Published on 25 September 2026
Cognitive behavioural therapy alone for stimulant use disorders: reading an odds ratio of 2.88 alongside its certainty
The essentials
This systematic review and meta-analysis (protocol registered in PROSPERO) included 9 randomised trials, 8 of them (849 participants) in the meta-analysis, comparing stand-alone cognitive behavioural therapy (CBT), without any other structured psychosocial intervention, with minimal treatment (usual care, waiting list, attention placebo) in people with a stimulant use disorder. Short-term abstinence (4 to 24 weeks) was more frequent with CBT: odds ratio 2.88 (95% CI 1.08 to 7.70), with substantial heterogeneity (I² 75.62%) and low GRADE certainty. Dropout was comparable and no adverse event attributable to CBT was reported. The authors conclude that CBT alone may increase short-term abstinence, and that this estimate should be interpreted with caution.
Context
According to the authors, there is no approved pharmacotherapy for stimulant use disorders; psychosocial interventions are the foundation of management, and contingency management has the strongest evidence, but its implementation faces organisational and ethical obstacles.
Earlier reviews often combined CBT with other interventions or compared it with usual care that contained active elements. This review isolates trials of stand-alone CBT against minimal treatment to estimate its own contribution. The pooled result is read together with its interval, its heterogeneity and its certainty.
The study at a glance
| Item | |
|---|---|
| Population | |
| People of any age with a stimulant use disorder (DSM or ICD criteria, or regular or dependent use); non-dependent recreational use excluded. In the included trials: cocaine (5), amphetamine (2), methamphetamine (2), all outpatient, 7 of 9 in Western countries, mean age most often between 30 and 40 years | |
| Intervention | |
| Stand-alone, structured, manualised CBT targeting stimulant use, without any other structured psychosocial intervention; individual or group, face-to-face or digital; from 2 to 48 sessions. Several trials took place in patients who were otherwise receiving methadone or diacetylmorphine | |
| Comparator | |
| Minimal treatment: usual care, no treatment (waiting list, assessment only) or attention placebo. In Table 1 of the publication: drug education, self-help booklet, waiting list, standard methadone, placebo added to background treatment | |
| Primary outcome | |
| Short-term abstinence from the targeted stimulant, binary outcome (abstinent or not), assessed at each trial’s own primary endpoint or at the first post-treatment follow-up (4 to 24 weeks); urine testing preferred, intention-to-treat analysis, participants lost to follow-up counted as not abstinent. Secondary outcomes: treatment dropout, adverse events | |
| Design | |
| Systematic review and meta-analysis (PRISMA 2020, PROSPERO protocol CRD420251012327); search of PubMed, Embase, PsycINFO and the Cochrane Library up to 15 May 2025, randomised trials published in English; 9 trials included, 8 in the meta-analysis (849 participants: 457 CBT, 392 controls); random-effects model (REML) | |
Quality check
| Item | Verdict |
|---|---|
| Registration and reporting | Registered, PRISMA 2020 |
| FindingPROSPERO protocol CRD420251012327 and PRISMA 2020 reporting. The analyses by region, by number of sessions, by verification method and the analysis without the zero-event trials are post hoc. | |
| Selection and extraction | Double reading, third reviewer if disagreement |
| FindingSelection and extraction by two independent reviewers, with a third reviewer in case of disagreement. For one trial, values were extracted from a figure using digitising software. | |
| Risk of bias | 4 of 8 trials at high risk |
| FindingCochrane RoB 2 tool. High risk for four trials, mainly because of missing data (domain 3) and, for two of them, an unblinded self-reported outcome (domain 4). | |
| Heterogeneity | Substantial, post hoc explanation |
| FindingI² 75.62%. The authors attribute it to two trials with no events in the control arm, in a post hoc analysis. | |
| Publication bias | Not detected, low power |
| FindingEgger’s test not significant (p = 0.28); with 8 trials, the authors stress that this result does not prove the absence of bias. | |
| Certainty of evidence | Low |
| FindingLow GRADE: downgraded one level for risk of bias and one level for imprecision. The authors did not downgrade for inconsistency. | |
