Published on 25 September 2026

Analysis · Treatment-resistant depression · Health economics

Pramipexole augmentation in treatment-resistant depression: what the PAX-D economic evaluation shows

▬ Publication The Lancet Regional Health - Europe · 2026 · 61: 101533 · Łaszewska et al. DOI: 10.1016/j.lanepe.2025.101533 PMID 41333541 Scientific 71 Editorial 56

Editorial collections A Study Under the Microscope

The essentials

The PAX-D trial randomised 150 adults with treatment-resistant depression, across nine UK National Health Service sites, to pramipexole or placebo added to their current treatment. This economic evaluation, conducted within the trial, covers 124 patients (61 on pramipexole, 63 on placebo) with complete data at baseline and at 12 weeks. From the health and social care perspective, the cost per quality-adjusted life year (QALY) gained is £5,069 at 12 weeks and £9,007 at 48 weeks, with a probability above 90% of being cost-effective at the £20,000 threshold. The confidence intervals of the ratios are wide (-£3,642 to £35,608 at 12 weeks, £2,219 to £27,258 at 48 weeks). The 48-week result depends on how missing data are handled: in the complete-case analysis (n=56), the cost per QALY rises to £20,840, the probability falls to 54% and the confidence interval of the QALY gain includes zero. Pramipexole has no authorisation for depression: this analysis describes efficiency within one health system, not a prescribing option.

Context

After several antidepressant failures, several options exist (switching or combining antidepressants, augmentation, ketamine, esketamine, neurostimulation), but according to the authors data on their cost-effectiveness are limited. Pramipexole, which enhances dopaminergic activity, was tested against placebo in the PAX-D trial; the 12-week clinical results were published separately (Browning et al., 2025), with adverse effects (nausea, headache, sleep disturbance). The authors identified no cost-effectiveness evaluation of pramipexole in treatment-resistant depression; their literature search ends in July 2025. The economic question is separate from the efficacy question: a treatment can work and still be too expensive, or cheap for a modest gain.

An economic evaluation conducted within a randomised trial benefits from randomisation, but it still depends on trial size, on the tariffs of the health system where it took place, and on missing data.

The study at a glance

Item
Population
Adults aged 18 or over, major depressive episode (DSM-5 criteria), QIDS-SR16 above 10, taking an antidepressant, with failure of at least two adequate antidepressant treatments; nine UK National Health Service sites; recruitment from January 2021 to May 2024
Intervention
Pramipexole added to the current antidepressant, started at 0.25 mg then titrated every 3 days up to 2.5 mg over 4 weeks (immediate-release form, one evening dose), with dose reduction possible in case of adverse effects
Comparator
Identical placebo augmentation
Outcomes
Primary outcome: incremental cost-effectiveness ratio (cost per QALY, EQ-5D-5L); secondary outcomes: full capability years (ICECAP-A) and capability-weighted life years (OxCAP-MH); health and social care perspective, then societal perspective; 12- and 48-week horizons; thresholds of £20,000 to £30,000 per QALY
Design
Cost-effectiveness evaluation conducted within a double-blind randomised trial (1:1 allocation); 124 patients analysed out of 150 randomised (intention-to-treat analysis of patients with complete data at 12 weeks); 2022/2023 costs in pounds sterling

Quality check

Item	Verdict
Methodological framework	Standards met
FindingNICE reference case (health technology evaluation manual); CHEERS 2022 checklist provided in the supplementary material.
Missing data	Reservations
FindingMultiple imputation by chained equations (25 imputations); missing data reach 22% (supplementary material), with a comparable profile across arms. The missing-at-random assumption cannot be verified, according to the authors. Discontinuation of study treatment differs between arms (20% on pramipexole versus 5% on placebo), whereas the authors judge study withdrawals to be comparable.
Uncertainty	Explored
FindingBootstrapping, acceptability curves, multiple sensitivity analyses.
Comparator	Placebo only
FindingNo comparison with another augmentation strategy.
Transferability	Limited
FindingTariffs and thresholds specific to the UK health system.
Conflicts of interest	Read the declarations
FindingFunding: National Institute for Health and Care Research, Efficacy and Mechanism Evaluation programme, with no role in design, data collection, analysis, interpretation or writing. Ties with industry (notably Compass Pathways, Otsuka, Janssen, Alto Neuroscience, Boehringer) are declared by several co-authors.

