Hostname: page-component-74656f9884-npj9w Total loading time: 0 Render date: 2026-11-07T13:20:57.385Z Has data issue: false hasContentIssue false
  • English
  • Français

Unpacking the role of self-reported compulsivity and impulsivity in obsessive-compulsive disorder

Published online by Cambridge University Press:  10 May 2017

Luisa Prochazkova ,
Linden Parkes Open the ORCID record for Linden Parkes [Opens in a new window] ,
Andrew Dawson ,
George Youssef ,
Gabriela M. Ferreira ,
Valentina Lorenzetti ,
Rebecca A. Segrave ,
Leonardo F. Fontenelle  and
Murat Yücel
Luisa Prochazkova
Affiliation:
Monash Institute of Cognitive and Clinical Neurosciences, School of Psychological Sciences, Monash University, Clayton, Victoria, Australia
Linden Parkes
Affiliation:
Monash Institute of Cognitive and Clinical Neurosciences, School of Psychological Sciences, Monash University, Clayton, Victoria, Australia
Andrew Dawson
Affiliation:
Monash Institute of Cognitive and Clinical Neurosciences, School of Psychological Sciences, Monash University, Clayton, Victoria, Australia
George Youssef
Affiliation:
Cognitive Neuroscience Unit, School of Psychology, Deakin University, Geelong, Victoria, Australia Centre for Adolescent Health, Murdoch Childrens Research Institute, Royal Children's Hospital, Parkville, Victoria, Australia
Gabriela M. Ferreira
Affiliation:
Monash Institute of Cognitive and Clinical Neurosciences, School of Psychological Sciences, Monash University, Clayton, Victoria, Australia D'Or Institute for Research and Education, Rio de Janeiro, Brazil
Valentina Lorenzetti
Affiliation:
Monash Institute of Cognitive and Clinical Neurosciences, School of Psychological Sciences, Monash University, Clayton, Victoria, Australia Department of Psychological Sciences, Institute of Psychology, Health and Society, University of Liverpool, Liverpool, United Kingdom
Rebecca A. Segrave
Affiliation:
Monash Institute of Cognitive and Clinical Neurosciences, School of Psychological Sciences, Monash University, Clayton, Victoria, Australia
Leonardo F. Fontenelle
Affiliation:
Monash Institute of Cognitive and Clinical Neurosciences, School of Psychological Sciences, Monash University, Clayton, Victoria, Australia Universidade Federal do Rio de Janeiro, Rio de Janeiro, Brazil D'Or Institute for Research and Education, Rio de Janeiro, Brazil
Murat Yücel*
Affiliation:
Monash Institute of Cognitive and Clinical Neurosciences, School of Psychological Sciences, Monash University, Clayton, Victoria, Australia
*
*Address correspondence to Murat Yücel, Brain and Mental Health Laboratory, Monash Institute of Cognitive and Clinical Neurosciences, c/o 770 Blackburn Road, Clayton, Victoria 3800, Australia. (Email: murat.yucel@monash.edu)
Rights & Permissions [Opens in a new window]

Abstract

Objective

We aimed to determine whether individuals with obsessive-compulsive disorder (OCD) and demographically matched healthy individuals can be clustered into distinct clinical subtypes based on dimensional measures of their self-reported compulsivity (OBQ–44 and IUS–12) and impulsivity (UPPS–P).

Methods

Participants (n=217) were 103 patients with a clinical diagnosis of OCD; 79 individuals from the community who were “OCD-likely” according to self-report (Obsessive-Compulsive Inventory–Revised scores equal or greater than 21); and 35 healthy controls. All data were collected between 2013 and 2015 using self-report measures that assessed different aspects of compulsivity and impulsivity. Principal component analysis revealed two components broadly representing an individual's level of compulsivity and impulsivity. Unsupervised clustering grouped participants into four subgroups, each representing one part of an orthogonal compulsive-impulsive phenotype.

Results

Clustering converged to yield four subgroups: one group low on both compulsivity and impulsivity, comprised mostly of healthy controls and demonstrating the lowest OCD symptom severity; two groups showing roughly equal clinical severity, but with opposing drivers (i.e., high compulsivity and low impulsivity, and vice versa); and a final group high on both compulsivity and impulsivity and recording the highest clinical severity. Notably, the largest cluster of individuals with OCD was characterized by high impulsivity and low compulsivity. Our results suggest that both impulsivity and compulsivity mediate obsessive-compulsive symptomatology.

Conclusions

Individuals with OCD can be clustered into distinct subtypes based on measures of compulsivity and impulsivity, with the latter being found to be one of the more defining characteristics of the disorder. These dimensions may serve as viable and novel treatment targets.

