Published on 17 September 2026
Lithium or Quetiapine for Augmentation: The First Twelve-Month Head-to-Head Trial
Health Technology Assessment · 2025; 29(12): 1 to 118 · Kerr-Gaffney et al.
DOI 10.3310/YQVF5347
PMID 40350798
Scientific 65
Editorial 78
Key findings
The LQD trial randomised 212 adults with treatment-resistant depression to lithium augmentation or quetiapine augmentation across six English NHS trusts, with 52 weeks of follow-up. On symptoms, quetiapine came out ahead of lithium: the area under the curve of differences was -68.36 (95% CI -129.95 to -6.76), p = 0.0296, which the authors translate into a gap of roughly two QIDS-SR points at one year. On the second co-primary outcome, time to treatment discontinuation, the difference did not reach significance: hazard ratio 0.72 (95% CI 0.47 to 1.09), p = 0.1196. The blinded-rater depression scale points the same way at 52 weeks, but does not survive the authors’ correction for multiple testing. The economic analysis concludes that quetiapine is more cost-effective, from both the NHS and the societal perspective. The trial remained unblinded and its symptom outcome was self-rated, which argues for describing a probable advantage rather than an established superiority.
Background
Once two antidepressants have failed, the next question is almost always the same: what do you add? Lithium and quetiapine both sit in the guidelines, backed by evidence from short trials run mostly against placebo. The main previous head-to-head comparison lasted six weeks and concluded non-inferiority. LQD was designed to fill that gap, funded publicly by the National Institute for Health and Care Research, award reference 14/222/02, with no industry funding for the trial itself. The individual conflicts of interest declared by the authors are, by contrast, numerous, including consultancy fees, grants and share holdings, and are detailed at the end of this analysis.
The choice, in the consulting room, is not decided by efficacy alone. Lithium requires regular blood monitoring, renal and thyroid vigilance, and patient education about dehydration. Quetiapine requires metabolic monitoring and carries a risk of sedation and weight gain. A trial that puts the two side by side over a full year answers a question that no eight-week trial can ask.
The study at a glance
| Element | Content |
|---|---|
| Population | Adults with treatment-resistant depression, NHS England |
| DetailDSM-5 criteria, 17-item Hamilton Depression Rating Scale score of at least 14 at screening, inadequate response to at least two antidepressant trials at an adequate dose in the current episode, current antidepressant at a therapeutic dose for at least six weeks. Psychosis and bipolar disorder excluded. N = 212, of whom 107 in the quetiapine arm and 105 in the lithium arm. Mean age 42.4 years, 54.2% female, 88.7% of patients of self-reported White ethnicity. | |
| Arms compared | Lithium augmentation versus quetiapine augmentation |
| DetailRandomisation 1 to 1 on the decision to prescribe, in variable-sized blocks via a centralised web-based service, stratified by region, depression severity and severity of resistance, across six NHS trusts in England. After the pre-prescribing safety checks, 95 of the 107 quetiapine patients received and started treatment, against 86 prescriptions and 84 initiations out of 105 in the lithium arm. | |
| Co-primary outcomes | Two joint outcomes |
| DetailDepressive symptoms over 52 weeks, measured weekly with the self-rated Quick Inventory of Depressive Symptomatology (QIDS-SR). And time to all-cause discontinuation of the assigned treatment. Each is tested at a two-sided 5% threshold, with no splitting of the type I error rate between the two. | |
| Design | Pragmatic, open-label, multicentre randomised controlled trial |
| DetailOxford level of evidence 1b for the design, an individually randomised phase 4 trial, with the caveat of an open-label design. Integrated economic analysis, from an NHS and Personal Social Services perspective and from a societal perspective. Registration ISRCTN16387615, protocol published in 2017, 118-page monograph freely available. | |
The results
The p-value for the symptom outcome. A result that clears the threshold without much room to spare, on an outcome completed by patients who knew which treatment they were taking.