Results

£5,069

Cost per QALY gained at 12 weeks (ΔQALY 0.012), health and social care perspective. At 48 weeks: £9,007 (ΔQALY 0.090).

Item	Value
12 weeks, health and social care	£5,069 per QALY
FindingΔQALY 0.012 (0.003 to 0.021); confidence interval of the ratio -£3,642 to £35,608; probability of being cost-effective above 90% at £20,000.
48 weeks, health and social care	£9,007 per QALY
FindingΔQALY 0.090 (0.036 to 0.144); confidence interval of the ratio £2,219 to £27,258; probability above 90% at £20,000.
48 weeks, societal perspective	Dominant on average
FindingCosts £255 lower on average in the pramipexole arm (interval -£1,621 to £1,111), with major imprecision; the authors recommend caution.
48 weeks, complete cases (n=56)	£20,840 per QALY
FindingΔQALY 0.038 (-0.050 to 0.126); confidence interval of the ratio -£166,588 to £207,354; probability of being cost-effective reduced to 54% at £20,000.
48 weeks, per protocol (n=103)	£9,065 per QALY
FindingΔQALY 0.092 (0.033 to 0.151); probability of being cost-effective of 88% at £20,000 (supplementary material).

Critical appraisal

Item	Verdict
Economic methodology	Sound
FindingEvaluation within a double-blind trial, NICE framework, multiple imputation, bootstrapping, acceptability curves, quality-of-life and capability measures (ICECAP-A, OxCAP-MH). The health economic analysis plan was published with the trial protocol.
Efficiency signal	Present
FindingPoint estimates of cost per QALY below the £20,000 threshold in the main analysis, with confidence intervals that exceed this threshold (up to £35,608 at 12 weeks, £27,258 at 48 weeks).
Fragility at 48 weeks	Needs qualifying
FindingUnstable complete-case results (QALY gain interval including zero), missing-at-random assumption that cannot be verified, differential treatment discontinuation, and a single participant accounting for 69% of the non-psychiatric hospitalisation costs in the pramipexole arm.
Transferability	Major reservation
FindingMolecule without authorisation for depression (could not be verified against the sources); UK tariffs and thresholds; placebo-only comparator.

Level of evidence

Scientific71/100

Editorial56/100

This is an economic evaluation conducted within a randomised trial; the level of evidence of 2 on the Oxford CEBM hierarchy is an editorial assessment, absent from the publication. Confidence is good in the methodological conduct and in the order of magnitude of efficiency within the UK health system, over the period of the trial.

It is low for anything beyond that frame: generalisation to another care system, duration beyond 48 weeks, and the 48-week result itself, which depends on how missing data are handled.

The colleague test

What an experienced colleague would say if you presented this study in two minutes, between two consultations.

“A serious economic evaluation, with a real efficiency signal in the UK health system. But this is a molecule with no authorisation for depression and costs I cannot transpose: a signal, not a prescribing instruction.”

Translation for practice: keep the method and the order of magnitude, and deduce neither a prescription nor reimbursement from it.

What you can do with this

What you can understand: a cost per QALY below the threshold says a treatment is efficient in a given system, at a given tariff, not that it is recommended.
What you can tell a patient: pramipexole has been studied as augmentation in treatment-resistant depression, the economic evaluation is favourable in the United Kingdom, but it has no authorisation for this indication.
What you can teach: how to read a cost per QALY together with its interval, its threshold, its horizon and the robustness of the complete-case analysis.
Practice setting in France: pramipexole has no marketing authorisation in depression in France and is not reimbursed for it there. Authorisation, marketing and reimbursement are three distinct matters, and the absence of reimbursement says nothing about the efficacy of the treatment. The same reasoning applies elsewhere: a treatment’s efficiency in one country does not amount to authorisation or coverage in another.
What you can watch for: trials comparing pramipexole directly with other augmentation strategies, which the authors call for, and any replication in another health system.

