Early suicidal ideation change and suicidality-associated emergency department utilization after ketamine or esketamine treatment in treatment-resistant depression

Abstract

Background:

Ketamine and esketamine ((es)ketamine) rapidly reduce depressive symptoms and suicidal ideation (SI) in treatment-resistant depression (TRD), but whether early SI improvement during the acute phase of the treatment predicts reduced suicide-associated emergency department (SAED) visits over the longer term remains unclear.

Objectives:

To examine whether early improvements in SI, measured using item 12 of the Quick Inventory of Depressive Symptomatology – Self Reported (QIDS-SR), following (es)ketamine treatment are associated with SAED visits over 6 months

Design:

Retrospective cohort study.

Methods:

Individuals with TRD who had available data on SI measured using the QIDS-SR during the acute phase, as well as data on SAED visits over 6 months, were included. Participants were classified based on changes in SI as resolved, partially improved, or unchanged/worsened using the QIDS-SR item 12 during acute treatment (up to six intravenous (IV) or eight intranasal (IN) treatments) and were followed for SAED visits.

Results:

Among 96 patients with SI data, 73 (76%) had baseline SI. Following acute-phase treatment, 36 (49%) achieved SI resolution, 14 (19%) partially improved, and 23 (32%) were unchanged/worsened. During the 6-month follow-up, 8 (11%) had post-acute SAED visits. While not statistically significant, patients with resolved SI had lower odds of post-acute ED visits (OR = 0.16, p = 0.12). Kaplan–Meier estimates of SAED visit-free survival across three SI change groups trended toward significance (p = 0.066), suggesting lower cumulative healthcare (ED) utilization in more favorable change groups. Secondary analysis among 57 patients who received IV ketamine and had baseline SI showed that, after two infusions, 33 (47.2%) demonstrated a reduction in SI. Early SI improvement (⩾50% reduction in SI) was associated with a lower likelihood of any post-treatment SAED visit at 6 months after adjusting for pre-treatment SAED (OR = 0.17, 95% CI: 0.02–0.98, p = 0.047), corresponding to numbers needed to treat of 4.3.

Conclusion:

Early SI changes may have prognostic implications, though larger studies are needed to confirm clinical significance.

Plain language summary

How early changes in suicidal ideation after ketamine or esketamine treatment relate to later emergency room visits for suicidality in people with treatment-resistant depression.

Depression is a common and serious illness, and people whose depression does not improve with standard treatments are at especially high risk for suicidal thoughts and crises. Ketamine and Esketamine ((es)ketamine) are newer treatments that can reduce depression and suicidal thoughts quickly, but it is not clear how early changes in suicidal thinking relate to future suicidal crises and emergency room use. In this study, we looked at adults who had tried at least two antidepressants in their current episode of depression without enough benefit and then received (es)ketamine as part of routine care at a specialized depression clinic. They completed brief questionnaires about their mood and suicidal thoughts before and after each (es)ketamine treatment, and we reviewed their medical records to see whether they had suicide-related emergency department visits in the six months after the initial course of (es)ketamine treatment.

Most patients reported suicidal thoughts at the start of treatment. After the acute course of (es)ketamine treatment, about half no longer reported suicidal thoughts and another group showed partial improvement. In simple analyses, patients whose suicidal thoughts improved or resolved were less likely to have suicide-related emergency visits over the next six months than those whose suicidal thoughts stayed the same or worsened.

However, when we took into account whether someone had needed suicide-related emergency care in the six months before starting (es)ketamine, early changes in suicidal thoughts were no longer the main predictor of later emergency visits. Instead, the history of prior suicide-related emergency visits was the strongest predictor of needing emergency care again.

These findings suggest that while early improvement in suicidal thoughts after (es)ketamine is encouraging, clinicians should pay particular attention to patients with a pattern of past suicide-related emergency visits and provide extra follow-up and support after treatment.

Keywords

esketamine health care utilization ketamine suicidality

Introduction

Depression is among the leading causes of disability worldwide and is strongly associated with an increased suicidal risk.¹ Individuals with treatment-resistant depression (TRD), defined by inadequate response to a minimum of two adequate antidepressant trials² and occurring in approximately 35% of patients, experience disproportionately higher morbidity, mortality, and healthcare utilization.³ Conventional monoaminergic antidepressants (including selective serotonin recapture inhibitors (SSRIs), serotonin and norepinephrine recapture inhibitors (SNRIs), Dopamine-Norepinephrine-Reuptake-Inhibitors (DNRIs), Tricyclic Antidepressants (TCAs), and Monoamine Oxidase Inhibitors (MAOIs), while effective in up to two-thirds of patients,⁴ take several weeks to achieve therapeutic benefit, leaving individuals vulnerable during periods of elevated suicidal risk, and carry the US FDA boxed warnings for increased suicidality in young adults under age 25.

In recent years, rapid-acting antidepressants such as intravenous (IV) ketamine and intranasal (IN) esketamine have emerged as promising treatments for TRD.^5
–8 Unlike conventional antidepressants, these N-methyl-D-aspartate receptor antagonists can produce rapid reductions in depressive symptoms and suicidal ideation (SI), often within hours to days of administration, though the strength of this anti-suicidal evidence differs between formulations: racemic IV ketamine has demonstrated SI reductions across multiple meta-analyses and RCTs,^9,10 whereas a recent comprehensive PRISMA meta-analysis of 87 studies found that the effect size of IN esketamine on suicidality was not significant at any time point.¹¹ These acute effects typically persist for only 7–14 days, requiring repeated treatments for sustained benefit.¹² Despite growing recognition of their acute anti-suicidal effects, assessing their long-term impact presents methodological challenges. Consequently, suicide-associated hospitalizations and emergency department (ED) visits have emerged as valuable proxy outcomes, offering objective measures.

