Multi-method assessment of suicidal thoughts and behaviors among patients in treatment for OCD and related disorders

Abstract

Obsessive-compulsive and related disorders (OCRDs) are associated with increased risk of suicidal thoughts and behaviors (STBs), yet research characterizing suicidality in OCRDs remains limited. A major challenge in assessing STBs is the reliance on explicit self-report. This study utilized multi-method assessment to examine changes in both implicit and explicit STBs in 31 adults receiving partial/residential treatment for OCRDs. Assessments were administered at admission and weekly during treatment. Approximately three-quarters of participants reported lifetime suicidal thoughts, with 16% reporting a prior suicide attempt. OCD severity was significantly correlated with lifetime suicidal thoughts, and was significantly higher for those with lifetime suicidal thoughts and prior attempts compared to those without. Implicit biases towards death were not associated with OCD severity, and did not predict explicitly endorsed STBs. This is the first study to measure both explicit and implicit STBs in adults with OCRDs. Limitations included small sample size and lack of racial/ethnic diversity. Given the majority had recent suicidal thoughts and one in six had a prior attempt, we emphasize the importance of STB assessment in OCD treatment settings.

1. Introduction

Obsessive compulsive disorder (OCD) is a chronic and often debilitating disorder marked by significant impairments in occupational and academic functioning, and diminished quality of life (Ledley & Pasupuleti, 2007; Ruscio et al., 2010; Yaryura-Tobias et al., 2000). The World Health Organization (2001) included OCD in the top 20 causes of illness-related disability for individuals 15–44 years old. Until recently, however, its associated risk for suicidality has received little attention. Prior research has documented significantly higher risk for both suicidal ideation (passive thoughts about wishing to be dead or active thoughts of killing oneself unaccompanied by preparatory behaviors; Posner et al., 2007) and suicide attempts among individuals with OCD compared to the general population, with OCD symptom severity positively correlated with suicidal thoughts and behaviors (STBs; suicidal ideation and suicidal acts, including active or passive suicidal ideation, suicide attempts, and preparatory behaviors for suicide attempts; Albert et al., 2019; Dell’Osso et al., 2018). Strikingly, a recent meta-analytic review suggested that approximately half of individuals with OCD have experienced suicidal ideation in their lifetimes, with one third reporting current suicidal ideation (Pellegrini et al., 2020). One study found that the most common risk factor for death by suicide among those with OCD was a prior attempt, and individuals with OCD were found to be more likely to have multiple attempts compared to the general population (Fernández de la Cruz et al., 2017).

However, accurate prevalence estimates for STBs among individuals with OCD remain unclear, with recent studies reporting widely ranging rates. Systematic reviews of adults with OCD have reported rates ranging from 26–74% for current suicidal ideation and 6–52% (Albert et al., 2018) and from 1–46% (Angelakis et al., 2015) for lifetime suicidal ideation. Possible explanations for the variability in these prevalence estimates include differences in sample characteristics (e.g., level of care, treatment history, symptom subtypes), assessment methods (e.g., self-report, clinician-rated), or operationalizations of suicidal ideation (e.g., passive vs. active ideation). Among a large international outpatient sample of individuals with OCD, 15% reported prior suicide attempts (Dell’Osso et al., 2018), and in a study of adults seeking intensive treatment for OCD (intensive outpatient, partial hospitalization, or residential treatment), nearly half reported current suicidal ideation (Hellberg et al., 2022). Overall, these findings highlight the need for increased research on STBs in individuals with OCD.

One major challenge in assessing, predicting, and preventing STBs is the reliance on explicit self-report. Ability and willingness to explicitly express suicidal thoughts and feelings are barriers to accurate assessment (Glenn et al., 2017; Nock et al., 2010), potentially due to feared consequences of disclosure (e.g., stigma, overreaction, involuntary hospitalization) or lack of insight into one’s own level of risk (Richards et al., 2019). Moreover, prior research has highlighted that clinicians’ assessments alone are little better than chance at predicting suicide attempts (Nock et al., 2022; Woodford et al., 2019). Busch and colleagues (2003) reported that approximately 80% of inpatients who died by suicide while in the hospital or immediately after discharge had previously denied suicidal ideation in their final communications, underscoring the critical need to more accurately assess risk. In a study of individuals who died by suicide across the US in 2015, intent to die by suicide was disclosed in only 22.4% of those without an identified mental health condition and 24.5% of those with a known mental health condition (Stone et al., 2018). Thus, a crucial need to refine methods of suicide risk assessment remains.

