Most of us trust, or at any rate hope, that the benefits of a drug our doctor prescribes will outweigh the side effects. Why else would we take it? We would probably be shocked to learn that most drugs don't do anything good for the majority of the people who use them. That's probably because we picture a simple cause-and-effect relationship, like antibiotics curing an infection. "But most chronic diseases involve a complex chain of biochemical interactions," says Dr. Jonathan St. George, assistant professor of emergency medicine at Weill Cornell Medical College. "The idea that you're going to take one drug that affects one pathway and dramatically change the course of the illness is just pie in the sky." The statistical measure that crystallizes this inconvenient truth is the NNT, or "number needed to treat" – that is, the number of people who have to take a drug in order for one person to benefit. There are plenty of popular drugs with NNTs over 50, and a drug with an NNT of five or fewer might fairly be considered a wonder drug – for instance, sumatriptan for migraines or steroids for kids with croup. "But if I told my patients that the drug I was prescribing them had only a 20 percent chance of working," St. George says, "they'd look at me like I was crazy."
The reason you've probably never heard of the NNT is that the pharmaceutical industry ignores it when marketing its wares to the public. According to Newman's website, thennt.com, which crunches the best available research data to arrive at NNTs for common tests and therapies, statins have an NNT of 60 – meaning 60 people would have to take a statin drug for five years to prevent one person from having a nonfatal heart attack. Not one heart attack death would be prevented. Picture a similar effect this way: a study in which a control group of 1,000 people taking no heart medication suffered 24 heart attacks over a five-year period, while the group on statins suffered 16. Because these numbers are small, even relatively minor differences between the incidence of heart attacks translate into an impressive-sounding difference, when you measure it as a percentage – the so-called relative risk. Now you've got the makings of a pharmaceutical ad campaign: "Statins reduce heart attacks by 33 percent."
It gets worse. Stanford epidemiologist John Ioannidis got the medical world's attention in 2005 with a journal article titled "Why Most Published Research Findings Are False." In it he notes that 80 percent of published drug studies are funded by the drug industry, and that some 30 percent of all drug studies are never published, presumably mostly the negative results that never enter into the final cost-benefit reckoning.
But, Hadler says, even if we were to take the research at face value – that a given drug has a statistically significant benefit when the NNT is, say, 50 or higher – the benefit is so small it's clinically meaningless. But fortunes are made from such microscopic benefits. The pharmaceutical companies can create blockbuster drugs by promoting meds that have shown benefit in a smaller, targeted population – say, statins for people who've already suffered a heart attack – to a larger, relatively healthier population, with the hope that the medication might be good for them, too. "Blockbuster drugs demand overtreatment," Hadler says. Beyond the side effects that the overtreated may suffer for no offsetting gain is what Newman calls the culture of the pill. "It's destructive to physicians," he says, "and to patients who believe, 'I can forget all the lifestyle stuff because I can take a pill and I'll be good.' "
Introduced in the States in the late 1980s, statins inhibit an enzyme that the liver uses to make cholesterol, in most people dropping that LDL number by between 30 and 50 percent. At a cost. Newman crunches the research figures and calculates that for every 50 people on statins, one will develop type 2 diabetes who otherwise would not have. The statistic that tells you what you need to know about the severity of a drug's side effect is the "number needed to harm." So if we're talking about diabetes risk, the NNH for statins is 50 – dose 50 people with a statin and you can expect to see one extra case of type 2 diabetes turn up.
The most common side effect of statins is muscle pain and weakness and, in severe cases, muscle breakdown. Here the NNH is 10 – 10 to treat, one to harm. Mental "fuzziness" and forgetfulness haven't been rigorously studied enough to generate an NNH, but enough anecdotal reports have come in that two years ago the FDA slapped statins with a cognitive safety alert.
So this past November, when a panel convened by the American Heart Association released its new guidelines on statins, you might have expected it would take a more conservative line on prescribing – that, given the possible side effects, they would want to prescribe the drug only to a more select group of patients for whom the benefits clearly outweigh the harms. But with statins you'd be wrong.
