The Effect of a Single Dose of Intravenous Ketamine on Suicidal Ideation: A Systematic Review and Individual Participant Data Meta-Analysis

Suicide is a major public health problem and rates have risen despite available treatments. Earlier research has shown that subanesthetic doses of ketamine produce rapid antidepressant effects and there is preliminary evidence that ketamine may reduce suicidal thoughts, but no meta-analysis had specifically examined ketamine's effects on suicidal ideation at the individual participant level across single-dose, controlled intravenous trials. Wilkinson and colleagues conducted a systematic review and individual participant data (IPD) meta-analysis to determine whether a single intravenous ketamine infusion reduces suicidal ideation more rapidly than control treatments (saline or midazolam) in patients who had some suicidal ideation at baseline. The analysis focused on explicit suicide-item scores from clinician-administered and self-report depression scales up to 1 week after infusion, and sought to examine whether any antisuicidal effect was independent of change in overall depressive symptom severity.

The investigators searched MEDLINE for clinical trials published between 1 January 2000 and 15 November 2016 using terms related to ketamine and suicidality, screened systematic reviews and references, and contacted study authors to obtain individual participant data. Eligible studies were single-dose intravenous ketamine comparison trials using saline or midazolam controls; multi-dose trials were excluded. Data were obtained for 10 of 11 eligible trials. Because three trials conducted under a single NIH protocol were treated as one study for modelling, the analytic count was k=8 for some analyses. The IPD set included only participants who had suicidal ideation at baseline. Suicidal ideation was defined a priori using single items from four scales: MADRS item 10 (score >=2) and HAM-D item 3 (score >=1) for clinician-administered measures, and QIDS‑SR item 12 (score >=1) and BDI item 9 (score >=1) for self-report measures. Investigators provided suicide-item scores, overall depression scores (with suicide items removed in some models), treatment assignment, and covariates (age, gender, race, inpatient versus outpatient status, diagnosis, concomitant psychotropic medication). Wherever available, data were collected at baseline and postinfusion days 1, 2, 3, and 7. The authors note that their method for handling missing data is described in an online supplement; that method is not fully reported in the extracted text. A one-stage hierarchical (mixed-effects, multilevel) general linear model was used, with participants nested within studies. Primary models tested group (ketamine versus control), time, and group-by-time interactions, with baseline suicidal ideation held constant. Additional models adjusted for baseline covariates and, in a final set of analyses, adjusted for concurrent change in overall depressive symptoms to assess whether ketamine's antisuicidal effects were independent of antidepressant effects; in these analyses the suicide items were removed from the composite depression score. Dichotomous outcomes (being free of suicidal ideation) were analysed separately for clinician-administered and self-report scales. Effect sizes were calculated as Cohen's d for mean change from baseline to each postinfusion time point, and statistical significance was set at p<0.05 (two-tailed). For crossover trials only the first treatment period was analysed to avoid carryover effects.

