Beating OCD and Learning Digitally (BOLD): An Open Trial of a Therapist-Supported, Self-Guided Exposure and Response Prevention App for Adults with OCD

Abstract

Exposure and response prevention (EX/RP) is a first-line, evidence-based treatment for obsessive-compulsive disorder (OCD), yet its widespread implementation remains limited. Major barriers to EX/RP access are the intervention’s intensity and lack of qualified providers. There is a growing need for scalable, technology-supported interventions that reduce treatment burden while preserving core therapeutic components. Therefore, we developed Beating OCD and Learning Digitally (BOLD), a self-guided EX/RP web application delivered with minimal therapist support. This small open trial (N=16) evaluated the acceptability, feasibility, and preliminary outcomes of BOLD in adults with OCD. We found that BOLD significantly reduced OCD symptoms with large effect sizes among treatment completers (d = −1.27). Half of participants completed the full intervention and reported high satisfaction and gains in self-efficacy. Two-thirds (62.5%) of treatment completers were classified as treatment responders. When compared to a matched sample receiving traditional EX/RP, BOLD showed comparable treatment outcomes. Time-effectiveness analysis indicated that BOLD required 73.39% less therapist contact time to achieve equivalent symptom reduction than traditional EX/RP. These findings highlight BOLD’s potential as a scalable, cost-effective intervention for OCD.

Researchers should refine BOLD by optimizing engagement and reducing barriers to participation and investigate its effectiveness in a full-power randomized controlled trial.

Introduction

Obsessive-compulsive disorder (OCD) is a chronic and often debilitating psychiatric condition affecting approximately 2–3% of the population (Ruscio et al., 2010). OCD is associated with substantial impairment across social, academic, and occupational domains (Huppert et al., 2009; Markarian et al., 2010) and psychiatric comorbidity such as anxiety and depression (Pinto et al., 2006; Sharma et al., 2021). Exposure and response prevention (EX/RP), a specialized form of cognitive behavioral therapy (CBT) that prioritizes the systematic confrontation of distressing thoughts, images, or situations while refraining from compulsive behaviors (Foa et al., 2012), is a first-line treatment for OCD with strong empirical support across many randomized controlled trials (RCTs) and meta-analyses (Abramowitz et al., 2003; Mao et al., 2022; Öst et al., 2015; Song et al., 2022).

Although EXRP is highly efficacious in treating OCD, it is not widely available. EX/RP is costly as it is typically delivered in 60- to 90-minute sessions across 15–18 weeks and requires highly trained clinicians (Foa et al., 2002). These demands limit scalability and lead to persistent disparities in treatment access, particularly in resource-constrained settings (Senter et al., 2021). Many individuals with OCD remain untreated or experience long treatment delays due to financial barriers and a shortage of trained clinicians, underscoring the urgent need for scalable, accessible interventions. Digital mental health interventions can increase treatment access to care and have showed some promising results (E. Andersson et al., 2012; Patel et al., 2018). Therefore, we developed Beating OCD and Learning Digitally (BOLD), a mobile EX/RP program that combines structured self-guided modules with low-intensity therapist coaching to increase access to the core principles of EX/RP while minimizing patient and clinician burden.

BOLD offers several advantages compared to other EX/RP delivery models (Table 1). First, BOLD introduces a lower-touch approach to EX/RP to address the scalability bottleneck of therapist time. Significant strides have been taken in this field in recent years, with multiple tests of digital EX/RP-based applications demonstrating significant efficacy in the treatment of OCD (Feusner et al., 2022; Gershkovich et al., 2021). Some, like nOCD, delivers EX/RP digitally with live and frequent licensed-therapist–administered EX/RP sessions supported by messaging and community features, while others, like GGRO, offers fully self-guided cognitive training without directly involving therapists (Roncero et al., 2019). Despite these advances, a need continues to exist for low-touch EX/RP. Therapist involvement has been shown to enhance engagement and increase effectiveness in digital interventions, including those targeting anxiety disorders (Lindner et al., 2018) and OCD (Gershkovich et al., 2021). Therapist support may be especially important in self-guided, digitally delivered EX/RP because EX/RP requires patients to repeatedly confront distressing content, a process that can lead to non-adherence or dropout (Simpson et al., 2021; Wheaton et al., 2016). However, reliance on therapist involvement limits scalability. Thus, BOLD was designed to deliver fully structured EX/RP content augmented by brief, structured therapist support to enhance accountability to exposure practice, reinforce learning, and promote greater fidelity within exposure practice, while reducing therapist burden.

Table 1.

