Characteristics and treatment outcome of suicidal depression: Two large naturalistic cohorts of depressed outpatients

Abstract

Objective:

The emergence of new drugs for managing suicidal ideation (e.g. ketamine) raises the question of whether suicidal depression (i.e. moderate to severe depression with concomitant suicidal ideation) is a specific depression phenotype. Therefore, this study characterized patients with suicidal depression (baseline clinical characteristics, suicidal ideation and depression evolutions, suicide risk) in two large cohorts of outpatients with depression.

Methods:

LUEUR and GENESE are two large, prospective, naturalistic cohorts of French adult outpatients with depression (Diagnostic and Statistical Manual of Mental Disorders, fourth edition, criteria), treated and followed up for 6 weeks. Depression severity was assessed with the Hospital Anxiety and Depression Scale, and suicidal ideation with the suicidal item of the Montgomery–Åsberg Depression Rating Scale. Patients with moderate or severe depression (Hospital Anxiety and Depression Scale–Depression subscale score >11) were selected and classified as without suicidal ideation (suicidal item of the Montgomery–Åsberg Depression Rating Scale <2), with moderate suicidal ideation (suicidal item of the Montgomery–Åsberg Depression Rating Scale [2; 3]) and with severe suicidal ideation (suicidal item of the Montgomery–Åsberg Depression Rating Scale ⩾4).

Results:

Baseline clinical features were more severe (e.g. higher anxiety and depression scores) in depressed patients with moderate/severe suicidal ideation. Depression remission after treatment was less frequent among patients with severe suicidal ideation. The risk of suicide attempt during the follow-up was threefold higher in patients with suicidal ideation among those 10% had persistent suicidal ideation.

Conclusion:

Suicidal depression could be a specific depression phenotype with more severe clinical characteristics, less frequent depression remission, suicidal ideation persistence and higher suicide attempt risk, despite antidepressant treatment. It seems that novel therapeutic strategies could be needed.

Keywords

Depression suicidal ideation suicidal depression suicide attempt antidepressant antisuicidal therapeutics

Introduction

With more than 800,000 deaths by suicide worldwide each year, and 20 times more suicide attempts (SAs) (World Health Organization 2019), suicidal behavior (SB) (i.e. completed suicide or SA) is a major public health problem. One way to prevent SB is to focus on suicidal ideation (SI) because SI is a major risk factor of future death by suicide (Franklin et al., 2017). Patients with a major depressive episode (MDE) (another important risk factor for SB) report SI in 45–70% of cases (Sokero, 2006). SI has been mainly considered a symptom or a consequence of a psychiatric disorder (Courtet and Nobile, 2020). Consequently, clinicians use psychotherapy (e.g. cognitive behavioral therapy [CBT], dialectical behavior therapy) (Ducasse et al., 2018) and non-SI-specific drugs (antidepressants, clozapine and lithium) to reduce SI and prevent SB (Courtet et al., 2020). Antidepressants represent the main drugs used in patients with MDE to reduce SI. Evidence on efficacy of antidepressant in suicidal patients is not convergent. On one hand, antidepressant reduces SI (notably by decreasing depression severity) (Weitz et al., 2014) and decreases SB risk by 40–60% (Rihmer and Gonda, 2013). On the other hand, depression remission after antidepressant treatment seems to be less frequent in depressed patients with SI or lifetime SA (Lopez-Castroman et al., 2016), and many of them report SI persistence (Seo et al., 2014). Moreover, most randomized controlled trials on antidepressants excluded suicidal patients (current SI and/or past history of SA) and a large part of studies were focused only on treatment-related emergence or worsening of SI. Similarly, clinical trials on CBT failed to show an improvement of suicidal outcome when CBT targeted mental illness, whereas CBT significantly reduced suicidal risk when it specifically targeted SI and SB rather than mental illness (Mewton and Andrews, 2016). Thus, further studies are needed to assess the course of pre-existing SI and the baseline clinical characteristics of depressed patients with SI (Courtet et al., 2020; Courtet and Nobile, 2020). Finally, some patients with MDE will never experience SI. Altogether, this raises the question of whether suicidal depression (i.e. moderate to severe depression with SI) could be a specific phenotype of depression needing specific treatments.

Depressed patients with SI have more severe baseline sociodemographic and clinical characteristics (e.g. more often single, higher depression and hopelessness severity, more lifetime SA), and less frequent remission of depression after antidepressant treatment (Lopez-Castroman et al., 2016; Trivedi et al., 2013; Zimmerman et al., 2018). SI course is related to, but separated from that of, depression. For instance, Batterham et al. (2019) recently described two trajectory types with different outcomes for SI and for depressive symptoms. They found that class membership for the two outcomes was quite independent. Another study showed that depressive symptomatology and its course were significantly different in suicidal and non-suicidal depressed patients (van Ballegooijen et al., 2019). About 20% of patients present persistent SI despite the improvement or remission of depression (Batterham et al., 2019; Nobile et al. 2021; Seo et al., 2014). Finally, growing evidence suggests SB has its own physiopathology (e.g. opioid system dysfunction) (Nobile et al. 2020), suggesting that suicidal depression could have a different physiopathology than depression without SI. A depressive episode with SI would thus need targeted treatments (van Ballegooijen et al., 2019).

To determine whether suicidal depression is a specific phenotype is important because promising new drugs are emerging for SI, particularly opiates (e.g. buprenorphine, tianeptine) (Nobile et al., 2018; Yovell et al., 2016) and the N-methyl-D-aspartate (NMDA) antagonists ketamine and esketamine (Molero et al., 2018). A systematic review and meta-analysis of individual participant data showed that a single infusion of ketamine reduces SI severity within 24 hours in more than half of patients and for up to 1 week (Wilkinson et al., 2018). Interestingly, ketamine effect on SI could partially be independent of its effects on mood (Ballard et al., 2014). In addition, specific CBTs targeting SB or SI are emerging since it has shown encouraging results (Mewton and Andrews, 2016). This is the opposite of the treatment as usual (TAU) that consists of treating depression with the hope to also reduce SI and SB. Interestingly, a recent study, comparing CBT targeting SI vs TAU in a military adult sample with SI, showed that patients treated with CBT were less at risk to make an SA during follow-up than those treated with TAU (Lee et al., 2020). The existence of a suicidal phenotype will promote the development of specific pharmacological treatments and psychotherapies targeting SI to widen the arsenal therapeutic for suicidal patients. Such results encourage the design of clinical trials to test the efficacy of these emerging drugs and specific psychotherapy in patients with suicidal depression.

We aimed to determine whether suicidal depression was a specific depression phenotype by characterizing patients with suicidal depression (baseline clinical characteristics, SI and depressive symptom changes over time, suicidal risk) identified in two large cohorts of outpatients with an MDE. We decided to conduct our analyses in the two independent cohorts separately because of the different eligibility criteria, (i.e. alcohol and substance dependence; co-treatment with mood stabilizer or antipsychotic) that could impact depression and SI courses. Moreover, this allowed observing whether main results in one cohort were replicated in the second cohort and thus to test the replication of our main hypothesis (i.e. suicidal depression could be a specific subtype of depression) in two large independent cohorts of depressed outpatients recruited throughout France.

