Published on 19 September 2026
After a successful course of ECT, which relapse-prevention strategies actually hold up?
The essentials
Twenty-eight studies, 11,119 patients, and a question every clinician asks without a trial that fully answers it: what comes next after a course of ECT that has worked. This systematic review catalogues the strategies that have been studied and ranks them by the strength of the evidence behind them. Two options stand out with the most consistent support: lithium, alone or combined with an antidepressant, and continuation ECT combined with pharmacotherapy. Psychotherapy emerges as a promising but underpowered avenue. No study meeting the inclusion criteria evaluated repetitive transcranial magnetic stimulation, ketamine or esketamine for this indication: the authors found only a four-case series and unpublished registered trials. They performed no statistical pooling, first because no new randomized data had appeared since the review by Youssef and McCall, then because of heterogeneity in designs and relapse definitions across studies. The review maps a field. It does not quantify an effect.
Context
A successful course of ECT does not resolve the question of follow-up care, it relocates it. Patients typically improve quickly, and the risk of relapse in the months that follow remains the main clinical difficulty of this treatment. The authors note that roughly half of patients relapse within twelve months of an effective course of ECT, most often within the first six months. The literature on the subject is scattered: a handful of older trials, many cohorts, relapse definitions that vary from one study to the next, and populations that mix patients in full remission with patients who only responded.
The value of this work therefore lies not in producing an estimate but in bringing order to the field. It answers a practical question: among everything that has been tried, what actually rests on solid ground?
The study at a glance
| Population | |
| Adults aged 18 and older with a unipolar major depressive episode, responders or in remission after acute ECT, with at least three months of follow-up. Patients with bipolar disorder and other primary diagnoses were excluded, as were mixed diagnostic groups whose unipolar subgroup was not analyzed separately. Total of 11,119 patients across 28 studies. | |
| Interventions covered | |
| Any relapse-prevention strategy: pharmacotherapy, including lithium and antidepressants, continuation or maintenance ECT, psychotherapy, and combinations of these approaches. | |
| Comparators | |
| Placebo, usual care, or a direct comparison between two strategies, depending on the included study. | |
| Outcomes | |
| Primary: relapse rate after successful ECT. Secondary: time to relapse, relapse risk expressed for example as a hazard ratio, and descriptive data. | |
| Design | |
| Systematic review of mixed study designs, with narrative synthesis and no meta-analysis. Protocol registered on PROSPERO under number CRD42023392575 on January 28, 2023, reported per PRISMA. Risk of bias assessed with RoB 2 for randomized trials and methodological quality with the MMAT, 2018 version, for all designs, by two independent reviewers. | |
| Selection | |
| PubMed, Embase, PsychInfo and the Cochrane Library searched up to March 1, 2025, with no publication-date limit. 17,115 records identified, 11,351 titles and abstracts screened after removing duplicates, 117 full texts assessed, 28 studies retained. Randomized trials, prospective and retrospective cohorts, and case series of more than ten cases were eligible; no case series was ultimately included. |
Results
The authors present their conclusions as an explicit gradation, from the best-supported strategies to the most experimental. This gradation is the article’s main product, and it is also what makes it usable in consultation.
