Obsessive-compulsive disorder presents in widely different forms, and clinicians have long suspected those variations respond differently to treatment. A new study tracking 152 OCD patients through an eight-week course of exposure and response prevention therapy found that most improved while maintaining their pre-treatment symptom profile. Hoarding-related symptoms proved significantly more resistant to change than other presentations.
The research, published in the British Journal of Clinical Psychology, used person-centred modelling to track how distinct OCD symptom clusters evolved across a full course of therapy. The findings suggest that patients with prominent hoarding symptoms may need targeted interventions beyond standard protocols.
OCD affects an estimated 2-3% of the population worldwide and is characterised by persistent intrusive thoughts and repetitive rituals performed to neutralise anxiety. The presentation is far from uniform. Some patients are dominated by contamination fears and washing rituals; others are paralysed by doubt and checking; still others accumulate objects compulsively or feel compelled to arrange them in exact patterns. These distinct symptom dimensions have been catalogued for decades, yet how they evolve at the individual level during therapy remained poorly understood.
Exposure and response prevention (ERP) is the gold-standard psychological treatment for OCD. The therapy works by having patients confront feared situations while refraining from their usual compulsive responses, gradually weakening the anxiety-relief cycle that maintains the disorder. Despite strong average outcomes, response rates vary considerably, and researchers have long sought predictors that could guide more personalised treatment planning. Seung Yun Baek of Sogang University in Seoul and colleagues set out to investigate whether distinct symptom profiles at treatment onset predict response. They also asked whether those profiles themselves shift across the therapy course.
The team drew on data from two randomised controlled trials originally funded by the US National Institute of Mental Health. The combined sample comprised 152 adults with a confirmed OCD diagnosis, with a mean age of 34.92 years and 42.1% female participants. Participants began treatment with a mean score of 26.13 on the Yale-Brown Obsessive-Compulsive Scale (Y-BOCS), placing the group firmly in the severe symptom range.
All 152 participants completed 17 twice-weekly ERP sessions over eight weeks. Assessments were conducted at baseline, midpoint (session eight, week four), and post-treatment (session 17, week eight). Symptom profiles were built from the Obsessive-Compulsive Inventory-Revised (OCI-R), a self-report measure covering six dimensions: washing, obsessing, ordering, checking, neutralising, and hoarding. Internal consistency for the OCI-R subscales ranged from .83 to .93. By post-treatment, the group’s mean Y-BOCS total had fallen from 26.13 to 17.57.
To identify distinct symptom subgroups and track their change over time, the researchers used two complementary statistical techniques. Latent profile analysis (LPA) is a person-centred approach that places individuals into classes based on their pattern of scores across multiple measures rather than analysing each variable in isolation. Latent transition analysis (LTA) then extends this by estimating the probability that a person moves from one class to another across successive time points. Together, the two methods allowed the team to ask not merely whether symptoms improved on average, but which distinct profiles existed and how stable those profiles remained across the eight weeks. Missing data were addressed using multiple imputation, with a formal test confirming the missing values were random rather than systematic (χ² = 36.95, p = .179).
At baseline the analyses identified four distinct symptom classes. The largest, Class 1, accounted for 53.3% of participants (n=81) and showed low scores across all six OCI-R dimensions. Class 2 (20.4%, n=31) was distinguished by elevated hoarding. Class 3 (17.1%, n=26) showed elevated neutralising scores, reflecting a pattern of performing mental acts, such as counting or silently repeating phrases, to counteract distressing thoughts. Class 4 (9.2%, n=14) combined very high hoarding with very high neutralising. This four-class structure was broadly consistent with prior OCD research showing that symptom dimensions cluster into meaningful subgroups.
Most patients remained in their baseline class as treatment progressed. Among those initially in Class 1, 84.5% were still in the corresponding class at midpoint, reflecting gradual overall improvement rather than any reorganisation of symptom structure.
Hoarding symptoms showed the least movement of any dimension across the treatment period. Patients in Class 2 (high hoarding) at baseline were markedly less likely than others to shift into lower-severity classes as therapy progressed. The authors noted that “hoarding symptoms in the context of OCD were particularly stable and less responsive to treatment.” This matters clinically because hoarding involves distinct psychological processes, including strong emotional attachment to objects and deeply held beliefs about their utility. Standard ERP protocols, designed primarily around contamination fears, may not adequately address these.
A notable finding was the emergence at midpoint of a symptom configuration absent at baseline. This class, labelled Class 4a, accounted for 17.1% of the sample (n=26) and was characterised by elevated neutralising scores. It drew members from across all four baseline classes, suggesting a transient mid-treatment state rather than a stable pre-existing subtype.
The researchers then examined what distinguished participants who transitioned into Class 4a at midpoint from those who remained in the lowest-severity group. Lower baseline avoidance behaviour emerged as a significant predictor. Mean baseline scores on the Y-BOCS avoidance subscale stood at 1.87 (SD 1.09), falling to 1.20 (SD 1.19) by post-treatment. Patients who had avoided fewer feared situations at the start may have engaged more readily with early exposure exercises. This could temporarily heighten the salience of neutralising cognitions before they too resolved with continued treatment.
The study carries several limitations. The sample of 152 participants is modest for latent profile analysis, and smaller classes may represent less stable statistical solutions. The constrained sample size also limited the number of predictors that could be included in the transition models, leaving questions about the role of comorbid diagnoses or medication variations unanswered.
The participant pool was predominantly non-Hispanic White and drawn from two specific trials conducted in the United States. All participants were receiving stable serotonin reuptake inhibitor (SRI) medication throughout the trial period, making it difficult to isolate the psychotherapy effects. Whether the same profile patterns would emerge in community samples, in unmedicated patients, or in more ethnically diverse populations remains to be tested.
The neutralising subscale of the OCI-R also carries weaker construct validity than its counterparts, which may affect the precision of neutralising-class distinctions. In addition, hoarding is now classified as a separate disorder in DSM-5 rather than an OCD symptom dimension. The hoarding findings here therefore describe hoarding within OCD specifically, not hoarding disorder as a standalone condition.
The authors propose that patients with prominent hoarding symptoms may benefit from augmentation strategies within the standard ERP framework. These might include additional motivational work, explicit targeting of saving cognitions, or hoarding-specific cognitive components added alongside the core protocol. Elevated neutralising at midpoint may signal a patient entering a productively unstable treatment phase, where mental rituals are actively being challenged but not yet resolved. As the paper notes, “most patients maintained their pretreatment profiles while gradually improving, whereas others showed notable shifts in symptom configurations.”
Replication in larger and more diverse samples will clarify how broadly these profile patterns generalise. Future work could also test whether beginning with hoarding-specific interventions before transitioning to standard ERP produces better outcomes for patients with prominent accumulation symptoms.
The study, “Changes in obsessive-compulsive symptom profiles in response to exposure and response prevention,” was authored by Seung Yun Baek, Hyunsik Kim, Michael G. Wheaton, Jeremy Tyler, Edna Foa, and Helen Blair Simpson.