Comparison of BOLD and Other EX/RP Delivery Models

Feature	BOLD	NOCD	GGRO	Traditional EX/RP
Therapist Time	3.5 hours (7 × 30 min)	~11–13 hours (2×/week 30-min sessions)	0 hours (fully self-guided)	15–25 hours (60–90 min/week)
Exposure Logging	In-app logging with notes/timestamps	In-app	Not included	Paper-based or verbal
Exposure Planning	In-app planner with therapist input + reminders	Therapist-guided	Not included	Therapist-led
Exposure Reminders	Automated SMS reminders + personalized therapist reminder emails	Not included	Not included	None or therapist-generated
Hierarchy Development	Interactive in-app builder + therapist-supported	Therapist-led	Not included	Therapist-guided
Structured Psychoeducation	On demand in-app modules	In-app modules	Gamified	Verbal or printed
Self-Monitoring Tools	Daily mood and OCD tracking	Included	Included	Minimal
Post-Exposure Processing Prompts	App-guided reflections + therapist check-in discussions	Not included	Not included	Therapist-led only
Mobile Access	Included	Included	Included	N/A
Asynchronous Use	Encouraged	Encouraged	Included	No
Clinician Level Required	Possibly deliverable by novice providers	Licensed clinicians	None	Licensed psychologist/therapist

Note: GGRO = Gamified GGRO by Stanford’s FREEDOM Center (Yoon et al.); NOCD = app-based teletherapy with live ERP therapists; data pulled from public sources and publications. BOLD = Blend of self-guided mobile EX/RP content with therapist support.

Second, BOLD distinguishes itself by digitizing the core behavioral mechanisms of EX/RP. For example, while GGRO primarily relies on cognitive training approaches to challenge maladaptive beliefs related to OCD rather than delivering structured behavioral exposure, BOLD was designed to digitize the active components of EX/RP, including hierarchy building, in vivo exposure, and response prevention, aligning digital intervention with the gold-standard first-line treatment for OCD. Traditional concepts of habituation were incorporated alongside more recent models of inhibitory learning to support a comprehensive behavioral change framework.

Third, built as a mobile-first web app to maximize accessibility via smartphone use, BOLD integrates advanced in-app features including tools for hierarchy building, SUDS tracking, exposure planning, ritual logging, as well as technical components such as in-app exposure logging and automated reminders, which were not widely available in other platforms like NOCD or GGRO. Beyond in-app prompts, BOLD incorporates SMS text messaging to remind users to log rituals, plan exposures, and complete scheduled exposures, a unique engagement strategy not available in other digital interventions. Together, these features reflect how EX/RP is done in real-world practice and can enhance engagement and adherence. To summarize, BOLD was developed to make core components of EX/RP more accessible, scalable, and easier to deliver with less therapist time.

The current study aimed to conduct an open trial to 1) evaluate the feasibility and acceptability of BOLD, 2) examine preliminary clinical outcomes of BOLD, and 3) collect qualitative feedback to inform future development of BOLD and related mobile mental health interventions. We hypothesized that 1) treatment would be highly acceptable and feasible, as measured by participant satisfaction, content completion, and patterns of app engagement, 2) participants would show large reductions in OCD severity from pre- to post-treatment, with more than 50% meeting criteria for treatment response, and 3) participants would provide concrete suggestions for improving usability, content clarity, and engagement strategies.

Methods

Study Design

Participants completed the BOLD protocol through this pilot evaluation of feasibility, acceptability, and preliminary clinical outcomes. Participants were enrolled following a phone screening and baseline eligibility assessment. OCD symptoms and general psychopathology were assessed at the baseline visit and the post-treatment visit. Modules were delivered over 14 weeks, allowing flexibility for two weeks of missed content. Therapist support was intentionally limited to seven 30-minute check-ins over the 14 weeks, totaling 3.5 hours of clinician time. Therapists were two experienced licensed psychologists with expertise in OCD, and two supervised clinical psychology doctoral students who had undergone thorough training in EX/RP, including completion of a 4-day workshop, a year of supervised experience, and subsequent training in how to adapt EX/RP to the BOLD app.

Inclusion and Exclusion Criteria

Inclusion criteria were (1) age between 18 and 65 years old, (2) a primary diagnosis of OCD as determined by the Diagnostic Interview for Anxiety, Mood, OCD, and Related Neuropsychiatric Disorders (DIAMOND), (3) possession of a mobile phone with internet access, and (4) able and willing to provide written informed consent. Exclusion criteria were (1) comorbid psychiatric conditions that elevated clinical risk or could confound results (e.g., bipolar disorder, psychotic disorders, or major depressive disorder with psychotic features), (2) current alcohol or substance use disorder, (3) current active suicidal ideation requiring a higher level of care as assessed via the Columbia Suicide Severity Rating Scale (C-SSRS) and clinical judgment, (4) suicide attempt within the past three years that required hospitalization or was deemed life-threatening, (5) serotonin reuptake inhibitor (SRI) dose changes within the past six weeks or SRI initiation within the past 12 weeks, (6) anticipated changes in psychiatric medication during the study period, and (7) concurrent engagement in psychotherapy targeting OCD, anxiety, or related conditions. Individuals who would have been required to pause ongoing evidence-based treatment to enroll were also excluded.

Procedures

Recruitment

Participants were recruited from across the United States, although most resided in the greater Philadelphia region. A small number recruited virtually from TrialX were located out of state (e.g., New Jersey, Illinois). Recruitment efforts were coordinated at the study headquarters in Philadelphia and included referrals through routine clinic flow at a university-affiliated outpatient center, outreach to providers at other mental health clinics, and the distribution of IRB-approved flyers on a university-affiliated clinical trial recruitment website, as well as at community clinics and public locations.

Screening

Initial pre-screening phone calls (approximately 30 minutes) were conducted by a trained research coordinator under the supervision of an OCD expert (Dr. A.G.G.). These calls included a review of inclusion/exclusion criteria and an overview of study procedures. Individuals who appeared eligible were scheduled for a remote baseline assessment conducted by bachelor’s or doctoral-level clinicians trained in study instruments. Baseline assessments included diagnostic interviews and self-report measures. Clinician-rated measures were audio-recorded for reliability coding, and the assessment team met weekly to review these recordings and maintain fidelity.