Keywords

CompulsivityimpulsivityOCDmachine learningsubtypes

Information

Type
Original Research
Information
CNS Spectrums , Volume 23 , Issue 1 , February 2018 , pp. 51 - 58
Copyright
© Cambridge University Press 2017 

Introduction

Obsessive-compulsive disorder (OCD) is a neuropsychiatric disorder affecting between 0.3 and 3.1% of the general population,Reference Fontenelle, Mendlowicz and Versiani 1 which places enormous personal, social, and economic burdens on society.Reference Hollander, Doernberg and Shavitt 2 Clinically, OCD can present with a variety of unwanted obsessions, compulsions, or an idiosyncratic mix of both. In addition, OCD can also be associated with other compulsive and impulsive tendencies, such as substance and gambling addiction.Reference Fineberg, Brown, Reghunandanan and Pampaloni 3 Historically, there has been a strong focus on the symptom phenotypes of OCD; however, given its heterogeneous presentation and high rates of relapse, it is important to understand OCD from the perspective of motivational phenotypes that may underlie the disorder. That is, to understand the precise processes and traits that mediate the dysfunctional neural circuits implicated in OCD (for a review, see Pauls et al.Reference Pauls, Abramovitch, Rauch and Geller 4 ) and its superficial symptomatology.Reference Gottesman and Gould 5 This approach may offer new avenues for progress with regard to intervention and prevention.Reference Cuthbert and Insel 6 – Reference Robbins, Gillan, Smith, de Wit and Ersche 9

OCD has traditionally been viewed as a disorder driven primarily by anxiety avoidance and excessive self-control.Reference Grassi, Pallanti and Righi 10 In recent years, however, a competing perspective has emerged that sees two phenotypes driving the behavioral features of OCD: compulsivity and impulsivity.Reference Fineberg, Chamberlain and Goudriaan 7 , Reference Robbins, Gillan, Smith, de Wit and Ersche 9 , Reference Fontenelle, Oostermeijer, Harrison, Pantelis and Yücel 11 Impulsivity is commonly defined as a predisposition to reacting to stimuli in a rapid and unplanned fashion with reduced concern of the potential consequences.Reference Fineberg, Chamberlain and Goudriaan 7 , Reference Chamberlain, Leppink, Redden and Grant 12 In contrast, current conceptualizations suggest that compulsivity reflects the repetitive performance of behaviors without adaptive function that are carried out habitually or in response to avoidance and fear of potential negative outcomes.Reference Fineberg, Chamberlain and Goudriaan 7 Traditional views posit that these phenotypes are diametrically opposed, whereas contemporary views argue that compulsivity and impulsivity may be more overlapping in nature,Reference Robbins, Gillan, Smith, de Wit and Ersche 9 , Reference Tavares and Gentil 13 suggesting that disorders typically characterized by clinically significant expression of one may also show maladaptive expression of the other.

OCD has been labeled the archetypal compulsive disorder.Reference Chamberlain, Leppink, Redden and Grant 14 It is therefore unsurprising that individuals with OCD differ from healthy controls (HCs) by showing compulsive responses across a number of tasks.Reference Fineberg, Chamberlain and Goudriaan 7 However, a growing body of evidence highlights the potential implication of impulsivity in OCD. For example, a number of studies examining trait impulsivity in OCD have found that OCD patients show higher “attentional,” “motor,” and “nonplanning” impulsivity than HCs,Reference Grassi, Pallanti and Righi 10 , Reference Benatti, Dell'Osso, Arici, Hollander and Altamura 15 – Reference Summerfeldt, Hood, Antony, Richter and Swinson 19 as well as increased “negative urgency.”Reference Zermatten and Van der Linden 20 Individuals with OCD, relative to HCs, tend to make risky decisions, favoring options that provide large initial rewards but ultimately lead to a disadvantageous outcome.Reference Grassi, Pallanti and Righi 10 , Reference Cavedini, Riboldi, D'Annucci, Belotti, Cisima and Bellodi 21 – Reference Zhang, Dong and Ji 28 This consistent pattern of irrational responding might reflect an exacerbated anticipation for reward,Reference Fontenelle, Oostermeijer, Ferreira, Lorenzetti, Luigjes and Yücel 29 or failure of a somatic marker to signal differences between advantageous and disadvantageous outcomes,Reference Cavedini, Zorzi and Baraldi 30 or both. Furthermore, OCD patients show a robust impairment in response inhibition.Reference Wright, Lipszyc, Dupuis, Thayapararajah and Schachar 31 – Reference Chamberlain, Fineberg, Blackwell, Robbins and Sahakian 34 Most strikingly, Figee et al.Reference Figee, van den Munckhof, Schuurman and Denys 35 found increased reward anticipation in the nucleus accumbens—a key region subserving reward, motivation, and impulsivityReference Basar, Sesia, Groenewegen, Steinbusch, Visser-Vandewalle and Temel 36 —in individuals with OCD, compared to HCs.Reference Knutson, Adams, Fong and Hommer 37 This growing body of evidence notwithstanding, a number of issues remain. First, some scattered studies have not found differences between individuals with (clinical or subclinical levels of) OCD and HCs on some measures of impulsivity.Reference Chamberlain, Leppink, Redden and Grant 14 , Reference Benatti, Dell'Osso, Arici, Hollander and Altamura 15 Evidence from our large sample should assist in interpreting whether these few studies are exceptional. Second, and more importantly, it is also unclear whether a potential impulsive subgroup would show more or less severe clinical outcomes than other more compulsively driven subgroups. Impulsive tendencies could plausibly compound or protect against compulsive tendencies. The scant evidence that speaks to this issue suggests that a potential impulsive subgroup might be especially vulnerable to worse prognosis,Reference Kashyap, Fontenelle and Miguel 24 but further investigation is needed.