A recent French hospital-based study¹³ examined 100 adults experiencing a major depressive episode who received one or two IV ketamine infusions within 1 week of a suicidal crisis. At 7 days, 61% were classified as responders, defined as having at least a 50% decrease in SI (measured by the Columbia Suicide Severity Rating Scale severity subscore) or a decrease from a score of 5 to 3. These responders had 75% lower odds of experiencing suicidal events (suicide attempts or hospitalization for SI) at 3 months.¹³ These findings suggest that early reduction in suicidality with ketamine may predict sustained reductions in suicidal behavior and healthcare utilization in the long term; however, replication is necessary to confirm their generalizability and clinical relevance. Our group previously reported that, in individuals with TRD, treatment with IV ketamine or IN esketamine was associated with a reduction in suicidality-associated ED (SAED) visits over a 6-month follow-up period.¹⁴ However, we did not examine the acute change in suicidality and its impact on suicidality-related healthcare utilization (SAED) at 6 months.

In the present study, leveraging data from the Ketamine Clinic at the Mayo Clinic Depression Center (MCDC),³ we examined the association between acute-phase improvements in suicidality following ketamine or esketamine treatment and subsequent ED visits for suicidality over 6 months in adults with TRD. Secondly, we aimed to replicate findings from the French study¹³ by examining whether 50% or greater reduction in suicidality after two IV ketamine treatments is associated with reduced SAED visits at 6 months. Identifying early predictors of suicidal behavior and healthcare engagement may help enhance risk stratification and optimize personalized post-treatment care for individuals with TRD.^5,15
–19

Methods

This is a secondary analysis of a naturalistic cohort of 97 patients with moderate-to-severe TRD from an ongoing study at the MCDC, where patients receive ketamine or esketamine as part of clinical care.²⁰ The study is IRB-approved (IRB #21-005927), and only patients who provided written research consent, prior to study initiation, are included. We followed the STROBE guidelines for cohort studies²¹ (see Supplemental Appendix).

Cohort details have been published previously.^{5,14,16,17,20,22} Briefly, adult patients with nonpsychotic unipolar or bipolar TRD were included as part of the clinical protocol. The diagnosis of a current major depressive episode was established by a fellowship-trained, mood-specialized psychiatrist through a comprehensive clinical interview and chart review in accordance with the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition criteria,²³ and failure to respond to at least two adequate antidepressant trials in the current episode was confirmed at this stage. Patients with psychiatric or medical contraindications were excluded prior to treatment initiation. Before treatment initiation, all cases were independently reviewed and approved by two separate ketamine clinic psychiatrists, providing an additional layer of diagnostic and eligibility verification. Participants had no active substance use disorder other than tobacco or caffeine. Psychiatric and somatic comorbidities were managed per evidence-based recommendations by treating clinician. Patients continued most of their medications during the acute treatment phase though benzodiazepine use was limited to less than 4 mg lorazepam equivalents, and patients were advised to withhold benzodiazepine and stimulants on the day of treatment.²⁴

Patients received IV ketamine or IN esketamine treatments during the acute phase, depending on their preference and insurance coverage, with the acute phase defined as up to six IV ketamine infusions according to our research-based clinical protocol, or eight IN esketamine treatments following the FDA-approved induction phase. IV ketamine was administered at the evidence-based recommended dose of 0.5 mg/kg, capped at 50 mg for patients weighing over 100 kg, and IN esketamine was administered strictly per FDA-approved doses (56 or 84 mg), as previously described.²⁰

Inclusion criteria: Individuals with TRD who had available data on SI measured using the 16-Item Quick Inventory of Depressive Symptomatology – Self Reported (QIDS-SR) item 12 during the acute phase, as well as data on SAED visits over 6 months.

Exclusion criteria: Missing data on QIDS-SR item #12 during the acute phase or missing 6-month SAED data.

SI and depressive severity were assessed using the QIDS-SR depression rating scale at baseline, before each treatment, and 24 h after each treatment during the acute phase.^5,25 The acute phase was defined as up to the first six IV treatments or the first eight IN treatments occurring within 44 days of the baseline treatment.²⁰ When the 24-h post-treatment QIDS-SR score was unavailable after the sixth IV or eighth IN treatment, the pre-treatment score from treatment 7 or 9, respectively, was used as a substitute. For participants assessed with the Montgomery-Asberg Depression Rating Scale, scores were converted to QIDS-SR equivalents using a validated Item Response Theory (IRT)-based crosswalk table,²⁶ which equates total scores across the two instruments based on psychometric modeling in a TRD sample, as described previously.²⁰ SI was assessed using QIDS-SR item 12 (0–3), with 0 indicating no SI and 1–3 indicating increasing severity from feeling life is not worth living to having specific plans or attempts.⁵

Participants with baseline SI data were classified into four groups based on the presence (item 12 score ⩾1) or absence (score = 0) of SI at baseline and at the end of the acute phase: (i) consistently asymptomatic—no SI at either timepoint; (ii) resolved—SI present at baseline but absent at the end of the acute phase; (iii) partially improved (PI)—a reduction in item 12 score that did not reach full resolution; and (iv) unchanged/worsened (UW)—no change or an increase in item 12 score. Change in SI was defined as the difference between baseline and end-of-acute-phase item 12 scores. One participant with a nominal decrease in item 12 score was classified as UW based on clinical review due to persistent clinically significant SI. For our analyses, we focused on individuals with SI present at baseline.