Treatment

The BOLD intervention was developed by experts in EX/RP in collaboration with software engineers and iterated based on feedback from other OCD specialists. The BOLD included 12 app-based modules that addressed core components of EX/RP in a sequential format as follows: Module 1: Psychoeducation and self-monitoring; Module 2: Exposure hierarchy development; Modules 3–11: Exposure and response prevention skills (e.g., imaginal exposure, in vivo exposure, troubleshooting ritual prevention, and related techniques); Module 12: Relapse prevention. Participants engaged with the modules independently via the web-app, which recorded engagement metrics (e.g., time in app, module completion, exposure logs). In addition to the web-app content, participants received minimal therapist support through seven 30-minute telehealth sessions conducted every two weeks. Therapists used these sessions to clarify content, personalize exposure assignments, troubleshoot barriers, and provide general support. Although they occasionally conducted exposures with patients when appropriate, the therapists’ primary role was to reinforce independently completed work and support treatment adherence. Asynchronous communication was also supported throughout the study period. Participants could contact their therapist via secure, HIPAA-compliant email and submit app-based forms or tracking logs as needed. Therapists responded to asynchronous messages within 24 to 48 hours, as is typical for other self-help interventions (G. Andersson & Titov, 2014; Schueller et al., 2017). Therapists monitored participants’ engagement between sessions and sent brief, personalized email reminders to encourage completion of exposures and app-based tasks when participants appeared to be falling behind.

Given the pilot nature of this study, we allowed for minor refinements to the BOLD platform during the trial, guided by participant feedback and therapist observation. Several edits were implemented to enhance engagement and reduce user frustration without altering core therapeutic content. These included adjusting the timing of automated reminder messages based on participant preferences, extending the inactivity timeout period to prevent premature logouts, and expanding the character limit in free-response fields to accommodate detailed reflections.

Protocol Registration and Deviation

The study protocol was pre-registered here: https://osf.io/56xta/files/osfstorage. Several minor deviations from the preregistered protocol occurred during the study. First, about half of the participants were provided access to the BOLD app and asked to complete the first module before their initial therapist check-in, rather than after, to streamline the first session and promote early engagement with the intervention content. Second, in a few cases, an additional (eighth) therapist check-in was offered to allow participants to complete content from the final scheduled session. Third, while the protocol specified a standard session agenda, therapist check-ins were occasionally adapted when participants had not completed the assigned exposure exercises or educational content, or if they had to miss a session due to logistical concerns. In such instances, therapists used the next session to review prior/missed material and support participants in re-engaging with the app-based content. These deviations were minor, pragmatic in nature, and aimed at maintaining participant engagement and treatment integrity. All changes were documented, and core intervention components were preserved across participants.

Treatment Fidelity

To ensure protocol adherence, the research team developed fidelity checklists aligned with the BOLD treatment manual (See Appendix for details). Two independent raters coded a sample of eight session recordings (one session randomly selected for each participant who completed the full intervention, or 17.4% of all sessions). Each checklist included 10 items rated on a 6-point Likert scale (0–5) assessing therapist adherence (i.e., delivery of required components) and competence (e.g., therapeutic engagement, flexibility, and alliance).

Materials

Participants completed a comprehensive battery of clinician-administered and self-report assessments at baseline and post-treatment to evaluate OCD symptom severity, psychiatric comorbidities, treatment response, and intervention acceptability, feasibility, and usability. Evaluator-rated assessments were administered independently by a licensed psychologist (A. G.), a postdoctoral fellow in clinical psychology (L. G.), or a postbaccalaureate research assistant (K. G.). The postdoctoral fellow and research assistant were trained and supervised by the licensed psychologist.

Clinician-Rated Measures

Diagnostic Interview for Anxiety, Mood, and OCD and Related Neuropsychiatric Disorders (DIAMOND; Tolin et al., 2018).

The DIAMOND is a structured diagnostic interview used to assess a broad range of DSM-5 anxiety, mood, obsessive-compulsive, and trauma-related conditions. The DIAMOND was administered at baseline to assess OCD and other psychiatric comorbidities. The DIAMOND demonstrates high diagnostic accuracy and strong interrater reliability (Tolin et al., 2018).

Yale-Brown Obsessive-Compulsive Scale – Second Edition (Y-BOCS-II; Storch, Rasmussen, et al., 2010).

The Y-BOCS-II is a gold-standard, semi-structured clinical interview used to assess the severity and type of obsessions and compulsions (Storch, Rasmussen, et al., 2010), with higher total scores indicating greater symptom severity. The Y-BOCS-II (Storch, Rasmussen, et al., 2010) has demonstrated strong interrater reliability, internal consistency, and sensitivity to treatment effects, making it widely used in clinical and research settings (Storch, Rasmussen, et al., 2010).

Clinical Global Impressions Scale (CGI; Guy, 2017).

The CGI-Improvement scale is an evaluator-rated assessment of overall improvement in psychiatric problems; in combination with the Y-BOCS-II, a rating of “much” or “very much improved” on the CGI is often used to define treatment response in OCD trials (Simpson et al., 2013; Storch, Lewin, et al., 2010).