One promising method may be implicitly assessing suicide risk. The measurement of implicit cognitions has been used to predict a diverse range of psychological symptoms (e.g., anxiety, depression, disordered eating; Werntz et al., 2016), and appears to have potential in the application of suicide assessment, although there has been a lack of research focused on implicit suicide cognitions in OCD (e.g., Nock et al., 2010; Teachman et al., 2019). The Implicit Association Test (IAT; Greenwald et al., 1998) measures the strength of implicit associations between a range of constructs without the need for participant verbalization or explicit awareness of thoughts. Nock and colleagues (2010) developed the IAT for Death/Suicide, in which individuals are asked to respond as quickly as possible in associating individual words related to themselves (e.g., “I,” “myself,” etc.) and others (“they,” “theirs,” etc.) with the category of either “death” or “life.” Higher implicit association of death and self has been associated with higher odds of making a future suicide attempt, hence providing a behavioral marker for risk of STBs. In emergency-based psychiatric care settings, the IAT has been found to predict suicide attempts in adults at 6-month follow-up (Nock et al., 2010; Tello et al., 2020). Similarly, several studies among adolescents in short-term psychiatric care have also found the IAT to predict STBs at follow-up time points (e.g., Brent et al., 2021; Glenn et al., 2019; Glenn et al., 2017). These results did not replicate, however, in a diagnostically heterogenous inpatient sample of adults hospitalized due to severe suicidal ideation or a recent suicide attempt (Rath et al., 2021). While the IAT has demonstrated utility in predicting STBs in acute generalist psychiatric settings, its use in assessing suicide risk among individuals with OCD, and with adults in non-acute treatment settings longitudinally, has yet to be investigated (Teachman et al., 2019).

The current study examined longitudinal assessments of implicit suicide-related cognitions and explicitly endorsed STBs in a sample of adults receiving partial/residential treatment for OCRDs. More specifically, the study aims were to: 1) identify the prevalence and clinical correlates of self-reported STBs in treatment-seeking adults with OCRDs, 2) examine implicit and explicit measures of STBs and the associations between them, and 3) investigate changes in STBs longitudinally during treatment, utilizing IAT scores. We hypothesized that the IAT would differentiate those with and without current and prior STBs, predict explicitly endorsed suicidal thoughts over time during treatment, and decrease over the course of treatment (showing stronger associations between self and life).

2. Method

2.1. Participants

Participants were 31 adults with 52% identifying as female, 45% male; and a write-in response for one participant (3%) was “none.” Of these participants, 87% identified as non-Hispanic White, 7% as Multiracial, 3% as Middle Eastern or North African, and 3% of Hispanic, Latino, or Spanish origin. Regarding sexual orientation, 87% identified as heterosexual, 7% as gay or lesbian, 3% bisexual, and 3% not listed. The mean age in the sample was 29.26 years old (SD = 7.58). All participants were admitted for partial/residential treatment for OCRDs from March 2019 to March 2020. As assessed by structured clinical interview (see Procedure), 90% had a primary OCRD diagnosis (OCD: n = 27; excoriation disorder: n = 1). Other primary diagnoses included generalized anxiety disorder (n = 1), post-traumatic stress disorder (n = 1), and panic disorder (n = 1). The three participants who did not have primary OCRD had a secondary diagnosis of an OCRD (OCD: n = 2; body dysmorphic disorder: n = 1). Mean Y-BOCS at program admission was 26.16 (SD = 5.38); reflecting moderate-severe OCD (Storch et al., 2015). See detailed demographic data in Table 1.

Table 1.

Demographic Characteristics

	M (SD) n (%)
Age	29.26 (7.58)
Gender
Female	16 (51.61%)
Male	14 (45.16%)
Not listed - participant’s write-in response “none”	1 (3.23%)
Race/Ethnicity
White	27 (87.10%)
Multiraciala	2 (6.45%)
Hispanic, Latino, or Spanish origin	1 (3.23%)
Middle Eastern or North African	1 (3.23%)
Education
Some high school	1 (3.23%)
High school graduate/GED	1 (3.23%)
Some college	5 (16.13%)
Bachelor’s degree	15 (48.39%)
Graduate or professional degree	9 (29.03%)
Sexual Orientation
Heterosexual	27 (87.10%)
Gay/lesbian	2 (6.45%)
Bisexual	1 (3.23%)
Not listed	1 (3.23%)
Primary DSM-5 Diagnosis
Obsessive compulsive disorder	27 (87.10%)
Excoriation disorder	1 (3.23%)
Posttraumatic stress disorder	1 (3.23%)
Generalized anxiety disorder	1 (3.23%)
Panic disorder	1 (3.23%)
Comorbid DSM-5 Diagnoses
Major depressive disorder	12 (38.71%)
Body dysmorphic disorder	5 (16.13%)
Excoriation disorder	5 (16.13%)
Posttraumatic stress disorder	5 (16.13%)
Persistent depressive disorder	3 (9.68%)
Social anxiety disorder	3 (9.68%)
Obsessive compulsive disorder	2 (6.45%)
Specific phobia	2 (6.45%)
Alcohol use disorder	1 (3.23%)
Avoidant/restrictive food intake disorder	1 (3.23%)
Bipolar I	1 (3.23%)
Cannabis use disorder	1 (3.23%)
Generalized anxiety disorder	1 (3.23%)
Hoarding disorder	1 (3.23%)
Opioid use disorder	1 (3.23%)
Panic disorder	1 (3.23%)
Trichotillomania	1 (3.23%)

Note.

