Common rituals in obsessive-compulsive disorder and implications for treatment: A mixed methods study

Abstract

Obsessive-compulsive disorder (OCD) is characterized by engagement in rituals that serve to obtain certainty and prevent feared outcomes. Exposure and response prevention is most effective when rituals are resisted, yet existing self-report measures of OCD limit identification of the full range of possible rituals, and little is known about how rituals might cluster together and predict worsened severity and poorer treatment outcomes. In a retrospective sample of 641 adult patients who received intensive OCD treatment, the current study used a mixed-methods approach to (a) identify and validate treatment provider-identified rituals using the Yale Brown Obsessive-Compulsive Scale, (b) identify clustering patterns of rituals, and (c) examine the impact of these clusters on severity and treatment outcomes. Sixty-two discrete rituals clustered into eight higher-order ritual clusters: Avoidance, Reassurance, Checking, Cleaning/Handwashing, Just Right, Rumination, Self-Assurance, and All Other Rituals. At admission, Reassurance predicted greater intolerance of uncertainty (IU) and Rumination predicted less OCD severity. Only one ritual cluster—Just Right—predicted treatment outcomes; patients with Just Right rituals had worse IU at discharge and significantly longer length of treatment (average 7.0 days longer). Clinical observation can identify more nuanced and individualized rituals than self-report assessment alone. Patients presenting with Just Right rituals may benefit less from treatment focused on harm avoidance and habituation; instead, treatment should be tailored to the idiosyncrasies of incompleteness and not just right experiences.

Introduction

Obsessive-compulsive disorder (OCD) is characterized by two primary symptoms: Obsessions, which are recurrent, intrusive, and distressing thoughts, images, or urges; and compulsions (“rituals” used interchangeably), which are repetitive behaviors or mental acts that the person feels driven to perform in response to the obsession (American Psychiatric Association, 2013). Commonly endorsed rituals among individuals with OCD include checking (72.8%), cleaning (59.2%), repeating (51.5%), ordering and arranging (50.0%), and counting (33.5%; Pinto et al., 2008). Underlying the urge to engage in most compulsive behavior is intolerance of uncertainty (IU), a dispositional characteristic found in all content dimensions of OCD involving negative cognitive appraisals (e.g., “uncertainty is dangerous”) and dysfunctional behavioral responses to uncertainty (e.g., perceived inability to move forward in the absence of certainty; Buhr & Dugas, 2002; Grayson, 2010; Pinciotti et al., 2021[a]). An individual may feel they cannot disengage from repeated checking, for example, until they perceive that they have obtained a sense of certainty regarding feared outcomes or doubts (Fourtounas & Thomas, 2016).

Compulsions can be conceptualized as efforts to escape or avoid obsession-related stimuli, thoughts, and uncertainty, yet maintain OCD symptoms over time because they prevent opportunities for corrective learning about the causes or consequences of feared outcomes (Blakey & Abramowitz, 2019). Moreover, it is compulsive rituals, not the obsessions themselves, that lead to dysfunction in relationships, employment, academics, and activities of daily living. For example, significant functional impairment may be caused by lengthy ritualistic showering and work tardiness, repeatedly checking that the stove burner is off before leaving the house and missing a social engagement, or requiring excessive reassurance from loved ones about feared outcomes and experiencing subsequent relationship discord.

OCD Rituals and Treatment Implications

The most well-studied and efficacious treatment for OCD is exposure and response prevention (ERP), a form of cognitive-behavioral therapy involving systematic confrontation with obsession-related stimuli (exposure) along with resisting urges to ritualize (response prevention) (e.g., Abramowitz, 2006). Findings regarding adherence to ERP suggest that while exposure to obsessional content is important for eliciting obsessions and distress, it is response prevention that best predicts and maximizes treatment gains. In particular, achieving 90% or better response prevention was associated with minimal OCD symptoms at post-treatment, whereas those with 75% response prevention had only an 18% likelihood of achieving success in treatment (Wheaton et al., 2016). Response prevention is necessary for corrective learning regarding the causes, consequences, or likelihood of the individual’s fears. While obsessions/intrusive thoughts cannot be controlled (and attempts to do so tend to increase the severity and frequency of them), they can be tolerated through the behavioral change that occurs with adequate response prevention, and a meaningful, functional life can still be lived despite the presence of disturbing obsessions. By resisting urges to obtain certainty through rituals, ERP is inherently effective in increasing tolerance to uncertainty, and these increases appear to partially explain improvements in OCD symptoms over the course of ERP (Pinciotti et al., 2020).

However, despite substantial evidence supporting ERP (e.g., Franklin et al., 2000; Reid et al., 2021), the overall attrition rate is estimated around 19% (Ong et al., 2016) and the research remains unclear as to whether certain symptom content dimensions are more or less amenable to treatment (Williams et al., 2013), perhaps because the standard categorization based on obsessional content does not provide enough nuance to the processes believed to underlie ERP. Namely, in ERP, the importance of content theme is typically de-emphasized by providers in favor of an emphasis on reducing and ultimately preventing engagement in OCD-maintaining rituals. However, it is not known whether certain rituals are more difficult to resist (i.e., poorer response prevention) and thus predictive of poorer treatment outcomes. For example, compulsions associated with feelings of incompleteness and a need to feel “just right” – observed in approximately 23% of individuals with OCD – predicted greater severity of OCD, global and social/occupational functional impairment, and lower quality of life (Sibrava et al., 2016), and incompleteness tends to linger after exposure and pharmacologic treatment (Foa et al., 1999). Similarly, rituals commonly co-occur, and these co-occurrence clusters may carry even more weight in predicting outcomes beyond the singular ritual.

Commonly used OCD measures that quantify symptom severity and treatment outcomes often either combine obsessions and compulsions into overarching content dimensions within the measure itself (e.g., Abramowitz et al., 2010), or are factor analyzed into content dimensions that combine both obsessions and compulsions (Bloch et al., 2008; Cervin et al., 2019; Miguel et al., 2008; Rosario-Campos et al., 2006; Torres et al., 2016). Combining obsessions and compulsions can be problematic because while some compulsions map neatly and intuitively onto respective obsessions—such as contamination obsessions and cleaning behaviors—compulsions often exist across content dimensions. Aggressive, sexual, religious, and somatic obsessions are all associated with checking compulsions, and symmetry obsessions are associated with ordering and arranging, counting, and repeating compulsions (Leckman et al., 1997). Even those rituals that do appear to map intuitively onto respective obsessions may include elements of multiple rituals – for example, handwashing according to very rigid rules or until it feels just right (e.g., Mathes et al., 2019). These nuances further convolute inconsistent findings regarding treatment effectiveness across content dimensions, suggesting that greater attention should be paid to the predictive validity of the rituals themselves. Standardized assessments of compulsions also inherently limit identification of the possible compulsions that may not neatly fit into pre-existing categories – such as those described under the catch-all category of “Other Compulsions” on the Yale-Brown Obsessive-Compulsive Scale (Y-BOCS; Goodman et al., 1989), implying that such compulsions are homogenous when this is likely not the case. A mixed-method approach to identifying rituals, such as using both the Y-BOCS self-report as well as clinical observation, provides more nuanced, case-specific identification of compulsions that extends beyond what the existing standardized measures can capture. This mixed-method approach is also more consistent with the approach used by providers in treatment in real time beyond what can be reported in more structured clinical studies.