| Funding and conflicts of interest | Public, no conflict declared |
| FindingGrant from the Ministry of Health and Welfare of the Republic of Korea (Mental Health related Social Problem Solving Project, RS-2024-00421277), with no role of the funder in the study. The authors declare no commercial or financial relationships that could be construed as a potential conflict of interest. | |
Results
| Item | Value |
|---|---|
| Short-term abstinence | OR 2.88 |
| Finding145/457 (31.7%) with CBT versus 42/392 (10.7%) with minimal treatment; 95% CI 1.08 to 7.70; p = 0.04; that is 150 more abstinent participants per 1,000 (from 8 to 373 more), according to the authors’ GRADE table. | |
| Heterogeneity | I² 75.62% |
| FindingSubstantial heterogeneity according to the authors. | |
| Certainty of evidence | GRADE low |
| FindingDowngraded for risk of bias and imprecision. | |
| Treatment dropout | OR 1.13 |
| Finding95% CI 0.67 to 1.91; 8 trials; 99/465 with CBT versus 77/389 with minimal treatment. These figures and the GRADE judgement for this outcome (low, imprecision and suspected publication bias) appear in the supplementary material. | |
| Adverse events | None reported |
| FindingNo trial reported an adverse event attributable to CBT; monitoring procedures varied across trials and were not specified in four of them. | |
| Analysis with per-protocol data | OR 3.39 |
| Finding95% CI 1.12 to 10.22. | |
| Analysis restricted to point abstinence at end of treatment | OR 5.55 |
| Finding95% CI 1.72 to 17.89; 5 trials. | |
| Analysis without the 4 trials at high risk of bias | OR 8.72 |
| Finding95% CI 1.45 to 52.63; I² 80.04%. | |
| Post hoc analysis without the 2 trials with zero events in the control arm | OR 2.22 |
| Finding95% CI 1.44 to 3.42; I² 0.00%. This result is identical to that of the non-Asia subgroup. | |
| Subgroup by stimulant | p = 0.05 |
| FindingMethamphetamine or amphetamine: OR 9.14 (1.69 to 49.35), 4 trials, 598 participants. Cocaine: OR 1.26 (0.42 to 3.72), 4 trials, 251 participants. Difference at the borderline of significance, which the authors judge partly artefactual. | |
| Subgroup by region (post hoc) | p < 0.001 |
| FindingAsia: OR 70.44 (9.57 to 518.42), 2 trials. Non-Asia: OR 2.22 (1.44 to 3.42), 6 trials. The authors see this as an artefact of the two zero-event trials, not a region-specific effect. | |
| Subgroups by format and number of sessions | No difference |
| FindingIndividual versus group: p = 0.29. Eight sessions or fewer (OR 3.07; 1.88 to 5.02) versus more than eight (OR 1.93; 0.19 to 19.41): p = 0.70. | |
| Longer-term outcomes | No conclusion |
| FindingFour trials provided follow-up beyond the planned period; the authors judge that the durations and outcomes are too disparate to allow any conclusion on durability. | |
Critical appraisal
| Item | Verdict |
|---|---|
| Conduct of the review | Solid |
| FindingRegistered protocol, double extraction, RoB 2, GRADE, limitations declared by the authors (English language, published trials, few trials). | |
| Heterogeneity | Substantial, post hoc explanation |
| FindingThe disappearance of heterogeneity after removal of two zero-event trials (one of which, Shakiba et al., rests on an assumed zero because no data were reported) relies on a post hoc analysis. It does not prove the absence of an effect modifier, and region and stimulant are confounded in these two trials. | |
| Precision | Needs qualifying |
| FindingThe lower bound of the interval (1.08) is close to 1 and p = 0.04; the authors judge the imprecision serious (few trials, few events, continuity correction). | |
| Comparator | Needs qualifying |
| FindingControls range from waiting list to placebo added to background treatment; several trials included patients receiving methadone or diacetylmorphine. The effect of CBT against another psychotherapy is not estimated, since the corresponding trials were excluded by design. | |
| Scope | Short term, mostly Western trials |
| FindingThe primary outcome covers 4 to 24 weeks; 7 of 9 trials come from Western countries and transposition to Asia is not established. | |
Level of evidence
This is a meta-analysis of randomised trials whose GRADE certainty is low, downgraded for risk of bias (4 of 8 trials at high risk) and for imprecision. The authors did not downgrade certainty for inconsistency, the heterogeneity being, in their view, linked to two zero-event trials.