Material and methods

Participants’ recruitment

LUEUR and GENESE are two large, prospective, naturalistic cohorts of more than 3000 French adult outpatients with a diagnosis of MDE according to the Diagnostic and Statistical Manual of Mental Disorders, fourth edition, (DSM-IV) criteria. These cohorts were created to study depression and SB. This study is not a post hoc analysis from another study (Courtet et al., 2014; Voegeli et al., 2016). Participants were recruited by general practitioners or psychiatrists at primary and psychiatric care clinical centers throughout France. The same physician assessed the patient at the first (baseline) and second visit (after at least 6 weeks). In both studies, non-inclusion criteria were younger than 18 years of age, bipolar disorder, schizophrenia, primary substance misuse, primary organic disease and pregnancy or lactation. Additional non-inclusion criteria in the GENESE study were non-Caucasian ethnicity, and alcohol and substance dependence. Current treatment with an antipsychotic or a mood stabilizer was an additional non-inclusion criterion in the LUEUR study.

In the GENESE cohort, all patients were treated with tianeptine (between 12.5 and 37.5 mg/day). Tianeptine was systematically prescribed as the first antidepressant medication or as a change of antidepressant molecule. In the LUEUR cohort, all patients were treated with antidepressants approved by the European Medicines Agency, although most patients received tianeptine. Of note, tianeptine was one of the most frequently prescribed antidepressants in France at the time of the trials (Gusmão et al., 2013). In these studies, the goal was not to assess the effectiveness of one particular antidepressant. In both cohorts, other concomitant treatments for current somatic problems or sleep or anxiety disorders were allowed, based on clinical judgment. These studies were performed according to the French regulatory guidelines and the current Good Clinical Practice, and were approved by local independent ethics committees. All patients were informed about the study aims and procedures and signed a written informed consent.

Clinical assessment

Demographic data, number of depressive episodes and lifetime history of SA were assessed at baseline. Physicians evaluated depression severity and SI at baseline and at week 6. Patients self-rated depression severity and SI at baseline, week 1, week 2 and week 6 in the LUEUR study, and at baseline, week 2, week 4 and week 6 in the GENESE study.

Depression severity was assessed with the French self-rated version of the Hospital Anxiety and Depression Scale (HADS). In addition to Anxiety (HADS-A) and Depression (HADS-D) subscales, an agitation subscore of the HADS was calculated (sum of items 1, 7 and 11) (Lopez-Castroman et al., 2020). This scale was chosen for its simplicity of use, good comprehensibility and robust psychometric properties, demonstrated also in outpatient groups (Demyttenaere et al., 2009). In this scale, depressive symptom severity is assessed independently of any SI/SB.

SI was evaluated using the suicidal item of the self-rated Montgomery–Åsberg Depression Rating Scale (MADRS) (item number 10, or MADRS-SI). The ratings range from 0 to 6 (0–1: enjoys life or takes it as it comes; 2–3: weary of life, only fleeting suicidal thoughts; 4–5: probably better off dead, suicidal thoughts are common and suicide is considered a possible solution, but without specific plans or intention; and 6: explicit plans for suicide when there is an opportunity, active preparations for suicide). A previous study showed that a single suicide item from a depression rating scale, either clinician-rated or self-reported, is a valid approach to assess SI compared with the Beck’s scale (Desseilles et al., 2012). This method was used in large clinical studies, such as the STAR*D (Zisook et al., 2009) and also in more recent studies (Ballard et al., 2017; Bernert et al., 2017; Nobile et al., 2018).

SA occurrence was assessed at the end of the study (week 6).

Hopelessness was assessed using the self-report version of the Beck Hopelessness Scale in the LUEUR cohort at baseline, week 1, week 2 and week 6. Impulsivity and sleep disturbance were assessed using the self-report version of the Plutchik’s questionnaire (Anon, n.d.) and the item on sleep of the MADRS, respectively, in the GENESE study at baseline, week 2, week 4 and week 6.

Definition of suicidal depression

In line with the inclusion criteria of recent clinical trials on antisuicidal agents (that used either Beck Depression Inventory; Hamilton Rating Scale for Depression; MADRS; Question B5 and B9 of the mini international neuropsychiatric interview (MINI) or Suicidal Ideation Scale) (Canuso et al., 2018; Grunebaum et al., 2018), only patients with moderate to severe depression were included (i.e. patients with HADS-D score >11 at baseline, according to the definition of the French High Authority of Health). Thus, 3969 patients (81.9%) from the LUEUR cohort and 2537 patients (71.1%) from the GENESE cohort were selected. The baseline MADRS-SI score was used to classify patients into three groups: (1) depression without SI (score <2), (2) depression with moderate SI (score between 2 and 3) and (3) depression with severe SI (score ⩾4), as done in previous studies (Lopez-Castroman et al., 2016; Reeves et al., 2008). A MADRS-SI score between 2 and 3 can be indicative of passive suicidal thoughts, and a score ⩾4 of active SI (Ballard et al., 2017).

Definition of response to treatment and remission of depression

Response was defined by a decrease by at least 50% of their baseline HADS-D score. Depression remission was defined by a final HADS-D score of ⩽7 (at week 6).

Definition of severe SI persistence and of SI remission

SI persistence was defined by a MADRS-SI score of ⩾4 at week 6 in patients who had a score of ⩾4 at baseline.

SI remission was defined by a MADRS-SI score of <2 at week 6 in patients who had a score of ⩾4 at baseline.

Lost to follow-up

Some patients were lost to follow-up in both cohorts. In the LUEUR cohort, 289 (7.3%) patients were lost to follow-up (7.3% of patients with depression without SI, 7.8% of patients with moderate suicidal depression and 6.1% of patients with severe suicidal depression). In the GENESE cohort, 71 (2.8%) patients were lost to follow-up (3.5% of patients with depression without SI, 1.9% of patients with moderate suicidal depression and 4% of patients with severe suicidal depression). These patients were included in the comparison of baseline sociodemographic and clinical data, but excluded from all other analyses.

Statistical analysis

Categorical variables were presented as percentages, and quantitative variables as means with standard deviation (SD). The group with depression without SI was chosen as reference for all analyses. The demographic and clinical characteristics of the three groups were compared using a multinomial logistic regression model adjusted for depression severity (HADS-D score) and the likelihood-ratio test (LRT). The adjustment on depression severity was previously used to assess the association between SI and other variables independently of depression (Perroud et al., 2010; Rogers et al., 2018; Zisook et al., 2011). Bonferroni correction was applied for post hoc analyses.

Curves to describe the HADS-D and MADRS-SI score changes during the follow-up were created, and a linear regression model was used to test whether intra-group score changes were significant. The slopes (β) and p values of the different curves are indicated in the graphs. A generalized least squares (GLS) linear regression model was used to test for global and longitudinal between-group differences at the depression level. First-order autoregressive temporal correlation structure was used to take account to the intra-individual correlation of the residual. Significances were established from LRT in the adjusted GLS model. Adjustment variables were chosen according to their association with depression score changes (variables associated with a value of p < 0.10 were included in the model). Patients with missing data for one of the adjustment variables were excluded from this analysis.

Treatment response and depression remission during the follow-up were compared using multivariate logistic regression model adjusted for variables associated with depression remission and treatment response.

Risk factors of SI persistence and predictive factors of SI remission at the end of the follow-up period in patients with severe SI at baseline were evaluated with a logistic regression. All variables associated with SI persistence and SI remission with a value of p < 0.10 were included in a multivariate model to find the strongest predictors. Adjusted odds ratios (ORs) and 95% confidence intervals (CIs) were estimated. To test whether SI changes were associated with changes in depression, anxiety, impulsivity (GENESE), sleep (GENESE) and hopelessness (LUEUR), their respective temporal variations were estimated from the slopes of simple linear regression models fitted on each patient. Then, the association between these variations was tested with analysis of variance (ANOVA) type III, first in an unadjusted model and then after adjustment for depression changes. The R² change was also examined to quantify how much of SI variance was explained by anxiety, impulsivity, sleep and hopelessness when the depression score changes were included in the model.