Limitations to existing dimensional approaches to OCD symptoms preclude the ability to pinpoint the frequency of compulsions, typical clustering of compulsions, and the impact this clustering may have on the effectiveness of OCD treatment. Given the implications of compulsions in treatment outcomes and functional impairment, more nuanced examination into the clustering and treatment impact of compulsions is needed. Using a mixed-method approach in a clinical sample, the current study sought to examine treatment provider-identified rituals for frequency, clustering, and impact on treatment trajectory and outcomes. It was hypothesized that these rituals would evidence expected relationships with Y-BOCS compulsions when relevant (e.g., between provider-identified rituals and compulsions explicitly listed on the Y-BOCS), yet this method would identify compulsions beyond what is listed in the Y-BOCS. These rituals were hypothesized to cluster together, suggesting common co-occurrences, and be uniquely predictive of OCD and IU severity at admission. Lastly, it was hypothesized that some ritual clusters would be less responsive to treatment over and above the effects of OCD severity, co-occurring conditions, and demographics (i.e., age, race, ethnicity, gender, and sex assigned at birth all examined for possible inclusion). For example, consistent with prior research, it was anticipated that ritual clusters associated with incompleteness/just right would be associated with greater OCD severity and worsened treatment outcomes.

Methods

Participants

The study was reviewed and approved by the Rogers Behavioral Health System Institutional Review Board. Participants included 641 adults diagnosed with OCD who received treatment in residential, partial hospitalization (PHP), and/or intensive outpatient (IOP) treatment programs for OCD and anxiety at Rogers Behavioral Health System in Oconomowoc, Wisconsin between June 3, 2015, and August 7, 2021. Only data from participant’s highest level of care was included, such that participants who stepped down through the continuum of care (e.g., from residential to PHP) were represented only in their residential stay. If participants had multiple encounters in their highest level of care (e.g., a patient completed PHP and then attended PHP at the hospital again in the future), data from their first encounter at their highest level of care was used. Participants ranged in age from 18 to 70 years (M = 29.05, SD = 11.28) and identified their race as 85.8% White (n = 550), 2.3% Asian (n = 15), 2.0% Black (n = 13), 0.8% American Indian/Alaskan Native (n = 5), 0.5% Native Hawaiian/Pacific Islander (n = 3), 1.9% multiple (n = 12), and 6.7% refused or were not recorded (n = 43). An additional 4.5% identified their ethnicity as Hispanic or Latin/x (n = 29). Participants for whom gender was available identified their gender as 50.0% cisgender woman (n = 167), 47.9% cisgender man (n = 160), and 2.1% transgender/nonconforming (n = 7). Sex assigned at birth for the sample was 52.1% female (n = 334) and 47.9% male (n = 307).

Assessment

Prior to admission, all patients completed a telephone interview assessing symptoms of OCD and anxiety. Licensed psychiatrists and/or psychologists with extensive training and expertise in OCD and anxiety reviewed all potential admissions screens and determined appropriateness for admission to a specialty OCD and anxiety treatment program, including the level of care that best fit the severity and needs of the patient.

At admission, patients completed an interview-based psychiatric evaluation performed by a licensed attending psychiatrist specializing in OCD to confirm the diagnosis of OCD and any co-occurring diagnoses based on the DSM-5, as well as battery of self-report assessments including the Y-BOCS–Self-Report and other self-report measures to corroborate comorbid diagnoses on a case-by-case basis. Diagnoses were updated in patient charts throughout treatment in cases where an additional diagnosis was identified after admission (e.g., personality disorder, posttraumatic stress disorder). Although some of these diagnoses were derived from structured, standardized clinical interviews, this level of data on a patient-by-patient basis was not made available.

Based on information gathered during initial treatment sessions as well as observed throughout treatment, the most current, frequent, and functionally impairing rituals were collaboratively identified by patients and their treatment providers and recorded in patient charting. An expectation of treatment in this program is that patients complete daily tracking of all urges to engage in each collaboratively agreed upon ritual (termed “bans” within treatment), including submits and resists to these ritual urges. Rituals are not representative of every single ritual in which the patient currently engages or has ever engaged in the past; instead, these rituals encompass the most severe and distressing behaviors at the time of treatment. Patients complete a battery of self-report assessments at admission, biweekly, and at discharge.

Treatment

The primary mode of treatment provided in all levels of care is cognitive behavioral therapy (CBT) with an emphasis on ERP. Specifically, the ERP model heavily emphasized in these programs was habituation, such that exposures were designed, selected, and implemented with a focus on subjective units of distress (SUDs) ratings (e.g., patients began treatment with moderately-rated exposures per SUDs and worked their way up their exposure hierarchy; within-trial exposures were discontinued when 50% habituation was achieved; and between-trial exposures were discontinued and replaced when peak SUDs ratings habituated by 50%). Additional therapeutic interventions include dialectical behavior therapy skills, cognitive restructuring, and recreational therapy. Ancillary treatments are provided secondarily on an as-needed basis, such as behavioral activation for depression or prolonged exposure for posttraumatic stress disorder. The IOPs and PHPs provide multidisciplinary care that includes individual, group, and family therapy, and medication management by psychiatrists. IOPs involve three hours of treatment four days a week and PHPs involve six hours of treatment five days a week. The residential treatment programs provide longer-term, 8-hours daily, multidisciplinary care that includes individual, group, and family therapy, medication management by psychiatrists, medical support by nursing staff, and dietary and substance use support as needed.

Measures

Demographics, Comorbid Psychiatric Conditions, and Treatment Variables.

Demographic variables, comorbid psychiatric conditions, and treatment variables were extracted from patient charts for inclusion as potential covariates. Demographics included age, race (1 = white, 0 = nonwhite), ethnicity (1 = Hispanic or Latin/x, 0 = non-Hispanic or Latin/x), gender (1 = cisgender, 0 = transgender/non-conforming), and sex assigned at birth (1 = female, 0 = male). Comorbid psychiatric conditions identified by psychiatrists during diagnostic evaluation were coded as 1 = present and 0 = absent for the following conditions: agoraphobia, body dysmorphic disorder (BDD), excoriation, feeding/eating, generalized anxiety, hoarding, mood, neurodevelopmental, other anxiety, panic, specific phobia, psychotic spectrum, sleep, social anxiety, substance use, tic, trauma or stressor-related, and trichotillomania disorders (see Table 1). Level of care was coded as 3 = residential, 2 = PHP, and 1 = IOP.

Table 1.

Comorbid psychiatric conditions of the sample

Comorbid Conditions	Percent	n
Agoraphobia	1.72%	11
BDD	2.18%	14
Excoriation	4.06%	26
Feeding/Eating	4.84%	31
Generalized anxiety	20.28%	130
Hoarding	0.16%	1
Mood	46.33%	297
Neurodevelopmental	11.23%	72
Other anxiety	4.21%	27
Panic	5.46%	35
Phobia	0.31%	2
Psychotic	0.62%	4
Sleep	0.62%	4
Social anxiety	14.51%	93
Somatic	1.56%	10
Substance use	5.62%	36
Tic	0.62%	4
Trauma/stressor	5.15%	33
Trichotillomania	1.25%	8

Note: Other anxiety disorders include other anxiety disorders (F41), other specified anxiety disorders (F41.8) and anxiety disorders, unspecified (F41.9).