| Outcome | Value |
|---|---|
| Symptoms over 52 weeks | Area under the curve of differences -68.36 (95% CI -129.95 to -6.76), p = 0.0296 |
| InterpretationIn favour of quetiapine. The metric cumulates the gap over the whole year and is not self-explanatory, but the authors translate it: at 52 weeks, mean QIDS-SR was 12.1 in the quetiapine arm versus 14.1 in the lithium arm, a two-point gap the authors regard as clinically meaningful between two active treatments. | |
| Time to discontinuation | Median 365 days versus 212 days, hazard ratio 0.72 (95% CI 0.47 to 1.09), p = 0.1196 |
| InterpretationNot significant. Among patients who actually started treatment, 38.9% discontinued under quetiapine versus 50.0% under lithium, or 34.6% and 40.0% when expressed against the randomised numbers. This outcome does not depend on blinding, and it does not confirm the first one. | |
| Cost-effectiveness | In favour of quetiapine in both perspectives |
| InterpretationFrom the NHS perspective, quetiapine dominates: a lower cost of £472.32 (95% CI -£1,111.12 to £166.47) and a gain of 0.074 QALYs, with a 0.99 probability of being the cost-effective option at the £20,000 per QALY threshold used by NICE. From the societal perspective, it costs more but remains cost-effective, with a probability of 0.91. The economic results inherit the same limitations as the clinical results they are derived from. | |
| Tolerability | 32 serious adverse events, 1 judged possibly related to treatment |
| InterpretationThese 32 events involved 18 patients, 15 events in the quetiapine arm and 17 in the lithium arm. The single event judged possibly related occurred under lithium. Weight, blood pressure and the side-effect rating scale did not differ between arms, but the specific adverse effects that led to discontinuation were not collected in detail, and a trial of this size cannot distinguish between the two tolerability profiles. | |
Critical appraisal
| Domain | Judgement |
|---|---|
| Randomisation | Low risk |
| FindingRandomisation in variable-sized blocks, sequence managed by a clinical trials unit’s web-based service, stratification on three factors. Trial registered as ISRCTN16387615, protocol published in 2017 ahead of the analysis, 118-page monograph freely available. Transparency here sits above the average for the field. | |
| Blinding and measurement | High risk |
| FindingThis is the point that governs the whole reading of the trial. An open-label trial whose main symptom outcome is completed by the patient stacks two sources of expectancy that pull in the same direction. Two safeguards exist, however: the MADRS and the Clinical Global Impression scale were rated by blinded assessors, and the statisticians were blinded too. The MADRS at 52 weeks points the same way, a 2.98-point difference in favour of quetiapine (95% CI -5.87 to -0.09, p = 0.0435), but no secondary outcome survives the authors’ correction for multiple testing at a 5% or 10% false discovery rate. | |
| Missing data | Mixed |
| FindingFor time to discontinuation, a single missing value, or 0.5% of the sample. For the QIDS-SR, 19.6% and 16.2% missing at 8 weeks, 25.2% and 39.0% at 26 weeks, 28.0% and 35.2% at 52 weeks, quetiapine then lithium, less than the 40% anticipated in the sample size calculation. Missingness is, however, substantial on several secondary outcomes and more frequent in the lithium arm, which opens a risk of measurement bias that the authors flag themselves. Mixed models under a missing-at-random assumption, with sensitivity analyses exploring departures from that assumption. | |
| Precision | Wide interval |
| FindingThe confidence bound closest to zero is -6.76, against a point estimate of -68.36. With two co-primary outcomes and only one reached, the robustness of the signal remains limited. The sensitivity analysis nonetheless shows the result only flips under an assumed decline of at least 0.2 QIDS-SR points in the quetiapine arm alone, an assumption the authors consider strong. | |
| Match between conclusion and data | Qualified |
| FindingThe plain-language summary is conditional, quetiapine may be the more effective option, and the limitations are stated, absence of blinding foremost among them. The scientific abstract is more assertive and speaks of a superiority of quetiapine on symptoms and on cost-effectiveness, whereas only one of the two co-primary outcomes was met. | |
Level of evidence
What can be taken as established: both strategies are usable over a year, with a high discontinuation rate in both arms, 38.9% under quetiapine and 50.0% under lithium among those who start treatment, which is in itself information worth sharing in a consultation. What can be taken as probable: a symptomatic advantage for quetiapine over twelve months, consistent between the self-rated outcome, the blinded-rated scale and social functioning. What cannot be taken as demonstrated: that this advantage is robust enough to overturn a first-line choice. One co-primary outcome reached and the other not, in an open-label trial with a self-rated primary outcome, with secondary outcomes that do not survive correction for multiple testing, amounts to a serious signal, not definitive proof.