SA risk during the follow-up was compared using a multivariate logistic regression model. Adjustment variables were chosen according to their association with SA during the follow-up. Sensitivity analyses were performed by excluding patients with lifetime SA (see Supplemental Material).

All multinomial logistic regression models were performed with SPSS software (version 23.0.0.2; IBM SPSS Statistics for Windows. Armonk, NY: IBM Corp), and all linear regression analyses and the GLS model were performed with R software (version 3.6). The significance level was set at p < 0.05.

Results

Comparison of the baseline sociodemographic and clinical data

In the LUEUR cohort, 671 patients had depression without SI (16.90%), 2292 had depression with moderate SI (57.75%) and 1006 had depression with severe SI (25.35%) (Table 1). In the GENESE cohort, 370 patients had depression without SI (14.58%), 1408 had depression with moderate SI (55.50%) and 759 had depression with severe SI (29.92%). About 20% of outpatients from both cohorts had depression with severe SI, 60% were women and the mean age was around 45 years.

Table 1.

Association between sociodemographic and clinical data and suicidal depression.

Variable	Suicidal depression (LUEUR)								Suicidal depression (GENESE)
		No		Moderate		Severe		No		Moderate		Severe
		671		2292		1006		370		1408		759		n	%
	n	%	n	%	Global adjusted p value	Post hoc analysis	n	%	n	%	n	%	Global adjusted p value	Post hoc analysis
Sex							0.98								0.41
Men	246	36.7	836	36.5	362	36.0			144	38.9	564	40.1	279	36.8
Women	425	63.3	1456	63.5	644	64.0			226	61.1	844	59.9	480	63.2
Age (years)	47.51 (14.32)		47.37 (13.70)		47.65 (14.04)		0.87		48.02 (14.64)		49.30 (14.58)		50.61 (14.82)		0.009***	S > N
Marital status							0.003	NS							0.003	NS
Single	99	14.9	407	17.8	163	16.3			62	16.8	255	18.2	137	18.1
Married	415	62.3	1246	54.6	508	50.6			236	64.0	817	58.2	405	53.5
Divorced	111	16.7	475	20.8	248	24.7			49	13.3	230	16.4	156	20.6
Widower	41	6.2	156	6.8	84	8.4			22	6.0	101	7.2	59	7.8
Children							0.69
Yes	513	77.1	1719	75.7	750	75.1
No	152	22.9	552	24.3	249	24.9
Living alone							0.003	S, M > N
Yes	191	28.9	812	35.8	391	39.5
No	471	71.1	1456	64.2	598	60.5
Study level							0.12								0.34
Junior secondary school	273	41.0	910	40.1	456	45.7			147	40.4	607	43.3	378	50.2
Senior secondary school	207	31.1	782	34.5	296	29.7			100	27.5	371	26.5	181	24.0
University	186	27.9	575	25.4	245	24.6			117	32.1	423	30.2	194	25.8
Professional activity							0.16								0.01	S > N
Working	364	57.5	1237	56.8	482	50.2			223	61.1	764	54.6	352	46.9
Unemployment	49	7.7	199	9.1	115	12.0			24	6.6	134	9.6	89	11.9
Retired	121	19.1	355	16.3	168	17.5			66	18.1	286	20.5	174	23.2
Sick leave	59	9.3	220	10.1	120	12.5			52	14.2	214	15.3	136	18.1
Other	40	6.3	168	7.7	75	7.8
Current MDE duration (months)							0.80								0.001	S > M, N
<2	430	65.3	1530	67.2	657	65.9			136	37.4	480	34.9	218	29.3
[2; 6]	203	30.9	673	29.6	302	30.3			153	42.0	583	42.4	302	40.6
⩾6	25	3.8	73	3.2	38	3.8			75	20.6	313	22.7	224	30.1
Cumulated MDE duration lifetime (weeks)									29.24 (23.84)		32.40 (26.85)		37.28 (30.19)		0.01	S > N
Number of MDE lifetime	1.58 (7.87)		1.48 (5.85)		2.42 (9.06)		0.03	S > M	2.22 (1.31)		2.74 (2.98)		3.19 (2.85)		<0.001	S > N
Age at first MDE (years)	36.21 (13.11)		34.73 (12.16)		33.60 (12.49)		0.03	S < N	36.68 (14.72)		36.54 (13.77)		35.54 (13.67)		0.70
Baseline HADS-A score	13.46 (3.41)		14.55 (3.22)		16.07 (3.23)		<0.001	S > M > N	13.52 (3.32)		14.14 (3.24)		15.46 (3.45)		<0.001	S > M > N
Baseline HADS-D score	15.04 (2.37)		15.99 (2.66)		17.79 (2.55)		<0.001*	S > M > N	14.76 (2.38)		15.34 (2.46)		16.70 (2.72)		<0.001**	S > M > N
Baseline HADS-Agitation score	7.24 (1.37)		7.11 (1.34)		7.10 (1.37)		0.07*		5.79 (1.58)		5.97 (1.57)		6.25 (1.86)		0.003**	S > M > N
Baseline HADS total score	28.50 (4.51)		30.54 (4.71)		33.85 (4.67)		<0.001*	S > M > N	28.28 (4.62)		29.47 (4.54)		32.16 (4.99)		<0.001**	S > M > N
Baseline Hopelessness scale score	10.76 (2.13)		11.22 (1.82)		11.42 (1.39)		<0.001	S > M > N
Baseline Impulsivity scale score									36.08 (5.73)		36.93 (6.11)		38.42 (6.72)		<0.001	S > M, N
Baseline MADRS-Sleep score									3.12 (1.58)		3.65 (1.34)		4.26 (1.42)		<0.001	S > M > N
Lifetime suicide attempts							<0.001	S > M > N							<0.001	S > M > N
Yes	36	5.4	387	17.0	373	37.1			17	4.7	119	8.7	177	23.9
No	635	94.6	1893	83.0	632	62.9			341	95.3	1245	91.3	565	76.1
Alcohol dependence							0.46
Yes	36	5.4	133	5.8	74	7.4
No	633	94.6	2151	94.2	927	92.6
Treatment instauration/change							<0.001	S > M, N							<0.001	S > M, N
Instauration	547	83.5	1806	80.9	674	69.4			311	84.5	1147	82.3	563	75.2
Change	108	16.5	427	19.1	297	30.6			57	15.5	247	17.7	186	24.8
Type of antidepressant							0.02	NS
Tianeptine	578	86.1	1933	84.3	789	78.4
SRIs	80	11.9	305	13.3	180	17.9
Others	13	1.9	54	2.4	37	3.7
Associated treatment							<0.001	S > M > N							<0.001	S > M > N
Yes	69	10.3	343	15.1	262	27.0			176	47.6	798	56.7	490	64.6
No	599	89.7	1922	84.9	709	73.0			194	52.4	610	43.3	269	35.4

MDE: major depressive episode; NS: not significant; SRIs: serotonin reuptake inhibitors; HADS: Hospital Anxiety and Depression Scale; HADS-A: HADS-Anxiety; HADS-D: HADS-Depression; MADRS: Montgomery–Åsberg Depression Rating Scale.*Non-adjusted; **Adjusted only for age; ***Adjusted only for depression severity.

The global p value was adjusted for depression severity in the LUEUR cohort, and for depression severity and age in the GENESE cohort.Bold values means that results are significant (p-value < 0.05).Shading means that the variable was not available in the cohort (since some variables were available in one cohort and not in the other).

In both cohorts, patients with severe SI were more likely to be living alone. In the GENESE cohort, patients with severe SI were older than patients with moderate or without SI.