Intolerance of Uncertainty Scale-Short Form (IUS-12; Carleton et al., 2007).

The IUS-12 is a 12-item self-report measure of emotional, cognitive, and behavioral reactions to uncertainty. Participants rate on a 5-point Likert scale how much they agree with each item (0 = not at all characteristic of me, 5 = entirely characteristic of me). Scores were summed to create a total score. The average IUS score in the current sample at admission was 38.28 (SD = 11.38) and at discharge was 31.88 (SD = 11.47). The IUS-12 has demonstrated strong internal consistency (α = .91), and in the current study, internal consistency was likewise strong (α = .96).

Length of Stay.

Length of stay represented the number of active treatment programming days for the patients’ first encounter at their highest level of care. Thus, for patients in residential care this included seven days/weekly until discharge, whereas the number of active treatment days for patients in PHPs and IOPs, this number reflected the number of days they attended programming (not including weekends), which equated to no more than five days/weekly, until discharge. The average length of stay for the current sample was 50.22 active treatment days (SD = 27.01) overall, including 55.47 active days (SD = 26.56) for residential, 28.11 active days (SD = 11.98) for PHP, and 24.24 active days (SD = 12.05) for IOP.

Rituals.

OCD rituals were extracted from free text fields in electronic patient charts. As part of treatment as usual, rituals are entered into each patient’s exposure hierarchy in a separate field to connote the “bans” that patients are to track daily. Patients complete daily tracking of all urges to engage in each identified ritual, including submits (i.e., completed the compulsion) and resists (i.e., resisted the compulsion) to these urges. Rituals are entered into charting based on the compulsive behaviors that are most current, frequent, and impairing for each patient, and can be collaboratively added throughout treatment based on clinical observations. For example, a patient may enter treatment unaware that they engage in mental review rituals, and then be assigned this ritual as a “ban” once identified by the treatment team and discussed collaboratively with the patient. This ritual then becomes a target for response prevention. Rituals are coded in analyses as overarching labels and a sum of rituals was computed. The range of the sum of rituals of the sample was 1 to 24, with an overall average number of 6.96 rituals (SD = 2.98) per patient, including an average 7.04 rituals (SD = 2.98) in residential, 6.41 rituals (SD = 2.87) in PHP, and 6.84 rituals (SD = 3.11) in IOP.

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

The Y-BOCS includes a checklist of obsessions and compulsions and a severity scale. Obsessions and compulsions on the checklist are divided into the following categories: Aggressive, contamination, sexual, hoarding/saving, religious, need for symmetry or exactness, somatic, and miscellaneous obsessions; and cleaning/washing, checking, repeating, counting, ordering/arranging, hoarding/collecting, and miscellaneous compulsions. The current study examined only currently endorsed compulsions in order to validate the coding of rituals when possible. Consistent with Leonard and Riemann (2012), 87.21% of the sample endorsed both obsessions and compulsions on the Y-BOCS, and the remaining 11.86% who did not self-report compulsions on the Y-BOCS had provider-identified rituals in charting. Thus, in total, 100% of the sample evidenced engagement in compulsions.

Following the checklist is a 10-item self-report measure of severity of obsessions and compulsions. The items measure time occupied by obsessions/compulsions, interference due to obsessions/compulsions, distress associated with obsessions/compulsions, resistance against obsessions/compulsions, and degree of control over obsessions/compulsions. Participants rate each item on a 5-point Likert scale, with higher numbers indicating greater severity. Total scores can range from 0-40 and researchers have suggested a cutoff of 16 for moderate OCD. The average Y-BOCS score in the current sample at admission was 25.56 (SD = 6.13), and at discharge was 16.66 (SD = 7.32). The self-report version of the Y-BOCS has demonstrated strong internal consistency and test-retest reliability (Steketee et al., 1996); in the current study, internal consistency was adequate (α = .86).

We report how we determined our sample size, all data exclusions (if any), all manipulations, and all measures in the study. To maintain confidentiality of patient data, raw data is not widely available. However, coding of analyses and output are available upon request. This study was not preregistered.

Data Analysis Plan

Rituals were extracted from patient charts and coded by the first author. Due to differences in subjective labeling of rituals across treatment providers and typographic and grammatical errors, the first author reviewed all 5,540 rituals extracted and created a coding scheme. Coding was conducted with a balance between categorizing and over-categorizing rituals; because one of the aims of the current study is to identify commonly occurring clusters of rituals, differently-labeled rituals were re-labeled and re-coded into overarching categories with other similar rituals only if it was clear that they fit into such categories, and that discrepancies in terminology were only semantic and likely provider- or patient-specific. Conversely, if rituals were likely related but functionally distinct (e.g., counting [possibly related to not just right experiences] versus good/bad numbers [possibly related to superstition]), they were left as distinct ritual categories (see Table 2 for examples of coding scheme based on exact extraction of rituals entered into patient charts by treatment providers). The coding scheme was then reviewed by the third author who served as clinical supervisor for the treatment programs sampled. Both first and third authors have years of expertise researching and treating OCD and anxiety and have worked as clinical providers within the treatment programs sampled, experience that also served to inform coding decisions. Discrepancies in coding decisions (primarily with respect to when to lump and split categories) were discussed between first and third author until a consensus was achieved.

Table 2.

Example coding scheme for rituals extracted from patient charts

Superstitious
Superstitious behaviors
Good luck behaviors
Superstitious behavior
listening signs
general superstitions
fingers crossed
superstitious beliefs
Signs
jinx
following signs
saying something will happen
Meaning making
Superstitions
knock wood
Repeating
Repeating
redo
Repeat
Redoing
retracing
repeating sentences
Repeating actions
Back-tracking
back tracking
repeating song lyrics
Repetition
Re-watching
resetting
restarting
refresh reset
backtracking
redoing repeating

Note. Words in bold represent the ritual label; all words underneath represent examples of rituals entered into patient charts that were coded into the ritual label.

Coding was validated against the Y-BOCS using binary flags in accordance with a coding scheme created by the first author. Y-BOCS compulsions were flagged if they clearly overlapped with rituals, and Y-BOCS compulsions were mapped onto one or more rituals where relevant (e.g., reassurance and confessing rituals both overlapped with the Y-BOCS item: “Need to tell, ask, confess”).

A sparse matrix was created of the rituals data, with binary flags indicating whether one of the keywords associated with the rituals categories appeared in any of a patient’s rituals. Data was organized with each patient representing one row, with flags for which rituals a patient endorsed. An agglomerative hierarchical cluster was built using the “cluster” package in R (Maechler et al., 2021). A hierarchical cluster is a heuristic-based algorithm that creates clusters based on patterns found in the data alone, not based on measures of probability or uncertainty. An agglomerative cluster is unique in that it works in a “bottom-up” fashion, with each ritual, in this case, starting as its own cluster. Thus, rituals that are depicted in the tree as nearest each other suggest high co-occurrence whereas rituals that are depicted further apart suggest lower co-occurrence. Clusters are then merged based on a distance metric until the entire hierarchy has been agglomerated. The algorithm used the row-wise data from each patient to model the frequency of ritual co-occurrence based on how many patients endorsed the same pattern of clusters. The clusters identified using this method were then included as factors in the multivariate regressions that follow (and coded as 1 = patient endorses ritual belonging to this cluster, 0 = patient does not endorse ritual belonging to this cluster).