Self-Report Measures

Obsessive-Compulsive Inventory – Revised (OCI-R) (Foa et al., 2002).

The OCI-R is an 18-item self-report questionnaire that evaluates the presence and severity of OCD symptoms across six domains: washing, checking, ordering, obsessing, hoarding, and neutralizing. Items are rated on a 5-point Likert scale. The OCI-R has demonstrated strong internal consistency, test-retest reliability, and convergent validity in both clinical and non-clinical populations (Abramovitch et al., 2021; Wootton, Diefenbach, et al., 2015).

Dimensional Obsessive-Compulsive Scale (DOCS; Abramowitz et al., 2010).

The DOCS is a 20-item self-report measure that assesses severity across the four established dimensions of obsessive-compulsive symptoms: contamination, responsibility for harm, unacceptable thoughts, and symmetry/perfectionism. The scale has demonstrated strong reliability, validity, and diagnostic sensitivity in clinical and research settings (López-Nicolás et al., 2021).

Florida Obsessive-Compulsive Inventory – Interviewer Version (FOCI-I; Storch et al., 2007):

The FOCI-I is a brief tool used to assess the presence and severity of obsessive-compulsive symptoms. It consists of two components: a symptom checklist and a five-item severity scale evaluating distress, time interference, functional impairment, resistance, and perceived control. The FOCI-I has demonstrated excellent interrater reliability, internal consistency, and convergent validity with other established OCD measures (e.g., Y-BOCS).

Patient Health Questionnaire-9 (PHQ-9; Kroenke et al., 2001).

The PHQ-9 is a widely used 9-item self-report tool for assessing depressive symptom severity and screening for major depressive episodes based on DSM criteria. The PHQ-9 has been validated extensively in both clinical and community samples (Beard et al., 2016; Levis et al., 2019).

Generalized Anxiety Disorder-7 (GAD-7; Spitzer et al., 2006).

The GAD-7 is a 7-item self-report measure designed to assess the severity of anxiety symptoms, including excessive worry, restlessness, and irritability over the past two weeks. The GAD-7 has demonstrated strong reliability and validity in both general and clinical populations (Johnson et al., 2019; Löwe et al., 2008), and has been shown to quickly assess anxiety severity broadly.

World Health Organization-Quality of Life (WHOQOL; Orley & Kuyken, 1994).

The WHOQOL is a cross-cultural quality of life assessment. It assesses well-being across physical, psychological, social, and environmental domains. The brief version of this measure has demonstrated strong psychometric properties within a variety of population groups, including psychiatric samples (Oliveira et al., 2016; The Whoqol Group, 1998).

General Self-Efficacy Scale (GSE; Schwarzer & Jerusalem, 2012):

The GSE is a 10-item measure that assesses individuals’ general sense of perceived self-efficacy. It was created to predict how individuals would cope with both daily inconveniences and stressful life events, and it has demonstrated strong reliability and cross-cultural validity (Luszczynska et al., 2005).

Acceptability & Usability Measures

Acceptability was assessed using the Client Satisfaction Questionnaire–Modified (CSQ-M; Attkisson & Zwick, 1982) and qualitative exit interviews and by the proportion of treatment completers.. The CSQ-M is a widely used self-report questionnaire assessing satisfaction with treatment, including perceived helpfulness, quality of care, and overall experience. The CSQ-M score ranges from 8 to 32, with higher scoring indicating greater satisfaction. Semi-structured exit interviews were conducted post-treatment using a standardized guide. Topics included usability, clarity of content, perceived benefits and barriers, suggestions for improvement, and experiences with therapist communication. The usability of the BOLD program was assessed with the System Usability Scale (SUS; Bangor et al., 2008; Lewis & Sauro, 2018). Its score ranges from 0 to 100, with a score of 68 or higher indicating at least “average” usability and satisfaction.

Engagement Measures

Intervention feasibility was evaluated using app-based metrics, including the number of modules completed, time spent on the app, and the number of app logins. These metrics were chosen in part based on other work in this area (Li et al., 2022; Meyerhoff et al., 2021; Schueller et al., 2017). Therapist utilization data included the number of messages sent and received, and the amount of time spent on the app per patient per week. These data were automatically collected by the app and monitored throughout the study.

Patient EX/RP Adherence Scale (PEAS; Simpson et al., 2010).

The PEAS is a 3-item scale that was used by BOLD therapists to assess participant adherence to EX/RP instructions between sessions, including quantity and quality of exposure practice and ritual prevention. This scale demonstrates strong interrater reliability and content validity, with higher scores predicting better treatment outcomes (Wheaton et al., 2016).

Statistical Analysis

To evaluate treatment effects, we conducted a series of mixed-effects models to examine change from pre- to post-treatment on all outcome measures. Time (baseline vs. post-treatment) was entered as a fixed effect, and participant was included as a random intercept to account for individual differences in baseline symptoms. Primary outcomes included OCD symptom severity as measured by clinician-rated (Y-BOCS-II) and self-report (OCI-R, FOCI-I, DOCS) scales. Secondary outcomes included global improvement (CGI-I), depressive symptoms (PHQ-9), anxiety symptoms (GAD-7), quality of life (WHOQOL-BREF), and self-efficacy (GSE). Effectiveness was measured by pre- to post-treatment change on OCD symptoms using Cohen’s d effect size estimate. To assess clinically meaningful response, we calculated the proportion of treatment completers (n=8) who demonstrated ≥35% reduction on the Y-BOCS-II from pre- to post-treatment (Farris et al., 2013; Mataix‐Cols et al., 2016).