In both cohorts, lifetime MDE number was higher, and lifetime history of SA and associated treatments (e.g. anxiolytics, hypnotics) were more frequent in patients with moderate or severe SI than in patients without SI. Moreover, the HADS-A and HADS-D scores and HADS total score were higher in patients with moderate and severe SI. Then, the levels of hopelessness (LUEUR cohort), impulsivity (GENESE cohort) and the MADRS-Sleep score (GENESE cohort) were higher in patients with moderate and severe SI than in patients without SI. It is important to point out that antidepressant type did not differ among patients’ group in the LUEUR cohort.

To identify differences related to suicidal phenotype, analyses to find variables associated with suicidal depression were adjusted for depression severity in both cohorts, and for age in the GENESE cohort.

When patients with lifetime history of SA were removed from the analysis, results did not change (except for sociodemographic data which were no longer different among groups) (Supplemental Table S1).

Depression score changes during the follow-up

In both cohorts, HADS-D scores significantly decreased in each group during the follow-up (Figure 1). In the LUEUR cohort, but not in the GENESE cohort, the HADS-D score decrease over time was different according to the patients’ groups (respectively, global p < 0.0001 for the unadjusted and adjusted models and global p = 0.07 for the unadjusted and adjusted models; Table 2).

Table 2.

Multivariate analysis of HADS-D total score changes during the follow-up.

Variables	Without adjustments			With adjustments
	Coefficients	95% CI	p value	Coefficients	95% CI	p value
LUEUR
Patients groups			<0.0001			<0.0001
Moderate suicidal depression	0.855	[0.545, 1.166]		0.825	[0.517, 1.133]
Severe suicidal depression	2.732	[2.378, 3.09]		2.594	[2.240, 2.947]
Period	−0.188	[−0.197, −0.179]	<0.0001	−0.188	[−0.197, −0.179]	<0.0001
Interaction between patient groups and period			<0.0001			<0.0001
Moderate suicidal depression: period	−0.013	[−0.023, −0.003]		−0.013	[−0.023, −0.003]
Severe suicidal depression: period	−0.028	[−0.039, −0.016]		−0.028	[−0.039, −0.016]
GENESE
Patients groups			<0.0001			<0.0001
Moderate suicidal depression	0.613	[0.201, 1.024]		0.529	[0.120, 0.937]
Severe suicidal depression	1.968	[1.521, 2.415]		1.756	[1.311, 2.202]
Period	−0.172	[−0.183, −0.161]	<0.0001	−0.172	[−0.183, −0.162]	<0.0001
Interaction between patient groups and period			0.07			0.07
Moderate suicidal depression: period	−0.007	[−0.019, 0.005]		−0.007	[−0.019, 0.005]
Severe suicidal depression: period	−0.014	[−0.027, −0.001]		−0.014	[−0.027, −0.001]

GLS: generalized least squares; MDE: major depressive episode; HADS-D: Hospital Anxiety and Depression Scale–Depression; CI: confidence interval.

GLS model (to compare depression score changes among groups and to evaluate whether the interactions between patients’ groups and depression score changes during the follow-up were significant) with 2314 patients for the GENESE cohort (223 patients were removed because of missing values*) and 3624 patients (345 patients were removed because of missing values*) for the LUEUR cohort and HADS total score (i.e. the outcome) during the follow-up; the variance-covariance matrix is an AR(1) process. Reference group: Patients without suicidal depression, study level (junior secondary school) and associated treatment (yes). The adjusted model was adjusted for age, study level, associated treatment and current MDE duration.

Missing values in variables for adjustment. This model cannot be performed with missing data in adjustment variables. All patients with at least two HADS-D measurements and complete data for the adjustment variables were included in this analysis.Bold values means that results are significant (p-value < 0.05).

Figure 1.

HADS-D score changes during the follow-up in the three groups of the (A) LUEUR and (B) GENESE cohorts.

Response to treatment (i.e. 50% reduction of the baseline HADS-D score) was not significantly different among groups in unadjusted models. However, when the analysis was adjusted for potential cofounders (listed in Table 3), there were significant differences at week 6 in the LUEUR cohort (OR = 0.77; 95% CI = [0.61, 0.97]; global p = 0.04) (Table 3) and a trend for an association in the GENESE cohort.

Table 3.

Response to treatment during the follow-up.

Suicidal depression	HADS-D score reduction by at least 50% (LUEUR)				Model 0		Model 1
	No		Yes
	n	%	n	%	OR [95% CI]	p value	OR [95% CI]	p value
Week 6	1684 (45.7)		1997 (54.3)
No	287	17	335	16.7	1	0.47	1	0.04
Moderate	950	56.4	1164	58.3	1.05 [0.88, 1.26]		0.94 [0.78, 1.14]
Severe	447	26.5	498	24.9	0.95 [0.78, 1.17]		0.77 [0.61, 0.97]
Suicidal depression	HADS-D score reduction by at least 50% (GENESE)				Model 0		Model 1
	No		Yes
	n	%	n	%	OR [95% CI]	p value	OR [95% CI]	p value
Week 6	1179 (47.9)		1285 (52.1)
No	166	14.1	191	14.9	1	0.25	1	0.08
Moderate	646	54.8	735	57.1	0.99 [0.78, 1.25]		0.97 [0.76, 1.25]
Severe	367	31.1	361	28	0.86 [0.66, 1.10]		0.78 [0.59, 1.04]

OR: odds ratio; CI: confidence interval; HADS-D: Hospital Anxiety and Depression Scale–Depression; MDE: major depressive episode; SA: suicide attempt.

Model 0: crude association. Model 1: adjusted for age, study level, lifetime SA, baseline HADS total score, current MDE duration, associated treatment.

Bold values means that results are significant (p-value < 0.05).

At week 6, depression remission (HADS-D score ⩽7) was observed in 53.5% (LUEUR) and 53.5% (GENESE) of depressed patients without SI, 50.1% (LUEUR) and 51% (GENESE) of patients with moderate SI and 39.7% (LUEUR) and 42.6% (GENESE) of patients with severe SI. Depression remission rates were significantly lower in patients with severe SI than in patients without SI in both cohorts (univariate analyses). This difference remained significant in the adjusted model in the LUEUR cohort (OR = 0.62; 95% CI = [0.44, 0.88]; global p = 0.01), but not in the GENESE cohort (OR = 0.80; 95% CI = [0.61, 1.05]; global p = 0.16) (Table 4). When patients with lifetime history of SA were removed from the analysis, the rate of depression remission remained lower in patients with severe SI than in the other two groups in the unadjusted and adjusted models (Supplemental Table S2).

Table 4.

Remission of depression in each group at week 6.

Variables	Depression remission				Model 0		Model 1
	No		Yes
	1913		1768
	n	%	n	%	OR [95% CI]	p value	OR [95% CI]	p value
Suicidal depression (LUEUR cohort)
No	289	15.1	333	18.8	1	<0.001	1	0.01
Moderate	1054	55.1	1060	60.0	0.87 [0.73, 1.04]		0.85 [0.64, 1.13]
Severe	570	29.8	375	21.2	0.57 [0.47, 0.70]		0.62 [0.44, 0.88]
	Depression remission
	No		Yes
	1260		1206
Suicidal depression (GENESE cohort)
No	166	13.2	191	15.8		<0.001		0.16
Moderate	676	53.7	705	58.5	0.91 [0.72, 1.14]		0.95 [0.74, 1.22]
Severe	418	33.2	310	25.7	0.65 [0.50, 0.83]		0.80 [0.61, 1.05]

OR: odds ratio; CI: confidence interval; HADS: Hospital Anxiety and Depression Scale; MDE: major depressive episode; SA: suicide attempt.