To examine the relationship between ritual clusters and admission severity (i.e., Y-BOCS and IUS), multivariate regressions were computed. Lastly, to examine the predictive validity of cluster rituals on treatment outcomes (i.e., discharge Y-BOCS, IUS, and length of stay), multivariate regressions were computed. Covariates in these models included demographics, comorbid psychiatric conditions, and treatment-related variables described above. All provided models are maximally parsimonious, such that only significant independent variables and covariates are included. Results were derived using stepwise backward selection.

Results

Ritual Frequency, Validation, and Clustering

In total, 62 unique rituals were identified and coded as 1 = present, 0 = absent for each patient. For each identified relevant compulsion on the Y-BOCS, a proportion was calculated for the number of patients assigned with the respective ritual who also endorsed the equivalent item on the Y-BOCS (see Table 3 for these proportions). 67.0% of patients had a ritual denoted in their chart that was also endorsed on the Y-BOCS, suggesting a moderate amount of agreement between patient self-reported compulsions and rituals assigned by treatment teams. As expected, of the 62 rituals categories identified for this analysis, only 27 of those categories are explicitly represented in the Y-BOCS, indicating that rituals identified through clinical observation and discussion provides more nuanced information beyond what is available on the Y-BOCS.

Table 3.

Frequency of rituals for the total sample and validity with relevant Y-BOCS compulsions

Ritual Category	Examples from chart	% endorsed (total)	N endorsed (total)	Y-BOCS compulsion(s) associated with ritual	% endorsed on Y-BOCS	n endorsed on Y-BOCS
Barriers	Barriers; using barriers; gloves; napkin; face mask	15.4%	99	(1) Other measures to prevent or remove contact with contaminants	72.7%	72
Changing clothes	Changing clothing; changing; laundry	2.0%	13	(1) Other measures to prevent or remove contact with contaminants	30.8%	4
Checking	Checking; checking physical items; physical checking; rechecking; reviewing	50.4%	323	(1) Checking locks/stoves/appliances, water faucets, emergency brake; (2) did not harm others; (3) did not make mistake	70.3%	227
Cleaning	Cleaning; wet wipes; Clorox wipes; Lysol; cleaning objects; wiping surfaces	36.8%	236	(1) Cleaning of household items or other inanimate objects	57.6%	136
Confessing	Need to confess; telling people things; admitting; warning others	11.5%	74	(1) Need to tell, ask, confess	74.3%	55
Counting	Counting; recounting	9.0%	58	(1) Counting compulsions	72.4%	42
Handwashing	Handwashing; washing hands; rewashing; washing wrists; hand sanitizing	33.2%	213	(1) Excessive or ritualized hand washing	78.4%	167
Hesitating	Hesitating; avoiding starting; pausing getting stuck; freezing	2.3%	15	(1) Pathological slowness	20.0%	3
Hoarding	Hoarding; saving items; collecting	3.1%	20	(1) Inspecting household trash and accumulating useless objects	75.0%	15
Just right	Just right; behavior done just right; just right feeling; just right thoughts; just right decisions; symmetry; perfecting; balancing; completeness	33.4%	214	(1) Can’t complete activity until just right	67.3%	144
List making	List making; note taking; documenting; mental list making	12.0%	77	(1) Excessive list making	44.2%	34
Mental review	Mental review; replaying thoughts; recall pictures; mental checking; replaying; rewinding	18.2%	117	(1) Mental rituals	59.0%	69
Mental rituals	Mental rituals; mental twitch; mental songs; mental repeating	4.8%	31	(1) Mental rituals	77.4%	24
Neutralizing	Neutralizing; replacing thoughts; cleaning thoughts; clearing mind; thought replacement; take back; cancelling	11.9%	76	(1) Mental rituals	68.4%	52
Ordering/arranging	Ordering arranging; rearranging; fixing; tidying; re-adjusting	8.7%	56	(1) Lines up clothes, canned goods, shoes in fixed order; (2) need for symmetry	73.2%	41
Praying	Praying; asking forgiveness; praying perfectly; bible verses; repenting; fixing relationship with God	8.3%	53	(1) Mental rituals	81.1%	43
Reassurance	Reassurance seeking; contacting PCP; emailing pastors; asking clarification; asking questions already asked	61.6%	395	(1) Asking for reassurance over and over; (2) need to tell, ask, confess; (3) repeats same questions	80.5%	318
Repeating	Repeating; redoing; retracing; backtracking; repeating sentences; repeating songs	19.2%	123	(1) Repeats same questions; (2) need to repeat routine activities	66.7%	82
Rereading	Rereading; reread	11.7%	75	(1) Rereading or rewriting	89.3%	67
Rewriting	Rewriting; rewrite; retyping	7.6%	49	(1) Rereading or rewriting	87.8%	43
Self-assurance	Self-assurance; affirmation; positive self-talk; reframing	23.6%	151	(1) Mental rituals	49.7%	75
Self-harm	Self-harm; punishment; scratching; mental self-harm; purging self-harm	1.7%	11	(1) Self-damaging behaviors	72.7%	8
Showering	Showering; rinsing	3.1%	20	(1) Excessive or ritualized showering, bathing, tooth brushing, grooming	80.0%	16
Somatic checking	Symptom monitoring; body checking; checking pulse; analyze bowel; checking health	3.3%	21	(1) Checking tied to somatic obsessions	38.1%	8
Somatic eyes	Staring; blinking; shutting eyes; physical rituals eyes; closing eyes	3.0%	19	(1) Rituals involving blinking or staring	57.9%	11
Superstitious	Superstitious behaviors; good luck behaviors; fingers crossed; signs; jinx; knock on wood	0.8%	5	(1) Superstitious behaviors	100.0%	5
Touch/tap/rub	Touch/tap/rub; tapping pen; foot tapping; touching things	7.2%	46	(1) Need to touch, tap, or rub	65.2%	30
Avoidance	Avoidance; fleeing; changing location; school avoidance	83.9%	538	--	--	--
Bingeing	Bingeing	1.1%	7	--	--	--
Calories	Counting calories	0.3%	2	--	--	--
Caretaking	Caretaking; caregiving	1.4%	9	--	--	--
Comparing	Comparing	4.3%	28	--	--	--
Distraction	Distractions; daydreaming; social media; keeping busy	3.1%	20	--	--	--
Eating	Compulsive eating; cutting food; grouping food	0.8%	5	--	--	--
Isolation	Isolating	16.2%	104	--	--	--
Masking	Masking; pretending; humor diffusion; joking; deflection	1.9%	12	--	--	--
Mind reading	Mind reading; checking people’s reactions and behaviors	0.6%	4	--	--	--
Need to know	Need to know	1.2%	8	--	--	--
Numbers	Numbers; odd numbers; even numbers; nines; safe numbers	0.3%	2	--	--	--
Over-analyzing	Over-analyzing; overthinking; calculations probability; evidence building	10.6%	68	--	--	--
Over-apologizing	Over-apologize; excessive apologizing; saying sorry	9.0%	58	--	--	--
Over-explaining	Over-explain; tangents; oversharing; qualifying overexplaining	5.3%	34	--	--	--
People pleasing	People pleasing; over-politeness; please thank excuse; thank you	7.5%	48	--	--	--
Phrases	Repeating phrases; mantras; chanting; mental sayings	3.1%	20	--	--	--
Picking	Picking; picking fingernails; skin picking; touching face; touching neck	13.3%	85	--	--	--
Preparing	Preparing; over-preparing; mental preparing; mental rehearsal; mental plan; making escape route	3.6%	23	--	--	--
Pulling	Pulling; hair pulling; eyebrow rubbing; touch hair; playing hair; searching feeling	5.5%	35	--	--	--
Researching	Researching; looking online; information seeking; googling health information	9.8%	63	--	--	--
Restricting	Restricting; restricting water, bathroom use, food	1.7%	11	--	--	--
Rumination	Rumination; mental rumination; verbal rumination	26.7%	171	--	--	--
Rushing	Rushing; impulse need to get something done immediately	0.5%	3	--	--	--
Safety behaviors	Safety behaviors; safety foods; safety person; safety objects	1.4%	9	--	--	--
Scanning	Scanning; looking; room scanning	2.3%	15	--	--	--
Somatic breath	Breathing; forced breathing; holding breath; inhale exhale; blowing; controlled breathing; yawning	1.9%	12	--	--	--
Somatic functions	Coughing; spitting; smelling; swallowing; nose clearing; cracking jaw; popping cracking	2.3%	15	--	--	--
Somatic mouth/throat	Mouth covering; throat noises; pursing lips; tongue biting	3.0%	19	--	--	--
Somatic movement	Movement; pacing; stomping; snapping; swaying; jumping; body movements	5.6%	36	--	--	--
Somatic shaking	Head shaking; hand shaking; leg shaking; nodding; fidgeting	2.0%	13	--	--	--
Somatic tensing	Flexing; clenching; tensing; hugging chest; bowel clenching; tightening	1.2%	8	--	--	--
Thought stopping	Thought stopping; push thoughts away; blocking thoughts; pushing; thought prevention	2.3%	15	--	--	--
Tracking	Keeping track; taking mental inventory; keeping tabs; mental tracking; taking mental pictures	1.9%	12	--	--	--
Worrying	Worrying; excessive worry thought; future-thinking; catastrophizing	5.0%	32	--	--	--

