs). Right now, doctors, nurses, and pharmacists--the health-care workers most likely to spot an ADR--don't have to report it to anyone. They should have to, and the information should go into a publicly available clearinghouse. (Drug and device makers already have to file mandatory forms with the FDA when they learn of patient problems, but those disclosures are often late and incomplete.)
Both of these fixes will help restore credibility to a system that has clearly lost it. Part of what has created the current paralysis, in fact, is that people feel betrayed--by regulators, by doctors, and most of all by Big Pharma.
But there is one crucial thing the FDA can do right now. Just ask Perry Cohen. The 59-year-old Marylander has been living with Parkinson's disease for the past decade. He holds a mathematics degree from Carnegie Mellon and an MBA and Ph.D. from MIT, worked for 20 years as a consultant to the NIH and private health plans, and advises a leading Parkinson's advocacy group. But ask him what's stalling the pace of discovery in the complex field of neurology, and Cohen trots out a glaringly simple answer: "The missing ingredient is the voice of the patient," he says. Including more patients in FDA deliberations would add much-needed urgency to the drug-approval process.
It wasn't until the late 1980s, after intense lobbying from the AIDS community, that the FDA began including patient advocates on drug-review panels like ODAC. But even after that reform, the patient--the person who is actually taking the risk and living with the pain and disability of disease--represents just one vote on a committee of nine to 14 experts.
The FDA's Office of Special Health Initiatives has a roster of 100 advocates in 40 diseases. Three staff members say they'd like to have more, but there's no money or staff to train them. Patients can, of course, wait for their turn to speak at the handful of FDA meetings that are open to the public, but they generally can't ask questions or have a back-and-forth with either presenters or panelists. So what's the point?
The point, says an OSHI staffer, "isn't for patients to bang their shoe on the table. These are scientific meetings."
Frank Burroughs shrugs. "Hey, I'll take my shoes off," he promises. "Just give me a seat at the table."
Research Associate: Doris Burke