Outcomes

The primary outcome was the occurrence of at least one SAED visit in the 6 months post-acute phase, identified through medical record review. The primary analysis assessed whether early SI change (during the acute phase) predicted subsequent SAED visits over the 6-month follow-up. We hypothesized that participants in the unchanged/worsened group would have higher rates of ED utilization compared with those in the improved and resolved groups.

For secondary analysis designed to approximate the French study's approach,¹³ we restricted our sample to patients receiving IV ketamine only and redefined early SI improvement as a ⩾50% reduction in QIDS-SR item 12 after 2 treatments. Using the same primary outcome (occurrence of at least one SAED visit in the 6 months post-acute phase), we assessed whether early SI improvement predicted subsequent SAED visits over the 6-month follow-up. Consistent with the primary analysis, we hypothesized that participants with early SI improvement would have lower rates of ED utilization compared with those without early improvement.

Statistical analysis

Baseline and clinical characteristics were compared by SAED status using Fisher's exact tests, trend tests, and Kruskal–Wallis tests as appropriate. All analyses were restricted to participants with baseline SI and compared resolved (R), PI, and UW SI change groups. Logistic regression models estimated odds ratios for any post-acute SAED visit, and quasipoisson regression estimated incidence rate ratios (IRR) for total SAED visits. Models were run both unadjusted and adjusted for pre-acute phase SAED utilization. Additional candidate covariates were evaluated but retained only if associated with outcomes. Numbers needed to treat (NNT) were derived from odds ratios and predicted probabilities obtained from the logistic regression models. Estimated SAED visit rates per 100 participants were calculated from the count models. Time to first SAED visit was compared using Kaplan–Meier curves.

In a secondary analysis designed to approximate the French study's approach, we restricted our sample to patients receiving IV ketamine and defined early SI improvement as a ⩾50% reduction in QIDS-SR item 12 after two treatments. In this reduced sample, Firth penalized logistic regression and bias-reduced Poisson models (brglm) were used to address sparse outcomes. All analyses were conducted using R version 4.4.1 (R Foundation for Statistical Computing, Vienna, Austria).

Results

Suicidality data were available for 96 patients, of whom 72 received IV ketamine and 24 received IN esketamine. At baseline, 76% (73/96) reported passive or active SI. During the 6-month follow-up period, eight patients with baseline SI had SAED visits. Among the 73 patients with baseline SI, 36 (49%) had resolved SI after the acute-phase (es)ketamine treatment, 14 (19%) were partially improved, and 23 (32%) were unchanged or worsened. Of the 22 patients without baseline SI, 21 (95%) remained free of SI while 1 (5%) developed new onset SI.

Table 1 presents baseline and acute-phase clinical characteristics stratified by post-treatment SAED visits among patients with baseline SI. Groups were similar in age, sex, BMI, employment, and primary diagnosis (96% major depressive disorder). Patients with post-acute phase SAED visits had shorter current episode duration (median 1.0 years (IQR 0.6–1.1) vs 3.0 years (IQR 0.8–8.0); p = 0.03) and higher rates of pre-treatment SAED visits (62% vs 15%; p = 0.008). No patients with pre-treatment SAED visits achieved remission, compared with 33% of those without pre-treatment SAED visits (p = 0.095). Seventy-three patients (76%) received subsequent (es)ketamine treatments during follow-up, with no differences in SAED visit rates (p = 0.39). There were no statistically significant differences in baseline medications, psychiatric comorbidities, and treatment characteristics between groups.

Table 1.

Clinical and sociodemographic characteristics.