Note. Percent and n (total) indicate the proportion of patients with the assigned ritual. Percent and n endorsed on Y-BOCS indicates the proportion of patients with the assigned ritual who endorsed the equivalent compulsion on the Y-BOCS. These numbers differ because not every assigned ritual has a clear corresponding compulsion on the Y-BOCS.

Eight ritual clusters were identified using agglomerative clustering (see Table 4 and Figure 1). The clustering method with the highest agglomerative coefficient (AC) was chosen, complete linkage with AC = 0.87. This coefficient describes the strength of the available clustering structure with a value between 0, indicating no structure, and 1, indicating clearer structure. The number of clusters was validated using the “clValid” package in R (Brock et al., 2008), by choosing the partition of clusters with the highest Dunn index. The Dunn index can lie between 0 and infinity, where a higher value indicates well-defined, well-separated clusters, and a low value indicates less definition between clusters. Calculated using Gower distance, the Dunn index for the eight-cluster partition was 0.74.

Table 4.

Cluster membership and frequencies

Cluster number	Percent (n)	Rituals
1	83.9% (538)	Avoidance
2	61.6% (395)	Reassurance
3	50.4% (323)	Checking
4	37.6% (241)	Cleaning/handwashing
5	33.4% (214)	Just right
6	26.7% (171)	Rumination
7	23.6% (151)	Self-assurance
8	92.5% (593)	All other rituals

Figure 1.

Agglomerative cluster map for endorsed rituals, where boxes indicate the partitioning of eight derived clusters.

Clusters within the tree, generally, indicate greater similarity—based on co-occurrence frequency—relative to rituals several nodes away from each other, and visualization confirms this (Figure 1). Cluster 8 (All Other Rituals) was the most difficult group of rituals to detect dissimilarity, and this is in part due to small sample size. However, patterns are evident in Cluster 8. For example, calories and numbers are one of the first pairs of rituals to join together, suggesting that these are more similar to each other than those joined together later. Barriers and isolation also join together quickly, as do list making and rewriting. Checking, in comparison, is the last ritual to join the hierarchy, indicating one of the highest dissimilarities with the other rituals. Cleaning and handwashing comprise the node with the longest branch away from the rest of the hierarchy, suggesting that these two have high similarity of co-occurrence yet are also highly dissimilar from the other rituals.

Severity at Treatment Admission by Ritual Clusters

Cluster 6 (Rumination) was negatively predictive of admission Y-BOCS, suggesting that individuals with rituals encompassing this cluster endorsed less severe OCD at admission, β = −1.30, p <.05 (see Table 5). Specifically, Cluster 6 was associated with an average admission Y-BOCS of 24.82 (SD = 6.47), 0.75 points lower than the average score for the sample. Among covariates, patients were more likely to have a higher Y-BOCS score at admission when enrolled at a higher level of care (β = 2.38, p <.001), or if they endorsed a larger sum of rituals (β = 0.25, p <.01). Patients were more likely to have a lower Y-BOCS at admission if they did not have a comorbid diagnosis of a generalized anxiety (β = −1.32, p <.05), psychotic spectrum (β = −12.67, p <.01), or social anxiety disorders (β = −1.91, p <.01).

Table 5.

Regressions predicting severity at treatment admission

Y-BOCS
Predictor	Estimate	SE	CI95%lower	CI95%upper	t	p
Intercept	18.20	1.21	15.81	20.58	15.00	<.001***
Level of care	2.38	0.39	1.61	3.15	6.08	<.001***
Sum of rituals	0.25	0.08	0.10	0.41	3.19	<.01**
GAD	−1.32	0.59	−2.49	−0.16	−2.23	<.05*
Psychotic	−12.67	4.12	−20.76	−4.58	−3.08	<.01**
Social anxiety	−1.91	0.67	−3.23	−0.58	−2.83	<.01**
Cluster 6	−1.30	0.55	−2.38	−0.21	−2.35	<.05*
IUS
Predictor	Estimate	SE	CI95%lower	CI95%upper	t	p
Intercept	27.54	1.72	24.16	30.93	15.99	<.001***
Age	0.08	0.04	0.01	0.16	2.10	<.05*
Sum of rituals	0.48	0.15	0.19	0.78	3.20	<.01**
GAD	2.55	1.14	0.32	4.78	2.24	<.05*
Mood	2.90	0.91	1.12	4.68	3.20	<.01**
Social anxiety	2.72	1.25	0.26	5.17	2.18	<.05*
Trauma/stressor	7.08	2.05	3.05	11.12	3.45	<.001***
Cluster 2	3.88	0.93	2.05	5.71	4.17	<.001***

Note. Cluster 2 = Reassurance, Cluster 6 = Rumination.