Model 0: crude association; Model 1: adjusted for age, study level, professional activity, current MDE duration, lifetime SA, associated treatment and treatment instauration/change in the GENESE cohort, and for the same variables plus marital status, age at first episode, number of MDE lifetime, alcohol dependence and baseline HADS total score in the LUEUR cohort.Bold values means that results are significant (p-value < 0.05).

SI changes during the follow-up

SI significantly decreased in all three patient groups during the follow-up (Figures 2 and 3). Nevertheless, SI (MADRS-SI score ⩾2) occurrence remained higher in patients with moderate and severe SI than in patients without SI in both cohorts. At week 6, 12.2% (GENESE cohort) and 7.4% (LUEUR cohort) of patients with severe SI still reported severe SI (MADRS-SI score ⩾4).

Figure 2.

MADRS-SI score changes during the follow-up in the three groups of the (A) LUEUR and (B) GENESE cohorts.

Figure 3.

MADRS-SI score changes during the follow-up in the three groups of the (A) LUEUR and (B) GENESE cohorts.

In both cohorts, the predictors of persisting severe SI at week 6 were longer MDE duration, lifetime history of SA, higher baseline MADRS-SI score and lower HADS score decrease, as well as lower decreases of Hopelessness scale (LUEUR cohort), Impulsivity scale and MADRS-Sleep scores (GENESE cohort) (Supplemental Table S3). In the multivariate model, the main predictors were higher baseline MADRS-SI score, lower improvement of the HADS-D score and lower decrease of the MADRS-Sleep score (Table 5).

Table 5.

Risk factors for persistent severe SI at the follow-up end in patients with severe suicidal depression at baseline (multivariate analysis).

Variable	LUEUR		GENESE
	OR [95% CI]	p value	OR [95% CI]	p value
Study level		0.16		0.12
Junior secondary school	1		2.29 [1.01, 5.18]
Senior secondary school	0.56 [0.17, 1.84]		1.53 [0.58, 4.05]
University	2.83 [0.53, 15.01]		1
Current MDE duration (months)		0.42		0.35
<2	1		1
⩾2	1.55 [0.53, 4.51]		1.20 [0.54, 2.68]
⩾6			1.69 [0.79, 3.61]
Cumulated MDE duration lifetime (weeks)			1.01 [0.99, 1.01]	0.24
Number of MDE lifetime	1.02 [0.94, 1.10]	0.70
Age at first MDE (years)	1.02 [0.98, 1.07]	0.33
MADRS-SI baseline	6.02 [2.62, 13.87]	<0.001	2.69 [2.08, 3.49]	<0.001
Decrease in HADS total score from baseline to week 6	1.56 [1.36, 1.80]	<0.001	1.16 [1.10, 1.21]	<0.001
Change in Hopelessness scale score from baseline to week 6	0.98 [0.78, 1.23]	0.88
Change in Impulsivity scale score from baseline to week 6			1.01 [0.95, 1.08]	0.713
Decrease in MADRS-Sleep score from baseline to week 6			1.38 [1.12, 1.69]	0.003
Lifetime SA		0.61		0.45
No	1		1
Yes	1.34 [0.43, 4.14]		1.28 [0.68, 2.41]
Treatment instauration/change		0.14		0.79
Change	0.44 [0.15, 1.30]		0.92 [0.50, 1.70]
Instauration	1		1
Associated treatment		0.32
No	1
Yes	1.75 [0.59, 5.25]

OR: odds ratio; CI: confidence interval; MDE: major depressive episode; SA: suicide attempt; MADRS-SI: suicidal item of the Montgomery–Åsberg Depression Rating Scale; HADS: Hospital Anxiety and Depression Scale.Bold values means that results are significant (p-value < 0.05).Shading means that the variable was not available in the cohort (since some variables were available in one cohort and not in the other).

SI remission at week 6 in patients with severe SI at baseline was predicted by shorter MDE duration; higher decrease in the first 2 weeks of depression, anxiety, hopelessness, impulsivity, sleep disturbances and SI; fewer lifetime SA; lower baseline MADRS-SI score; and instauration rather than a change of antidepressant treatment in both cohorts (Supplemental Table S4). In the multivariate model, fewer lifetime SA and higher early decrease of the MADRS-SI score were the only predictive factors of SI remission that remained significant for both cohorts (Table 6).

Table 6.

Predictors of SI remission (MADRS-SI score <2) at the follow-up end in patients with severe suicidal depression at baseline (multivariate analysis).

Variable	LUEUR		GENESE
	OR [95% CI]	p value	OR [95% CI]	p value
Sex		0.27
Men	1
Women	0.81 [0.56, 1.18]
Age			1.03 [1.00, 1.06]	0.04
Study level		0.007
Junior secondary school	1
Senior secondary school	0.73 [0.48, 1.11]
University	1.60 [1.02, 2.51]
Professional activity				0.32
Working			1
Unemployment			0.82 [0.37, 1.84]
Retired			0.42 [0.17, 1.05]
Sick leave			0.84 [0.44, 1.62]
Current MDE duration (months)		0.30		0.001
<2	1		1
⩾2	0.78 [0.53, 1.15]		1.09 [0.60, 1.99]
⩾6	0.58 [0.21, 1.56]		0.37 [0.19, 0.70]
Number of MDE lifetime			0.96 [0.87, 1.06]	0.96
Decrease in HADS total score from baseline to week 2	1.13 [1.08, 1.17]	<0.001	1.01 [0.96, 1.06]	0.74
Change in Hopelessness scale score from baseline to week 2	1.03 [0.94, 1.12]	0.52
Change in Impulsivity scale score from baseline to week 2			1.03 [0.97, 1.10]	0.36
Decrease in MADRS-Sleep score from baseline to week 2			1.15 [0.93, 1.42]	0.21
Decrease in MADRS-SI score from baseline to week 2	2.22 [1.76, 2.80]	<0.001	1.67 [1.28, 2.18]	<0.001
Lifetime SA		0.03		0.02
No	1.52 [1.04, 2.23]		1.93 [1.09, 3.42]
Yes	1		1
Treatment instauration/change		0.14		0.37
Change	1.37 [0.90, 2.09]		0.78 [0.44, 1.36]
Instauration	1		1
Associated treatment		0.03
No	1.61 [1.06, 2.43]
Yes	1

OR: odds ratio; CI: confidence interval; MDE: major depressive episode; SA: suicide attempt; MADRS-SI: suicidal item of the Montgomery–Åsberg Depression Rating Scale; HADS: Hospital Anxiety and Depression Scale; SI: suicidal ideation.Bold values means that results are significant (p-value < 0.05).Shading means that the variable was not available in the cohort (since some variables were available in one cohort and not in the other).

The analysis of variables that explained SI variation over time showed that 23% (LUEUR) and 25% (GENESE) of SI variances were explained by changes in the HADS-D score, 1.2% (LUEUR) and 1% (GENESE) by changes in the HADS-A score, 4.5% (GENESE only) by changes in the sleep score and less than 1% by changes in the Hopelessness (LUEUR) and Impulsivity (GENESE) score (Table 7). When considering only patients with severe SI at baseline, depression score changes explained 39% (GENESE) and 46% (LUEUR) of SI variance (Supplemental Table S5).

Table 7.

Linear regression analysis of MADRS-SI score variation in all patients.

Variables	Beta (SD)Crude	p value	Beta (SD)Fully adjusted	p value	% of variance explained by the variable	R² fully adjusted
LUEUR
Anxiety	0.1213 (0.0042)	<0.001	0.0439 (0.0057)	<0.001	1.2	0.2513
Hopelessness	0.0666 (0.0065)	<0.001	0.0259 (0.0058)	<0.001	<1	0.2513
GENESE
Anxiety	0.1262 (0.0051)	<0.001	0.0256 (0.0074)	0.0005	1	0.3106
Impulsivity	0.0727 (0.0036)	<0.001	0.0104 (0.0043)	0.02	<1	0.3106
Sleep	0.3673 (0.0147)	<0.001	0.2042 (0.0160)	<0.001	4.5	0.3106

MADRS-SI: suicidal item of the Montgomery–Åsberg Depression Rating Scale.Bold values means that results are significant (p-value < 0.05).