Characteristic	No post-ketamine suicide-ED visits (N = 65)	Post-ketamine suicide-ED visit (N = 8)	Total (N = 73)	p Value
Demographics
Age in years	48.0 (39.5, 55.1)	47.0 (43.5, 57.5)	48.0 (39.5, 55.7)	0.64^a
Sex				0.42^b
Male	16 (25%)	3 (38%)	19 (26%)
Female	49 (75%)	5 (62%)	54 (74%)
BMI	28.2 (23.8, 33.5)	27.0 (24.7, 30.5)	28.2 (23.9, 32.1)	0.54^a
Employed (at baseline)				0.93^b
Unemployed	8 (12%)	1 (12%)	9 (12%)
Employed	38 (58%)	4 (50%)	42 (58%)
Disability due to depression	9 (14%)	1 (12%)	10 (14%)
Homemaker/retired/student	10 (15%)	2 (25%)	12 (16%)
Diagnosis				1.00^b
MDD	62 (95%)	8 (100%)	70 (96%)
BP-II	3 (5%)	0 (0%)	3 (4%)
Duration of depressive episode (years)	3.0 (0.8, 8.0)	1.0 (0.6, 1.1)	2.0 (0.8, 7.0)	0.03 ^a
Type of treatment				0.68^b
IV	50 (77%)	7 (88%)	57 (78%)
Intranasal	15 (23%)	1 (12%)	16 (22%)
Acute phase
SI response (based on QIDS-SR Q 12)				0.02 ^c
Resolved	35 (54%)	1 (12%)	36 (49%)
Partially improved	12 (18%)	2 (25%)	14 (19%)
Unchanged/worsened	18 (28%)	5 (62%)	23 (32%)
QIDS-SR (at baseline)	18.0 (16.0, 20.0)	20.5 (19.5, 21.2)	18.0 (16.0, 21.0)	0.10^a
% Change in QIDS-SR (acute phase)	57.9 (26.3, 70.0)	40.8 (30.5, 45.2)	52.4 (26.7, 68.8)	0.11^a
Remitters	21 (32%)	0 (0%)	21 (29%)	0.095^b
Maintenance	52 (80%)	5 (62%)	57 (78%)	0.36^b
Total suicide-ED visits (pre-ketamine)				0.008 ^b
No pre-ketamine Suicide-ED Visit	55 (85%)	3 (38%)	58 (79%)
⩾1 Pre-ketamine suicide-ED visit	10 (15%)	5 (62%)	15 (21%)
Comorbidities
PTSD	15 (23%)	0 (0%)	15 (21%)	0.19^b
GAD anxiety	44 (68%)	7 (88%)	51 (70%)	0.42^b
Fibromyalgia/chronic pain	12 (18%)	0 (0%)	12 (16%)	0.34^b
OCD	3 (5%)	0 (0%)	3 (4%)	1.00^b
Eating disorder	3 (5%)	1 (12%)	4 (5%)	0.38^b
Borderline personality disorder	5 (8%)	1 (12%)	6 (8%)	0.52^b
History of SUD	9 (14%)	1 (12%)	10 (14%)	1.00^b
Medication/treatment
Current number of psychotropics	3.0 (3.0, 4.0)	3.5 (3.0, 5.2)	3.0 (3.0, 4.0)	0.12^a
SSRI	19 (29%)	4 (50%)	23 (32%)	0.25^b
SNRI	23 (35%)	3 (38%)	26 (36%)	1.00^b
TCA	5 (8%)	2 (25%)	7 (10%)	0.17^b
MAOIs	1 (2%)	0 (0%)	1 (1%)	1.00^b
Antipsychotic	16 (25%)	2 (25%)	18 (25%)	1.00^b
Mirtazapine	4 (6%)	0 (0%)	4 (5%)	1.00^b
Bupropion	21 (32%)	1 (12%)	22 (30%)	0.42^b
Stimulant	13 (20%)	0 (0%)	13 (18%)	0.34^b
Trazodone	17 (26%)	3 (38%)	20 (27%)	0.68^b
Gabapentin	5 (8%)	3 (38%)	8 (11%)	0.04 ^b
Z-drugs	10 (15%)	2 (25%)	12 (16%)	0.61^b
Benzo	24 (37%)	5 (62%)	29 (40%)	0.25^b
ECT	17 (26%)	2 (25%)	19 (26%)	1.00^b
TMS	16 (25%)	1 (12%)	17 (23%)	0.67^b

Continuous data are reported as the median with interquartile range, and categorical data are reported as counts (n) with percentages. Statistically significant p values (<0.05) are highlighted in bold.

Kruskal–Wallis rank sum test.

Fisher’s exact test for count data.

Trend test for ordinal variables.

BMI, Body Mass Index; BP-II, bipolar disorder type II; ECT, electro convulsive therapy; ED, Emergency Department; GAD, generalized anxiety disorder; IV, Intravenous; MAOIs, monoamineoxidase inhibitor; MDD, major depressive disorder; OCD, obsessive compulsive disorder; PTSD, post-traumatic stress disorder; QIDS-SR, Quick Inventory of Depressive Symptoms – Self Reported; SI, suicidal ideation; SNRI, serotonin and norepinephrine recapture inhibitor; SSRI, selective serotonin recapture inhibitor; SUD, substance use disorder; TCA, tricyclic antidepressant; TMS, transcranial magnetic stimulation.

In unadjusted analyses, patients in the resolved SI change group had a lower likelihood of any post-acute SAED visit (OR = 0.10, 95% CI: 0.01–0.70, p = 0.04) and fewer total visits, though the latter was borderline non-significant (IRR = 0.11, 95% CI: 0.00–0.69, p = 0.052). After adjusting for pre-treatment SAED utilization, associations with SI were attenuated and no longer significant (ANOVA p ⩾ 0.21; Table 2). Episode duration was considered as a candidate covariate but was not retained in final models. Point estimates suggested a trend toward lower post-treatment SAED utilization among patients with more favorable SI change, though these associations did not reach statistical significance (OR = 0.50, 95% CI: 0.6–3.09, p = 0.61 for PI; OR = 0.16, 95% CI: 0.01–1.23, p = 0.12 for resolved SI). Based on model-estimated probabilities, the corresponding NNT was 18.6 for PI and 10.7 for resolved group. The estimated post-acute treatment SAED visit rate was 81% lower for the resolved group compared to UW, though this difference did not reach statistical significance (IRR = 0.19, 95% CI: 0.01–1.38, p = 0.17). The corresponding rates per 100 patients were 10.4 (95% CI 2.5–42.9) in the UW group versus 2.0 (95% CI: 0.2–18.2) in the resolved group, an absolute difference of 8.4 visits per 100 participants over 6 months (Figure 1(a)).

Table 2.

Association of acute-phase change in suicide ideation with post-treatment SAED visits (N = 73).