^***p < .001

^**p < .01

^*p < .05

^†p < .10

Cluster 2 (Reassurance) was positively predictive of IUS at admission, suggesting that individuals with rituals encompassing this cluster endorsed greater intolerance of uncertainty, β = 3.88, p<.001 (see Table 5). Specifically, Cluster 2 was associated with an average admission IUS of 40.28 (SD = 10.72), 2.00 points higher than the average score for the sample. Among covariates, higher IUS score at admission was associated with patients who were older (β = 0.08, p <.05), endorsed a greater sum of rituals (β = 0.48, p <.01), had a comorbid diagnosis of generalized anxiety (β = 2.55, p <.05), mood (β = 2.90, p <.01), social anxiety (β = 2.72, p <.05), or trauma/stressor-related disorders (β = 7.08, p <.001).

Treatment Outcomes by Ritual Clusters

No ritual cluster predicted discharge Y-BOCS score. Patients were more likely to have a higher Y-BOCS score at discharge if they had a higher admission score (β = 0.44, p <.001), were enrolled at a higher level of care (β = 1.28, p <.01), or had a comorbid diagnosis of BDD (β = 6.02, p <.001). Lower Y-BOCS score at discharge was predicted only by a comorbid diagnosis of a panic-related disorder (β = −3.60, p <.01; see Table 6).

Table 6.

Regressions predicting treatment outcomes

Y-BOCS
Predictor	Estimate	SE	CI95%lower	CI95%upper	t	p
Intercept	1.93	1.52	−1.04	4.91	1.28	.20
Admission Y-BOCS	0.44	0.05	0.35	0.53	9.70	<.001***
Level of care	1.28	0.46	0.37	2.19	2.75	<.01**
BDD	6.02	1.92	2.25	9.80	3.14	<.001***
Panic	−3.60	1.18	−5.92	−1.28	−3.05	<.01**
IUS
Predictor	Estimate	SE	CI95%lower	CI95%upper	t	p
Intercept	8.07	1.60	4.92	11.22	5.04	<.001***
Admission IUS	0.53	0.04	0.46	0.61	14.75	<.001***
Sum of rituals	0.27	0.14	−0.01	0.54	1.92	.06†
Trauma/stressor	5.26	1.80	1.71	8.80	2.91	<.01**
Cluster 5	2.71	0.86	1.02	4.40	3.15	<.01**
Length of Stay
Predictor	Estimate	SE	CI95%lower	CI95%upper	t	p
Intercept	−9.33	4.99	−19.13	0.46	−1.87	.06†
Level of care	17.43	1.63	14.23	20.62	10.71	<.001***
Sum of rituals	1.41	0.33	0.75	2.06	4.23	<.001***
Cluster 5	5.98	2.11	1.84	10.13	2.83	<.01**

Note. Cluster 5 = Just Right.

^***p < .001

^**p < .01

^*p < .05

^†p < .10

Only one ritual cluster was found to significantly predict IUS at discharge. Cluster 5 (Just Right, β = 2.12, p <.05) predicted significantly higher discharge IUS, suggesting that individuals with rituals encompassing this cluster endorsed greater IU at discharge. Cluster 5 was associated with an average discharge IUS of 34.03 (SD = 11.40), 2.14 points higher than the average score for the sample. Among covariates, patients were more likely to have a higher IUS score at discharge when they had a comorbid mood (β = 1.92, p <.05) or trauma/stressor-related disorder diagnosis (β = 8.04, p <.001; see Table 6).

Lastly, Cluster 5 (Just Right) was also found to predict significantly longer length of stay, β = 5.93, p <.01. Specifically, Cluster 5 was associated with an average length of stay of 57.22 days (SD = 27.88), 7.00 days longer than the average length of stay for the sample. Longer length of stay was also associated with higher level of care (β = 17.43, p <.001) and endorsing a larger number of rituals (β = 1.41, p <.001; see Table 6).

Discussion

The current study utilized a mixed-method approach to examine commonly occurring and co-occurring OCD rituals, and their impact on symptom severity and treatment outcomes. Replicating prior research (Leonard & Riemann, 2012), 100% of the sample engaged in OCD compulsions, including nearly 12% who endorsed no compulsions on the Y-BOCS checklist but ultimately were able to collaboratively identify rituals with their treatment provider. Examination and coding from patient charts indicated 62 discrete rituals defined as patients’ most current, frequent, and impairing compulsions. As expected, these rituals extended beyond the compulsion options available on the Y-BOCS self-report, with 35 additional rituals identified without having a clear association with a respective Y-BOCS compulsion. This suggests that clinical observation, in tandem with gold-standard self-report measures of OCD, will provide more nuance and individuation to patient conceptualization and treatment. It also suggests that the essential phenomenology of OCD involves both obsessions and compulsions, as opposed to the current wording the DSM-5 (American Psychiatric Association, 2013) which states that a diagnosis of OCD can be made with the presence of either obsessions or compulsions.

Among rituals with clear associations with respective Y-BOCS compulsion(s), validity—as defined by endorsing the respective Y-BOCS associated with the assigned ritual—was moderate, indicating a moderate level of insight and agreement among patients and treatment providers into their compulsive behaviors. Interestingly, some rituals exhibited much greater validity (e.g., 100% agreement on superstitious behaviors; 89% agreement on rereading; 87% agreement on rewriting; 81% agreement on praying; 80% agreement on showering; 80% agreement on reassurance seeking, etc.), whereas others exhibited less validity (e.g., 44% agreement on list making; 38% agreement on somatic checking; 31% agreement on changing clothes), suggesting that some compulsions are more easily identified at treatment admission by patients as present and impairing, whereas others may be more covert, automatic, or associated with less insight into the compulsive and impairing nature of the behavior (which perhaps explains the DSM-5’s diagnostic criteria described above). This finding highlights the need for treatment providers to work with patients to build insight into their compulsive behavior over the course of OCD treatment, including, at a foundational level, simply identifying that certain behaviors are compulsive and may be important treatment targets for response prevention. To ensure mental rituals are not overlooked, clinicians should ask about these rituals explicitly and provide examples to increase patient awareness and understanding as needed.

Results of the agglomerative hierarchical cluster analysis suggested eight higher-order ritual clusters that defined the sample, including: Avoidance, Reassurance, Checking, Cleaning/Handwashing, Just Right, Rumination, Self-Assurance, and All Other Rituals. Most patients had rituals encompassing All Other Rituals (92%) and Avoidance (84%), the majority had rituals encompassing Reassurance (62%) and Checking (50%), and approximately one-quarter to one-third of the sample had rituals encompassing Just Right (33%), Rumination (27%), and Self-Assurance (24%). These clusters are consistent with Pinto et al. (2008) in which checking, cleaning, repeating, and ordering/arranging were the most commonly endorsed compulsions, respectively. The current study builds upon findings by Pinto and colleagues (2008) by including avoidance, the most commonly identified ritual in the current sample. Although there is no debate as to the excessive avoidance behaviors characterizing OCD (e.g., Gillan et al., 2014; Starcevic et al., 2011), the current study emphasizes the need to build awareness of this frequent and impairing behavior and monitor it appropriately throughout treatment.