At week 6, among all patients with SI at baseline (MADRS-SI ⩾2), SI (but not depression) was still present in 46 (1.5%) and 115 (5.6%) patients, whereas depression (but not SI) was still present in 900 (39.2%) and 434 (21.2%) patients in the LUEUR and GENESE cohorts, respectively. Among the patients with severe SI at baseline (MADRS-SI ⩾4), SI (but not depression) was still present in 22 (2.4%) and 56 (7.9%) patients, whereas depression (but not SI) was still present in 178 (19.2%) and 93 (13.2%) patients in the LUEUR and GENESE cohorts, respectively.

SA during the follow-up

During the follow-up, 68 (1.9%) patients in the LUEUR cohort and 15 (0.6%) in the GENESE cohort reported SA. The risk of SA during the follow-up was higher for patients with severe SI. In the LUEUR cohort (not applicable in the GENESE cohort due to the too small effective of SA), the number of patients with SA during the follow-up was significantly higher in patients with severe SI than in the other two groups in unadjusted and adjusted models (OR = 3.17; 95% CI = [1.80, 5.57], p < 0.001 for the adjusted model, Table 8).

Table 8.

SA occurrence during the follow-up in the three groups (binary logistic regression).

Variables	SA during the follow-up				Model 0		Model 1
	No		Yes
	3567		68
	n	%	n	%	OR [95% CI]	p value	OR [95% CI]	p value
Suicidal depression (LUEUR cohort)
No/Moderate	2664	74.7	26	38.2	1	<0.001	1	<0.001
Severe	903	25.3	42	61.8	4.77 [2.91, 7.82]		3.17 [1.80, 5.57]
	SA during the follow-up
	No		Yes
	2452		15
Suicidal depression (GENESE cohort)
No/Moderate	1726	70.4	8	53.3		NA		NA
Severe	726	29.6	7	46.7

SA: suicide attempt; OR: odds ratio; CI: confidence interval; HADS: Hospital Anxiety and Depression Scale.

Model 0: crude association. Model 1: adjusted for marital status, professional activity, alcohol dependence, lifetime SA and baseline HADS total score.Bold values means that results are significant (p-value < 0.05).

Discussion

To our knowledge, this is the first study that characterized specifically suicidal depression in two large cohorts of depressed outpatients. Our results strengthen the hypothesis that suicidal depression could be a specific phenotype of depression, especially in the presence of severe SI. Indeed, baseline clinical features were more severe in patients with suicidal depression, and not only the severity of the current depressive episode. Interestingly, some of these features are considered to increase suicidal risk: greater number of lifetime MDEs (Trivedi et al., 2013), higher levels of anxiety and depression (Hoertel et al., 2018), hopelessness (Hawton et al., 2013), impulsivity (Coryell et al., 2018), more sleep disturbance (Bernert et al., 2015), more frequent history of SA (Hawton et al., 2013) and higher number of associated treatments and antidepressant treatment changes for the current MDE. These differences mostly persisted after adjusting for depression severity, suggesting that SI might help to define a more homogeneous type of depressive illness. Moreover, a gradient of SI severity was observed. Clinical features were more severe in depressed patients with moderate SI than in patients without SI, and in patients with severe SI than in patients with moderate SI.

Second, depression changes over time were different in the three groups of patients. Although depression severity decreased more in severe suicidal depressed patients than in the two other groups (significant difference of the slope with a higher slope for patients with severe SI, as shown by the GLS), this reduction was not sufficient to obtain depression remission at the end of the follow-up. Indeed, analysis of the depression changes during the 6 weeks of treatment showed that the number of responders to antidepressant treatment was lower. Remission at the end of the follow-up was less frequent in patients with severe SI than in the other two groups. Lower depression remission rate despite a greater decrease of depression severity was previously observed in a population of elderly depressed patients (Bogers et al. 2017). Despite the important decrease of depressive symptomatology in suicidal patients after antidepressant therapy initiation, these patients might have more persisting ‘residual’ symptoms. By assessing depression changes in different ways, our study strengthens previous results suggesting that in patients with depression at high risk of suicide (patients with current severe SI and/or lifetime SA), depression course is different than in non-suicidal patients and with a lower remission rate at the end of the follow-up (Lopez-Castroman et al., 2016; Pompili et al., 2010). This is not a real surprise when considering that patients with SI/SB are regularly excluded from clinical trials of antidepressants, and consequently, the real efficiency of antidepressants in this population remains unclear (Courtet and Nobile, 2020; Iltis et al., 2020).

The percentage of patients with SI during the follow-up remained higher in the group with severe SI, and the risk of SA during the follow-up was also threefold higher in this group. Moreover, SI remission (MADRS-SI score <2) was observed in 56.2% (LUEUR) and 47.6% (GENESE) of patients in the severe SI at week 6, and in 10% (GENESE) and 25% (LUEUR) at week 2 of treatment. Although not directly comparable, it is interesting to observe that with two new potential antisuicidal agents, SI decreased in about 60% of patients at day 1 with ketamine (Grunebaum et al., 2018) and in about 54% of patients at day 2 with esketamine (Canuso et al., 2018).

In our study, SI persistence was observed in 10% of patients with severe SI, as previously shown (Kasckow et al., 2016). Importantly, these patients were more at risk of SA during the follow-up than those without SI persistence. The best predictors of SI persistence were high baseline MADRS-SI score, lower decrease of HADS scores and sleep disturbances. Patients with severe SI who respond less to treatment could thus be at higher risk of SI persistence and SA. Therefore, it is primordial to strengthen the follow-up of patients with SI, sleep disturbances and no depression improvement after antidepressant therapy (McCall et al., 2019).

Furthermore, greater early SI decrease (during the first 2 weeks of treatment) and absence of lifetime history of SA were the two main predictors for SI remission in patients with severe SI at baseline in both cohorts. This suggests that the best predictors of SI remission are linked to the patients’ suicidal status rather than to their depression status. Therefore, SI decrease might be related not only to a decrease of depression severity. Indeed, depression changes explained about 25% of SI variation in the entire population and 45% when including only patients with severe SI. This result reinforces the hypothesis that patients with severe SI are different. Nevertheless, depression changes over time explained only less than 50% of SI variation, suggesting that other variables may interact with SI (i.e. sleep that explained almost 5% of SI variance). Only targeting depression is not sufficient to decrease SI as shown by SI persistence despite depression remission in some patients. Thus, depression improvement does not ensure SI resolution (Seo et al., 2014).

There are some discrepancies in results between both cohorts, mainly differences in remission of depression rates between patient’s groups according to studies. This may be explained by the two main differences in exclusion/inclusion criteria between cohorts. Indeed, patients on antipsychotics or mood stabilizers could be included in the GENESE, but not in the LUEUR cohort. It is known that antipsychotic drugs in combination with antidepressants and lithium can improve the efficacy of treatments for depression and SI in suicidal patients (Smith and Cipriani, 2017; Vermeulen et al., 2019). Therefore, the treatment might have influenced depression changes between cohorts. Moreover, alcohol abuse or dependence was an exclusion criterion in the GENESE but not in the LUEUR cohort; such comorbidity may impact the course of depression and SI.