The conduct of the review is rigorous. Confidence is limited on the size of the effect, whose interval is wide, on its extension to cocaine (non-significant subgroup) and on its durability, which was not assessed quantitatively.
The colleague test
What an experienced colleague would say if you presented this study in two minutes, between two consultations.
“CBT alone is an accessible option when provision is limited: short-term abstinence is more frequent, with an odds ratio of 2.88, and dropout is comparable. But the interval is wide, the certainty is low, and the effect on cocaine has not been shown. I offer it and I measure the outcome.”
Translation for practice: CBT can be offered as an accessible option, with systematic monitoring of outcomes, without guaranteeing its effect.
What you can do with this
- What you can understand: an odds ratio of 2.88 with a lower bound of 1.08, an I² of 75.62% and low certainty indicates an association with more short-term abstinence, to be interpreted with caution.
- What you can tell a patient: CBT may increase the chances of short-term abstinence, without yet knowing whether the effect lasts, and no adverse effect specific to CBT was reported in these trials.
- What you can do: offer manualised CBT, in particular when contingency management is not available, agreeing measurable goals with the patient (abstinence, ideally verified by urine testing) and reassessing at a set date.
- What you can teach: read an odds ratio together with its interval, heterogeneity and certainty, and notice that heterogeneity explained by a post hoc analysis remains a hypothesis.
- Practice setting in France: manualised CBT is available in addiction care, support and prevention centres (CSAPA), and there is no pharmacological authorisation for these disorders. The reasoning applies wherever no medication is authorised for this indication.
- The course of action is set out in the NICE decision tree for depression in adults.
Frequently asked questions
Is CBT effective for stimulant use disorders?
It may increase short-term abstinence, according to the authors: odds ratio 2.88 (95% CI 1.08 to 7.70), with substantial heterogeneity and low certainty. The longer-term effect is not established.
What does an I² of 75.62% mean?
That a large share of the variability between trial results is not explained by chance. The authors attribute it to two trials with no events in the control arm: without them, heterogeneity disappears and the odds ratio is 2.22 (1.44 to 3.42), but this analysis is post hoc.
Is there a medication for these disorders?
According to the authors, there is no approved pharmacotherapy. Contingency management is the best-supported psychosocial approach, but it is difficult to implement in many settings.
Is CBT alone better than another psychotherapy?
The review cannot say: the comparator is minimal treatment, and trials comparing CBT with other psychotherapies were excluded.
Do these results apply to all stimulants?
Not in a demonstrated way. The effect is significant for methamphetamine or amphetamine (OR 9.14; 1.69 to 49.35) but not for cocaine (OR 1.26; 0.42 to 3.72); the difference is at the borderline of significance (p = 0.05) and the authors judge it partly artefactual.
Annotated bibliography
Source study. Kim J, Kwak J, Jeong H, Kim NJ, Lee S-Y, Kim Y, Kim J, Han S, Chun H-r, Park KJ, Lee S-B, Kim G, Lee HK, Yim HW. Efficacy of cognitive behavioral therapy for stimulant use disorders: a systematic review and meta-analysis. Frontiers in Psychiatry. 2025;16:1695702. Published 10 November 2025. https://doi.org/10.3389/fpsyt.2025.1695702. PMID: 41293198. Funding: Mental Health related Social Problem Solving Project, Ministry of Health and Welfare of the Republic of Korea (grant RS-2024-00421277), with no role in design, analysis or writing. Conflicts of interest: the authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. Supplementary material: the text and tables of the supplementary material (appendices A to I, including the GRADE table for the dropout outcome) were consulted; supplementary figures S1 to S6 were not consulted, and no data appearing only there are used here.
Editorial collections
Tags
Verified on 25 September 2026 against the full text of the publication and its supplementary material where available. This analysis underwent an independent double reading. The English version was checked for conformity on 25 September 2026, against the figures of the French version and against the source. How we verify what we publish
This analysis is intended for healthcare professionals. It does not constitute a prescribing recommendation and does not replace individual clinical judgment.
Analysis from Psychiatry Evidence Base, evidence-based psychiatry, explained with rigor.