Agglomerative hierarchical cluster analysis expands upon current literature by allowing further examination into the co-occurrence of ritual clusters based upon the relative length of branches and sequence in which rituals join the hierarchy. In particular, cleaning and handwashing were found to both have a high degree of co-occurrence while also having the highest degree of dissimilarity from the other rituals, suggesting intuitively that these cleaning behaviors are likely to occur in tandem but, interestingly, are conceptually unique from all other rituals. Similarly, checking, as the last ritual to join the hierarchy, evidences high dissimilarity to other rituals, also suggesting that checking rituals are conceptually unique from other rituals. Conversely, rituals such as counting calories and numbers, use of barriers and isolation, and list-making and rewriting each join together quickly, suggesting a high propensity for co-occurrence and therefore high degree of conceptual relatedness.

The high degree of co-occurrence of cleaning and handwashing, coupled with the high degree of dissimilarity from other rituals, may be because they are the two rituals most clearly and consistently linked to a single obsessional theme (i.e., contamination). Whereas other rituals may be more likely to present across obsessional themes and thus have a less defined pattern of expected co-occurrence with other rituals, cleaning and handwashing are likely to map onto contamination concerns in virtually all cases (even if present within the context of other obsessions, such as identity-related obsessions). Checking, one of the most commonly identified rituals in OCD, is a compulsion notably lacking in content specificity, as it is observed across several content dimensions (Leckman et al., 1997). This is likely why Checking emerged as its own unique cluster without other co-occurrences, unlike cleaning and handwashing. Yet checking is unique because while it aims to reduce uncertainty, prevent harm (Rachman, 2002), or diminish not-just-right feelings (Coles & Ravid, 2016), repeated checking has paradoxically been found to increase uncertainty and doubt and reduce accuracy and confidence in memory (van den Hout & Kindt, 2003). The paradoxical effect of checking may explain why this ritual emerged as a unique, highly dissimilar behavior relative to other OCD rituals.

Taken together, findings highlight the clinical limitations of the Y-BOCS checklist, a self-report tool frequently used in clinical samples to identify target compulsions. Although program-specific nuances and patient individual differences likely informed the labeling of certain compulsions in this sample, the current study found that fewer than half (43%) of the rituals deemed most distressing and impairing in a clinical population were explicitly represented on the Y-BOCS checklist. Improvements to the Y-BOCS checklist were made in an updated version (Storch et al., 2010), including the important addition of ritualized avoidance, however the heterogeneity of OCD requires that clinicians do not limit their identification of compulsions to standardized self-report measures alone because doing so will guarantee that compulsions—even those that are most distressing and impairing—will be missed.

Severity

We were not surprised to find that a greater number of different rituals was predictive of greater OCD and IU severity at admission. Indeed, a higher number of identified rituals indicates engagement in a greater number of current compulsions that are deemed severe and impairing by patients or treatment providers. Only two particular ritual clusters were predictive of severity at admission: Rumination was associated with less severe OCD symptoms and Reassurance was associated with more severe IU. Rumination about unwanted intrusive thoughts maintains the distress associated with such thoughts and urges to neutralize them (Kollárik et al., 2020), and is broadly associated with elevations in anxiety and distress (e.g., Blagden & Craske, 1996). Thus, findings from the current study appear to contradict previous literature on the implications of rumination. It is possible that patients at treatment admission are not considering rumination as a mental ritual and therefore not considering the amount of time they are ruminating, the impact of rumination, etc. when completing the Y-BOCS severity items, whereas consideration of this behavior as a ritual may have resulted in higher Y-BOCS severity scores. It is worth noting that the difference in Y-BOCS scores between individuals with reassurance-seeking rituals and the sample’s average was less than one point, suggesting that replication of this finding in future research is needed to prevent over-interpretation of a potentially spurious finding.

Although it is argued that IU is likely to underlie most rituals, as they are all designed to obtain certainty, reassurance seeking is perhaps one of the more explicit certainty-seeking behaviors (Kobori et al., 2012). Reassurance seeking is a behavior intended to reduce distress associated with uncertainty and encompasses asking questions or making statements repeatedly, asking unanswerable questions, asking questions in order to obtain a desired or absolute answer, or indefinitely pursuing information without making a decision or drawing a conclusion (Anxiety Disorders Center, 2016). The provision of reassurance is one of the most common methods of family accommodation (Peris et al., 2008), and family accommodation is associated with more severe functional impairment over and above OCD severity (Storch et al., 2010). Thus, individuals who excessively seek reassurance may experience a paradoxical self-fulfilling cycle in which their need to obtain (unobtainable) certainty drives the reassurance seeking behavior, which, when reinforced, strengthens the perceived need to obtain certainty (i.e., greater intolerance of uncertainty) and increases reliance on reassurance seeking.

Outcomes

Treatment outcome was defined in three different ways: Discharge OCD and IU severity, and length of stay in intensive treatment. Despite no differences in OCD severity or IU at admission, Just Right was the only ritual cluster that predicted treatment outcome, associated not only with more severe IU at discharge, but significantly longer length of stay in treatment – an additional seven days longer compared to individuals without this set of rituals. In intensive treatment, length of stay is weighted equally—if not more heavily—as an indicator of treatment response compared to OCD severity and IU because at intensive levels of care, less variability in outcome scores exists. As opposed to outpatient treatment, which may conclude when symptoms are minimal or in remission, when the client steps up to a higher level of care, or when the client drops out due to any number of other reasons, intensive treatment tends to have lower dropout rates and discharge often occurs when the patient makes treatment gains expected for that level of care. For example, an individual in residential treatment would not remain in treatment until complete symptom remission is obtained but would rather discharge once symptoms are in the moderate range and/or functionality is improved, with expectations to step down to PHP treatment for continued care. Thus, that Just Right was associated with an additional week of intensive treatment has substantial treatment and financial implications.

Just Right may be associated with delayed treatment effectiveness for several reasons. Just right behaviors often stem from feelings of incompleteness and “not just right experiences” (NJREs). Incompleteness and NJREs are correlated with greater OCD severity overall, including greater compulsion severity, more comorbidities, worsened functional impairment, higher rates of unemployment and disability, lower quality of life, and manifests in symmetry/exactness obsessions and ordering/arranging, hoarding, and washing compulsions (Fornés-Romero & Belloch, 2017; Sibrava et al., 2016). Compared to non-clinical individuals, incompleteness and NJREs in individuals with OCD were reported as being more disturbing, harder to suppress, and initiated more urges to ritualize (Fornés-Romero & Belloch, 2017). It has been argued that conventional ERP sometimes over-emphasizes the harm avoidance model of OCD over incompleteness, perhaps explaining why the treatment approach does not work optimally for some individuals with just right compulsions (Schwartz, 2018). A meta-analysis found that OCD treatment explicitly tailored to incompleteness yields modest improvement and suggests a greater need for providers to understand and build competence in treating incompleteness (Schwartz, 2018).

Notably, intolerance of uncertainty improved less over treatment in patients with Just Right rituals compared to those without these rituals in the current sample despite evidencing no differences in intolerance at treatment admission, suggesting that treatment may not have effectively addressed the idiosyncrasies of just right-specific intolerance of uncertainty. For example, patients with Just Right rituals may not have found relief from their negative appraisals of NJRE uncertainty (e.g., “What if I feel off forever?”) or were less able to accept the presence of NJRE sensations and resist engaging in just right rituals that would otherwise eliminate these sensations. The habituation model of ERP, used exclusively in the treatment programs sampled, may not be the best model for addressing Just Right rituals and intolerance of uncertainty. Patients may struggle to habituate to these sensations and may misunderstand, within a habituation framework, that the goal of treatment is to eliminate them. Perhaps instead an inhibitory learning approach (e.g., Jacoby & Abramowitz, 2016) may be better suited for Just Right rituals, wherein the role of habituation is deemphasized in favor of acceptance of uncomfortable sensations and emotions (e.g., NJRE) and expectancy violation (e.g., “I cannot move on with my day until I get rid of this sensation”). Future research may wish to examine the comparative effectiveness of these two ERP approaches on Just Right and other ritual clusters, as the presence of certain rituals may be an important determining factor in treatment approach selection.

