The Placebo Response in Anxiety Trials: How Large Is It and What Drives It?

When a new anxiety medication is tested, both the drug group and the placebo group typically improve. Often the placebo group improves substantially. Understanding what drives this improvement — and how large it is — matters for reading any treatment result honestly.

How large is the placebo response?

In generalized anxiety disorder (GAD) trials, placebo response rates are consistently high. A meta-analysis of 28 randomized placebo-controlled GAD trials found that the mean placebo response rate was about 38–42%, depending on how response was defined [1]. That means in many trials, four in ten people given an inert pill improved enough to be counted as responders.

The response on the drug was higher, but not by as much as many people expect. The advantage of the drug over placebo was often in the range of 10–15 percentage points. A treatment that “works” in a trial may work by a modest margin above a substantial baseline.

What is in the placebo response?

The term “placebo effect” is sometimes used loosely, but several distinct things contribute to improvement in a placebo group.

Expectation. People who believe they are receiving an effective treatment may feel and report less anxiety. This is a real psychological effect, not mere pretending. A dismantling trial in irritable bowel syndrome (a condition with parallels to anxiety in its reliance on self-report) separated the effects of assessment alone, a sham treatment and a sham treatment given with a warm therapeutic interaction. Each additional component added to the improvement, with the warm interaction contributing the most [3].

Therapeutic contact. Trial participants see a clinician regularly, have their symptoms taken seriously, and receive structured monitoring. This is more clinical contact than many anxious people receive in routine care.

Natural fluctuation. Anxiety waxes and wanes. People tend to enter trials when their symptoms are at a high point (that’s often what prompts them to seek treatment), so some improvement would happen regardless.

Regression to the mean. Relatedly, if people are selected because their scores exceed a threshold, some of those high scores reflect temporary fluctuation. On re-measurement, scores tend to drift back toward the average.

A Cochrane review tried to separate the genuine placebo effect from natural course by comparing placebo groups with no-treatment groups across 202 trials of various conditions. For subjective outcomes like pain and nausea, placebo interventions had a moderate effect. For observer-rated outcomes, the effect was smaller [4]. This suggests that expectation matters more for self-reported symptoms, which is exactly what anxiety questionnaires measure.

Why the placebo response has grown

Research on antidepressant trials has documented a trend: the placebo response rate has risen over time [6]. The same trend appears in anxiety trials. A meta-analysis found that later publication year predicted higher placebo response in GAD trials [1].

Several explanations have been proposed:

• Trial design changes. More clinic visits, longer trials and more attentive monitoring all increase contact, which itself helps.
• Participant selection. As the number of trials has grown, recruiting has become harder. Some participants enroll in multiple trials; others respond to advertisements and may have milder symptoms or different motivations than people seeking treatment in a clinic.
• Broader definitions. Diagnostic criteria and inclusion criteria have broadened over time, potentially enrolling people who are more likely to improve on their own.
• Reduced drug-placebo differences. Higher placebo response makes it harder to demonstrate a drug effect, which may be why an increasing proportion of trials “fail” (find no significant difference between drug and placebo) [5].

What this means for reading results

The placebo response doesn’t mean treatment is useless. It means the specific effect of a treatment — the part attributable to the active ingredient rather than everything else — is the difference between the treatment and placebo groups, not the total improvement in the treatment group.

When a press release says “70% of patients responded to the new drug,” the relevant comparison is what happened in the placebo group. If 42% responded to placebo, the specific benefit is closer to 28 percentage points. For how to evaluate whether that difference is large enough to matter, see understanding effect sizes.

Questions to ask about placebo in a trial

1. What was the response rate in the placebo group, and how does it compare with the drug group?
2. How was “response” defined, and was the same threshold used for both groups?
3. How many clinic visits did participants in both groups receive?
4. Were participants allowed to know which group they might be in (was blinding maintained)?
5. Was there a no-treatment group, allowing separation of placebo effects from natural improvement?

What remains uncertain

We don’t yet have good ways to predict who will respond to placebo and who needs specific treatment. Some researchers have explored whether high placebo responders should be identified and excluded from trials, but this raises ethical and practical problems. What drives the placebo response in anxiety specifically, as opposed to pain or depression, has received less direct study. And whether the therapeutic contact in a trial could be delivered more efficiently as part of routine care — capturing some of the “placebo” benefit without a trial — is an open question.

Reading treatment results. The placebo response doesn’t mean your improvement on a treatment isn’t real. It means a proper evaluation requires knowing how much of that improvement would have happened without the specific treatment. If you’re considering a change in treatment, talk to your prescriber. If anxiety is affecting your daily life, a GP or mental health professional can help. If you are in crisis or thinking about harming yourself, contact emergency services or a crisis line now: in the US, call or text 988; in the UK and Ireland, Samaritans on 116 123.