To further examine the effectiveness of BOLD as compared to traditional EX/RP, we conducted propensity score matching (PSM) to compare the BOLD sample and a naturalistic sample of patients who received standard 60–120 min EX/RP at the OCD and anxiety clinic where the study was conducted. Propensity scores were estimated using a logistic regression model including age, gender, and baseline OCD symptom severity (OCI-R score). We compared several matching procedures, including nearest neighbor, full, and optimal matching, and evaluated balance using standardized mean differences and Love plots. Optimal matching was selected because it yielded best matching outcomes. We compared the effectiveness (indicated as the post-treatment OCD symptoms) between the current sample and the matched naturalistic sample after controlling for the baseline symptom severity. We also compared the time effectiveness, indicated as therapist contact time per unit of symptom improvement, between the two samples. As a sensitivity analysis, we repeated analyses using the full unmatched naturalistic sample.

Feasibility and acceptability were examined by evaluating: 1) ratings on the CSQ-M (mean rating of at least 3/4 on each Likert scale item) and SUS (score of at least 68), 2) number of eligible participants who complete the trial (expected ≥ 70%), 3) number of modules completed by participants, 4) therapist time spent per week per patient, and 5) number of messages sent and received. All quantitative analyses were conducted in R version 4.4.3.

Qualitative Analysis

Acceptability and feasibility of the BOLD intervention were also examined through thematic analysis of semi-structured exit interviews, guided by Braun and Clarke’s (2006) six-phase framework, including 1) transcription, 2) code generation, 3) developing themes across codes, 4) reviewing themes, 5) defining and naming themes, and 6) producing a report (Braun & Clarke, 2006). Interviews were transcribed using Rev, de-identified, and analyzed in MaxQDA. A hybrid inductive–deductive approach was used to develop and refine the codebook. Deductive codes were drawn from the interview guide and study aims (e.g., feedback on usability, perceived impact, therapist communication), while inductive codes emerged through close reading of the transcripts. Two independent coders analyzed a subset of three transcripts to assess interrater reliability. Coding proceeded iteratively, and thematic saturation was reached after coding 7 of the 8 interviews. Remaining transcripts were coded independently with frequent calibration. Sentence-level coding agreement ranged from 70.51% to 80.97%, with an average overlap of approximately 70%. Discrepancies were resolved through consensus meetings with the research team. Subcodes were grouped into preliminary themes, which were reviewed and refined to capture participants’ lived experiences with the BOLD program.

Results

Participant Characteristics

Of the 35 individuals who completed the phone screening, 21 consented and completed the baseline assessment (Figure 1). Sixteen participants were eligible for the intervention and included in the final analytic sample. The sample consisted of primarily female participants (n=13; 81.3%) and White participants (n=10; 62.5%), with an average age of 30.38 years (SD = 10.61, range = 20–63). Participants demonstrated moderate OCD symptoms at baseline (Y-BOCS-II M = 24.14, SD = 5.45).

Figure 1.

CONSORT Flow Diagram Modified for Non-Randomized Pilot Trial Design

Feasibility, Acceptability, and Adherence

Of the 16 enrolled participants, 8 (50%) completed the full BOLD intervention. Reasons for dropout included a preference for traditional treatment in the specialty clinic associated with the study (n=2), insufficient time and availability to complete the program (n=2), and unknown (n=4). No significant differences were found between treatment completers and non-completers in terms of demographic or baseline characteristics (ps > .05), except for scores on the WHOQOL, which were higher among treatment completers (p =.002).

Engagement metrics for the 8 treatment completers are summarized in Table 2. Among the treatment completers, participants attended an average of 5.75 therapist sessions (SD = 1.53, median = 6.5; range = 2–8) and reported completing an average of 72.1% of assigned homework modules. Across the treatment period, therapists sent an average of 2.63 messages per participant (e.g., reminders), and participants sent an average of 1.31 messages (e.g., questions, requests for support). Half of treatment completers (n=4) indicated that they would have preferred additional sessions. Participants who completed the post-treatment visit (n=9) reported high satisfaction with the intervention (CSQ-M M = 30.25/32) and rated the platform as highly usable (SUS M = 81.56/100).

Table 2.

BOLD Program Engagement Metrics

Variables	Post-treatment (n=8)
	M	SD
Number of completed modules	11.75	0.46
Number of therapist interactions	5.75	1.53
Messages Sent by Therapist	2.63	1.50
Messages Sent by Patient	1.31	1.78
CSQ (range: 8–32; higher scores better)	30.25	2.05
SUS (range: 0–100; higher scores better)	81.56	20.48
Treatment adherence (range: 1–7, higher scores better)	4.98	1.74

Fidelity scores for adherence of all participants ranged from 48 to 50 out of 50 (M = 49.3, SD = 0.6), with two sessions scoring 48 because they had less motivational content compared to other sessions, an anticipated pattern given the structure and focus of Sessions 1 and 7. Therapist competence was uniformly rated at the maximum score across all coded sessions. These findings support high treatment fidelity and consistency across therapists.

