The Prevalence and Incidence of Suicidal Thoughts and Behavior in a Smartphone-Delivered Treatment Trial for Body Dysmorphic Disorder: Cohort Study

Digital mental health interventions (DMHIs) can expand access to care but often exclude individuals with suicidal thoughts and behaviors (STB), potentially limiting generalizability and perpetuating barriers to care. Meta-analyses indicate 60%-90% of DMHI trials for depression/anxiety exclude participants based on suicide risk, including current ideation, recent ideation, or historical suicidality. BDD populations have particularly elevated STB prevalence. Given difficulties predicting STB and scarce reporting on in-trial STB incidence, this study examined, in an RCT of smartphone-delivered CBT for BDD with relatively permissive STB criteria (excluding only past-month active ideation), (1) lifetime prevalence of STB and (2) incidence and predictors of STB during the 12-week trial, alongside a description of suicide risk mitigation procedures.

Prior reviews show high exclusion rates for suicide risk in DMHI trials: about one-third exclude current suicidal ideation, 70% exclude recent-past ideation, and ~15% exclude based on any history of suicidality. Among prior BDD-focused DMHI trials (excluding the current RCT), most used STB-related exclusions (eg, past-week/month active ideation or any past suicide attempt). BDD is associated with high rates of suicidal ideation, attempts, and suicide; thus, overly restrictive criteria reduce applicability. Prediction of STB is difficult and few trials report detailed in-trial STB incidence or adverse events, complicating evidence-based eligibility decisions.

Design and participants: Randomized waitlist-controlled trial with 80 U.S. adults (≥18 years) with primary BDD. Exclusions: current psychotherapy, ≥4 prior CBT sessions for BDD, recent psychotropic changes (<2 months), acute/active suicidal ideation in the past month (C-SSRS ideation severity ≥2), inability to use a smartphone app, current severe substance use disorder, current severe major depressive disorder, lifetime bipolar or psychotic disorder. The sample was predominantly female (84%), non-Hispanic (88%), White (71%), mean age 27 (SD 9.6). IRB-approved; informed consent obtained. Intervention: Perspectives BDD smartphone CBT app (12 weeks) with coach guidance (two calls: onboarding and mid-treatment; asynchronous in-app messaging). Modules included psychoeducation, cognitive restructuring, exposure and response prevention, mindfulness/attentional retraining, values/self-esteem/self-compassion, relapse prevention. Waitlist participants received the app after 3 months; their weekly data while on waitlist were included. Assessments: Blinded, doctoral-level evaluators administered at baseline: BDD-YBOCS (BDD severity), MINI (DSM-5 diagnoses), C-SSRS (lifetime and past-30-day suicidal ideation/intent/behavior). Self-report at baseline, week 6, week 12: QIDS-SR, CGI-BDD; weekly brief in-app surveys included QIDS-SR item #12 (thoughts of death/suicide). PHQ-2 assessed core depression symptoms; CGI-BDD item captured perceived change (1 very much improved to 7 very much worse). Risk mitigation procedures: Screening excluded very severe depression (QIDS-SR ≥21) and active/acute past-month ideation (C-SSRS ideation severity ≥2); flagged cases received clinician follow-up within 24 hours and referral if indicated. During trial, app displayed crisis resources (911, national hotline); participants provided an emergency contact; PI could withdraw participants if risk or deterioration warranted higher care. Weekly monitoring: CGI-BDD rating of 6 or 7 triggered alerts to clinician/coach with clinician follow-up within 24 hours; clinical deterioration required two consecutive weeks of CGI-BDD 6/7 plus PI judgment. Suicidality monitoring: QIDS-SR item #12 >0 prompted an in-app pop-up with resources; >1 triggered alerts to clinician/coach; clinician follow-up within 24 hours included structured risk assessment (eg, C-SSRS probes), safety planning, and referral as needed. Outcomes and data handling: Primary outcomes were STB severity and suicide attempts. Weekly QIDS-SR item #12 responses across treatment and waitlist periods were aggregated: 1440 possible observations; 1043 observed; 397 (27.5%) missing. Three dichotomous outcomes were created: (1) any death/suicide-related thought (item ≥1), (2) several death/suicide-related thoughts in any week (item ≥2), (3) increased severity relative to baseline week (any subsequent item > baseline). Descriptive statistics summarized prevalence and incidence. Predictors: Bivariate logistic regressions with (a) lifetime suicidal ideation severity (C-SSRS) or (b) lifetime suicide attempt (yes/no). Multivariate logistic regressions included baseline covariates: birth sex, age, sexual orientation (Straight/Heterosexual; LGB; Other), lifetime C-SSRS ideation severity, lifetime suicide attempt, QIDS-SR total excluding item #12, BDD-YBOCS. Analyses used R 4.4.2.

