Could It Be That Antidepressants Do Little More Than Placebos
No prescription for happiness
Could it be that antidepressants do little more than placebos
By Thomas J. Moore, 10/17/99
America's love affair with antidepressant drugs is peculiar indeed. The big three - Prozac, Zoloft and Paxil - rank among the six best-selling drugs of any kind, and more than 130 million prescriptions were written for all antidepressants last year, according to IMS Health, a consulting firm. But these drugs are much less effective than many consumers and doctors believe. The newer antidepressant drugs post only marginal advantages over placebos in clinical trials for major depression, and cause frequent and unpleasant side effects.
Today's antidepressant drugs should be less appealing than the mood-altering drugs that preceded them: A century ago, so many medicines were laced with opium that we imported the equivalent of 15 doses to every man, woman, and child in the country. No sooner had opium been curbed with early narcotics laws that cocaine became popular. It was the original stimulant in Coca-Cola and the darling of Sigmund Freud.
Starting in the 1930s, barbiturates became the drug of choice and eventually more than 30 different drugs were on the market. In the 1950s came the next boom, amphetamines, which were prescribed for everything from depression to weight loss. This was followed by Valium and Librium and the tranquilizer decade.
All these drugs altered moods and perception in a fundamentally pleasant way. They produce outright mood elevation (opium), heightened energy and concentration (amphetamines), or provided a calming effect (barbiturates and tranquilizers). It is not hard to imagine why people wanted to take them. However, by no stretch of the imagination are antidepressant drugs pleasant to take for most patients. Consider Prozac, the drug that began the antidepressant boom because it supposedly eliminated many of the side effects that limited the popularity of its predecessors .
Prozac, according to Eli Lilly, causes rashes or hives in 7 percent of the people who take it, insomnia in 20 percent, sedation in 13 percent, tremors in 10 percent, nausea in 23 percent, diarrhea in 12 percent, and loss of appetite in 12 percent. Because it was not well studied during testing, sexual dysfunction was understated in the official ranking. Depending on definitions, from 10 to 70 percent find their sex lives diminished. Still, Prozac is by no means uniquely unpleasant. While there are differences, a similar pattern of frequent adverse effects can be found among all the popular antidepressants. Psychiatrists, of course, know this. Here's what an important text, Synopsis of Psychiatry, recommends to doctors: "It is almost always good practice to explain side effects in detail but to emphasize that they are signs the drug is working." But is the drug really working? Is it producing important benefits in relieving depression - benefits great enough to offset the side effects? Eli Lilly, its manufacturer, declares that Prozac is, indeed, "significantly more effective than placebo." But how much more effective?
According to the official record at the Food and Drug Administration, the benefits are surprisingly similar to the effects of a placebo.
To get FDA approval, manufacturers conduct FDA-supervised clinical trials in which some patients get a placebo and others the active drug. Neither patient nor investigator is supposed to know who is taking the drug. Lilly had conducted 10 such clinical trials for Prozac, according to FDA records. However, in six of these trials no measurable overall difference could be detected between those treated with Prozac and those who got the placebo. Prozac was usually ahead slightly, but within a margin that often could be explained by chance.
One trial deemed successful was based on such a small group of participants that only eight patients taking Prozac completed it. The FDA discarded another ''successful'' trial because no one could explain why one investigator got so much better results than others giving the same drug to similar patients.
If one combines all the Prozac studies in FDA files, it becomes clear that nearly 90 percent of the improvements reported by patients taking Prozac were also reported by the patients taking placebos.
Prozac is not the only antidepressant to post marginal benefits in clinical trials. Failure to produce a measurable effect is a routine event in the testing of drugs for depression. Serzone, for example, was tested in eight clinical trials, and produced a measurable benefit in only two. Another drug, Paxil, was tested six times in identical protocols and failed to produce a measurable effect in three of them.
It's possible that a drug effect existed in these many failed experiments that was too small to be measured without a bigger study. However, such a tiny effect is also too small to be of clinical interest.
It's also possible that the drugs really don't work in depression and that the small effects seen are the results of other flaws in the clinical trials.
So how did so many doctors and patients get persuaded that these drugs are so much more effective than the impartial and official scientific record demonstrates? Here are some of the most important reasons:
Striking anecdotes. We all love those stories about people whose lives were transformed by a drug. Some may describe a rare but real benefit. In other cases, spontaneous improvement gets mistaken for a drug effect. Peter Kramer filled an entire bestseller,Listening to Prozac, with fetching anecdotes he concedes do not portray the usual effects. But how many readers absorbed this fine print?
Misleading statistics. Many scientific reports and medical textbooks cite response rates that are inflated in one of several ways. In most studies, a responder is someone whom the investigator judged to be much improved during one weekly visit. But included as responders were people who improved in one visit but later dropped out either because of adverse effects, lack of efficacy, or both. Another approach is to ignore those who drop out. One published trial of Effexor proclaimed that 90 percent of "completers" had responded even though 36 percent abandoned the drug for adverse reactions, lack of effect or other problems before the trial ended.
Selective publication of trial results. A recent federal study of 315 published clinical trials of 29 antidepressant drugs noted that every one which identified a sponsor came from a drug company. While the FDA requires companies to submit failed trials, reports of drugs with no effect seldom make it to the medical journals.
For example, one published trial of Serzone with a marginal result matches the FDA records. But an identical trial with no measurable drug effect did not get published, according to the federal study. How many drug companies are going to publish a study that makes their drug look bad?
Doctors' overestimates of benefits. A few studies submitted to the FDA included the patient's evaluation as well as the doctor's opinion. In most cases, patient reports indicated no difference between drug and placebo, even when doctors thought they observed improvement.
For example, in two identical trials of Paxil, medical investigators thought more Paxil patients improved, compared with those on a placebo. But according to the patients' own evaluations, there was no difference between treatment and placebo. In a third identical study, neither evaluation showed a drug effect. Because doctors want to help people, it leads them to believe drugs are working better than they do.
Misleading psychoactive effects. Even though the measured effects on depression are marginal, these are still powerful psychoactive drugs that change how people feel in many ways. Many patients report a lack of emotion or being more detached from problems and everyday events that otherwise might trouble them. Others get jittery, lose a lot of weight or break into sweats. If changes do occur for the better, the drug may mistakenly be credited.
Giving the wrong credit for improvement. Over six weeks of treatment, a majority of patients do improve, according to the trials conducted for FDA approval. As measured by the Hamilton Depression scale, the scores drop about 40 percent.
While far from a cure, this is a marked improvement. But as noted, patients getting a placebo improved by a nearly similar amount. That means the improvement comes for some other reason. What a pity so many give undeserved credit to the drug, and not to their own efforts.
These factors - along with the billions spent on savvy drug marketing - partly explain why millions of people and thousands of doctors came to embrace a family of psychoactive drugs with marginal benefits and numerous adverse effects.
But it is still troubling to discover how easily we ended up spending billions of dollars on these questionable drugs. Maybe before our next love affair with a new group of psychoactive drugs, doctors and patients alike will take a more critical look at what they actually accomplish.