Patient 1 wanted to kill someone. Normally even-tempered, the 63-year-old man found himself awaking with an uncontrolled anger and the desire to smash things. His violent impulses started after he began taking the cholesterol-lowering statin Lipitor and they vanished within two days of quitting the drug. Patient 2 developed a short fuse after he started on Zocor*, another popular statin. The 59-year-old felt an impulse to kill his wife, and once tried, unsuccessfully, to do so. His violent tendencies subsided within a few weeks of stopping Zocor. Patient 3, a 46-year-old female, became unusually irritable while taking Lipitor, repeatedly blowing up at her husband for no reason. Like the others, her uncharacteristic behavior disappeared after she quit taking statins.
Physician Beatrice Golomb at the University of California-San Diego has collected thousands of anecdotes like these through her website, Statineffects.com, and she’s convinced that these drugs—taken by one in four Americans over the age of 45—can provoke severe irritability and violence among a tiny subset of users.
While it might seem crazy to blame bad behavior on a drug that’s prescribed for your heart, such effects are not unprecedented. Violent behavior has been linked to at least 31 other medications, such as tobacco cessation aids and antidepressants, and the notion that lowering cholesterol might make someone violent or aggressive has some scientific basis. Monkeys put on cholesterol-reducing diets become more aggressive, and numerous studies have linked low or lowered cholesterol to violent behavior in people, too. For instance, one study compared the cholesterol measurements of nearly 80,000 Swedes who’d enrolled in a health-screening project against police records and found that violent criminals had significantly lower cholesterol levels than noncriminals. Golomb points to low serotonin levels, which are also associated with low cholesterol, as a possible cause.
Still, the link between statins and behavior has been widely rejected by the medical community. “Golomb is a nut case,” says cardiologist Steven Nissen of the Cleveland Clinic, a vocal advocate for drug safety who refuses to take money from drug companies. None of the randomized, controlled trials done on statins have turned up this side effect, he says, and absent some rigorous evidence, the link between statins and violence is nothing more than speculation. “I’m not a tool for industry, but I also don’t want to scare people away from life-saving drugs because of fringe ideas,” he says, before urging me not to write this story at all.
There’s no question that Golomb’s idea is fringe, and from a public health standpoint, the benefits of statins—which can cut the risk of a second heart attack by about one-third—almost certainly outweigh whatever small risks of violent behavior they might, in theory, pose, at least for people with existing heart disease. (The benefits for people without heart disease are less clear.) Regardless, the overwhelming majority of people who take statins don't become violent, nor do they try to kill their spouses. Statins have prevented thousands of deaths, and it may be the case that Golomb’s theory about how they affect mental health is total bunk.
But right or wrong, this work highlights a worrisome problem. The Food and Drug Administration's system for tracking drugs' side effects is simply not equipped to detect rare but serious problems. By definition, these would be so uncommon as to slip by the initial studies that are used to create pharmaceutical warning labels. Such studies typically involve just a few thousand patients, so any response that affected a tiny sliver of the population would scarcely show up. Once a new drug hits the market, it's up to consumers and their doctors to report any unusual side effects to the FDA, and if the side effect were completely unexpected (as in the case of violent behavior), not all of these sufferers would think to consider whether the new symptom might be related to the drug.
When you’re prescribing a pill to treat a life-threatening disease, you might not worry too much about whether it carries a very rare side effect. Even if someone were to become irritable or violent while taking a powerful antibiotic, that would be a small price to pay for curing a potentially fatal infection. The infrequent side effect becomes more of an issue when you're talking about blockbuster drugs that are used in preventive medicine, however. Such medications are prescribed to millions of people throughout the country, as a way of warding off disease before it ever shows up. Even if a very small subset of these patients developed an unusual side effect—statin-related aggression or otherwise—then we might be talking about thousands of victims who were never that sick to begin with.
Nissen himself calls the FDA’s approach to dealing with drug side effects “woefully inadequate.” Once a drug has been approved for sale, the FDA looks out for so-called “adverse drug events” (ADEs) through a computerized system called MedWatch. Patients, doctors, and other health care professionals voluntarily submit ADE reports, but a 2002 Journal of the American Medical Association report estimated that more than 90 percent never get reported, since the system is voluntary and there are no real incentives for anyone to share their experiences. It's also not clear what happens to information once it's entered in the database. An FDA representative told me that staff members routinely monitor the database and they have an internal system in place for deciding when to look into a particular ADE, but there are no publicly stated rules about what it takes to trigger a formal government investigation.
Side effects are inherently difficult to pin down. In a previous Slate column on whether birth-control pills cause weight gain, I explained how easy it is for people to misattribute a symptom to a drug they’re taking. If you happen to develop a problem soon after starting on a new medicine, you’ll have no trouble deciding what’s to blame. The people who reported aggressive or violent behavior to Golomb after taking statins may have been affected by their prescription drugs, or they may simply have had an unrelated episode that happened to coincide with a new treatment. Golomb classifies a behavior as being “linked” to the drug if it began after the drug was initiated, went away when the drug was withdrawn, and reappeared if the drug was taken again. These criteria make sense, but they, too, can provide only suggestions, not definitive proof. Behavior is subjective and malleable, and inherently difficult to measure. Golomb recalls one patient who insisted, over his wife’s objections, that he wasn’t more irritable while taking statins. His wife just happened to be more irritating during the time when he’d been taking them.
More convincing evidence might come from a case control study in which people who showed violent behavior over a given time period are compared with similar people who didn't. If more of the violent folks were taking statins, that would be another warning sign. Even stronger would be a double-blind study in which violence-prone subjects were dosed with either a statin or a placebo. In any case, Golomb herself has conceded that the evidence to support her claim is a little shaky. In a case series published in 2004 in QJM, she notes that the link might be just a chance association. Seven years later, she's collected enough anecdotes to convince her that the link is real, and she's now recruited a geneticist at the University of British Columbia to look for genetic variants that could make people susceptible to becoming violent on statins. If a genetic link can be found, it would offer another hint that the problem exists in the first place, as well as a way for doctors to screen their patients before prescribing the drug.
What about the FDA? The key to finding rare side effects is to create a database large enough to spot them, and the administration could potentially do this by requiring drug companies themselves to monitor for side effects after their drugs go on sale. The FDA already asks manufacturers in some cases to conduct phase IV trials, which track the drug once it’s on the market, but they could make these trials mandatory for all new drugs, with a set deadline and stiff penalties for those who don’t comply. (A 2004 editorial in JAMA stated that fewer than half of the postmarketing studies that drug companies promised to conduct were actually completed, and many of them weren’t initiated in the first place.) The FDA could also require companies to send detailed surveys to a large cross-section of patients taking a drug to monitor for side effects. As electronic medical records become more common, these could also be scoured for conditions that turned up after a patient began a new drug.
The consensus view among doctors holds that statins pose no mental-health risks to consumers, and for the moment there's not much reason for most people to worry about becoming violent while taking the drug. But history has shown that many important drug side effects are slow to gain recognition—a 2002 JAMA study showed that only half of all serious drug side effects are detected within seven years of the drug’s approval. No one but Golomb seems to be tracking behavioral side effects associated with statins. Without the benefit of a more systematic national monitoring system, we don't have much else to go on.
Correction, Nov. 8, 2011: The original article misspelled the name of the drug Zocor. (Return to the corrected sentence.)