Screening and exclusions: Of 107 screened, 27 (25.2%) were not included; 4 (3.7%) due to acute past-month suicidal ideation; 1 (0.9%) due to acute depression. Risk alerts: During the treatment phase (first 3 months for immediate treatment; first 6 months for waitlist), 12 risk alerts were triggered by 11 participants (≈10 alerts for elevated death/suicide-related thoughts; 2 for increased clinical severity). Clinician follow-up occurred within hours in all but one case; all assessed cases were judged low imminent risk, with safety plans as needed. One participant, unreachable and with increased severity, was withdrawn and provided referrals. Baseline lifetime STB (C-SSRS): 65.0% (n=52) ever wished to die; 41.3% (n=33) had nonspecific active suicidal thoughts; 37.5% (n=30) active thoughts without plan/intent; 20% (n=16) active thoughts with some intent but no plan; 13.5% (n=11) active thoughts with intent and a specific plan; 10% (n=8) lifetime suicide attempt; 16% (n=13) any suicidal behavior; 30% (n=24) nonsuicidal self-injury. Past month: 25% (n=20) endorsed wish to die; no more severe ideation due to exclusion criteria. Incidence during trial (weekly QIDS-SR item #12; N=80): Any death/suicide-related thought (“life feels empty or wonder if worth living” or higher): 34 (42.5%). Several death/suicide-related thoughts in any week (item ≥2): 10 (12.5%). New onset: 3 (3.8%) for item ≥1; 2 (2.5%) for item ≥2. Increased severity relative to week before baseline: 20 (25.0%); increased relative to lifetime: 1 (1.3%). No suicide plans reported, no daily detailed thoughts, and no suicide attempts. Withdrawn for worsening symptoms or suicidal thoughts: 1 (1.3%). Predictors of in-trial thoughts: Bivariate models—lifetime suicide attempt predicted any death/suicide-related thought (OR 11.00, 95% CI 2.14–59.14; P<.01); lifetime ideation severity predicted any thought (OR 1.61–1.76 across outcomes; P≤.006). In multivariate models adjusting for age, sex, sexual orientation, BDD and depression severity, only baseline suicide-thought severity remained significant for any thought; predictive performance modest to good (e.g., PPV 0.91, NPV 0.75, AUC 0.83 for best model). Overall AUCs ranged ~0.63–0.83. Included vs excluded at screening did not differ significantly in baseline depression severity or prevalence of past-week death/suicide-related thoughts. Missing weekly data were 27.5% (397/1440).

Including participants with most forms of past STB in a light-touch, smartphone-delivered CBT trial for BDD was feasible and safe when paired with structured, low-burden, ongoing monitoring and clinician follow-up protocols. Despite high lifetime STB prevalence in this BDD sample, in-trial reports of death/suicide-related thoughts were common but not acute or frequent; no suicide plans or attempts occurred. Prior suicide attempt and higher baseline suicidal ideation severity were associated with reporting in-trial death/suicide-related thoughts, but multivariate prediction remained only modest, underscoring the limits of short-term STB prediction. The study supports the use of frequent, brief, high–face validity assessments (e.g., weekly single-item probes) with clear escalation thresholds and rapid clinician outreach to manage risk while maintaining trial generalizability.

It is feasible to conduct RCTs of DMHIs for BDD that include participants with moderate suicide risk histories, provided eligibility criteria are thoughtfully defined and robust, low-burden monitoring and risk mitigation procedures are in place. Overly strict exclusion criteria may reduce risk but harm generalizability and access; overly lenient criteria may compromise safety. Trials should balance these trade-offs via trained screening, ongoing monitoring, tailored mitigation protocols, and transparent communication with participants. Future research should refine and validate brief risk assessments for digital settings, test alternative eligibility thresholds, enhance monitoring precision without undue burden, and report STB-related adverse events and safety procedures to inform best practices.

Approximately 27.5% of weekly QIDS-SR item #12 responses were missing, limiting precision of incidence estimates. Weekly suicidality was assessed with a single item whose response options mix constructs (death-related thoughts vs suicidality; frequency vs severity), potentially reducing measurement precision and risking misclassification. Exclusion of participants with past-month active suicidal ideation may have been overly restrictive, limiting generalizability and potentially perpetuating barriers to care; findings may not extend to higher-acuity populations. Light-touch monitoring and alert thresholds, while pragmatic, may require adjustment in studies with less restrictive criteria.