In this study, we apply k-means clustering to trait measures of compulsivity and impulsivity to regroup a large sample of individuals with OCD. Our aims were to determine whether (1) orthogonally related components of compulsivity and impulsivity could be used to separate groups of OCD patients, and (2) these phenotypes are linked to clinical severity. To address the first aim, we separated our sample into four clusters using self-report measures of compulsivity and impulsivity. Specifically, we expected to find a cluster of OCD individuals who score low on both compulsivity and impulsivity, a cluster who score high on compulsivity but low on impulsivity, a cluster who score low on compulsivity but high on impulsivity (i.e., the reverse of the previous cluster), and a cluster who score high on both. With regard to the traditional anxiety-avoidance model in OCD,Reference Grassi, Pallanti and Righi 10 we expected to find the largest proportion of individuals with OCD falling into the high compulsivity and low impulsivity cluster. Furthermore, we predict that the cluster scoring high in both clinical phenotypes should demonstrate the highest clinical severity.Reference Kashyap, Fontenelle and Miguel 24

Methods

Subjects

The total sample consisted of 217 age- and sex-matched participants (female=108; mean age=35.70 years; age range=18–77): 103 individuals clinically diagnosed with OCD; 79 “OCD-likely” individuals; and 35 HCs. All participants gave written informed consent to participate in study protocols approved by the relevant Australian and Brazilian ethics committees. Clinically diagnosed individuals participated in the present study at one of two clinical institutes as part of larger studies: the Monash Institute of Cognitive and Clinical Neurosciences–Melbourne (n=36) and the Institute of Psychiatry of the Federal University of Rio de Janeiro (n=67). Recruitment for both sites was primarily from patient clinics local to each site, as well as media advertisements placed in the broader community. A primary diagnosis of OCD was determined by the patient version of the Structured Clinical Interview for the DSM-IV (SCID–IV). If patients had other comorbid diagnoses thought to be more severe or to underpin their OCD, they were excluded from the study. OCD-likely individuals were recruited online through the Amazon Mechanical Turk community. “OCD-likely” was defined as a score of 21 or higher on the Obsessive-Compulsive Inventory–Revised (OCI––R), an 18-item self-report instrument that determines the severity of OCD symptoms in subclinical populations.Reference Foa, Huppert and Leiberg 38 HCs were part of a larger study at the Melbourne site. HCs were screened using the Mini International Neuropsychiatric Interview (MINI; DSM–IV) as well as an in-house structured interview designed to assess lifetime history of neurological illness, and were excluded if they presented with any past or present psychiatric or neurological illness.

The definitions of impulsivity and compulsivity outlined in the introduction have a strong focus on behavior; however, here we were interested in examining the underlying motivational phenotypes that drive these behaviors. As such, all participants completed a 30-minute computerized battery of self-report questionnaires that putatively assessed compulsivity and impulsivity. To assess compulsivity, we used the Intolerance of Uncertainty Scale-Short Form (IUS–12) Reference Carleton, Norton and Asmundson 39 and the Obsessive Beliefs Questionnaire-44 (OBQ–44). 40 The OBQ–44 measures belief domains linked to OCD using four subscales: perfectionism, importance and control of thoughts, responsibility, and overestimation of threat. The IUS–12 measures responses to uncertainty, ambiguous situations, and the future using two subscales: prospective intolerance to uncertainty, and inhibitory response to uncertainty. Obsessive beliefs and intolerance to uncertainty are hallmarks of OCD and are predictive of symptomatology.Reference Tolin, Abramowitz, Brigidi and Foa 41 , Reference Gentes and Ruscio 42 Thus, the IUS–12 and OBQ–44 likely capture critical motivational elements of compulsivity in OCD. To assess impulsivity, we used the well-validated UPPS–P.Reference Cyders, Smith, Spillane, Fischer, Annus and Peterson 43 , Reference Whiteside, Lynam, Miller and Reynolds 44 The UPPS–P measures several dimensions of impulsivity: negative urgency, lack of premeditation, lack of perseverance, sensation seeking, and positive urgency. Extended descriptions of the compulsive/impulsive measures can be found in supplementary material. Finally, the OCI––RReference Foa, Huppert and Leiberg 38 and the Beck Depression Inventory (BDI)Reference Beck, Steer and Brown 45 were included to measure symptom severity.

Data analysis

For assumption checks please refer to the supplementary material.