Predictors	Any SAED visit						Total SAED visits
	Unadjusted			Adjusted			Unadjusted			Adjusted
	OR	95% CI	p	OR	95% CI	p	IRR	95% CI	p	IRR	95% CI	p
SI response groups*
Partially improved	0.60	0.08–3.31	0.58	0.50	0.06–3.09	0.47	0.82	0.15–3.41	0.79	0.74	0.14–3.15	0.70
Resolved	0.10	0.01–0.70	0.04	0.16	0.01–1.23	0.12	0.11	0.00–0.69	0.052	0.19	0.01–1.38	0.17
Any pre-treatment SAED visit		6.54	1.28–39.21	0.03		5.98	1.44–33.77	0.02

Unchanged or Worsened group is the reference group.

CI, Confidence interval; IRR, incidence rate ratio; OR, odds ratio; SAED, Suicidality-Associated Emergency Department; SI, suicidal ideation.

Statistically significant values (p < 0.05) are highlighted in bold.

Figure 1.

(a) Model-based adjusted mean SAED visits per 100 persons with 95% CI. (b) Kaplan–Meier curves depicting SAED visit-free probability over page months across the R, PI, and UW groups.

To examine whether the association between SI response and post-treatment SAED was independent of global antidepressant response, we conducted supplementary analyses adjusting for change in modified QIDS score (excluding the SI item). Change in depressive symptom severity alone was not associated with post-treatment SAED (OR = 1.01, 95% CI: 0.86–1.19, p = 0.925). Furthermore, the protective association observed for SI resolution in the primary model was not attenuated by additional adjustment for depression change, suggesting the effect is not attributable to global antidepressant response. The distribution of modified QIDS stratified by SAED status and the additional analyses are presented in Supplemental Appendix Tables 1 and 2.

The Kaplan–Meier curves (Figure 1(b)) demonstrate no statistically significant differences in SAED visit-free survival between the groups over 6 months (p = 0.066), although point estimates suggested lower cumulative SAED utilization among more favorable SI change groups.

Secondary analysis

Among the 72 individuals who received IV ketamine, 57 had baseline SI. After two infusions, 33 (47.2%) demonstrated reduction in SI. Early SI improvement was associated with a lower likelihood of any post-treatment SAED visit after adjusting for pre-treatment SAED (OR = 0.17, 95% CI: 0.02–0.98, p = 0.047), corresponding to a NNT of 4.3. In analyses of visit counts, early reduction in SI was associated with a lower post-treatment SAED visit rate, although this association did not reach statistical significance (IRR = 0.20, 95% CI: 0.04–1.10, p = 0.06). Model-estimated post-treatment SAED visit rates per 100 patients were 13.1 among those without early reduction in SI compared with 2.6 among those with early reduction, an absolute difference of 10.5 SAED visits per 100 patients over 6 months (Figure 2(a)). The Kaplan–Meier curves (Figure 2(b)) demonstrate a statistically significant difference in SAED visit-free survival between the groups over 6 months (p = 0.01).

Figure 2.

Among the 57 IV ketamine patients who had SI baseline by those who achieved reduction in suicidal ideations after two infusions. (a) Model-based adjusted mean SAED visits per 100 persons with 95% CIs. (b) Kaplan–Meier curves depicting SAED visit-free probability over 6 months.

Discussion

Identifying early predictors of suicidal behavior following acute-phase (es)ketamine treatment is essential for guiding clinical decisions and safety planning. This analysis examined whether SI changes during acute treatment predict SAED utilization over the subsequent 6 months. Among patients with TRD and baseline SI who completed acute-phase IV ketamine or IN esketamine, approximately half achieved complete SI resolution while an additional 19% experienced partial improvement.

In our sample, SI change showed a graded association with SAED visit risk in unadjusted analyses, with patients achieving resolution or improvement less likely to have subsequent visits. However, after accounting for pre-treatment SAED history, SI change did not independently predict post-acute SAED utilization. Instead, pre-treatment SAED visits emerged as the dominant predictor, with patients demonstrating substantially higher risk of subsequently seeking emergency care for suicidality, a pattern that remained consistent whether examining visit occurrence or frequency during the 6-month follow-up. This finding aligns with prior research demonstrating that baseline use of emergency care robustly predicts future utilization in psychiatric populations.^27,28

This pattern indicates that ED utilization following acute-phase (es)ketamine treatment may be shaped more by enduring patient-level vulnerabilities than by early symptom response. Prior SAED visits likely capture a constellation of risk factors that persist beyond pharmacological intervention, including limited access to outpatient mental health services, fragmented care coordination, chronic psychosocial instability, and entrenched patterns of crisis-driven help-seeking behavior developed through repeated exposure to acute care settings. The attenuation of the SI-SAED association after adjusting for baseline ED utilization suggests that the apparent protective effect of acute phase symptom improvement may partly reflect underlying differences in care access, social support, or other unmeasured confounders rather than a direct causal pathway. This finding parallels well-established suicide research demonstrating that prior suicide attempts are among the strongest predictors of future attempts and completed suicide,²⁹ suggesting that patterns of suicidal crisis and help-seeking behavior show similar persistence over time. However, these mechanisms remain speculative, and prospective studies incorporating measures of outpatient care engagement, social determinants of health, and structural barriers to ongoing psychiatric treatment are needed to clarify these relationships and identify modifiable targets beyond acute symptom management.