Emerging research continues to find an association between symmetry/just right symptoms and trauma exposure, perhaps because symmetry/just right compulsions serve to restore a sense of completeness and predictability following trauma (Cromer et al., 2007; Pinciotti & Fisher, 2022; Pinciotti et al., 2021[b]). Patients with OCD and trauma have less symptom remission over time (Tibi et al., 2020) and may be less likely to benefit from behavioral and pharmacologic treatment (Gershuny et al., 2008). Thus, it is likely that for some patients in the current sample, just right rituals developed following trauma, and the compounded effect of trauma and just right rituals are inherently more challenging and complex to treat.

No ritual cluster nor sum of rituals, controlling for covariates, were significant predictors of discharge OCD severity, although sum of rituals was predictive of a longer length of treatment stay. This finding suggests, intuitively, that patients struggling with a greater number of rituals during their treatment will require a greater dose of ERP, particularly with respect to response prevention, in order to achieve the same reduction in OCD symptoms.

Constraints on Generality and Future Directions

The current study has limitations to acknowledge which may encourage future research. With respect to sample, participants included in this study were primarily white and cisgender, and likely to be of higher socioeconomic status given that the sample was derived from a population of patients enrolled in intensive treatment. This limitation is important to correct in future research given findings that racial and gender marginalization may impact the presentation and severity of OCD (e.g., Pinciotti et al., 2022; Wilson & Thayer, 2020). Inclusivity in clinical research derived from treatment-seeking samples hinges on intentional, sweeping efforts to make these programs more accessible and inclusive for all patients with OCD.

The unstandardized method of identifying and recording patient rituals in charting represents both a strength and weakness of the current study. This approach is a strength because it extends examination of OCD rituals beyond what is listed on existing assessment measures and offers a more nuanced understanding of the precise rituals that are currently most impactful to the individual patient. This approach is limited, however, in that the language used to define observed behaviors may not generalize to other treatment sites. For example, a patient may present to different treatment providers and use a label for their rituals that differs slightly between providers, despite the behavior being the same. It is also contingent on providers’ knowledge and awareness of patients’ most impairing behaviors, allowing the possibility that some more overt rituals may not identified if patients are withholding, or insight is low. Additionally, the subjective nature in which these rituals were labeled by treatment providers and then coded by the authors limits generalizability to other clinical and research samples; although the moderate to high validity found in comparison to Y-BOCS suggests that despite some human subjectivity in labeling and coding decisions, many of the rituals identified are largely consistent with what is endorsed in standardized, self-report measures.

It is our hope that these ritual categories may be used to flexibly inform future research and clinical work, with the understanding that slight semantic differences would be expected across different settings, programs, and providers. In fact, rigidly adhering in future application to only the exact labels identified in this specific sample would contradict this study’s emphasis on the need for nuanced, individualized identification of rituals beyond predetermined checklists. Rather, we encourage future researchers and clinicians to use this list of ritual categories as a working guide, not as a comprehensive list. Thus, although it is possible that rituals identified in the current study that replicate in future research could inform the creation of a more comprehensive rituals checklist beyond the Y-BOCS self-report checklist for clinicians and researchers, creation of such a checklist should be done thoughtfully and flexibly, with proper continued acknowledgement that use of predetermined checklists alone is clearly not sufficient to identify patient’s most distressing and functionally impairing rituals.

Although the sample included only patients with an OCD diagnosis who received treatment in an intensive program primarily focused on OCD, as expected most patients also have comorbid disorders that may have been treated concurrently (e.g., social anxiety secondary to OCD). As a result, the “bans” identified by treatment providers may have included some safety behaviors that were not OCD rituals (e.g., avoidance of public speaking due to social anxiety may have just been coded as avoidance). Without context regarding each individual patient’s presenting problem, it is not possible to use the available data to ascertain which rituals may relate to a secondary or tertiary disorder. This is further complicated by the lack of standardized diagnostic interviews to corroborate diagnoses documented by licensed psychiatrists. Although it is possible that diagnoses documented in patient charting may not have fully captured the entirety of a patient’s clinical presentation, it is unlikely that an obsessive-compulsive or anxiety-based disorder causing clinically significant distress or impairment would not be documented as this would be a relevant and integral treatment target in this setting. Similarly, with the treatment program’s primary focus on OCD/ERP, and with participants being limited in the current sample to only those with an OCD diagnosis, it can be assumed that most rituals identified are specific to OCD. Future research using more standardized methods of assessment and coding, such as a more fine-grained approach to including only rituals that can be clearly traced to OCD, would provide greater reliability to current findings.

The available data was broad enough to inform the partitioning of a small but interpretable number of ritual clusters. In a cluster analysis, balancing cluster specificity with the ability to explain the partition is ideal, and these data facilitate this. However, to truly understand whether patient outcomes can be predicted based on non-negligible features (e.g., demographics, social determinants of health), more diverse data is needed, and future research is needed to answer this open question. It is an encouraging step forward to understand which groupings or categories of rituals predict treatment outcomes and length of stay. However, the findings based on these data alone do not account for variation at the individual level, but rather are based on the patterns of rituals with higher frequencies. A larger sample of data is needed to further examine the patterns of rituals belonging to Cluster 8; whereas the other clusters characterized outcomes based on one to two rituals, Cluster 8 comprised 54 rituals, making it a similarly non-specific category as the “Other Compulsions” item on the Y-BOCS. It is quite possible that further clusters could be elucidated from Cluster 8 with more data.

Future research may seek to test the validity of the explanations provided in this paper for worsened treatment outcomes in individuals with Just Right rituals. Specifically, it is worth examining whether intensive treatment that is better tailored to incompleteness/NJREs may yield better outcomes with respect to individuals with just right rituals in intensive programs. Perhaps, as proposed by Schwartz (2018), differences in treatment outcomes between individuals with harm avoidance-related rituals and incompleteness-related rituals will cease to exist once the latter is better understood and addressed. Studies examining the relationship between trauma and OCD may wish to identify the function of just right rituals within the context of trauma and whether these behaviors develop as a trauma response, as previously hypothesized, which may inform a trauma-focused treatment approach. Lastly, future research may wish to numerically examine response prevention specific to rituals or ritual clusters to determine whether current findings regarding severity and treatment outcomes can be traced to increased ritual-specific difficulties with response prevention as theorized here.

Public Significance Statement.

This study suggests that mixed-method approaches to identify rituals in obsessive-compulsive disorder (OCD) provide more nuanced and individualized clinical information. OCD rituals were found to cluster together, suggesting high co-occurrences, and patients with Just Right rituals experienced delayed and reduced benefit from cognitive behavioral therapy.