Compulsive and impulsive phenotypes of OCD

To investigate whether participants with OCD could be regrouped using the constructs of compulsivity and impulsivity, we performed unsupervised clustering on the subscales of the UPPS–P, OBQ–44, and IUS–12. Principal-component analysis (PCA) was first used to reduce the questionnaire data down to two leading principal components, which explained 52.7% of the variance in the data. A two-component (i.e., a compulsivity component and an impulsivity component) solution was selected primarily to be consistent with the theoretical framework of the study. These two components carried eigenvalues of 4.21 and 2.20, respectively. While a third PC may also have been retained due to having an eigenvalue greater than 1 (eigenvalue=1.10), it explained only an additional 9% of variance and contained loadings from only two subscales of the OBQ-44. Thus, to facilitate interpretability, we only retained two components. Additionally, as some subscales showed variance inflation factors of greater than 3, PCA also served to reduce multicollinearity, and subsequent overrepresentation, during clustering.

Primary analysis: compulsive and impulsive clustering of individuals with OCD

Using the PCA results from the above step, we clustered the clinical and OCD-likely participants by grouping individuals together with similar scores on our compulsive and impulsive phenotypes using k-means clustering.Reference Hastie, Tibshirani and Friedman 46 K-means clustering requires that the number of clusters drawn from the data be defined a priori. Given our primary aim to investigate the overlapping constructs of compulsivity and impulsivity, we extracted four clusters from the data. Participants within each k-means cluster should exhibit homogenous scores on the compulsive and impulsive phenotypes. To examine whether these groupings provide clinical utility in OCD, we compared the clusters' OCD severity using the OCI––R. See the supplementary materials for validation against different clustering methods and of the optimal number of clusters.

Secondary analysis: Compulsive and impulsive clustering of individuals with OCD and HCs

To capture a broad spectrum of OCD severity, we included OCD-likely participants in our primary analysis, which introduced the influence of subclinical OCD tendencies into our sample. To extend the spectrum of severity even further toward normality,Reference Cuthbert and Insel 6 , Reference Cuthbert and Kozak 47 we repeated our PCA and clustering analyses including 35 HC participants and examined the proportions of HCs and OCDs (both clinical and likely) that fell into each of the four clusters extracted from the data using k-means clustering.

Results

Sample characteristics

OCD and OCD-likely individuals had significantly higher (p<0.001) OCI–R scores (M=31.1) than HC participants (M=4.82). The OCD, OCD-likely, and HC samples did not differ on age or gender (p>0.05). Pearson correlations between total OCI–R scores and the questionnaire measures can be found in Supplementary Table 1.

Compulsive and impulsive phenotypes of OCD

Our primary goal was to evaluate whether orthogonally related constructs of compulsivity and impulsivity, measured using the OBQ–44 and IUS–12, as well as the UPPS–P, respectively, could distinguish symptom severity in a sample of OCD and OCD-likely participants. PCA revealed two components that putatively represent an individual's level of compulsivity and impulsivity (Table 1). The internal consistency for the compulsivity and impulsivity components were α=0.76 and 0.70, respectively. Table 1 illustrates how the measures of compulsivity and impulsivity used loaded onto respective compulsive and impulsive components.

Table 1

Pattern Matrix from principal component analysis showing variable loadings on the Compulsive and Impulsive components for a sample of OCD participants (N=182). The coefficients are displayed in descending loading order on the first component.

Note. OCI-R=Obsessive-Compulsive Inventory-Revised; OBQ-44=Obsessive Beliefs Questionnaire-44; IUS-12=Intolerance of Uncertainty Scale-12; UPPS-R+P=UPPS-R+Positive Urgency Impulsive Behaviours Scale; Extraction Method: PCA. Rotation Method: Premix with Kaiser Normalization.

Compulsive and impulsive clustering of OCD

We employed unsupervised clustering methods to identify subgroups of OCD participants based upon their levels of compulsivity and impulsivity. When extracting four clusters using k-means, the results identify four groups of OCD participants that mapped onto our predicted compulsive/impulsive phenotype (Figure 1). Cluster 1 contained 30 participants scoring low on both compulsivity and impulsivity (hereafter referred to as low/low); cluster 2 had 41 participants scoring moderately on compulsivity and low on impulsivity (hereafter referred to as high/low); cluster 3 contained 71 participants low on compulsivity and high on impulsivity (hereafter referred to as low/high); and cluster 4 contained 40 participants high on both compulsivity and impulsivity (hereafter referred to as high/high). It is important to note that low and high scores on the PCs represents a score relative to the average of the sample. Thus, “low” on compulsivity (or impulsivity) may not represent qualitatively “normal” levels of compulsivity, but rather lower compared to other OCDs in this sample.

Figure 1

K-means clustering reveals four distinct groupings of OCD participants based on measures of compulsivity and impulsivity. Clustering yields a subgroup of OCD participants low on both compulsivity and impulsivity (Low/Low), a more classical OCD subgroup (High/Low), an impulsive subgroup (Low/High), and a subgroup high on both (High/High).