Treatment Effectiveness & Time-effectiveness

Mixed-effects models revealed significant reductions in OCD symptoms from pre- to post-treatment across clinician- and self-report measures (Table 3). Specifically, we found significant decreases on the Y-BOCS-II (p < .001, d = −1.27), FOCI-I (p < .001, d = −1.23), and DOCS scores (p = .005, d = −0.89) with large effect sizes after completing the BOLD intervention. OCI-R scores decreased marginally significantly (p = .051, d = −0.45). Global clinical severity also improved significantly (CGI-I, p = .002). Among treatment completers (n=8), five participants (62.5%) achieved treatment response, indicated as a ≥35% reduction in OCD symptoms on the Y-BOCS-II (Farris et al., 2013; Mataix‐Cols et al., 2016). No statistically significant changes were observed in depressive symptoms (PHQ-9), anxiety (GAD-7), general self-efficacy (GSE), or quality of life (WHOQOL-BREF).

Table 3.

Mixed Effects Models for BOLD Treatment Outcomes

Variables	Pre-treatment (n=16)		Post-treatment (n=8)		Estimated Difference	SE	d	p
	M	SD	M	SD
YBOCS-II	24.14	5.45	15.12	6.58	−7.92	0.79	−1.27	<.001***
OCI-R	29.75	12.27	21.12	15.79	−7.67	3.43	−0.45	.051
FOCI-II	25.38	7.00	13.00	8.11	−11.01	2.14	−1.23	<.001***
DOCS	27.62	10.37	17.75	11.31	−9.49	2.45	−0.89	.005**
CGI	4.75	0.68	3.38	1.51	−1.31	0.32	−0.80	.002**
GAD7	9.12	6.21	5.38	4.78	−3.26	2.08	−0.55	.150
PHQ9	6.31	3.77	5.12	4.09	−0.29	0.75	0.04	.710
GSE	29.69	3.44	32.38	3.62	2.69	1.52	0.31	.090
WHOQOL	91.25	11.43	93.50	7.23	−0.53	3.49	−0.19	.888

Note: Treatment Responder: ≥ 35% reduction in YBOCS-II.

PSM identified a naturalistic sample of 16 patients receiving traditional EX/RP who were matched to the BOLD sample on age, gender, and baseline OCD symptoms (OCR-I scores). No significant differences were found between the BOLD and the traditional EX/RP sample in post-treatment OCD symptom severity after controlling for the baseline symptom severity, suggesting comparable effectiveness of BOLD and traditional EX/RP. Time-effectiveness analyses indicated that traditional EX/RP required an average of 94.29 minutes of therapist time to achieve a one-point reduction in OCD symptoms, whereas BOLD required only 25.09 minutes (Table 4), suggesting that BOLD can save 73.39% of therapist time. Sensitivity analysis comparing the BOLD sample to the full, unmatched traditional EX/RP sample yielded consistent results. In the whole naturalistic sample, therapists spent an average of 102.79 minutes to achieve a one-point reduction in OCD symptoms.

Table 4.

Time-effectiveness Results

	Bold (n = 16)	Traditional Ex/RP (Matched naturalistic sample; n=16)	Traditional Ex/RP (Unmatched naturalistic sample; n=224)
	Mean (SD)	Mean (SD)	Mean (SD)
Baseline	28.56 (11.87)	28.25 (13.05)	22.94 (13.47)
Post-treatment	20.50 (15.72)	10.17 (5.71)	10.28 (8.74)
Within-group change	6.88 (10.38)	10.50 (10.62)	10.49 (12.02)
Therapist contact time per patient (mins)	172.5 (45.83)	990.0 (532.81)	1078.32 (483.64)
Therapist contact time per OCI-R point (mins)	25.09	94.29	102.79

Qualitative Feedback

Qualitative analysis of exit interviews revealed five major themes (Table 5):

Table 5.

Participant Experiences with BOLD Program: Identified Themes

Theme	Description	# Segments	% of Total Segments	Representative Quote
Learning About OCD and Treatment	Participants described gaining new knowledge about OCD and differences between treatment strategies.	26	11.56%	“I didn’t know about the difference between imaginary exposures and in vivo. I didn’t know there was a difference, but that was really cool to learn about.”
Navigating the App: Convenience, Challenges, and Autonomy	Reflections on app flexibility, ease of use, reminders, and frustration with bugs or technical issues.	61	27.11%	“I liked the flexibility... the app was helpful as far as logging things because it made me think more about the rituals I did or the exposures.”
Appreciation for Therapist Support	Emphasis on feeling supported by the therapist, especially during difficult moments or when participants needed grounding.	43	19.11%	“He was really good at guard railing and sort of reeling me in when I was becoming ungrounded or sort of forgetting to do certain things.”
Making the Treatment Fit: Structure, Feedback, and Preferences	Reflections on how the structure of treatment (e.g., app vs therapist sessions) supported different aspects of learning and behavior change.	69	30.67%	“I think they were equally helpful, but on the app, I was able to think about my hierarchy more than I would be able to in an hour-long therapy session.”
Growth and Self-Efficacy	Descriptions of increased skills, confidence, and independence in managing OCD symptoms.	26	11.56%	“Before this, I had no tools in my toolbox to deal with my OCD, but now that I do, I feel like I can handle it on my own.”