Our secondary analysis findings converged with those of the French study¹³ despite substantial methodological divergence. Their cohort comprised inpatients treated during acute suicidal crises who received one to two infusions with 3-month follow-up, whereas our sample consisted of outpatients receiving IV ketamine for TRD who completed full acute-phase treatment courses with 6-month surveillance. Despite these differences, both investigations observed comparable early response rates (approximately 50%), and our findings corroborate the protective association between early SI reduction and decreased suicidality-related healthcare utilization. Notably, SAED-free survival differed significantly between early SI responders and non-responders in the secondary analysis, replicating the French study's protective survival effect, a distinction that was not statistically significant in the primary analysis examining end-of-acute-phase SI changes. While this pattern may suggest that the timing of SI assessment (after two treatments vs at completion of the acute phase) could influence the strength of association with subsequent outcomes, this observation is constrained by the methodological differences between the two analysis and should be interpreted in light of the small sample size and low event rate. As such, this finding remains hypothesis-generating and requires replication in larger prospective studies.

The divergence between the primary and secondary analyses may reflect pharmacological heterogeneity between the two formulations included in the primary analysis. As noted, the anti-suicidal evidence base differs between IV racemic ketamine and IN esketamine, and the inclusion of esketamine-treated patients (n = 24) in the pooled primary analysis may have attenuated the predictive association between early SI improvement and subsequent SAED utilization. The secondary analysis, restricted to IV ketamine treatment, allowed to examine a more pharmacologically homogeneous sample, and yielded a statistically significant association after adjustment for pre-treatment SAED utilization (OR = 0.17, p = 0.047). While we cannot formally attribute the attenuation in the primary analysis to esketamine inclusion given the small sample sizes involved (n = 24), this possibility warrants acknowledgment.

Furthermore, supplementary analyses suggest that the association between early SI resolution and reduced post-treatment SAED visits may be specific to suicidal ideation change rather than reflecting a global antidepressant response, as change in depressive symptom severity independent of suicidal ideation did not predict post-treatment SAED utilization. However, given the small sample size and low event rate, these findings should be considered hypothesis-generating pending replication in larger cohorts.

More broadly, the impact of (es)ketamine on suicidality may not be fully captured by acute symptom measures alone. Accumulating observations suggest that effects of (es)ketamine may extend to longer-term clinical outcomes, including patterns of emergency healthcare utilization.^30,31 The sixfold increased risk associated with prior SAED visits—independent of treatment response—underscores that acute pharmacological intervention addresses only one component of a complex clinical picture. If sustained reductions in crisis-driven care can be reliably achieved in high-risk populations through integrated interventions combining (es)ketamine with enhanced care coordination and outpatient engagement, this could have meaningful implications for healthcare costs and resource allocation in TRD. However, such possibilities remain speculative and require prospective investigation with health economic endpoints and systematic assessment of care utilization patterns, outpatient treatment adherence, and the sustainability of symptom improvement beyond the acute phase.

The finding that individuals with post-acute treatment SAED visits tended to have shorter depressive episodes is consistent with prior research demonstrating an inverse relationship between illness duration and suicide attempts among mood disorder patients presenting to EDs.³² This pattern may reflect more acute, volatile illness phenotypes, less developed coping mechanisms in earlier-stage illness, or differences in help-seeking behaviors—mechanisms that warrant further investigation to inform risk stratification and safety planning following acute-phase (es)ketamine treatment.

Limitations

This study has several important limitations. The single-center, observational design limits generalizability and precludes causal inferences about the relationship between SI improvement and SAED utilization. SAED visits were ascertained through medical record review within our health system; patients seeking emergency care at facilities not using our integrated electronic health record system would not have been captured, potentially underestimating true rates. The small sample size and low event rate, with only eight SAED events occurring among patients with baseline SI, limited statistical power and model stability, reducing the precision and reliability of effect estimates. The wide CIs and borderline p-values observed throughout reflect these constraints. Taken together, these findings should be interpreted with caution and considered hypothesis-generating rather than confirmatory, pending replication in larger prospective cohorts with higher event rates.

Suicidality was assessed using QIDS-SR Item 12, which reflects the information routinely available in naturalistic clinical settings where comprehensive suicidality rating scales are rarely administered systematically outside of formal research protocols. While this approach has inherent constraints, it captures real-world clinical practice and provides a pragmatically meaningful measure of SI change. Nevertheless, several implications merit consideration. The 0–3 ordinal scale may not register all clinically meaningful shifts in suicidal thinking, potentially introducing some misclassification across SI change groups. Self-report of suicidal ideation may also be subject to underreporting bias in outpatient settings. Additionally, QIDS-SR Item 12 does not capture key dimensions of suicidality such as intent, frequency, or controllability, and a post-treatment score of 0 does not preclude residual clinically significant SI not detected by this item. Finally, measurement imprecision in the exposure variable likely introduces attenuation bias, suggesting that the true associations between early SI improvement and SAED utilization may be stronger than those observed. These considerations underscore the value of future prospective studies incorporating validated, multidimensional suicidality instruments to build on the present findings.

Relatedly, the ⩾50% SI reduction threshold used in the secondary analysis, while chosen to approximate the French study's methodology, carries inherent interpretive challenges when applied to a coarsely scaled ordinal variable ranging from 0 to 3. For instance, a reduction from 2 to 1 and a reduction from 1 to 0 both satisfy this criterion yet represent meaningfully different clinical realities. We note, however, that the French study applied an analogous threshold to the C-SSRS severity subscore, also an ordinal scale, suggesting this is a shared limitation of the replication approach rather than one unique to our study. The primary analysis, which used clinically anchored absolute categories, is not subject to this concern.