Assessing the robustness of our clustering solution using the OCD and OCD-likely groups, we also found that: (1) two-step cluster analysis confirmed the four-cluster solution as optimal, with the silhouette measure of cohesion and separation equal to 0.5Reference Zhang, Ramakrishnan and Livny 48 ; (2) a four-cluster solution was also supported by convergence between Ward's and k-means methods (Cramer's V=0.74, p<0.001); (3) cluster membership was not driven by whether participants were diagnosed with OCD or whether they were OCD-likely (Cramer's V=0.10, p=0.065); and (4) clusters did not differ in terms of gender (χ2=2.11; p=0.065), age (F(3, 178)=1.380, p=0.250), or depression (Cramer's V=9.185; p=0.105).

OCD severity across subgroups

There was a significant effect of cluster membership on OCD severity (F(3, 178)=19.91, p<0.001). Post-hoc t tests indicated that the high/high OCD cluster had higher scores on the OCI–R (M=39.23, SD=12.27) than all other clusters (ps<0.05; Figure 1), whereas the low/low OCD cluster had lower scores on the OCI–R than all the other clusters (M=21.60, SD=8.90, all ps<0.001). There was no difference between the high/low OCD cluster (M=32.88, SD=9.84) and the low/high OCD cluster (M=30.04, SD=9.48, p=0.937). Interestingly, apart from the obsessing subscale, which was significantly higher in the high/high OCD group compared to the other three clusters (p<0.008), there were no significant differences in OCI–R subtypes across the four clusters, suggesting that the clusters were not driven by classical OCD subtypes.

Figure 2

K-means clustering reveals four distinct groupings of HC and OCD participants based on measures of compulsivity and impulsivity. Clustering yields a subgroup of participants low on both compulsivity and impulsivity (Low/Low), a compulsive subgroup (High/Low), an impulsive subgroup (Low/High), and a subgroup high on both (High/High).

Compulsive and impulsive clustering of individuals with OCD and HCs

Having established that a self-reported compulsive/impulsive phenotype can be used to separate OCD participants to reveal significant differences in disorder severity, we wanted to explore what influence introducing a HC group had on the clustering solution. The above PCA and k-means analyses were repeated, this time including 35 HCs. This time the compulsivity and impulsivity components accounted for 58.2% of the overall variance. Again, a clear compulsive/impulsive phenotype emerged. Table 2 illustrates how specific compulsive and impulsive measures loaded onto the principal components.

Table 2

Pattern Matrix from principal component analysis showing variable loadings on the Compulsive and Impulsive components for a sample of OCD and HC participants (N=217). The coefficients are displayed in descending loading order on the first component.

Note. OCI-R=Obsessive-Compulsive Inventory-Revised; OBQ-44=Obsessive Beliefs Questionnaire-44; IUS-12=Intolerance of Uncertainty Scale-12; UPPS-R+P=UPPS-R+Positive Urgency Impulsive Behaviours Scale; Extraction Method: PCA. Rotation Method: Premix with Kaiser Normalization.

Results of k-means clustering revealed a qualitatively similar pattern of compulsive/impulsive clusters in the data (Figure 2). Specifically, the low/low, high/low, low/high, and high/high clusters were all recaptured when adding the HC participants. Notably, 83.3% of the healthy controls fell into the low/low cluster, making up approximately half of the participants in that cluster (Figure 3). The high/high cluster was made up exclusively of OCDs, and the high/low and low/high clusters were predominantly composed of OCDs. Again, as before, multiple clustering techniques converged on a four-cluster solution and clusters did not differ in terms of gender (χ2=2.78; p=0.42). There was an effect of age (F(3, 213)=2.75, p=0.043) but post-hoc analyses showed no age differences between any two clusters (all ps<0.524).

Figure 3

Proportions of HC and OCD subjects within each of the four subgroups from the k-means clustering solution presented in Figure 2. The majority of the HC participants are within the Low/Low cluster.

There was a significant effect of cluster membership on OCD severity (F(3, 213)=55, p<0.001). Post-hoc t tests indicated that the high/high cluster showed the highest OCI–R score (M=38.39, SD=11.74) compared to the other groups (all ps<0.001). Again, the low/low cluster recorded the lowest OCI–R scores (M=11.51, SD=11.30) (all ps<0.001). There was no difference between the high/low OCD cluster (M=28.26, SD=11.50) and the low/high OCD cluster (M=29.59, SD=9.72, p=0.45).

Discussion

Using unsupervised clustering methods applied to questionnaire data, we grouped participants with OCD into four distinct subgroups based on their levels of impulsivity and compulsivity. We expected that the constructs of impulsivity and compulsivity would relate orthogonally to one another. Accordingly, OCD participants were delineated on the basis of high levels of interindividual compulsivity (i.e., high/low), impulsivity (low/high), or both (high/high), as well as low levels of both (low/low). These phenotypic differences in OCD participants were also linked to differences in OCD severity.