^a=participants who endorsed multiple race and ethnicity groups were grouped into the “multiracial” category. GED = General Educational Development.

Participants were recruited during their first week of treatment as part of a larger trial of cognitive bias modification (see Falkenstein et al., 2022); all procedures were approved by the Institutional Review Board. Inclusion criteria were: ≥ 18 years old, able to complete a 20 minute computer task, and ≥ 131 on the Obsessive Beliefs Questionnaire-44 [1 SD above the mean of the non-clinical sample reported in the original validation paper (OCCWG, 2005)]. Exclusion criteria were: acute symptoms of mania and/or psychosis, diagnosis of an autism spectrum disorder, diagnosis of a psychotic disorder, concurrent electroconvulsive therapy, or history of traumatic brain injury. Consistent with admission criteria for the treatment program, participants could not meet the program’s exclusionary criteria of active suicidal ideation that was severe enough to impair one’s ability to engage in the treatment program. Thus, no patients with severe, active suicidal ideation at admission were included in the study.

2.2. Measures

2.2.1. Suicidal thoughts and behaviors

Columbia-Suicide Severity Rating Scale (C-SSRS; Posner et al., 2011).

This was administered as an interviewer-rated assessment of lifetime and recent (past month) STBs by independent evaluators who were research assistants, doctoral student trainees, and doctoral-level staff trained and supervised by a licensed clinical psychologist. The suicidal ideation severity scale comprised five dichotomous questions of increasingly severe ideation for each time period: (1) wishing to be dead, (2) non-specific active suicidal ideation, (3) suicidal ideation with a method yet no intent, (4) suicidal ideation with some intent and no specific plan, and (5) suicidal ideation with a specific plan and intent to kill oneself. This measure is scored on a 6-point scale with a final score ranging 0 to 5 according to the most severe suicidal ideation endorsed. Suicidal behaviors were assessed dichotomously and include actual suicide attempts, interrupted suicide attempts, aborted suicide attempts, other preparatory acts (e.g., collecting pills, writing suicide note), and nonsuicidal self-injury. It shows good convergent and divergent validity psychometric properties for assessment of suicide risk (Posner et al., 2011).

Death/Suicide Implicit Association Test (IAT; Nock et al., 2010).

This is a computerized categorization task which aims to measure the implicit associations respondents hold about death/suicide and life. IATs have traditionally been used to measure bias for or against particular groups of people (Greenwald et al., 1998). The Death/Suicide version asks participants to associate words related to death to themselves and to others. It was administered via Inquisit computer software (Inquisit 5, 2016), which calculated d-scores from the measure (Greenwald et al., 2003). Positive d-scores reflect stronger associations between self and death (‘Me-Death’ and ‘Not Me-Life’), whereas negative d-scores reflect stronger associations between self and life (‘Me-Life’ and ‘Not Me-Death’). The IAT has convergent validity with other validated measurements of STBs (e.g., Harrison et al., 2014).

Depressive Symptom Index Suicidality Subscale (DSI; Joiner et al., 2002; von Glischinski et al., 2016).

This is a widely used self-report measure for the detection of suicidal ideation. DSI item #1 was used in the study to assess suicidal ideation frequency and was dichotomized to determine whether an individual endorsed any suicidal ideation [i.e., reported 1 (sometimes I have thoughts of killing myself), 2 (most of the time I have thoughts of killing myself), or 3 (I always have thoughts of killing myself)] or did not endorse suicidal ideation (i.e., reported 0, I do not have thoughts of killing myself) over the past week. The DSI has good internal consistency, convergent validity, discriminant validity, and sensitivity to change (Stanley et al., 2021).

2.2.2. Diagnostic characterization

Yale-Brown Obsessive Compulsive Scale (Y-BOCS; Goodman et al. 1989).

The Y-BOCS is considered the “gold-standard” assessment of OCD symptom severity, with 10 items rated on a 5-point Likert-type scale from 0 (not at all/none) to 4 (extreme); scores range from 0–40. The Y-BOCS was administered by trained research staff and doctoral-level students under the supervision of a licensed clinical psychologist. This interviewer-rated measure has demonstrated both good interrater reliability for the total score (r = .95; Goodman et al., 1989) and sensitivity to treatment effects (DeVeaugh-Geiss et al., 1989).

Structured Clinical Interview for DSM-5 Disorders (SCID-5; First et al., 2015).

The SCID-5 is a structured clinical interview that assesses for Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5; American Psychiatric Association, 2013) diagnoses, administered by trained research staff and doctoral-level students at admission under the supervision of a licensed clinical psychologist. It has shown good test-retest reliability, internal consistency, and predictive validity (Shankman et al., 2017).