The study combined participants receiving IV ketamine and IN esketamine (due to small number of events), which differ in pharmacokinetics, treatment schedules, and patient selection factors (insurance coverage, patient preference), potentially introducing unmeasured heterogeneity. We lacked data on critical mediating factors including psychosocial interventions, outpatient care engagement, and social determinants of health—all of which likely influenced outcomes. Our proposed mechanisms linking episode duration to help-seeking patterns remain speculative, as distress tolerance, coping strategies, and therapeutic alliance were not directly measured.

Finally, our findings reflect patients with TRD who completed acute-phase (es)ketamine treatment and may not generalize to those with non-treatment-resistant depression, patients who discontinued treatment prematurely, or settings with different treatment delivery models. Despite these constraints, the study's naturalistic design and rigorous characterization of a real-world TRD population receiving (es)ketamine treatment provide clinically relevant insights into predictors of post-acute emergency utilization.

Conclusion

In this naturalistic cohort of patients with TRD receiving acute-phase (es)ketamine treatment, early reductions in suicidal ideation following initial treatments were associated with lower rates of suicide-related emergency department visits in unadjusted analysis; however, this association was no longer statistically significant after accounting for prior SAED visits. A history of suicide-related emergency visits prior to treatment emerged as the strongest predictor of subsequent emergency utilization and attenuated the association between acute-phase changes in suicidal ideation and later outcomes. These findings highlight the importance of identifying patients with established patterns of emergency care utilization and providing enhanced outpatient support to help reduce crisis-driven healthcare use following acute-phase treatment. Future prospective studies incorporating comprehensive suicidality assessments, measures of outpatient care engagement, social determinants of health, and health economic endpoints are needed to clarify mechanisms linking treatment response to long-term outcomes and identify modifiable targets for reducing crisis-driven healthcare utilization in this vulnerable population.

Supplemental Material

sj-docx-1-tpp-10.1177_20451253261450747 – Supplemental material for Early suicidal ideation change and suicidality-associated emergency department utilization after ketamine or esketamine treatment in treatment-resistant depression

Supplemental material, sj-docx-1-tpp-10.1177_20451253261450747 for Early suicidal ideation change and suicidality-associated emergency department utilization after ketamine or esketamine treatment in treatment-resistant depression by Liliana Patarroyo-Rodriguez, Vanessa K. Pazdernik, Jennifer L. Vande Voort, Simon Kung and Balwinder Singh in Therapeutic Advances in Psychopharmacology

Supplemental Material

sj-pdf-2-tpp-10.1177_20451253261450747 – Supplemental material for Early suicidal ideation change and suicidality-associated emergency department utilization after ketamine or esketamine treatment in treatment-resistant depression

Supplemental material, sj-pdf-2-tpp-10.1177_20451253261450747 for Early suicidal ideation change and suicidality-associated emergency department utilization after ketamine or esketamine treatment in treatment-resistant depression by Liliana Patarroyo-Rodriguez, Vanessa K. Pazdernik, Jennifer L. Vande Voort, Simon Kung and Balwinder Singh in Therapeutic Advances in Psychopharmacology

Footnotes

Acknowledgements

None.

Author’s note

Previous presentation: Part of the data was presented as a panel at the Annual meeting of the American College of Neuropsychopharmacology, Nassau, Bahamas (January 14, 2026).

Declarations

ORCID iD

Balwinder Singh

Supplemental material

Supplemental material for this article is available online.

AI tools disclosure

During the preparation of this manuscript, the authors used open AI ChatGPT 5 to edit and condense text. After using this tool, the authors reviewed and modified the content as pertinent and take full responsibility for the content of the manuscript.

References

Anderson

Crawford

Fava

, et al. Depression - understanding, identifying, and diagnosing. N Engl J Med. 2024; 390(17): e41.

McIntyre

Alsuwaidan

Baune

, et al. Treatment-resistant depression: definition, prevalence, detection, management, and investigational interventions. World Psychiatry 2023; 22(3): 394–412.

Marques

Ozerdem

Kung

, et al. Next-step treatment options for treatment-resistant depression: insights from the mayo clinic depression center panel. J Clin Psychiatry 2026; 87(1): 25cs16066.

Rush

Trivedi

Wisniewski

, et al. Acute and longer-term outcomes in depressed outpatients requiring one or several treatment steps: a STAR*D report. Am J Psychiatry 2006; 163(11): 1905–1917.

Singh

Vande Voort

Pazdernik

, et al. Treatment-resistant depression patients with baseline suicidal ideation required more treatments to achieve therapeutic response with ketamine/esketamine. J Affect Disord 2024; 351: 534–540.

Patarroyo-Rodriguez

Cavalcanti

Vande Voort

, et al. The use of ketamine for the treatment of anhedonia in depression. CNS Drugs 2024; 38(8): 583–596.

Seshadri

Prokop

Singh

Efficacy of intravenous ketamine and intranasal esketamine with dose escalation for Major depression: a systematic review and meta-analysis. J Affect Disord 2024; 356: 379–384.

Singh

Ketamine and esketamine for depression in daily practice: opportunities and challenges. J Clin Psychopharmacol 2024; 44(5): 451–455.