Although specific treatments are lacking for suicidal patients, new promising drugs (such as ketamine or esketamine) are currently tested. Ketamine reduces SI in depressed patients (Grunebaum et al., 2018). Importantly, ketamine acts rapidly, unlike classical antidepressants, and reduces SI in at least half of the patients in the 24 hours following its administration (Ballard et al., 2014; Grunebaum et al., 2018). Therefore, it could be used as interventional treatment in patients with depression and severe SI to reduce the immediate risk of suicide and because antidepressants do not seem to affect SI in this group of patients. Development of specific psychotherapies to target SI is also justified in this context.

Our study has some limitations. First, a single item of the MADRS was used to define suicidal depression. However, self-report has been suggested to better catch SI than clinician reports (Czyz et al., 2016), and using a single item is a valid approach for SI evaluation (Desseilles et al., 2012). Second, the follow-up lasted only 6 weeks. However, this is the usual time needed to observe a response to antidepressant treatment. Furthermore, the interval between SI onset and SA is often very short and it is primordial to act on SI rapidly (Courtet et al., 2020; Glenn and Nock, 2014). Moreover, a longer follow-up could have increased the number of patients lost to follow-up.

Our study has some strengths. Suicidal depression was characterized in two large cohorts of outpatients with depression and with few individuals lost to follow-up (<10%). Almost all the results were replicated in the two cohorts (more than 2000 patients in each cohort), reinforcing their validity. Finally, these cohorts are representative of the population of depressed patients because they were recruited throughout France by general practitioners or psychiatrists and were outpatients. Interestingly, about 20% of patients included in the two cohorts met the inclusion criteria of clinical trials on new medications, leaving a large space for offering new therapeutic options to these patients.

Our results suggest that suicidal depression could be a specific phenotype of depression with more severe clinical characteristics, less frequent clinical remission of depression, SI persistence and higher risk of SA, despite antidepressant treatment. This suggests that SI may be a marker of severity of the depressive illness and of persistence of the suicide risk. Then, clinicians should implement personalized care with specific strategies and continuous monitoring of suicide risk in patients presenting with suicidal depression. This phenotype mainly concerns patients with severe SI because the characteristics of patients with moderate SI were closer to those of patients without SI. Our results contribute to the hypothesis that depression and SI are related but could also have independent physiopathology because remission of depression is not always associated with remission of SI. Moreover other factors might explain SI changes over time. Antidepressant drugs do not seem to be sufficient in patients with suicidal depression. These patients should be closely monitored. Combined treatments with antidepressants (pharmacological or non-pharmacological) and novel therapeutic strategies could be needed.

Supplemental Material

sj-docx-1-anp-10.1177_00048674211025697 – Supplemental material for Characteristics and treatment outcome of suicidal depression: Two large naturalistic cohorts of depressed outpatients

Supplemental material, sj-docx-1-anp-10.1177_00048674211025697 for Characteristics and treatment outcome of suicidal depression: Two large naturalistic cohorts of depressed outpatients by Bénédicte Nobile, Emilie Olié, Jonathan Dubois, Sebastien Guillaume, Philip Gorwood and Philippe Courtet in Australian & New Zealand Journal of Psychiatry

Footnotes

Author Contributions

B.N. performed statistical analyses and wrote the paper. J.D. contributed to the statistical analysis. P.G., S.G., E.O. and P.C. contributed to the data interpretation and to manuscript writing. All authors have contributed to the manuscript and have accepted the final version of the paper.

Declaration of Conflicting Interests

The author(s) declared the following potential conflicts of interest with respect to the research, authorship and/or publication of this article: The authors have interests in relation with one or more organization that could be perceived as a possible conflict of interest in the context of the subject of this manuscript. The relationships are summarized as follows. B.N., E.O. and J.D. report no financial relationships with commercial interests. S.G. received honoraria or research or educational conference grants from Bristol-Myers Squibb, Otsuka, Servier, Lundbeck, AstraZeneca and Janssen. P.G. reports no shares; has paid positions at University of Paris Descartes & Hospital Sainte-Anne; is on the advisory board at AstraZeneca, Janssen, Servier and Wyeth; and has no other involvement. P.C. reports no shares; has paid positions at University of Montpellier and CHU Montpellier; is on the advisory board at Janssen; and received honoraria or research or educational conference grants from Otsuka, Lundbeck, Exeltis, Pfizer and Janssen.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship and/or publication of this article: This work was supported by a research grant from Servier that had no involvement in the study design, organization, analysis or preparation for publication.

ORCID iD

Bénédicte Nobile

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author, B.N. The data are not publicly available due to restrictions (their containing information that could compromise the privacy of research participants).

Supplemental Material

Supplemental material for this article is available online.

References

Anon (n.d) The Measurement of Emotions, 1st Edition. Available at: www.elsevier.com/books/the-measurement-of-emotions/plutchik/978-0-12-558704-4 (accessed 19 April 2019).

Ballard

Ionescu

Vande Voort

, et al (2014) Improvement in suicidal ideation after ketamine infusion: Relationship to reductions in depression and anxiety. Journal of Psychiatric Research 58: 161–166.

Ballard

Snider

Nugent

, et al (2017) Active suicidal ideation during clinical antidepressant trials. Psychiatry Research 257: 303–308.

Batterham

Spijker

BAJ

Mackinnon

, et al (2019) Consistency of trajectories of suicidal ideation and depression symptoms: Evidence from a randomized controlled trial. Depression and Anxiety 36: 321–329.

Bernert

Kim

Iwata

, et al (2015) Sleep disturbances as an evidence-based suicide risk factor. Current Psychiatry Reports 17: 554.

Bernert

Luckenbaugh

Duncan

, et al (2017) Sleep architecture parameters as a putative biomarker of suicidal ideation in treatment-resistant depression. Journal of Affective Disorders 208: 309–315.

Bogers

ICHM

Zuidersma

Boshuisen

, et al (2017) The influence of thoughts of death and suicidal ideation on the course of depression in older depressed patients. International Journal of Geriatric Psychiatry 32: 882–891.

Canuso

Singh

Fedgchin

, et al (2018) Efficacy and safety of intranasal esketamine for the rapid reduction of symptoms of depression and suicidality in patients at imminent risk for suicide: Results of a double-blind, randomized, placebo-controlled study. The American Journal of Psychiatry 175: 620–630.

Coryell

Wilcox

Evans

, et al (2018) Aggression, impulsivity and inflammatory markers as risk factors for suicidal behavior. Journal of Psychiatric Research 106: 38–42.

10.

Courtet

Nobile

(2020) Inclusion of suicidal individuals in research studies. The Journal of Clinical Psychiatry 81: 20com13276.

11.

Courtet

Lopez-Castroman

Jaussent

, et al (2014) Antidepressant dosage and suicidal ideation. JAMA Internal Medicine 174: 1863–1865.

12.

Courtet

Nobile

Guillaume

, et al (2020) An urgent need for rapid anti-suicidal drugs. French Journal of Psychiatry 1: 1–4.

13.

Czyz

Horwitz

King

(2016) Self-rated expectations of suicidal behavior predict future suicide attempts among adolescent and young adult psychiatric emergency patients. Depression and Anxiety 33: 512–519.

14.

Demyttenaere

Verhaeghen

Dantchev

, et al (2009) ‘Caseness’ for depression and anxiety in a depressed outpatient population: Symptomatic outcome as a function of baseline diagnostic categories. The Primary Care Companion to the Journal of Clinical Psychiatry 11: 307–315.

15.

Desseilles

Perroud

Guillaume

, et al (2012) Is it valid to measure suicidal ideation by depression rating scales? Journal of Affective Disorders 136: 398–404.

16.