From 153 citations, 11 eligible trials were identified; individual participant data were obtained for 10 trials. After treating three NIH trials conducted under a single protocol as one study for modelling purposes, the analyses were based on data from k=8 study units. The pooled IPD comprised 298 patients from the 10 trials, of whom 167 met the prespecified criteria for baseline suicidal ideation and were included in the primary analyses. Included patients had higher baseline depressive severity and were more likely to be taking concomitant psychotropic medications than those without baseline suicidal ideation. Within the included sample, baseline characteristics did not differ significantly between ketamine and control groups on age, gender, diagnosis, inpatient status, concomitant medications, or baseline MADRS score. Clinician-administered measures: Using the pooled clinician-administered suicide items (MADRS and HAM‑D), ketamine reduced suicidal ideation significantly more rapidly than control treatments, with a significant group-by-time interaction (total N=167, k=8; chi-square=50.6, p<0.001). This benefit persisted after adjustment for baseline covariates and no baseline variable significantly moderated the effect. Effect sizes (Cohen's d) for clinician-administered outcomes were moderate to large: day 1 d=0.85 (95% CI=0.53–1.17), day 2 d=0.85 (95% CI=0.52–1.17), day 3 d=0.67 (95% CI=0.35–0.99), and day 7 d=0.61 (95% CI=0.27–0.94). A greater proportion of ketamine-treated patients were free of suicidal ideation at days 1, 2, 3, and 7 by clinician ratings, with over half of participants reporting no suicidal ideation across postinfusion time points. The number needed to treat (NNT) for being free of suicidal ideation ranged from 3.1–4.0 for days 1–7. Meta-analytic heterogeneity was low. Self-report measures: On pooled self-report items (QIDS‑SR and BDI), ketamine also reduced suicidal ideation more rapidly than controls overall (N=144, k=8; chi-square=45.5, p<0.001). When examined separately, the QIDS‑SR showed a significant group-by-time interaction (chi-square=32.5, p<0.001) but the BDI did not (chi-square=8.34, p=0.08); sample sizes were smaller for each instrument. Self-report effect sizes were moderate to large: day 1 d=0.73 (95% CI=0.38–1.07), day 2 d=0.84 (95% CI=0.49–1.19), day 3 d=0.63 (95% CI=0.28–0.98), and day 7 d=0.48 (95% CI=0.12–0.83). Ketamine was associated with a significantly greater proportion free of suicidal ideation at days 1–3 by self-report but not at day 7; NNTs were approximately 3.2–5.0 for days 1–3 and 9.6 at day 7. Correlation and independence analyses: Changes in suicidal ideation and overall depressive symptom severity were strongly correlated at all time points. For clinician-administered measures, adjusted r-squared values for the association between change in suicidal ideation and change in depression severity ranged from about 0.37–0.46 across days 1–7 (all p<0.001). For self-report measures, adjusted r-squared values were lower but significant across the same time points. After adjusting for concurrent change in depressive symptoms (with suicide items removed from the composite), the time-by-treatment interaction remained significant for clinician-administered outcomes (chi-square=10.84, p=0.028) and for self-report outcomes (chi-square=13.19, p=0.010). Clinician-administered differences between groups remained significant at all postinfusion time points; self-report differences remained significant at days 1–3 but not at day 7. Durability and responder persistence: Among patients who had resolution of suicidal ideation by 24 hours (clinician-administered), 86.0% of ketamine-treated patients remained free of suicidal ideation at 1 week versus 52.9% of controls (chi-square=7.98, p=0.005). Using self-report measures, 89.2% of ketamine patients versus 42.9% of controls maintained resolution at 1 week (chi-square=12.1, p<0.001). Control condition comparison: When saline alone was the comparator (N=50 saline, N=93 ketamine), clinician-administered effect sizes were larger (day 1 d=0.90; day 2 d=0.90; day 3 d=0.82; day 7 d=0.69). When midazolam was used as the comparator (N=24 midazolam, N=93 ketamine), effect sizes were more modest (day 1 d=0.62; day 2 d=0.69; day 3 d=0.34; day 7 d=0.41). The linear mixed model did not detect a statistically significant difference between midazolam and saline control groups, but numbers for the midazolam arm were relatively small.

Wilkinson and colleagues interpret these pooled IPD results as evidence that a single intravenous ketamine infusion produces rapid reductions in suicidal ideation among patients who had some degree of suicidal thinking at baseline, with moderate to large effects observed within 24 hours and benefits persisting up to 1 week. The authors emphasise that change in depressive symptom severity was strongly correlated with change in suicidal ideation, but that ketamine's effect on suicidal ideation remained significant after adjusting for concurrent changes in mood symptoms; they therefore propose a partially independent antisuicidal effect rather than complete pseudospecificity. The findings are placed in clinical context by comparison with other treatments: for example, a higher proportion of patients were free of suicidal ideation at 24 hours after ketamine than in an open‑label ECT study at 1 week, but the investigators caution that populations and treatment contexts differ and that ECT produces sustained resolution with repeated treatments. The authors note that most included trials recruited patients with mood disorders and treatment resistance and generally excluded patients with imminent suicide risk, substance use disorders, or psychotic disorders, which limits generalisability. Key limitations acknowledged by the study team include modest overall sample sizes for some instrument-specific analyses (which may explain the null finding for the BDI), reliance on single-item measures of suicidal ideation rather than multidimensional suicidality assessments, short follow-up (≤7 days) that precludes conclusions about durability beyond 1 week, and the possibility of functional unblinding in saline-controlled trials because of ketamine's psychoactive effects. The analysis also cannot address whether reductions in suicidal ideation translate into reductions in suicidal behaviour or deaths. The authors recommend future research to address these gaps: trials recruiting diagnostically diverse and higher-risk samples, longer follow-up and repeated-dosing protocols, direct assessment of suicidal behaviours, exploration of combinations with established somatic or psychotherapeutic interventions, and studies clarifying optimal patient selection, dosing frequency, monitoring, and safety.

Using individual participant-level data from single-dose, controlled intravenous ketamine trials that included only participants with baseline suicidal ideation, the investigators found that a single ketamine infusion led to rapid reductions in suicidal thinking in more than half of patients within 24 hours and showed benefits up to 1 week. While the results suggest ketamine as a promising rapid-acting intervention for suicidal ideation, the authors stress that the evidence remains preliminary. They call for further research on long-term efficacy and safety, effects on suicidal behaviour, optimal clinical protocols, and careful weighing of ketamine's risks against its potential benefits before routine clinical implementation.