Frequently asked questions

Is pramipexole cost-effective in treatment-resistant depression?

In the UK health system, yes in the main analysis: £5,069 per QALY at 12 weeks and £9,007 at 48 weeks, below the usual thresholds. At 48 weeks, the complete-case analysis is less favourable (£20,840, probability of 54%, very wide intervals). The authors conclude that pramipexole is cost-effective, within this health system.

What is a QALY?

It is a quality-adjusted life year: a year in full health counts as 1, a year with impaired health counts as less. The cost per QALY is used to compare very different treatments.

Can this result be applied in France?

Not directly. The tariffs and thresholds are British, and pramipexole has no authorisation for depression in France.

Why is the 48-week result considered fragile?

Because 22% of data are missing (partly because participants recruited after September 2023 had shorter follow-up), because treatment discontinuation differs between arms (20% versus 5%), and because the analysis restricted to complete cases (n=56) changes the result. The authors nevertheless judge that the sensitivity analyses lead to similar conclusions.

Does the lack of reimbursement mean the treatment is ineffective?

No. Authorisation, marketing and reimbursement answer different questions, and a coverage decision does not say that a treatment is ineffective.

Annotated bibliography

Source study. Łaszewska A, Helter T, Baldwin A, Cleare AJ, Cowen PJ, Evans J, Huys QJM, Kurkar M, Lewis AC, Nixon N, Rastogi A, Watson S, Geddes JR, Browning M, Simon J. Cost-effectiveness of pramipexole augmentation for acute phase and maintenance therapy of treatment-resistant depression compared to placebo augmentation: economic evaluation of the PAX-D randomised controlled trial. The Lancet Regional Health - Europe. 2026;61:101533 (published online 17 November 2025). https://doi.org/10.1016/j.lanepe.2025.101533. PMID: 41333541.

Funding. National Institute for Health and Care Research (NIHR), Efficacy and Mechanism Evaluation programme (16/127/17); additional support from the NIHR Oxford Health Biomedical Research Centre, the Office for Life Sciences and the NIHR Mental Health Translational Research Collaboration. The funder had no role in design, data collection, analysis, interpretation or writing.

Conflicts of interest, as declared. AJC: grants (ADM Protexin, Beckley Psytech), consulting fees (Compass Pathways, Otsuka, Janssen), payment for presentations (Compass Pathways, Otsuka, Viatris, Medscape), president of the International Society for Affective Disorders. PJC: NIHR grants (EME programme). QJMH: support from the NIHR UCLH Biomedical Research Centre and the EME programme, grants (Wellcome Trust, NIHR, Carigest SA, Koa Health), consulting fees and options (Aya Technologies, Alto Neuroscience). ACL: salary funded by the NIHR EME programme. NN: grants (Wellcome Trust, NIHR, BlueSkeye AI), consulting fees (NICE Scientific), meeting attendance funded by Compass Pathways. MB: grants (NIHR and others, Wellcome Trust, Medical Research Council), consulting fees (Engrail Therapeutics, Jansen Research, Boehringer, CHDR, P1vital, Alto Neuroscience, Empyrean Therapeutics), former employee of P1vital. JS: NIHR support, grants (WWTF, European Commission, NIHR, FWF), payments or honoraria from the European Brain Council. The other authors declare no conflicts of interest.

Supplementary material. The supplementary file (appendix tables and figures) was provided. The tables were consulted after conversion to text; the content of appendix tables 3 and 4 was not converted and the figures (images) were not consulted. No data appearing only there are used here, apart from tables 9 to 11 read as text (sensitivity analyses: complete cases, per protocol).

Editorial collections

A Study Under the Microscope