2.3. Procedure

Treatment at the partial/residential program has a focus on exposure and response prevention (ERP; Foa et al., 2005, Franklin et al., 2000), an effective first-line treatment for OCRDs (Olatunji et al., 2013) and acceptance and commitment therapy (Hayes et al., 2012). Patients receive individual and symptom-specific group therapy, including up to 4 hours of ERP daily, meetings with a psychiatrist, family therapist, and behavioral therapist. Mean length of stay in this sample was 60.16 days (SD = 25.99).

Diagnostic assessments were conducted within the first two weeks of treatment using the SCID-5 for 83% of the participants. Due to program scheduling conflicts, the remaining participants were assigned diagnoses by their program behavioral therapist.

Baseline assessment at program admission consisted of the IAT, DSI, and interviewer-administered C-SSRS and Y-BOCS. DSI was administered weekly through week 4; IAT was administered again at week 2 and week 4; C-SSRS and Y-BOCS were administered again at week 4 and week 8 (follow-up). All admission and weekly assessments (up to week 4) were completed in-person with some week 8 assessments completed by phone. Due to the naturalistic setting and range of durations of program admission in the treatment program, there was some variability in when the assessments occurred. Study recruitment occurred prior to the COVID-19 pandemic, however the study was ongoing during its onset and consequently, two participants discharged suddenly due to limitations on patient care and clinical research, and one did not complete the week 4 or week 8 assessments. Additionally, due to the varying duration of program admission, we allowed the timing of the assessments to vary as needed (e.g., delayed by 1 week). Participants were compensated $30 for completing assessments through week 4, and $15 for the week 8 assessment.

2.4. Analytic Plan

For Aim 1, to identify the prevalence and clinical correlates of self-reported STBs in treatment-seeking adults with OCRDs, we examined frequencies of explicitly endorsed STBs. Pearson correlation coefficients were calculated to examine the association between self-reported and interviewer-rated STBs, as well as the association between implicit suicidal thoughts (IAT d-score) and OCD symptom severity at program admission and week 4. Independent-samples Welch’s t-test was used to compare the severity of OCD symptoms at program admission between those with and without a lifetime history of suicidal ideation and those with and without a prior suicide attempt.

For Aim 2, to examine implicit and explicit STBs and the associations between them, Pearson correlations examined associations between interviewer-rated suicidal ideation and weekly self-reported suicidal ideation. To determine the extent to which baseline implicit suicidal thoughts at program admission was associated with measures of interest, Pearson correlation coefficients were calculated among IAT d-score and self- and interviewer-rated suicidal ideation. Independent-samples Welch’s t-tests were used to compare implicit suicidal thoughts (IAT d-score) at program admission between those with and without a lifetime history of suicidal ideation and those with and without a prior suicide attempt.

For Aim 3, to investigate changes in STBs longitudinally during treatment, utilizing IAT scores, we used a linear mixed-effects model with time entered as a fixed effect and per-person random intercepts. A mixed-effects logistic regression model with repeated measures was used to determine the effect of time on self-reported suicidal ideation (DSI Item #1 binary variable) across treatment, controlling for IAT d-score at program admission. To determine whether the IAT would significantly predict explicit suicidal ideation over time, we entered time, admission IAT d-score, and their interaction as fixed effects, with per-person random intercepts. Statistical analyses were performed in R using the “lme4” package (Bates et al., 2015).

Given that results reported in this manuscript assessed exploratory hypotheses with an underpowered sample, no steps were taken to control family-wise error rates.

3. Results

3.1. Aim 1

3.1.1. Explicit measurement of STBs in lifetime.

Based on the C-SSRS lifetime assessment, 77.42% (n = 24) of participants endorsed having experienced passive suicidal ideation (i.e., wishing to be dead), 64.52% (n = 20) endorsed non-specific active suicidal ideation (i.e., wanting to end one’s life without method, plan, or intent), 58.06% (n = 18) reported suicidal ideation with a method and no intent, 48.39% (n = 15) reported suicidal ideation with some intent and without a specific plan, and 32.26% (n = 10) reported suicidal ideation with specific plan and intent. With regard to lifetime suicidal behaviors, 16.13% (n = 5) reported prior (actual) attempts, 6.45% (n = 2) reported interrupted attempts, 29.03% (n = 9) reported aborted attempts, and 29.03% (n = 9) reported preparatory suicidal acts or behaviors. Additionally, 48.39% (n = 15) reported nonsuicidal self-injurious behaviors.

3.1.2. Explicit measurement of STBs in past month.

When asked about experiences in the past month before admitting to the program, 61.29% (n = 19) endorsed having had passive suicidal ideation, 16.13% (n = 5) endorsed non-specific active suicidal ideation, 9.68% (n = 3) endorsed active suicidal ideation with a method and no intent, and 0% endorsed active suicidal ideation with a method and any intent; suicidal behaviors were not endorsed in the past month by any participants.