Ducasse

Jaussent

Arpon-Brand

, et al (2018) Acceptance and commitment therapy for the management of suicidal patients: A randomized controlled trial. Psychotherapy and Psychosomatics 87: 211–222.

17.

Franklin

Ribeiro

Fox

, et al (2017) Risk factors for suicidal thoughts and behaviors: A meta-analysis of 50 years of research. Psychological Bulletin 143: 187–232.

18.

Glenn

Nock

(2014) Improving the short-term prediction of suicidal behavior. American Journal of Preventive Medicine 47: S176–S180.

19.

Grunebaum

Galfalvy

Choo

T-H

, et al (2018) Ketamine for rapid reduction of suicidal thoughts in major depression: A Midazolam-Controlled Randomized Clinical Trial. The American Journal of Psychiatry 175: 327–335.

20.

Gusmão

Quintão

McDaid

, et al (2013) Antidepressant utilization and suicide in Europe: An ecological multi-national study. PLoS ONE 8: e66455.

21.

Hawton

Casañas

Comabella

Haw

, et al (2013) Risk factors for suicide in individuals with depression: A systematic review. Journal of Affective Disorders 147: 17–28.

22.

Hoertel

Blanco

Olfson

, et al (2018) A comprehensive model of predictors of suicide attempt in depressed individuals and effect of treatment-seeking behavior: Results from a National 3-Year Prospective Study. The Journal of Clinical Psychiatry 79: 17m11704.

23.

Iltis

McCall

Deria

(2020) Suicidality, depression, and the FDA: Health inequities and the ethical conduct of research. The Journal of Clinical Psychiatry 81: 19m13050.

24.

Kasckow

Youk

Anderson

, et al (2016) Trajectories of suicidal ideation in depressed older adults undergoing antidepressant treatment. Journal of Psychiatric Research 73: 96–101.

25.

Lee

Bryan

Rudd

(2020) Longitudinal suicide ideation trajectories in a clinical trial of brief CBT for U.S. Military personnel recently discharged from psychiatric hospitalization. Psychiatry Research 293: 113335.

26.

Lopez-Castroman

Jaussent

Gorwood

, et al (2016) Suicidal depressed patients respond less well to antidepressants in the short term. Depression and Anxiety 33: 483–494.

27.

Lopez-Castroman

Jaussent

Gorwood

, et al (2020) Activation syndrome induced by the antidepressant tianeptine and suicidal ideation: Evidence from a large depressed outpatient sample. Progress in Neuro-Psychopharmacology and Biological Psychiatry 97: 109762.

28.

McCall

Benca

Rosenquist

, et al (2019) Reducing suicidal ideation through insomnia treatment (REST-IT): A randomized clinical trial. The American Journal of Psychiatry 176: 957–965.

29.

Mewton

Andrews

(2016) Cognitive behavioral therapy for suicidal behaviors: Improving patient outcomes. Psychology Research and Behavior Management 9: 21–29.

30.

Molero

Ramos-Quiroga

Martin-Santos

, et al (2018) Antidepressant efficacy and tolerability of ketamine and esketamine: A Critical Review. CNS Drugs 32: 411–420.

31.

Nobile

Dubois

Aouizerate

, et al (2021) Characterization of depressed bipolar patients with current suicidal ideation. Australian and New Zealand Journal of Psychiatry 55: 289–304.

32.

Nobile

Jaussent

Gorwood

, et al (2018) Tianeptine is associated with lower risk of suicidal ideation worsening during the first weeks of treatment onset compared with other antidepressants: A naturalistic study. Journal of Psychiatric Research 96: 167–170.

33.

Nobile

Lutz

P-E

Olié

, et al (2020) The role of opiates in social pain and suicidal behavior. Current Topics in Behavioral Neurosciences 46: 197–210.

34.

Pompili

Baldessarini

Tondo

, et al (2010) Response to intravenous antidepressant treatment by suicidal vs. nonsuicidal depressed patients. Journal of Affective Disorders 122: 154–158.

35.

Perroud

, et al (2010) History of suicide attempts among patients with depression in the GENDEP project. Journal of Affective Disorders 123: 131–137.

36.

Reeves

Batra

May

, et al (2008) Efficacy of risperidone augmentation to antidepressants in the management of suicidality in major depressive disorder: A randomized, double-blind, placebo-controlled pilot study. The Journal of Clinical Psychiatry 69: 1228–1236.

37.

Rihmer

Gonda

(2013) Pharmacological prevention of suicide in patients with major mood disorders. Neuroscience & Biobehavioral Reviews 37: 2398–2403.

38.

Rogers

Stanley

Hom

, et al. (2018) Conceptual and empirical scrutiny of covarying depression out of suicidal ideation. Assessment 25: 159–172.

39.

Seo

H-J

Jung

Y-E

Jeong

, et al (2014) Persistence and resolution of suicidal ideation during treatment of depression in patients with significant suicidality at the beginning of treatment: The CRESCEND study. Journal of Affective Disorders 155: 208–215.

40.

Smith

Cipriani

(2017) Lithium and suicide in mood disorders: Updated meta-review of the scientific literature. Bipolar Disorders 19: 575–586.

41.

Sokero

(2006) Suicidal ideation and attempts among psychiatric patients with major depressive disorder. Journal of Clinical Psychiatry 64: 1094–1100.

42.

Trivedi

Morris

Wisniewski

, et al (2013) Clinical and sociodemographic characteristics associated with suicidal ideation in depressed outpatients. Canadian Journal of Psychiatry/Revue Canadienne De Psychiatrie 58: 113–122.

43.

van Ballegooijen

Eikelenboom

Fokkema

, et al (2019) Comparing factor structures of depressed patients with and without suicidal ideation, a measurement invariance analysis. Journal of Affective Disorders 245: 180–187.

44.

Vermeulen

van Rooijen

van de Kerkhof

MPJ

, et al (2019) Clozapine and long-term mortality risk in patients with schizophrenia: A systematic review and meta-analysis of studies lasting 1.1–12.5 years. Schizophrenia Bulletin 45: 315–329.

45.

Voegeli

Ramoz

Shekhtman

, et al (2016) Neurotrophin genes and antidepressant-worsening suicidal ideation: A prospective case-control study. The International Journal of Neuropsychopharmacology 19: pyw059.

46.

Weitz

Hollon

Kerkhof

, et al (2014) Do depression treatments reduce suicidal ideation? The effects of CBT, IPT, pharmacotherapy, and placebo on suicidality. Journal of Affective Disorders 167: 98–103.

47.

Wilkinson

Ballard

Bloch

, et al (2018) The effect of a single dose of intravenous ketamine on suicidal ideation: A systematic review and individual participant data meta-analysis. The American Journal of Psychiatry 175: 150–158.

48.

World Health Organization (2019) Suicide in the World: Global Health Estimates. Geneva: World Health Organization.

49.

Yovell

Bar

Mashiah

, et al (2016) Ultra-low-dose buprenorphine as a time-limited treatment for severe suicidal ideation: A randomized controlled trial. The American Journal of Psychiatry 173: 491–498.

50.

Zimmerman

Balling

Chelminski

, et al (2018) Understanding the severity of depression: Which symptoms of depression are the best indicators of depression severity? Comprehensive Psychiatry 87: 84–88.

51.

Zisook

Trivedi

Warden

, et al (2009) Clinical correlates of the worsening or emergence of suicidal ideation during SSRI treatment of depression: An examination of citalopram in the STAR*D study. Journal of Affective Disorders 117: 63–73.

52.

Zisook

, et al. (2011) Effect of antidepressant medication treatment on suicidal ideation and behavior in a randomized trial: An exploratory report from the combining medications to enhance depression outcomes study. The Journal of Clinical Psychiatry 72: 1322–1332.

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

0.07 MB