Feng

Chen

Zeng

, et al. Efficacy of single and repeated ketamine administration for suicidal ideation in adults with psychiatric disorders: a meta-analysis. Prog Neuropsychopharmacol Biol Psychiatry 2025; 136: 111152.

10.

Shen

Gao

, et al. A meta-analysis of the effects of ketamine on suicidal ideation in depression patients. Transl Psychiatry 2024; 14(1): 248.

11.

Fountoulakis

Saitis

Schatzberg

AF.

Esketamine treatment for depression in adults: a PRISMA systematic review and meta-analysis. Am J Psychiatry 2025; 182(3): 259–275.

12.

Elmosalamy

Tarikogullari

Patarroyo-Rodriguez

, et al. Intravenous ketamine versus esketamine for depression: a systematic review and meta-analysis. Ther Adv Psychopharmacol 2025; 15: 20451253251394127.

13.

Pastre

Chancel

Malestroit

, et al. Early response to ketamine for suicidal crisis reduces suicidal events at 3 months. J Clin Psychiatry 2025; 86(3): 25m15829.

14.

Patarroyo-Rodriguez

Pazdernik

Vande Voort

, et al. Ketamine reduces suicidality-associated emergency department utilization in patients with treatment-resistant depression: a 6-month mirror-image study. J Clin Psychiatry 2025; 86(4): 25m15941.

15.

Parikh

Vande Voort

Yocum

, et al. Clinical outcomes in the biomarkers of ketamine (Bio-K) study of open-label IV ketamine for refractory depression. J Affect Disord 2024; 348: 143–151.

16.

Patarroyo-Rodriguez

Pazdernik

Vande Voort

, et al. Hypersomnia as a predictor of response to intravenous ketamine/intranasal esketamine in treatment resistant depression. J Affect Disord 2024; 349: 272–276.

17.

Cavalcanti

Pazdernik

Vande Voort

, et al. Baseline perceived stress as a predictor of ketamine/esketamine treatment response in treatment-resistant depression. J Affect Disord 2025; 383: 283–289.

18.

Murata

Madaj

Capan

, et al. High baseline plasma anthranilic acid predicts remission upon acute-series ketamine infusion for treatment-resistant depression. Biol Psychiatry Glob Open Sci 2025; 5(4): 100503.

19.

Penaluna

Vande Voort

Bobo

, et al. Utilizing depression symptom-based phenotypes to explore ketamine treatment response in major depression: the Bio-K multicenter trial. J Affect Disord 2025; 385: 119414.

20.

Singh

Kung

Pazdernik

, et al. Comparative effectiveness of intravenous ketamine and intranasal esketamine in clinical practice among patients with treatment-refractory depression: an observational study. J Clin Psychiatry 2023; 84(2): 22m14548.

21.

von Elm

Altman

Egger

, et al. Strengthening the reporting of observational studies in epidemiology (STROBE) statement: guidelines for reporting observational studies. BMJ 2007; 335(7624): 806–808.

22.

Singh

Pazdernik

Kung

, et al. Long-term outcomes in patients with treatment-refractory depression receiving intravenous ketamine and intranasal esketamine: an observational study. J Clin Psychiatry 2025; 86(3): 25m15831.

23.

American Psychiatric Association. Diagnostic and statistical manual of mental disorders: DSM-5-TR. 5th edn, text revision ed. Washington, DC: American Psychiatric Association Publishing, 2022.

24.

Frye

Blier

Tye

SJ.

Concomitant benzodiazepine use attenuates ketamine response: implications for large scale study design and clinical development. J Clin Psychopharmacol 2015; 35(3): 334–336.

25.

Rush

Trivedi

Ibrahim

, et al. The 16-Item Quick Inventory of Depressive Symptomatology (QIDS), clinician rating (QIDS-C), and self-report (QIDS-SR): a psychometric evaluation in patients with chronic major depression. Biol Psychiatry 2003; 54(5): 573–583.

26.

Carmody

Rush

Bernstein

, et al. Making clinicians lives easier: guidance on use of the QIDS self-report in place of the MADRS. J Affect Disord 2006; 95(1–3): 115–118.

27.

Gentil

Huynh

Grenier

, et al. Predictors of emergency department visits for suicidal ideation and suicide attempt. Psychiatry Res 2020; 285: 112805.

28.

Niedzwiecki

Sharma

Kanzaria

, et al. Factors associated with emergency department use by patients with and without mental health diagnoses. JAMA Netw Open 2018; 1(6): e183528.

29.

Bostwick

Pabbati

Geske

, et al. Suicide attempt as a risk factor for completed suicide: even more lethal than we knew. Am J Psychiatry 2016; 173(11): 1094–1100.

30.

Pan

Gorenflo

Davis

, et al. Suicidal ideation following ketamine prescription in patients with recurrent major depressive disorder: a nation-wide cohort study. Transl Psychiatry 2024; 14(1): 327.

31.

Harding

Zhdanava

Teeple

, et al. Health care resource use and medical costs among patients with major depressive disorder and acute suicidal ideation or behavior initiated on esketamine nasal spray or traditional treatments in the United States. Clin Ther 2025; 47(3): 189–195.

32.

Xie

Wang

Han

, et al. Suicidality and its associated factors among mood disorder patients in emergency department in China: a comparative study using propensity score matching approach. Transl Psychiatry 2023; 13(1): 372.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.79 MB

0.07 MB