3.1.3. Explicit Measurement of STBs during treatment.

Self-reported suicidal ideation (DSI) had a significant, medium-to-large correlation with interviewer-rated suicidal ideation (C-SSRS) at week 1 (r = .44, p < .05) and a significant, medium-to-large correlation at week 4 (r = .45, p <.05; see Table 2). During the study, one of the 31 (3.2%) participants was hospitalized for suicidal ideation.

Table 2.

Means, standard deviations, and correlations.

Variable	M	SD	1	2	3	4	5	6	7	8	9	10	11	12	13	14
1. OCD Severity at Admission (Y-BOCS)	26.16	5.38	1
2. OCD Severity at Week 4 (Y-BOCS)	18.77	4.52	.22	1
3. OCD Severity at Week 8 (Y-BOCS)	18.70	6.17	.47**	.50**	1
4. Highest Lifetime Suicidal Ideation at Admission (C-SSRS)	2.87	2.05	.50**	.50**	.42*	1
5. Highest Recent Suicidal Ideation at Admission (C-SSRS)	0.87	0.92	.26	.26	.04	.61**	1
6. Highest Recent Suicidal Ideation at Week 4 (C-SSRS)	0.77	1.14	.14	.43*	.18	.41*	.64**	1
7. Highest Recent Suicidal Ideation at Week 8 (C-SSRS)	0.86	1.22	.31	.38*	.29	.53**	72**	72**	1
8. Self-reported Suicidal Ideation at Week 1 (DSI)	0.48	0.57	−.19	.11	−.15	.23	.44*	.29	.21	1
9. Self-reported Suicidal Ideation at Week 2 (DSI)	0.45	0.57	−.06	.30	−.02	.31	.43*	.59**	.55**	.64**	1
10. Self-reported Suicidal Ideation at Week 3 (DSI)	0.45	0.57	−.06	.31	.02	.37*	.56**	.60**	.56**	.74**	90**	1
11. Self-reported Suicidal Ideation at Week 4 (DSI)	0.37	0.49	−.02	.31	.07	.14	.14	.14	.30	.56**	75**	.70**	1
12. Implicit Suicidal Ideation at Admission (IAT)	−0.38	0.36	−.05	−.00	.04	−.03	.10	.10	−.02	.17	−.03	.16	.23	1
13. Implicit Suicidal Ideation at Week 2 (IAT)	−0.36	0.37	.30	.18	.27	.37*	.16	.10	.03	.27	.24	.29	.39*	51**	1
14. Implicit Suicidal Ideation at Week 4 (IAT)	−0.27	0.45	.18	.08	.14	.10	.08	−.04	.05	.12	.23	.22	.31	.58**	74**	1

Note. Y-BOCS = Yale-Brown Obsessive Compulsive Scale; C-SSRS = Columbia-Suicide Severity Rating Scale; DSI-SS = Depressive Symptom Index-Suicidality Subscale; IAT = Implicit Associations Test. M and SD are used to represent mean and standard deviation, respectively.

^*p < .05.,

^**p < .01.

3.1.4. Suicidal ideation and OCD symptoms.

OCD symptoms at program admission (Y-BOCS) were significantly higher [t(21.48) = −2.60, p = .017, d = .96] for participants who endorsed a history of suicidal ideation (M = 27.85, SD = 5.01) compared to those who denied a history of suicidal ideation (M = 23.09, SD = 4.81), as assessed by the C-SSRS. Additionally, levels of current OCD severity had a large, significant correlation (r = .50, p < .01) with lifetime suicidal ideation (C-SSRS), and a small-to-medium, non-significant correlation (r = .26, p = .158) with past-month suicidal ideation (C-SSRS). See Figure 1. In terms of implicit suicidal ideation, OCD symptoms at program admission were not correlated with IAT scores at program admission (r = −.05, p = .789) or with IAT scores at week 4 (r = .185, p = .337). Additionally, OCD symptoms at week 4 were not correlated with IAT scores at week 4 (r = .080, p = .682).

Figure 1.

Average OCD Symptoms by Lifetime Suicidal Ideation and Attempts

3.1.5. Suicide attempts and OCD symptoms.

At program admission, OCD symptoms were significantly higher [t(7.37) = −2.40, p = .046, d = .93] for participants who endorsed a suicide attempt history (M = 30.20, SD = 3.83) compared to those who denied a suicide attempt history (M = 25.38, SD = 5.33). See Figure 1.

3.2. Aim 2

3.2.1. Is the IAT associated with current, interviewer-rated and self-rated STBs?

At each IAT assessment time point (admission, week 2, and week 4), IAT scores were not significantly correlated with self-reported suicidal ideation assessed within the same time frame (r’s = .17 to .31, p-values ranged from .13 to .36). A small significant correlation was found between IAT score at week 2 and self-reported suicidal ideation at week 4 (r = .39, p = .045). Correlations between self-reported suicidal ideation and IAT scores at all other time points were small-to-medium, non-significant (r’s = −.03 to .30, p-values ranged from .12 to .85). At program admission, IAT score had a small, non-significant correlation with past-month interviewer-rated suicidal ideation (r = .10, p = .611). IAT score at week 4 was not significantly correlated to interviewer-rated suicidal ideation at week 4 (r = −.04, p = .834).

