A success rate can look reassuring while leaving important people outside the calculation: those who stopped attending, left treatment early, or could not be reached afterward. Their experiences are not all the same. But a report that does not explain their absence may describe the easiest patients to measure rather than everyone the service treated.
Key findings
19.7%: weighted estimate of adult-psychotherapy dropout in a 2012 meta-analysis.
669 studies and 83,834 clients: the evidence included in that review.
18.7% to 20.7%: the 95% confidence interval around the pooled estimate. This is historical research, not a 2026 national rate or a prediction for every clinic. [1]
44,957 patients: a 2026 NHS study illustrating that planned discharge and clinical recovery can move in different directions. [2]
Almost one in five, but not one universal rate
The interval describes uncertainty around the pooled estimate, not the range of results across all studies. Source: Swift and Greenberg, 2012. [1]
The review found variation by dropout definition, setting, diagnosis, and other study characteristics. Its value is in showing that premature discontinuation is substantial and measurable, not in supplying one timeless number for every treatment.
Its complement is not a recovery rate. Remaining in treatment does not establish that symptoms improved. Equally, leaving an episode early does not explain what happened next.
Four numbers that should never be merged
| Measure | What it shows | What remains unknown |
|---|---|---|
| Attendance | Appointments attended | Whether treatment helped |
| Completion | A defined episode was completed | Whether recovery occurred |
| Clinical outcome | Change on a specified health measure | What caused the change without a suitable study |
| Follow-up response | Who provided later information | How nonrespondents were doing |
A service may perform well on one measure and less well on another. That is not a contradiction. It is a reason to report them separately instead of calling the most favorable percentage a general success rate.
A family should also ask whether the measures describe the same people. A completion count from one year and an outcome survey from a different group do not automatically form a single patient pathway.
Engagement and recovery can move apart
A February 2026 observational study examined 44,957 patients across seven NHS Talking Therapies services. Physical long-term conditions were associated with higher odds of planned discharge but lower odds of reliable recovery. [2]
The finding does not prove that a physical condition caused a particular result. It demonstrates why a person staying engaged and a person meeting a clinical endpoint are distinct outcomes.
For private care, the implication is practical. A patient attending every session may still need reassessment, a different treatment component, or help with another condition. Completion deserves attention, but it cannot stand in for the clinical result.
Not every early ending means the same thing
A person may transfer, agree that another approach is more appropriate, stop for practical reasons, or decline further treatment. These are possible categories to investigate, not explanations to assign without evidence.
A useful record distinguishes an accepted transfer from unexplained loss of contact. It documents the reason where known and recognizes that the clinician’s view of completion may differ from the patient’s.
The solution is not to relabel every early departure as a planned ending. That would protect a percentage at the expense of useful information. Nor should every departure automatically be described as deterioration.
Annual totals do not necessarily track one group
NHS England’s 2025/26 annual report notes that referral, access, and completion totals do not all relate to the same referral cohort. Dividing one total by another can therefore fail to measure the dropout rate readers assume it represents. [3]
The same rule applies to private reporting. Does the report follow everyone admitted during a period, everyone discharged, or a fixed group observed over time? Each design can be useful, but the starting population must be clear.
A graphic showing stages of care should follow compatible people and periods. Otherwise, an apparent loss between stages may reflect different reporting groups rather than patients dropping out.
Missing follow-up is unknown, not a convenient result
Counting everyone lost to follow-up as recovered is misleading. Declaring every missing response a failure is also not an observed result.
A careful report states what was measured and examines how reasonable assumptions about missing outcomes affect the conclusion. The method should not be changed afterward to produce the most attractive percentage.
In a small private cohort, a few missing observations can materially change a rate. Exact counts and clear limitations may be more informative than an extra decimal place. Reporting also needs to protect people who might be identifiable in small subgroups.
For international clients, follow-up needs a practical plan
A person treated abroad may return home before follow-up is due. That creates a logistical question, not evidence of recovery or relapse. The pathway should specify who makes contact and what the receiving clinician has agreed to do.
For someone moving between the UK, the United States, Europe, or the Gulf, arrangements should reflect the relevant professional and legal requirements. An assumption that remote care works identically everywhere is not a care plan.
Financial resources may support travel and flexible appointments without removing every barrier to participation. Ask what arrangement the patient can realistically use, how consent is handled, and what happens after a missed contact.
Better engagement must not become pressure to stay
Reducing avoidable dropout is a reasonable service aim. It should not pressure someone to remain in a program regardless of clinical need or preference.
NICE recommends discussing treatment preferences, previous experiences, and barriers to engagement in depression care. That is more useful than assuming a person who leaves lacks motivation. [4]
Questions about timing, communication, expectations, and treatment fit can identify improvements. The patient’s explanation should inform the assessment. A retention target should support consent-based care rather than override it.
What families should see before admission
| Disclosure | Why it matters |
|---|---|
| Completion and dropout definitions | Prevents unlike rules from appearing comparable |
| Counts at each stage | Shows who contributes to each percentage |
| Reason for ending | Separates transfers, agreed endings, and lost contact |
| Clinical status at ending | Distinguishes attendance from improvement |
| Follow-up coverage | Shows how much of the later picture was observed |
| Missing-data approach | Explains uncertainty in the result |
The admissions discussion should also explain review points and circumstances in which the plan might change. Financial terms belong in a clear agreement, not in the clinical definition of successful completion.
A family member or payer may help, but that does not automatically create unrestricted access to clinical information. The person receiving care should know how concerns can be raised and how decisions will be made.
This checklist is a reporting proposal. It is not an audit of private providers or a THE BALANCE retention claim.
What the numbers cannot tell us
The historical 19.7% estimate is not the current rate for every therapy. The 2026 association does not establish causation. Neither supplies a private-clinic comparison. A proper provider evaluation would need a defined cohort, reconciled records, transparent exclusions, and consistent outcome measures.
The bottom line
Ask who is absent from the success rate. A service that explains its missing data may offer a more useful evidence record than one displaying a higher percentage without the underlying counts. The objective is appropriate treatment and honest follow-up, not keeping every patient in the original program at any cost.
For journalists
Key number: 19.7% weighted dropout in a 2012 review of 669 adult-psychotherapy studies.
Current comparison: a 2026 NHS study found that engagement and recovery could differ for patients with physical long-term conditions.
Suggested attribution: THE BALANCE analysis of published psychotherapy and outcome-reporting research.
Methodology and sources
This narrative analysis preserves the historical review’s date and confidence interval. It combines that evidence with a separate observational study and official guidance. No new pooled dropout estimate or provider score was calculated.
- Swift and Greenberg. Premature discontinuation in adult psychotherapy: a meta-analysis. 2012.
- Verbist and colleagues. Physical long-term conditions and psychological treatment outcomes in NHS Talking Therapies. 2026.
- NHS England. Talking Therapies annual report, 2025/26.
- NICE NG222. Depression in adults: treatment and management.