The committee recommended new guidelines that, if faithfully followed, would, according to Brown University–affiliated cardiologist Dr. Barbara Roberts, put 44 percent of American men over the age of 40 on statins. ("The American Heart Association has been the death star for years," Newman fumes.) Because the evidence that statins confer significant heart protection by lowering LDL was so weak, the committee took a new tack, recommending the drugs for anyone at even a moderately elevated – for any reason – risk of heart attack, a 7.5 percent risk over the next 10 years, a figure calculated by the committee's own formula.
An op-ed piece in the New York Times the next day, co-written by the horrified editor of JAMA Internal Medicine, pointed out that people on statins in this new, broader group would have, according to her calculations, an NNT of 140 – 140 to treat, one to benefit – without there being an overall reduction in death or life-threatening illness. "Statins give the illusion of protection to many people," Dr. Rita Redberg wrote, "who would be much better served by simply walking an extra 10 minutes a day." For the sake of comparison, a study published last year in the
New England Journal of Medicine found that going on the Mediterranean diet, heavy on olive oil, nuts, and beans, had an NNT of 61 – for every 61 people on the diet, one was spared a heart attack, a stroke, or death. That's not a great number for a drug, but for a diet it is. If millions of people ate this way, a lot of people benefit and no one gets hurt – an NNH of zero.
How could some of the most eminent cardiologists in the country have fallen into this statistical rabbit hole? Welch points to the myopia of medical specialists in general: "They don't want to miss anyone who might conceivably benefit from diagnosis and treatment, and what they don't see is the harm that this strategy produces."
Then there is the money factor. More than half the doctors on the committee have received compensation from the pharmaceutical industry in the form of speaking and consulting fees and research subsidies. They recused themselves from voting on the committee recommendations only if they "felt" there was a conflict of interest, according to one committee member quoted on WebMD. But even harsh critics of the cozy relationship between mainstream academic medicine and the pharmaceutical industry tend not to impute bad faith to the doctors. The bias in favor of drug therapy has, they say, been internalized. "They're mostly true believers," Newman says. Says Roberts, author of The Truth About Statins, "They've drunk the Kool-Aid."
Blood Pressure Meds
Almost a third of American adults suffer from high blood pressure, the majority in the mild-hypertension, 140/90 to 159/99 camp. And most of them are on antihypertensive drugs, which is why, according to one analysis of data from the American Heart Association, last year they spent about $32.1 billion on meds and doctor visits, nearly 1 percent of the nation's health care bill. The catch, and you saw this one coming, is that while the risk of heart attack and stroke goes down when blood pressure drops in response to changes in diet or exercise or handling stress, when you use drugs to treat mild hypertension to get the same reduction, nothing comparably good happens. And according to thennt.com, the drugs have an NNH of 12: For every 12 treated, one will suffer from side effects that, depending on the type of drug, include fatigue, dehydration, and sexual dysfunction.
For men on the cusp of high blood pressure, with a systolic reading in, say, the high 140s or 150s, there is no one-size-fits-all rule as to whether they should be on the meds if they can't bring the numbers down themselves. Welch's advice: Go to the government's online heart-risk calculator at heart.org and see how much risk reduction you'll likely get from dropping your blood pressure a certain number of points. And buy a home blood-pressure monitor. "You get real positive feedback when you exercise and your blood pressure falls," he says.
Antibiotics (for Upper Respiratory Infections)
One-fifth of the antibiotics prescribed in the U.S. are for upper respiratory infections. The patient walks into the doctor's office or the urgent care clinic with a nasty case of bronchitis or sinusitis and, most of the time, walks out with a scrip. This scenario seems impervious to the fact that most of the infections are viral, not bacterial, and that the antibiotics are worthless against them. In study after study, the drugs have been found to do precious little good, at best shortening the duration of a symptom like a cough by a day or so, and at the risk of building up antibiotic resistance in cases where the drugs may really be needed. Doctors know all of this, but, as Hadler says, "in this system, it takes me 20 seconds to prescribe a drug and 20 minutes to explain to the patient why they don't need it."