3.2.2. Is the IAT associated with prior STBs?

Implicit suicidal ideation (IAT at program admission) did not differ [t(4.55) = −.36, p = .737, d = 0.25] between participants who had experienced a suicide attempt in their lifetimes [n = 5; (M = −.33, SD = .55)] compared with participants who had no attempt history [n = 26; (M = −.42, SD = .33)]. Additionally, implicit suicidal ideation did not differ [t(29.00) = .06, p = .954, d = 0.02] between participants who explicitly endorsed lifetime suicidal ideation [n = 20; (M = −.41, SD = .43)] compared to those who denied lifetime suicidal ideation on the C-SSRS [n = 11; (M = −.40, SD = .23)].

3.3. Aim 3

3.3.1. Longitudinal changes in STBs

We hypothesized that baseline IAT at program admission would predict increased self-reported suicidal ideation (DSI #1) overall during the first four weeks of treatment. We also hypothesized IAT scores would improve over time. There was no significant main effect of IAT [β = 2.47, z(120) = .81, p = .419] or time [β = −.17, z(120) = −1.07, p = .284], nor a baseline IAT by time interaction [β = .74, z(73) = .10, p = .918] in predicting self-reported suicidal ideation. Contrary to hypotheses, IAT scores did not decrease during treatment, [β = 0.05, SE = 0.03, t(103) = 1.53, p = .130].

4. Discussion

This study advances the growing body of literature about the prevalence of STBs among adults with OCD and related disorders. We focused on both explicitly reported STBs, as well as implicit suicide-specific cognitions. Additionally, we aimed to investigate the magnitude of associations between these implicit cognitions and explicitly endorsed suicide risk. Finally, we sought to examine these implicit cognitions longitudinally in a treatment setting among patients seeking treatment for OCRDs.

4.1. Explicit measurement of suicidal thoughts

4.1.1. Ideation.

When measuring explicitly reported suicidal ideation in our real world treatment context, we found that approximately three-quarters of participants at admission to the program reported having experienced suicidal ideation at some point in their lifetimes, with over half experiencing active suicidal ideation. For suicidal ideation in the past month prior to admission, over half of participants reported experiencing suicidal ideation with 16% experiencing active suicidal ideation. Our findings are comparable to a different sample of participants seeking intensive outpatient, partial, and residential treatment for OCD (N = 101), in which 62% of patients endorsed lifetime suicidal ideation, and 12% reported active past-month suicidal ideation (Storch et al., 2017). In both the study by Storch and colleagues and the current study of participants seeking intensive treatment for OCRDs, the rates of suicidal ideation reported at program admission were higher than those in a recent meta-analysis which found approximately half of individuals with OCD experience suicidal ideation (Pellegrini et al., 2020). In the current study, we also found OCD severity was significantly higher for participants with lifetime suicidal ideation. We found these numbers quite high in light of the fact that the treatment program does not admit patients with severe, active suicidal ideation. Taken together, this emphasizes the particular importance of assessing suicidal ideation in patients seeking intensive levels of care for OCRDs and tailoring treatment accordingly.

4.1.2. Attempts.

A concerning finding, given that prior attempts are the single most important risk factor for suicide for both adults with OCD and the general population (Fernández de la Cruz et al., 2017; World Health Organization, 2021), is that 16% of study participants had made a prior suicide attempt in their lifetimes. This stands in striking contrast to the general population, in which 2.7%−5% of individuals have reported a prior suicide attempt in their lifetimes (Kessler et al., 1999; Nock et al., 2008), and underscores the importance of assessing prior attempt history at admission to higher levels of care for OCRDs. Our findings suggest STBs are more likely as levels of OCD severity increase; we found support for our hypothesis that OCD symptom severity would be significantly higher for those with an attempt history than those without. This is in line with prior studies suggesting that higher OCD symptom severity is associated with increased STBs (Albert et al., 2019; Hellberg et al., 2022; Storch et al., 2017).

4.2. Associations between implicit & explicit STBs

In addition to measuring the prevalence of explicitly endorsed STBs, we examined the relationship between implicit suicidal cognitions and explicitly endorsed STBs at program admission. In contrast to our hypotheses, implicit suicidal cognition did not distinguish between participants with and without a history of explicit lifetime suicidal ideation, nor did it distinguish those with and without attempts prior to program admission. Our results replicate findings by Rath and colleagues (2021) in which the IAT did not distinguish between lifetime or recent suicide attempts or suicidal ideation in either a low risk or high risk sample. The IAT did not generally correlate with interviewer-rated or self-reported scores of suicidal ideation across the program, with the exception of IAT scores at week 2 and self-reported suicidal ideation at week 4. These results suggest that implicit suicidal cognition is not necessarily associated with prior STBs, although further study of suicidal ideation that is endorsed explicitly within the same timeframe is needed.