Traditionally, OCD has been viewed as the archetypal compulsive disorder. Contemporary views have argued that OCD may also encompass issues with impulsivity and may therefore be alternatively conceptualized as a behavioral addiction.Reference Grassi, Pallanti and Righi 10 , Reference Fontenelle, Oostermeijer, Harrison, Pantelis and Yücel 11 , Reference Fineberg, Potenza and Chamberlain 49 However, this hypothesis remains controversial,Reference Abramovitch and McKay 50 with some researchers reporting higher impulsivity in OCDs relative to controls,Reference Grassi, Pallanti and Righi 10 , Reference Kashyap, Fontenelle and Miguel 24 while others report no differenceReference Tavares and Gentil 13 or lower impulsivity in OCD.Reference Stein, Hollander, Simeon and Cohen 51 Rather than argue for one driver over another, our results suggest that both impulsivity and compulsivity mediate obsessive-compulsive symptomatology. Within our sample of OCD participants, there are subsets of individuals characterized by elevated levels of impulsivity with or without elevated compulsivity. Notably, the largest cluster of participants in our results were characterized by high impulsivity and low compulsivity relative to the sample mean. This suggests that a “one-size-fits-all” approach to understanding OCD phenomenology (e.g., OCD as an anxiety-avoidance disorder vs. a behavioral addiction) is likely to be overly simplistic.

Fontenelle et al.Reference Fontenelle, Oostermeijer, Harrison, Pantelis and Yücel 11 postulated that OCD behaviors may become more “impulsive” in nature with chronicity, suggesting impulsivity as a potential endpoint for some OCD patients. While chronicity has been linked to increased severity of OCD,Reference Visser, van Oppen, van Megen, Eikelenboom and van Balkom 52 our results suggest that the relationship with impulsivity is not so clear. We found that OCD severity differed across our clusters, with the low/low cluster showing the lowest OCD severity and the high/high cluster showing the highest OCD severity, suggesting that an interaction between the phenotypes may be linked to worse prognosis. However, we also found that, while OCD severity was higher for the high/low and the low/high clusters relative to the low/low cluster, severity did not differ between the two. A more comprehensive understanding of the role of impulsivity in relation to OCD severity requires longitudinal data not present in the current study. Nevertheless, our results point toward the clinical importance of understanding an OCD patient's impulsivity and demonstrate how traditional diagnostic categories can mask the underlying motivations driving OCD behaviors.

Our finding that OCD severity links to a compulsive/impulsive phenotype may have potential implications for the management of OCD treatment in clinical practice. For instance, specific psychotherapeutic strategies to manage impulsivity may be added to exposure and response prevention of OCD patients. We have previously suggested that medication targeting impulsive behaviors (such as naltrexone, topiramate, and methylphenidate, among others) may be of theoretical help to some OCD patients.Reference Fontenelle, Oostermeijer, Harrison, Pantelis and Yücel 11 However, there is mixed data in relation to the efficacy of these drugs in OCD, with some reports of improvement,Reference Sandyk 53 , Reference Berlin, Koran and Jenike 54 no change,Reference Amiaz, Fostick, Gershon and Zohar 55 or even worsening of OCD symptoms.Reference Insel and Pickar 56 These heterogeneous outcomes may be ascribed to specific patient profiles and assessment of baseline impulsive features (e.g., severity of compulsive symptomsReference Berlin, Koran and Jenike 54 ) and may help us to determine the phenotype that would be best targeted by an anti-impulsive medication in OCD.

Limitations

One limitation of this study is the inclusion of only self-report measures of compulsivity and impulsivity. Compulsivity and impulsivity are both heterogeneous constructsReference Robbins, Gillan, Smith, de Wit and Ersche 9 , Reference Grant and Kim 57 and are unlikely to be fully covered by self-report alone. Thus, future work should investigate whether our findings replicate when including behavioral tasks either instead of self-report measures or alongside them. However, our results demonstrate that self-report measures of the compulsivity and impulsivity phenotypes are sensitive enough to distinguish clinically relevant differences in OCD subgroups. This has strong clinical utility given the convenient and cost-effective nature of self-report data compared to behavioral tasks, which often require training and practice to be administered successfully. Finally, the fact that approximately half of the low/low cluster was composed of OCD participants when including healthy control participants may be ascribed to the fact that our OCD sample includes patients who were in different stages of treatment.

Conclusions

In summary, we used objective data-driven methods applied to self-report data to show that individuals with OCD can be sub-grouped based on their levels of compulsivity (OBQ–44 and IUS–12) and impulsivity (UPPS–P), which in turn links to differences in clinical severity. We found that individuals with high levels of impulsivity (and relatively lower compulsivity) have comparable clinical severity to those with high compulsivity (i.e., the stereotypical OCD presentation). Furthermore, we showed that the combination of high levels of compulsivity and impulsivity is linked to the highest clinical severity.