1. Learning About OCD and EX/RP: Many participants reported an increased understanding of OCD and EX/RP concepts through the app content. The ability to revisit psychoeducational materials was noted as particularly useful in reinforcing learning.

2. Navigating the App: Participants reported both strengths and limitations of the BOLD app. Helpful features included SUDS logging, reminder texts, and exposure tracking tools that promoted daily engagement and reflection. Common challenges included login difficulties, session time-outs, and inconsistent reminders. Recommendations included preset response buttons and editable hierarchies.

3. Appreciation for Therapist Support: Therapist involvement was described as central to the BOLD experience. Participants valued the therapist’s technical and emotional support, which contributed to a sense of individualized care.

4. Making the Treatment Fit: Most participants found the blend of app-based work and therapist guidance beneficial, although preferences varied regarding session length and frequency. Some expressed a desire for more frequent or in-person contact, while others appreciated the flexibility and convenience of the current model.

5. Growth and Self-Efficacy: Participants described increased insight, skill development, and confidence in managing OCD symptoms. Many reported applying EX/RP tools in daily life and feeling more empowered to confront fears independently. Therapist collaboration was often cited as a catalyst for sustained progress and self-efficacy.

Discussion

This pilot study evaluated the feasibility, acceptability, and preliminary outcomes of Beating OCD and Learning Digitally, a mobile app-based EX/RP program supported by brief therapist check-ins. The treatment completers reported high satisfaction and usability of BOLD and demonstrated strong adherence and engagement with the app. BOLD significantly reduced OCD symptoms with large effect sizes while requiring substantially less therapist contact time to achieve equivalent symptom reduction than traditional EX/RP. These findings suggest that BOLD is a low-intensive, cost-effective intervention in adults with OCD. Nevertheless, only half of participants completed the full BOLD protocol, highlighting the need to investigate and address potential barriers to participation. Qualitative feedback from participants reveals effective ingredients and features of BOLD (e.g., independent exposure practice, therapist support), which can inform future refinements. Consistent with emerging evidence supporting blended digital EX/RP approaches (Lundström et al., 2022; Wheaton et al., 2016), BOLD can expand access to evidence-based care while minimizing clinician burden.

This study provided preliminary evidence for the effectiveness of BOLD in adults with OCD. Participants showed significant reductions in OCD symptoms across clinician-rated and self-report measures (Y-BOCS-II, FOCI-I, DOCS) with large effect sizes. When compared to a matched naturalistic sample receiving traditional EX/RP, BOLD yielded comparable treatment effectiveness and required 73.39% less therapist contact time to achieve equivalent symptom reduction than traditional EX/RP, suggesting that BOLD is highly cost-effective. The observed decline in Y-BOCS-II scores in the current sample (24.14 to 15.12) is similar to reductions reported in traditional EX/RP protocols (27.2 to 13.0 in Simpson et al., 2013). While BOLD’s within-group effect sizes (d =1.27) is slightly lower than in-person EX/RP, it is superior to many fully self-guided CBT trials (Kenwright et al., 2005; Wootton, Dear, et al., 2015), suggesting that BOLD effectively bridges the gap between intensive therapy and scalable, unguided tools.

Moreover, among participants who completed treatment, 62.5% achieved a clinically meaningful response (defined as ≥35% reduction on the Y-BOCS-II), a rate comparable to those reported in meta-analyses of standard EX/RP (Öst et al., 2015). Notably, when considering the full intent-to-treat sample (N=16), the response rate was lower (31.25%) given the high attrition rate, suggesting that while BOLD is effective for those who engage, attrition significantly reduces its overall impact. These results highlight the high efficiency of blended care models like BOLD for delivering evidence-based OCD treatment with reduced clinician burden.

This study suggests that BOLD is both feasible and usable, though retention remains a challenge. The treatment completers reported high satisfaction (CSQ-M = 30.25/32), which is comparable to or higher than satisfaction rates often reported in face-to-face outpatient settings (Attkisson & Greenfield, 1996). The treatment completers also reported high usability and engaged meaningfully with the app content, completing over 70% of assigned exercises. By comparison, other digital mental health interventions often show much lower adherence: a meta-analysis across RCTs found that over 70% of users failed to complete all modules, with adherence (dose received) averaging only ~61% of prescribed content (Gan et al., 2021). Real-world engagement in widely used, unguided mental health apps is strikingly low, with only 3–4% of users remaining active at 15–30 days post-download (Baumel et al., 2019). In contrast, BOLD’s engagement rates compare favorably, reflecting stronger sustained user involvement typical of therapist-supported or minimally guided digital interventions.

However, attrition was a notable limitation of the current pilot study, with only 50% of participants completing the full BOLD protocol. This rate is higher than anticipated and higher than that observed in some other digital EX/RP studies (e.g., 29% in Lundström et al., 2022) and standard in-person EX/RP (15–18%; Ong et al., 2016). This highlights the inherent tension between efficiency and engagement in digital mental health interventions. Future digital EX/RP interventions may benefit from additional features to reduce dropout, for example, a flexible support model in which therapist support start minimal but increases in frequency for participants who show lower engagement or explicitly request additional support.