4.3. Longitudinal implicit measurement of suicidal thoughts

To learn more about implicit suicide-specific cognitions in a sample of patients with OCRDs longitudinally, we examined whether these cognitions changed throughout the first four weeks of treatment. We found that IAT scores in our sample did not significantly change over time. At admission, IAT scores were consistent with those of other clinical samples (e.g., Brent et al., 2021; Glenn et al., 2017; Sohn et al., 2021). For instance, based on IAT scores, participants at program admission scored similarly to a sample of adolescents receiving outpatient psychiatric treatment with past-year suicidal ideation (Glenn et al., 2019). In the present study, participant IAT scores appear to reflect low suicide-specific cognitions overall, given that IAT scores above zero have been associated with an increased likelihood of suicide attempts (Sohn et al., 2021). It is important to note that a recent meta-analysis highlighted that the IAT may be a helpful adjunctive tool in the prediction of past and present STBs, but should not be used in isolation (Sohn et al., 2021).

In addition to examining IAT scores over time, we attempted to further investigate longitudinal suicidal thoughts by examining self-reported suicidal thoughts over the same time period, and did not find support for our hypothesis that an implicit bias towards death at admission would predict self-reported suicidal thoughts in these first four weeks of treatment. There are several possible explanations for this finding, including that the sample did not explicitly endorse suicidal thoughts during this timeframe, with the mean response during these first four weeks on DSI #1 between “I do not have thoughts of killing myself” and “sometimes I have thoughts of killing myself.” Other explanations could be due to a number of study limitations detailed below, including the lack of effect, or the sample being underpowered.

4.4. Limitations

The present study had several limitations. First, the sample was limited to 31 participants, as recruitment ended abruptly due to COVID-19, thus impacting the ability to fully power the analyses. This is particularly the case given prior studies examining the IAT among individuals with STBs finding small-to-medium or even negligible effects (e.g., Cohen’s d’s = −0.02 −.44) in larger sample sizes (e.g., Nock et al., 2010, N = 157; Tello et al., 2020, N = 165). Power analyses were not conducted prior to the study for the analyses included in the present study as it was part of a larger project. Future research should replicate and extend current findings to a well-powered sample.

Second, the test-retest reliability of the Death/Suicide IAT is unclear as it remains undetermined the extent to which practice/exposure to the task may affect IAT scores, especially in a longitudinal design (i.e., a residential program). Regardless, these findings appear to suggest that practice effects were not problematic on the IAT in this sample as scores did not appear to change over time. An additional consideration is that prior studies of adults in a psychiatric inpatient program, and adolescents in psychiatric residential and inpatient programs, have found low-to-moderate test-retest reliability for the IAT (Glenn et al., 2017; Millner et al., 2019; Rath et al., 2021). However, it is difficult to discern whether this is reflective of the reliability of the IAT when assessed longitudinally versus participants’ actual changes in implicit associations with death over the course of hospitalization and follow-up.

Third, the diversity of our sample was limited in multiple ways, with a majority of non-Hispanic White, college-educated, heterosexual participants seeking treatment for an OCRD in the Northeastern United States. Various barriers to treatment, including the cost of the program and/or taking time off work, lack of means of transportation, fears of stigmatization, and lack of knowledge about the program likely contributed to decreased representation in our sample, which consisted of patients presenting for partial or residential treatment. Our findings may not be generalizable to individuals with other identities and backgrounds given that the manifestation of OCD symptoms varies across racial and ethnic groups, and sexual orientation (e.g., Pinciotti et al., 2023; Wheaton et al., 2013; Wilson & Thayer, 2020).

In particular, cultural stressors associated with racial and ethnic identities may hold a unique relationship with STBs. For example, studies have found an increased risk of suicide attempts in Hispanic females compared to other populations (Omary, 2020), as well as Black adolescents, representing one of the largest population increases in the prevalence of suicide attempts over the last thirty years (Xiao et al., 2021). Research on discrimination has suggested that race-based discrimination is significantly associated with increased suicidal ideation (Goodwill et al., 2019). Given that our sample is primarily non-Hispanic White, we might expect that a more racially and ethnically representative sample would endorse different levels of suicidal ideation and suicide attempts.

Fourth, all participants were in partial/residential treatment, and this could affect the findings in several ways. The study focuses on passive rather than active suicidal ideation, as active suicidal ideation that would impair one’s ability to engage in the treatment program is an exclusionary criterion for admission to the treatment program and therefore participants with severe, active suicidal ideation are not included in this sample. This suggests that our results could reflect lower rates of past-month active STBs at program admission than is representative of the general population seeking treatment for OCRDs. However, given that our program involves intensive residential and partial treatment, results could also reflect higher rates of STBs compared to individuals seeking OCRD treatment in lower levels of care, such as outpatient treatment.