Conflicts of Interest

The authors hereby declare that they have no financial or other relationships relevant to the subject of this article.

Disclosures

Luisa Prochazkova, Rebecca A. Segrave, and Murat Yücel are supported by the David Winston Turner Endowment Fund. Murat Yücel is also supported by an National Health and Medical Research Council senior research fellowship (grant no. APP1117188). Linden Parkes and Andrew Dawson are supported by Australian postgraduate awards. Leonardo F. Fontenelle is supported by the D'Or Institute for Research and Education, the Conselho Nacional de Desenvolvimento Científico e Tecnológico (#308237/2014-5), and the Fundação de Amparo à Pesquisa do Estado do Rio de Janeiro (#203590). George Youssef, Gabriela Ferreira, and Valentina Lorenzetti have nothing to disclose.

Supplementary Material

To view supplementary materials for this article, please visit https:/doi.org/10.1017/S1092852917000244

Footnotes

†

These authors contributed equally.

References

1. Fontenelle, LF, Mendlowicz, MV, Versiani, M. The descriptive epidemiology of obsessive-compulsive disorder. Prog Neuropsychopharmacol Biol Psychiatry. 2006; 30(3): 327–337; Epub ahead of print Jan 18.Google Scholar
2. Hollander, E, Doernberg, E, Shavitt, R, et al. The cost and impact of compulsivity: a research perspective. Eur Neuropsychopharmacol. 2016; 26(5): 800–809; Epub ahead of print Feb 10.CrossRefGoogle ScholarPubMed
3. Fineberg, NA, Brown, A, Reghunandanan, S, Pampaloni, I. Evidence-based pharmacotherapy of obsessive-compulsive disorder. Int J Neuropsychopharmacol. 2012; 15(8): 1173–1191; Epub ahead of print Jan 9. Available at: https://academic.oup.com/ijnp/article-lookup/doi/10.1017/S1461145711001829.CrossRefGoogle Scholar
4. Pauls, DL, Abramovitch, A, Rauch, SL, Geller, DA. Obsessive-compulsive disorder: an integrative genetic and neurobiological perspective. Nat Rev Neurosci. 2014; 15(6): 410–424.CrossRefGoogle ScholarPubMed
5. Gottesman, II, Gould, TD. The endophenotype concept in psychiatry: etymology and strategic intentions. Am J Psychiatry. 2003; 160(4): 636–645; Available at: http://ajp.psychiatryonline.org/doi/pdf/10.1176/appi.ajp.160.4.636.CrossRefGoogle ScholarPubMed
6. Cuthbert, BN, Insel, TR. Toward the future of psychiatric diagnosis: the seven pillars of RDoC. BMC Med. 2013; 11(1): 126 Available at: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3653747/pdf/1741-7015-11-126.pdf.Google Scholar
7. Fineberg, NA, Chamberlain, SR, Goudriaan, AE, et al. New developments in human neurocognition: clinical, genetic, and brain imaging correlates of impulsivity and compulsivity. CNS Spectr. 2014; 19(1): 69–89; Available at: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4113335/pdf/nihms605414.pdf.Google Scholar
8. Gillan, CM, Kosinski, M, Whelan, R, Phelps, EA, Daw, ND. Characterizing a psychiatric symptom dimension related to deficits in goal-directed control. Elife. 2016; 5: 1–24; Available at: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4786435/.Google Scholar
9. Robbins, TW, Gillan, CM, Smith, DG, de Wit, S, Ersche, KD. Neurocognitive endophenotypes of impulsivity and compulsivity: towards dimensional psychiatry. Trends Cogn Sci. 2012; 16(1): 81–91; Epub ahead of print Dec 10, 2011.Google Scholar
10. Grassi, G, Pallanti, S, Righi, L, et al. Think twice: impulsivity and decision making in obsessive-compulsive disorder. J Behav Addict. 2015; 4(4): 263–272; Available at: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4712760/.Google Scholar
11. Fontenelle, LF, Oostermeijer, S, Harrison, BJ, Pantelis, C, Yücel, M. Obsessive-compulsive disorder, impulse control disorders and drug addiction: common features and potential treatments. Drugs. 2011; 71(7): 827–840.Google Scholar
12. Chamberlain, SR, Leppink, EW, Redden, SA, Grant, JE. Are obsessive-compulsive symptoms impulsive, compulsive or both? Compr Psychiatry. 2016; 68: 111–118; Epub ahead of print Apr 15.CrossRefGoogle ScholarPubMed
13. Tavares, H, Gentil, V. Pathological gambling and obsessive-compulsive disorder: towards a spectrum of disorders of volition. Rev Bras Psiquiatr. 2007; 29(2): 107–117; Available at: http://www.scielo.br/pdf/rbp/v29n2/a05v29n2.pdf.Google Scholar