| Strategy | Strength of evidence according to the authors |
|---|---|
| Antidepressant combined with lithium, first line | Most consistent evidence base |
| FindingThis is the most extensively documented option, with data from both older trials and large registries. The authors’ grading table draws on seventeen pharmacotherapy studies, including two randomized trials and five cohorts specifically addressing lithium. The article’s own abstract, meanwhile, cites ten studies favoring lithium: the two counts do not match. Consistency does not mean certainty; a substantial share of this evidence is observational. | |
| Continuation ECT combined with pharmacotherapy | Most consistent evidence base |
| FindingFour randomized trials and one prospective cohort show lower relapse rates than pharmacotherapy alone. The authors place it second line, after failure of the first option, or as an immediate choice when relapse risk is high or a relapse has been severe. They note its logistical burden and resource cost. | |
| Continuation ECT without structured pharmacotherapy | Emerging lead |
| FindingOne randomized trial places it on a par with pharmacotherapy, while one retrospective cohort finds no benefit. The evidence base is too thin to settle the question. | |
| Psychotherapy, as an adjunct | Emerging lead |
| FindingOnly two studies: a randomized trial of 60 responders and a prospective cohort of 8 patients. The findings are encouraging but scarce. The level of evidence does not allow this strategy to be ranked alongside the first two. | |
| Long-term spaced maintenance ECT | Limited data |
| FindingTwo retrospective cohorts and one prospective cohort suggest a possible benefit, which the authors judge more likely after age 65. The strategy is used in practice without the literature specifying how. | |
| Repetitive transcranial magnetic stimulation, ketamine and esketamine | Preliminary and experimental |
| FindingNo study met the inclusion criteria. The authors cite only a four-case series of magnetic stimulation used as a bridge after ECT, with response maintained for 6 to 13 months under a concurrent antidepressant, and three unpublished registered trials. This silence in the literature says nothing about the efficacy of these approaches for relapse prevention after ECT; it says that the question has not yet been addressed in a usable study design. | |
A few numerical results are cited by the authors for illustration, not pooled. In the three randomized trials with a placebo arm, relapse rates under placebo range from 67% to 84% over follow-up periods of 18 to 24 weeks. The historic 2001 trial, conducted in 84 patients in remission and followed for 24 weeks, reports 39% relapse under a combination of a tricyclic antidepressant and lithium, 60% under that antidepressant alone, and 84% under placebo. A registry study of 7,350 patients reports a relative reduction in readmissions of about 16%, corresponding to a hazard ratio of 0.84, among patients who had filled a lithium prescription after ECT, with no suicides in that group. These figures come from individual studies with different methods and populations: they illustrate the direction of the signal, they do not add up.
Quality control
| Point checked | Verdict |
|---|---|
| Protocol registered before the analysis | Yes |
| FindingPROSPERO registration CRD42023392575 on January 28, 2023, reported per PRISMA, data extraction begun on April 29, 2025. The search strategy, the list of included studies and the extraction tables are provided as supplementary material. | |
| Risk-of-bias assessment | Dual tool |
| FindingRoB 2 for randomized trials and MMAT for all designs, which is appropriate for a mixed body of evidence. Recount of the supplied assessment grids: of the 13 randomized trials, 6 are at low risk of bias, 5 raise some concerns and 2 are at high risk. Across the 28 studies, MMAT scores break down as 6 studies at 5 out of 5, 12 at 4 out of 5, 5 at 3 out of 5 and 5 at 2 out of 5. | |
| Statistical pooling | Absent |
| FindingNo meta-analysis. The authors give two successive justifications: no new randomized data since the review by Youssef and McCall, then heterogeneity in designs and relapse definitions. The result is the same either way: the article produces no pooled estimate and no summary confidence interval. | |
| Nature of the included body of evidence | About half observational |
| FindingCount from the extraction tables: 13 randomized trials, 8 non-randomized prospective cohorts and 7 retrospective cohorts. No case series was included, although the protocol allowed them above ten cases. The ranking of strategies therefore reflects the volume of available literature as much as the quality of the study designs. | |
| Definition of relapse | Heterogeneous |
| FindingThe included studies do not define relapse in the same way: rating-scale thresholds, clinical judgment, rehospitalization, suicide or resumption of ECT, depending on the study. They also mix patients in full remission with patients who only responded. These are two clinical situations with different baseline risk, and in a responder, relapse in the strict sense cannot occur, since remission was never reached. | |
| Publication bias | Not tested |
| FindingNo formal test, neither a funnel plot nor Egger’s test, which follows logically from the absence of pooling. The authors did, however, search trial registries and the grey literature, and they themselves conclude that there is a risk of publication bias because of registered results that were never published, particularly for newer interventions. This is therefore not a finding of an absence of bias. | |
| Internal consistency of the figures | Internal discrepancies |