Finally, given that this study occurred within the context of a real world treatment setting, participants could have been concerned that disclosure in the explicit measures of STBs would affect their treatment, as evidenced by prior research (Richards et al., 2019). As such, it is possible that participants’ reports of active suicidal ideation may have reflected their comfort in disclosure, and not necessarily their subjective feelings, reiterating why we are interested in implicit measures of suicidality. Barriers to disclosure represent a critical issue for research on suicide more broadly, with potential consequences including stigma, legal and structural implications, involuntary hospitalization, loss of privacy, and welfare checks from police, which may particularly impact people of color, who experience higher risks of harmful outcomes from police interactions (Blanchard & Farber, 2020; Sheehan et al., 2019; Vitiello & Moseley, 2021). In order to continue improving our understanding of STBs, it is critical that we continue developing new assessment methods and studying barriers to disclosure across real-world settings.

4.5. Future Directions

In addition to OCD itself, related disorders share risk for STBs and warrant further study. In the current study, our sample was recruited from a naturalistic program for disorders related to OCD that are likely to benefit from similar treatment plans, thus 13% of our sample had related disorders. A recent systematic review and meta-analysis of people with OCRDs (excluding OCD) found that the highest lifetime suicide attempt rate was among those with body dysmorphic disorder (35%), followed by hoarding disorder (24%) and body-focused repetitive behaviors (trichotillomania and excoriation; 14%; Pellegrini et al., 2021). These high rates of lifetime attempts in OCRDs reinforce the importance of assessment at the outset of treatment in not only OCD, but related disorders as well. Given the substantial risk of suicide in OCRDs, it is also critical to gain an understanding about the mechanisms that account for the associations between OC symptoms and STBs in order to facilitate and enhance prevention and intervention strategies.

Future studies examining the association between clinical and demographic characteristics associated with suicide risk in OCD populations will be important in optimizing assessments and interventions at the outset of treatment. Research should examine the prevalence and correlates of STBs in more diverse samples, including racial and ethnic minority populations. In order to increase representation, future researchers should consider recruiting participants from community samples, in addition to contributing to efforts to increase access to the existing treatment programs they recruit from. Increasing access to these programs could include connecting with community health clinics, leveraging social media, publicizing payment and insurance options, and translating program materials to other languages.

Another future direction involves evaluation of specific symptom subtypes in relation to STBs, given the considerable heterogeneity of OCD symptom presentations. For instance, recent research suggests that unacceptable thoughts and responsibility for harm subtypes are more strongly associated with STBs compared to other symptom dimensions (e.g., Khosravani et al., 2021). There is evidence to suggest that within unacceptable thoughts, the severity of sexual/religious obsessions in particular is significantly related to increased STBs (Torres et al., 2011; Velloso et al., 2016). In addition to the heterogeneity within OCD itself, related disorders also share risk for STBs and warrant further study.

To our knowledge, this is the first treatment study involving the use of an IAT in OCD. As mentioned in the review by Teachman and colleagues of implicit cognition and psychopathology (2019), there have not yet been any IATs developed focusing on implicit OC-relevant cognition. Given the potential predictive and clinical utility of IATs that has been shown with studies such as predicting treatment outcomes in panic disorder (Teachman et al., 2008) or the likelihood of relapse in depression (Van Tuijl et al., 2018), this may be beneficial within OCD and related disorders as well.

4.6. Conclusion

This is the first study to measure both explicit and implicit suicidality in a sample of adult patients with OCRDs, using an interviewer-rated measure, self-report, and a measure of suicides-pecific implicit cognitions. The importance of assessing for both lifetime and recent suicidal ideation at program admission in this population is highlighted by these findings, given that the majority had experienced recent suicidal ideation and one in six patients had experienced a prior attempt at the time of admission. The associations we found between OCD symptoms, suicidal ideation, and suicide attempts reflect the heavy burden that many people with OCRDs face, such as feeling inflated responsibility for harm, having unacceptable thoughts, and the general tendency to become isolated from valued relationships and activities. It appears that once patients progressed through the treatment program, explicitly endorsed suicidal ideation did not worsen.

Future studies with larger, more diverse samples in treatment settings are an imperative next step in determining whether the IAT can prove useful in conjunction with other measures of suicidal ideation to provide a new framework to assess suicidality in adults seeking treatment for OCRDs. This is crucial given the difficulties in assessing suicide risk with methods that do not rely on explicitly endorsed risk.

Highlights.

• Suicidal thoughts and behaviors remain understudied in OCD

• Examined implicit and explicit suicidality in adults receiving treatment for OCD

• 77% reported lifetime suicidal thoughts; 16% reported prior suicide attempt

• OCD severity was higher for those with lifetime suicidal thoughts and prior attempts

• Implicit biases towards death did not predict explicitly endorsed suicidality