Therapy Dropout Rates in Anxiety Treatment: What the Numbers Mean

Treatment for anxiety can work. But a substantial number of people who start treatment don’t finish it, and this fact has important implications for how we read evidence and think about treatment in practice.

How common is dropout?

A review of attrition in anxiety treatment RCTs found that roughly 15–25% of participants leave before completing the treatment phase [1]. A broader meta-analysis across all adult psychotherapy found an average dropout rate of about 20% [3]. For CBT specifically, a meta-analysis estimated that about one in five people discontinue prematurely [4].

Medication trials show similar or higher attrition, often 25–35%, with side effects as a major driver. Some of this is captured in discontinuation studies, where the gap between “started” and “completed” is a key but often underreported number.

These are trial figures. In routine clinical practice, where there is less structure, less monitoring and more real-world barriers, dropout rates are typically higher [2]. The treatments shown to work in trials may be working in a somewhat selected group of people who manage to complete them.

Why it matters for reading evidence

When a study reports that 60% of CBT completers responded to treatment, it sounds encouraging. But if 20% dropped out before completing, the real question is: what happened to them? Several scenarios are possible:

They improved early. Some people leave because they feel better and decide they don’t need more sessions. If this is common, dropout inflates the denominator of non-responders.

They didn’t improve or got worse. If people who weren’t benefiting disproportionately dropped out, the completer analysis overstates efficacy because it’s only counting the people who stuck around — a form of survivorship bias.

Practical barriers. Cost, scheduling, distance and life events cause dropout regardless of treatment response. These dropouts may be randomly distributed with respect to outcome.

In practice, the second scenario — differential dropout biased toward non-responders — is a serious concern. Intent-to-treat (ITT) analyses, which include all randomized participants, partially address this by using the last available data or modeling missing outcomes. But ITT analyses make assumptions about the missing data that may or may not be correct. For how this affects long-term follow-up, the problem compounds over time.

What predicts dropout

Several factors consistently predict treatment discontinuation:

Severity. More severely affected individuals are somewhat more likely to drop out, though the relationship is not strong or consistent across studies.

Comorbidity. People with comorbid conditions — particularly depression, substance use or personality disorders — have higher dropout rates.

Early response. People who don’t see early improvement (within the first 3–4 sessions of therapy or the first 2–4 weeks of medication) are more likely to leave. This is rational from the patient’s perspective but creates a systematic bias in the evidence.

Treatment factors. Exposure-based components of CBT can be distressing, and some people discontinue because the treatment itself is uncomfortable. Side effects drive medication dropout. The therapeutic relationship matters — poor alliance predicts dropout.

Practical barriers. In clinical practice, cost, transport, time off work and childcare are significant drivers that don’t appear in most trial analyses.

Dropout and the intention to treat

The distinction between efficacy and effectiveness is partly about dropout:

Efficacy — does the treatment work under ideal conditions? This is typically what trials measure, often emphasizing completers.

Effectiveness — does the treatment work in the real world, including for people who struggle to complete it? This requires accounting for dropout.

A treatment can be efficacious (it works for people who complete it) without being effective for a significant proportion of people who start it. Understanding this gap is important for realistic expectations.

Questions to ask about dropout in a study

1. What percentage of participants completed the treatment?
2. Was the primary analysis intent-to-treat or completer-only?
3. Were reasons for dropout reported?
4. Did the study compare completers with dropouts on baseline characteristics?
5. How does the trial dropout rate compare with what happens in routine clinical practice?

What remains uncertain

Why some people drop out and others persist is not well predicted by baseline measures. Whether shorter or more flexible treatments could reduce dropout without sacrificing efficacy is an important practical question with limited evidence. Whether dropout is better understood as treatment failure, rational decision-making, or system failure (barriers to access) probably varies between individuals and needs more nuanced study.

If you’ve stopped treatment. Leaving therapy doesn’t mean you’ve failed. People stop for many reasons — some valid, some worth reconsidering. If you stopped because it wasn’t working, a different therapist or approach might help. If practical barriers got in the way, online options exist. If anxiety is affecting your daily life, talk to a GP or mental health professional. 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.