Ketamine vs Electroconvulsive Therapy for Major Depressive Episode: A Systematic Review and Meta-analysis

Electroconvulsive therapy (ECT) is regarded as a criterion-standard antidepressant for major depressive episode (MDE) but its use is constrained by stigma, cognitive adverse effects, and uneven availability. Ketamine, an N‑methyl‑D‑aspartate (NMDA) receptor antagonist, has emerged as a rapidly acting antidepressant with a different adverse‑effect profile, and several randomised clinical trials (RCTs) have directly compared ketamine and ECT in adults with MDE. A prior meta-analysis concluded that ECT was superior to ketamine for antidepressant effect but found no clear differences in cognitive or serious adverse events; the present study seeks to update and extend that quantitative synthesis. Menon and colleagues conducted a systematic review and meta-analysis to compare ketamine and ECT in patients with MDE. Beyond post‑treatment depression ratings, the investigators aimed to synthesise trial-defined response and remission rates, the number of treatment sessions required to achieve response or remission, and adverse events, thereby informing clinical and research decision‑making about the relative roles of ketamine and ECT.

Menon and colleagues registered a protocol and performed a systematic review and meta-analysis limited to parallel‑group RCTs published in English that directly compared ketamine and ECT in adults with MDE. Searches of MEDLINE, ScienceDirect, and Google Scholar were conducted from database inception to 31 May 2022 using MeSH and free‑text terms for depression, bipolar depression, ECT, and ketamine. The full search strategy and additional materials were reported in supplements to the paper. The primary outcome was depression rating at approximately 1 week after treatment end point. Secondary outcomes included study-defined response and remission rates, the number of treatment sessions required to attain response and remission, and adverse effects. Effect sizes were pooled using random‑effects meta-analysis. Continuous outcomes were summarised as standardised mean differences (SMD; Hedges’ g) with 95% CIs, and categorical outcomes as risk ratios (RR) with 95% CIs. The investigators chose a random‑effects model because they anticipated clinical and methodological heterogeneity across settings and populations. Trial-level risk of bias was appraised with version 2 of the Cochrane risk‑of‑bias tool for RCTs. Two reviewers independently performed the search, screening, data extraction, and quality assessment, with disagreements resolved by consensus or by consulting additional senior authors. The study team attempted to obtain additional data from original trial authors; only two trial teams provided the requested data.

Five parallel‑group RCTs met the inclusion criteria, yielding a pooled sample of 278 participants. In the primary analysis of depression ratings at about 1 week posttreatment (5 RCTs; n = 278), there was a nonsignificant trend favouring ECT over ketamine (SMD, -0.39; 95% CI, -0.81 to 0.02; I2 = 45%). A sensitivity analysis that excluded three trials judged to have questionable methods and reporting produced a statistically significant advantage for ECT on the same primary outcome: SMD, -0.45 (95% CI, -0.75 to -0.14; I2 = 6%; 2 RCTs; n = 211). For study‑defined response, pooled results from three trials (n = 229) favoured ECT (RR, 1.27; 95% CI, 1.06–1.53; I2 = 0%). For study‑defined remission, two trials (n = 211) again favoured ECT (RR, 1.43; 95% CI, 1.12–1.82; I2 = 0%). Analyses of the number of treatment sessions required to reach response (2 RCTs; n = 139) and remission (2 RCTs; n = 119) yielded SMDs that numerically favoured ECT (response: SMD 0.68, 95% CI -0.80 to 2.16, I2 = 87%; remission: SMD 0.57, 95% CI -0.65 to 1.79, I2 = 78%), but confidence intervals were wide and estimates were not statistically significant. Posttreatment cognitive scores showed little difference between groups (SMD, 0.13; 95% CI, -0.57 to 0.83; I2 = 1%; 2 RCTs; n = 34). Adverse‑event comparisons indicated fewer headaches (RR, 0.38; 95% CI, 0.18–0.79; I2 = 73%; 3 RCTs; n = 230) and less musculoskeletal pain (RR, 0.25; 95% CI, 0.15–0.42; I2 = 0%; 3 RCTs; n = 230) with ketamine versus ECT. Conversely, dissociative symptoms were substantially more common with ketamine (RR, 6.45; 95% CI, 1.92–21.64; I2 = 32%; 3 RCTs; n = 230). Risk‑of‑bias assessment identified multiple concerns: four trials had high or unclear risk related to randomisation processes; all five trials were at high risk for deviations from intended interventions because of lack of blinding of participants and personnel; three trials had high risk due to differential missing outcome data; two trials had high risk in outcome measurement due to nonblinded raters; and two trials were judged at high risk for selective reporting because they were unregistered or retrospectively registered.

Across five RCTs (N = 278), ECT was not associated with a statistically significant advantage over ketamine for depression scores at about 1 week in the primary pooled analysis, but a sensitivity analysis excluding trials with questionable methods showed a significant and small‑to‑medium advantage for ECT. The investigators emphasised that effect sizes were modest and that heterogeneity was small to moderate in the principal analyses. Menon and colleagues highlighted that ECT produced significantly higher response and remission rates than ketamine, whereas there was no clear evidence that ECT led to faster onset of response or remission. Cognitive outcomes did not differ appreciably between treatments in the limited trials reporting these data. Adverse‑event profiles differed: ketamine was associated with fewer headaches and musculoskeletal pains but with markedly more dissociative symptoms. The authors positioned their findings as broadly consistent with prior meta‑analytic work but suggested that the superiority of ECT over ketamine may be smaller than previously reported. They therefore recommended that a trial of ketamine could reasonably be attempted before ECT in patients with MDE, while acknowledging that this guidance rests on a small evidence base drawn from few trials and participants. Key limitations acknowledged by the study team included the small number of trials and participants, multiple sources of risk of bias across included RCTs, heterogeneity in methods and reporting, and incomplete availability of trial data. The investigators also noted uncertainty about differential effectiveness in bipolar depression versus major depressive disorder and urged that future trials report outcomes separately for these diagnostic subgroups to enhance the interpretability of future meta‑analyses.