Peter D Kramer

In Practice


No, Placebo Response Rates Are Not on the Rise

A new study undercuts the notion that antidepressant effects are placebo effects

Posted Oct 13, 2016

Katefreedman/Wikimedia Commons
Source: Katefreedman/Wikimedia Commons

Commonly, patients who respond to an antidepressant ask whether they’re merely experiencing a placebo effect. What they may not know is that the placebo has fallen on hard times. Progressively, steadily, depression researchers have seen a weakening in the case for classic placebo effects, the argument that when antidepressants work, the engine is hopeful expectation.

The latest damage comes from a reassessment of trends in drug research. 14 years ago, researchers at Columbia made note of a disturbing pattern. In antidepressant trials, placebo response rates had been rising steadily, year by year, decade by decade. Where, in 1980, perhaps 20 percent of participants on dummy pills showed substantial improvement, by 2000, the rate was 30 percent. Later, response rates of 35 or 40 percent became common.

What was causing the change? For observers who put stock in classic placebo effects, the answer was simple. As antidepressants gained acceptance—as more people came to believe that they work—the mere fact of taking pills stimulated improvement arising from faith in their powers.

But what if placebo response rates have not been on the rise after all? We might conclude that increased acceptance of antidepressants does not result in greater efficacy for dummy pills. We would want to say that the classic placebo response, the one grounded in hopeful expectations, is not powerful after all.

Sure enough, an international team of researchers has just provided a new, more detailed and comprehensive analysis of antidepressant studies. It finds that, at core, placebo response rates have been stable for 25 years and more.

Yes, in the raw data, dummy pill effects look to have been on the rise—but that’s because of changes in the way the trials have been conducted. Comparing like to like—looking at similar studies—the impact of dummy pills has not budged.

Here are some details. In drug trials’ control arms (where patients get dummy pills), between 1978 and 1990 response rates rose steadily. In 1991, just when new medications, like Prozac and Zoloft, became popular—which is to say, just as the public became more aware of antidepressants—the curve flattened. Again, when we look individually at each sort of trial, recent studies show no more powerful placebo effects than ones conducted ten or twenty years ago.

The new overview detected two factors that predict higher response rates in people not receiving medication: longer trials and multi-center trials. More weeks of participation mean more robust responses. And single-site trials show lower placebo response rates than do trials conducted in many clinics simultaneously.

Why? In the course of trials that last longer, more patients will enjoy spontaneous remissions. And as time spent with staff members increases, so will implicit psychotherapy effects. Earlier commentators on this sort of finding concluded, “Rather than patient expectancy, what appeared to influence treatment response was the amount of therapeutic contact patients received.”

As for more numerous test sites, using them is an indicator that participants have been hard to recruit. No single clinic attracts enough patients. Also, commercial trials, run for drug companies, are likelier to be multi-site than are trials run by scholars in academic settings. “Multi-site” may be a proxy for poor quality.

Here’s what I think happened. In the 1970s and1980s, the National Institute of Mental Health modeled standards for adequate emotional support in antidepressant trials. As researchers, guided by ethics panels, adopted those standards, response rates for patients on placebo rose.

After 1991, what changed was the quality of patient samples. It deteriorated. People with engaged physicians were getting treatment and recovering. As recruitment for trials became difficult, socially marginalized patients played a greater role, and commercial centers came to dominate drug assessment.

That’s what I concluded in my recent book, Ordinarily Well: “[D]rug trials don’t show much in the way of classic placebo effects. The rise in placebo responses over the years is more likely due to the supportive factors in drug trials…and increasing problems with enrollment.”

The new finding—no upward trend in placebo responses—is unexpected and certain to be contested. Meanwhile, it stands as a rebuke to a popular narrative. By that account, drug effects had been hyped, expectations soared, and the inflated hopes were reflected in rising placebo response rates.

The new publication suggests that two factors have caused changes in the outcomes researchers observe: increasing emotional support and decreasing ability to recruit representative patients. Pharma’s intrusion into the evaluation process appears to have distorted the assessment process as well.

What has disappeared from the equation is the classic placebo effect. The notion that depression is highly placebo responsive, and more particularly the notion that antidepressant effects are basically placebo effects, has become ever less plausible and ever harder to defend.