| FindingSeveral discrepancies remain between the text, the tables and the appendices of the publication. The database-by-database counts in the flow diagram total 17,505 records against the 17,115 stated. The sum of the sample sizes across the 28 studies gives 11,120 patients against the 11,119 stated. The narrative text cites 15% relapse under imipramine where the table and the extraction appendix indicate 18%, and 35% relapse in one combination trial where those same sources indicate 47.1%. The neuromodulation section states four studies where the table lists only three. None of these discrepancies changes the ranking of strategies, but they are reason enough not to cite any figure from this article without checking the primary study. | |
| Funding and conflicts of interest | None declared |
| FindingThe publication states that no funding was received for this work and declares no conflicts of interest. Declarations verified against the published text. | |
Critical appraisal
| Domain | Judgment |
|---|---|
| Comprehensiveness and grading | Solid |
| FindingEvery strategy the review sets out to cover is addressed, and the grading is honest: the authors explicitly distinguish what is consistent, what is emerging, what is limited and what remains experimental, rather than presenting everything at the same level. | |
| Absence of a pooled estimate | Major limitation |
| FindingWithout pooling, there is no summary effect size, no summary confidence interval, and no number needed to treat. The only estimates available are those of individual studies, reproduced as such, for example a hazard ratio of 0.21, 95% confidence interval 0.05 to 0.84, p = 0.027, for lithium in a prospective cohort of 47 patients in remission. The reader gets a ranking, not a magnitude. This is what places the level of evidence of this work below that of a meta-analysis of trials. | |
| Reported sample size | Misleading if taken alone |
| FindingThree registry studies account for 9,710 of the 11,119 patients, or 87%. A sample of this size does not improve protection against the confounding and selection biases inherent to observational data, and registries record neither dosage, nor adherence, nor relapses that do not lead to hospitalization. | |
| Identification of gaps | Useful |
| FindingThe absence of eligible studies on transcranial magnetic stimulation and on ketamine for this indication is a finding in its own right, and it is correctly presented as a gap rather than as an unfavorable signal. The authors also flag ongoing registered trials, which clearly distinguishes an unexplored field from one that has been explored without result. | |
Level of evidence
Confidence is high in the description of the field: the search was registered, systematic and well documented, and the ranking of strategies matches that of earlier reviews the authors themselves cite, the 2013 review by Jelovac and colleagues, the 2014 review by Youssef and McCall, the 2014 review by Brown and colleagues, and the 2021 review by Lambrichts and colleagues. Confidence is low for any quantitative translation. An article that pools nothing cannot support a sentence of the form “this strategy reduces the risk by this much.”
What is demonstrated: that the eligible literature concentrates on lithium and on continuation ECT, and that it contains no usable study of transcranial magnetic stimulation or ketamine for this indication. What is suggested: that these two leading options are indeed the most effective, a hypothesis strengthened by their convergence with earlier reviews and with the included randomized trials. What amounts to expert opinion: the proposed ordering between strategies, which rests on a synthesis judgment rather than a statistical comparison, as well as the first-line status given to the antidepressant-lithium combination, which the authors justify as much by feasibility as by the data.
The colleague test
What an experienced colleague would say if shown this study in two minutes, between two consultations.
This is a good map of the terrain. Lithium and continuation ECT with medication, those are the two pillars, and psychotherapy is coming up behind them. What strikes me is what is missing: no magnetic stimulation, no ketamine, nobody has published that trial yet. But there is not a single summary number in this article that I can quote to a patient.
Translation for practice: this article is useful for setting an order of options and for not forgetting one, not for putting a number on a benefit in front of a patient. For a number on continuation ECT, go back to the randomized trials themselves.
What you can do with this
- Set an explicit order of options after a successful course of ECT, rather than defaulting to whatever treatment the patient is already on.
- Reconsider lithium in a patient who has just responded to ECT, taking into account renal and thyroid function and the feasibility of biological monitoring. The authors themselves make their first-line recommendation conditional on this monitoring and on tolerability.
- Do not conclude that transcranial magnetic stimulation or ketamine are ineffective in this situation: the literature has not evaluated them, which is different.
- Keep in mind that the included studies mix full remission with partial response. A patient still symptomatic at discharge is not in the same risk situation as one who is doing well.
- Weigh the cognitive burden when deciding whether to continue ECT sessions: the authors note that available reviews do not find consistent cognitive decline, but that the quality of the data is low and most studies relied on a screening test now considered insufficiently sensitive.
- Verify the availability, in your own country, of the molecules cited in the reference literature before transposing a published protocol.
Frequently asked questions
Why did the authors not perform a meta-analysis?