Qualitative feedback highlighted participants’ perceptions of the intervention’s active ingredients, including gains in OCD knowledge, self-efficacy, and empowerment through independent exposure practice, which align with core theoretical tenets of EX/RP. The program’s self-directed design fostered autonomy and personal agency, while its structure, combining app-based content with brief therapist support, balanced flexibility with therapeutic momentum. Qualitative feedback suggests that therapist support, though minimal, was viewed by participants as critical for sustaining motivation, troubleshooting obstacles, and reinforcing learning. These findings align with prior work underscoring the importance of therapist presence in digital mental health interventions (Schueller et al., 2017; Simpson et al., 2012; Wheaton et al., 2016). A unique strength of BOLD is that it integrates guided processing and expectancy violation through both modules and therapist interactions instead of solely delivering exposure instructions without supporting deeper reflection. Overall, these results point to the value of pairing digital interventions with clinician support to enhance engagement and fidelity.

Practical Implications

From a design standpoint, BOLD’s mobile delivery format and structure offer a unique contribution to the digital EX/RP landscape. On-demand psychoeducation, interactive hierarchy development, real-time exposure tracking, and moment-to-moment self-monitoring are useful features for applying EX/RP skills in daily life. In addition, BOLD includes a built-in planning and reminder system, sending automated exposure prompts and goals via text message, which may enhance follow-through and continuity between sessions. Participants also engaged with structured post-exposure processing prompts aimed at reinforcing learning, promoting expectancy violation, and supporting reflection, a feature not consistently present in other digital EX/RP platforms. Based on participant feedback, future versions of BOLD may also benefit from: (1) personalization algorithms to tailor exposures; (2) adaptive scheduling; (3) expanded asynchronous messaging to promote adherence; 4) addressing technical issues.

From an implementation science perspective, BOLD represents a promising model for delivering gold-standard OCD care in a resource-efficient manner. The program was explicitly designed to minimize demands on highly trained clinicians, addressing a critical barrier in EX/RP dissemination (Senter et al., 2021). Requiring 73.39% less therapist contact time, BOLD is particularly well-suited for stepped-care systems and resource-limited settings. For example, patients with mild or moderate symptoms could start with BOLD and step up to traditional EX/RP if needed while those completing outpatient therapy could use BOLD to maintain treatment gains. Also, participants in our study reported using the app most often during evenings and weekends, periods when traditional therapy is typically unavailable, indicating that asynchronous tools can extend care beyond standard clinical hours without increasing therapist burden. A key next step will be evaluating whether novice clinicians, such as bachelor’s- or master’s-level providers, can deliver BOLD with comparable efficacy to expert providers, further expanding its scalability.

Limitations and Future Directions

Several limitations must be acknowledged. First, this study did not include a control or comparison condition; findings should therefore be interpreted as preliminary and subject to expectancy and other nonspecific effects. Additionally, the pilot sample was small, relatively homogenous, and recruited from a single academic medical center, which further limits generalizability. Self-selection bias could have further influenced enrollment and outcomes, as individuals who opted into a digital program may differ from the broader OCD population in terms of motivation or comfort with technology. Second, long-term follow-up data were not collected, leaving questions about the durability of gains unanswered.

Third, while our higher attrition rate may reflect challenges inherent to self-guided digital care, it remains to be a threat to study validity. Given this and the other limitations mentioned above, the evidence for BOLD’s effectiveness should be considered with caution. These limitations are common to early-stage pilot work and underscore the need for a fully powered RCT to rigorously evaluate BOLD’s efficacy and attrition. To address attrition in future trials, researchers need to provide patients with psychoeducation about digital care models and their equivalence to in-person treatment, as well as a hybrid onboarding process to increase patients’ confidence and motivation in the digital intervention. Identifying predictors of response and dropout, such as baseline symptom severity, app usage patterns, or therapist contact frequency, could help further refine BOLD.

Fourth, although we conducted a time-effectiveness analysis, cost effectiveness remains unclear. Given the substantial cost of the development and maintenance of the digital interventions, it is important to perform formal cost-effectiveness analyses to examine BOLD’s economic value relative to standard EX/RP and other digital treatments. Finally, although BOLD incorporated various technical features to enhance engagement (e.g., automated reminders), their unique contributions to adherence and outcomes remain unclear. Isolating and evaluating the role of these features can inform which components most effectively enhance engagement and clinical outcomes.

Conclusions

This pilot study offers encouraging initial evidence that BOLD, a brief therapist-supported digital EX/RP intervention, is feasible, acceptable, and associated with substantial improvements in OCD symptoms. By maintaining fidelity to core treatment principles while minimizing therapist burden, BOLD offers a scalable approach for expanding access to gold-standard OCD care. Importantly, our findings underscore the critical role that even minimal clinician support may play in enhancing adherence, learning, and overall treatment engagement in digital interventions. Continued refinement and rigorous testing of digital EX/RP interventions will be critical to determining their role in stepped-care models and their potential to extend evidence-based treatment to individuals who might otherwise remain untreated. Because attrition was higher than expected in this study, another key feature in the next steps of this work will be to enhance engagement with the program. Future work should evaluate BOLD in a larger, more diverse sample, assess its integration into health care systems, and determine its optimal placement within stepped-care frameworks for OCD.

Data Availability:

Data will be made available upon request. The study protocol was pre-registered here: https://osf.io/56xta/files/osfstorage.