And the symmetry applies both ways. The absence of a difference in time to discontinuation does not prove the two strategies are equivalent on that measure: the trial was only powered to detect a large gap, 70% versus 50% of patients still on treatment at one year, and the observed gap was smaller than that.
The colleague test
What an experienced colleague might say if you described this trial to them in two minutes, between two consultations.
“At last someone has compared them over a full year, that alone is significant. That said, open-label trial, patient-completed questionnaire, and the hard outcome, stopping treatment, does not move. The blinded-rated scale points the same way, but it does not hold up under the statistical correction. I would still choose based on the patient’s profile and on what they will accept to be monitored for.”
Translated for practice: the trial tips one way without settling the question, and it makes the choice better informed. Above all, it gives a usable order of magnitude for the one-year discontinuation rate, a figure worth mentioning to the patient when deciding.
What clinicians can do with this
- Tell the patient, from the outset, that roughly four in ten patients on quetiapine and one in two on lithium stop their augmentation within the year among those who start it. This information changes the conversation.
- Do not switch first-line choice on the strength of this trial alone. The signal in favour of quetiapine is consistent across measures, but it rests on an open-label design and is not accompanied by a significant advantage in treatment retention.
- Keep weighing the usual factors: history, metabolic comorbidities, renal and thyroid function, tolerance for sedation, and capacity to sustain blood monitoring.
- Check the authorisation status before prescribing. In the UK context of this trial, extended-release quetiapine is the only atypical antipsychotic licensed for augmentation in treatment-resistant depression, a reminder that licensing, market availability and reimbursement are three separate questions, each of which should be checked against the rules in force wherever you practise.
- Remember that the trial’s economic case is calculated within the NHS, in pounds sterling and against the £20,000 per QALY threshold used by NICE. It does not transfer as is to other healthcare systems.
Frequently asked questions
Should quetiapine now be preferred over lithium?
Not on this basis alone. One of the two co-primary outcomes was met, the other was not. The blinded clinician-rated assessment points the same way at one year, but it does not survive the correction for multiple testing, and the trial was open-label.
Why was the trial not blinded?
Because the trial is pragmatic: the authors chose an open-label design to reflect real-world practice as closely as possible. Lithium requires repeated blood level monitoring, with a target of 0.4 to 1.0 mmol/L early in treatment, which makes blinding over twelve months difficult to sustain in practice. The trade-off is the one just described.
What should be taken from the discontinuation rate?
That it is high in both arms, and it may be the single most directly usable result of the trial. Adverse effects and an insufficient clinical response are the two main reasons behind it. An augmentation that is not sustained is not an augmentation.
Does the cost-effectiveness result apply outside the NHS?
Not directly. Drug prices, monitoring arrangements, hospitalisation costs and willingness-to-pay thresholds all differ between healthcare systems. The method is informative, the figure itself is not transferable.
Annotated bibliography
Source study. Kerr-Gaffney J, Zenasni Z, Goldsmith K, Yaziji N, Jin H, Colasanti A, Geddes J, Kessler D, McAllister-Williams RH, Young AH, Barrera A, Marwood L, Taylor RW, Tee H, Cleare AJ, for the LQD Study Group. Clinical and cost-effectiveness of lithium versus quetiapine augmentation for treatment-resistant depression in adults: LQD a pragmatic randomised controlled trial. Health Technology Assessment, 2025, volume 29, issue 12, pages 1 to 118. DOI 10.3310/YQVF5347. PMID 40350798, record established in the PubMed registry on 1 September 2026. Funded by the National Institute for Health and Care Research, Health Technology Assessment programme, award reference 14/222/02, with no industry funding for the trial. Registration ISRCTN16387615. Protocol published in 2017 in BMC Psychiatry by Marwood et al. Conflicts of interest declared by the authors: AstraZeneca and GSK shareholdings for the first author, consultancy or speaker fees from several pharmaceutical companies paid to two of the senior authors, two authors who are employees and shareholders of COMPASS Pathways, and multiple public and charitable grants for the others.
Editorial collections
Tags
This analysis is intended for healthcare professionals. It does not replace individual clinical judgement or current guidelines, and does not constitute therapeutic advice.
Psychiatry Evidence Base. Evidence-based psychiatry, explained with rigour.