They give two reasons. The first, stated in the methods, is that no new randomized data had been identified since the systematic review by Youssef and McCall. The second, stated in the discussion, is heterogeneity in study designs and relapse definitions. The included studies genuinely do not measure the same thing in the same way: designs range from randomized trials to national registries, and populations mix response with remission. Pooling this evidence would have produced a single figure that no one could say what it actually measured.
Is lithium the best option?
It is the option with the most consistent evidence, which is not the same thing. Much of that evidence is observational, and no direct comparison with the other strategies is pooled here. The authors place it first line in combination with an antidepressant, citing its feasibility relative to continuation ECT as well as the data. The decision remains individual.
What about a patient with bipolar disorder?
This review excludes them, on the grounds that the course, pathophysiology and prevention recommendations differ from those of unipolar depression. Its conclusions do not apply to them, and it would be incorrect to extend them. The question of relapse prevention after ECT in bipolar disorder remains open in this work.
Can psychotherapy be offered as a follow-on treatment?
The authors classify it as an emerging lead, based on a single randomized trial of 60 responders and a prospective cohort of 8 patients. That is not much. Nothing prevents offering it in combination, provided it is not presented as equivalent to the best-documented options. The authors note that cognitive behavioral therapy had to be adapted for this population, with more structure, repetition and written materials, to account for the cognitive difficulties that follow ECT.
What does the absence of studies on ketamine mean here?
Only that no study meeting the inclusion criteria was found for this specific indication. Trials are registered and not yet published, which the authors note. This is neither a negative result, nor a contraindication, nor an argument for or against its use in other situations.
Annotated bibliography
Source study. Rovers JJE, van Zeijl NT, Tendolkar I, Dols A, van Eijndhoven PFP. Systematic review on relapse-prevention strategies following successful electroconvulsive therapy for major depressive disorder. BJPsych Open 2026;12:e37, pages 1 to 14. DOI 10.1192/bjo.2025.10946 · PMID 41530963. Received October 28, 2025, revised December 4, 2025, accepted December 5, 2025. Open access under a Creative Commons Attribution 4.0 licence.
Earlier reviews cited by the authors. Jelovac A, Kolshus E, McLoughlin DM. Relapse following successful electroconvulsive therapy for major depression: a meta-analysis. Neuropsychopharmacology 2013;38:2467–2474. Youssef NA, McCall WV. Relapse prevention after index electroconvulsive therapy in treatment-resistant depression. Annals of Clinical Psychiatry 2014;26:288–296. Brown ED, Lee H, Scott D, Cummings GG. Efficacy of continuation/maintenance electroconvulsive therapy for the prevention of recurrence of a major depressive episode in adults with unipolar depression: a systematic review. Journal of ECT 2014;30:195–202. Lambrichts S, Detraux J, Vansteelandt K, Nordenskjöld A, Obbels J, Schrijvers D, et al. Does lithium prevent relapse following successful electroconvulsive therapy for major depression? A systematic review and meta-analysis. Acta Psychiatrica Scandinavica 2021;143:294–306. These are the works that supply the numerical estimates the present review does not produce: one ranks, the others measure.
The two individual studies cited above. Sackeim HA, Haskett RF, Mulsant BH, Thase ME, Mann JJ, Pettinati HM, et al. Continuation pharmacotherapy in the prevention of relapse following electroconvulsive therapy: a randomized controlled trial. JAMA 2001;285:1299–1307. Brus O, Cao Y, Hammar Å, Landén M, Lundberg J, Nordanskog P, et al. Lithium for suicide and readmission prevention after electroconvulsive therapy for unipolar depression: population-based register study. BJPsych Open 2019;5:e46. The figures reproduced above come from the review and its extraction table, and are best checked against the original publications.
Regulatory context. This review is Dutch, and it does not document the regulatory status, in the reader’s own country, of the molecules it discusses, lithium in particular. Marketing authorization, actual commercial availability and reimbursement are three distinct questions: a molecule described in the literature of one country, France among others, is not necessarily available through all three of these channels elsewhere. Readers should check these three points for their own jurisdiction before transposing a protocol drawn from this review.
Editorial collections
Tags
Verified on August 29, 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 September 19, 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.