14. Chamberlain, SR, Leppink, EW, Redden, SA, Grant, JE. Are obsessive-compulsive symptoms impulsive, compulsive or both? Compr Psychiatry. 2016; 68: 111–118; Epub ahead of print Apr 15.Google Scholar
15. Benatti, B, Dell'Osso, B, Arici, C, Hollander, E, Altamura, AC. Characterizing impulsivity profile in patients with obsessive-compulsive disorder. Int J Psychiatry Clin Pract. 2014; 18(3): 156–160.CrossRefGoogle ScholarPubMed
16. Boisseau, CL, Thompson-Brenner, H, Caldwell-Harris, C, Pratt, E, Farchione, T, Barlow, DH. Behavioral and cognitive impulsivity in obsessive-compulsive disorder and eating disorders. Psychiatry Res. 2012; 200(2–3): 1062–1066.CrossRefGoogle ScholarPubMed
17. Ettelt, S, Ruhrmann, S, Barnow, S, et al. Impulsiveness in obsessive-compulsive disorder: results from a family study. Acta Psychiatr Scand. 2007; 115(1): 41–47.CrossRefGoogle ScholarPubMed
18. Sohn, SY, Kang, JI, Namkoong, K, Kim, SJ. Multidimensional measures of impulsivity in obsessive-compulsive disorder: cannot wait and stop. PLoS One. 2014; 9(11): e111739 Available at: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4221112/pdf/pone.0111739.pdf.CrossRefGoogle ScholarPubMed
19. Summerfeldt, LJ, Hood, K, Antony, MM, Richter, MA, Swinson, RP. Impulsivity in obsessive-compulsive disorder: comparisons with other anxiety disorders and within tic-related subgroups. Pers Individ Dif. 2004; 36(3): 539–553.CrossRefGoogle Scholar
20. Zermatten, A, Van der Linden, M. Impulsivity in non-clinical persons with obsessive-compulsive symptoms. Pers Individ Dif. 2008; 44(8): 1824–1830.Google Scholar
21. Cavedini, P, Riboldi, G, D'Annucci, A, Belotti, P, Cisima, M, Bellodi, L. Decision-making heterogeneity in obsessive-compulsive disorder: ventromedial prefrontal cortex function predicts different treatment outcomes. Neuropsychologia. 2002; 40(2): 205–211.Google Scholar
22. Cavedini, P, Zorzi, C, Piccinni, M, Cavallini, MC, Bellodi, L. Executive dysfunctions in obsessive-compulsive patients and unaffected relatives: searching for a new intermediate phenotype. Biol Psychiatry. 2010; 67(12): 1178–1184; Epub ahead of print Apr 8.Google Scholar
23. da Rocha, FF, Alvarenga, NB, Malloy-Diniz, L, Corrêa, H. Decision-making impairment in obsessive-compulsive disorder as measured by the Iowa Gambling Task. Arq Neuropsiquiatr. 2011; 69(4): 642–647; Available at: http://www.scielo.br/pdf/anp/v69n4/a13v69n4.pdf.CrossRefGoogle ScholarPubMed
24. Kashyap, H, Fontenelle, LF, Miguel, EC, et al. “Impulsive compulsivity” in obsessive-compulsive disorder: a phenotypic marker of patients with poor clinical outcome. J Psychiatr Res. 2012; 46(9): 1146–1152.Google Scholar
25. Kim, HW, Kang, JI, Namkoong, K, Jhung, K, Ha, RY, Kim, SJ. Further evidence of a dissociation between decision-making under ambiguity and decision-making under risk in obsessive-compulsive disorder. J Affect Disord. 2015; 176: 118–124.Google Scholar
26. Kodaira, M, Iwadare, Y, Ushijima, H, et al. Poor performance on the Iowa Gambling Task in children with obsessive-compulsive disorder. Ann Gen Psychiatry. 2012; 11(1): 1–7; Available at: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3508952/pdf/1744-859X-11-25.pdf.Google Scholar
27. Starcke, K, Tuschen-Caffier, B, Markowitsch, HJ, Brand, M. Dissociation of decisions in ambiguous and risky situations in obsessive-compulsive disorder. Psychiatry Res. 2010; 175(1–2): 114–120.Google Scholar
28. Zhang, L, Dong, Y, Ji, Y, et al. Trait-related decision making impairment in obsessive-compulsive disorder: evidence from decision making under ambiguity but not decision making under risk. Sci Rep. 2015; 5: 17312 Available at: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4658550/.Google Scholar
29. Fontenelle, LF, Oostermeijer, S, Ferreira, GM, Lorenzetti, V, Luigjes, J, Yücel, M. Anticipated reward in obsessive-compulsive disorder. J Clin Psychiatry. 2015; 76(9): e1134–e1135; Available at: http://www.psychiatrist.com/jcp/article/Pages/2015/v76n09/v76n0908.aspx.Google Scholar
30. Cavedini, P, Zorzi, C, Baraldi, C, et al. The somatic marker affecting decisional processes in obsessive-compulsive disorder. Cogn Neuropsychiatry. 2012; 17(2): 177–190.Google Scholar
31. Wright, L, Lipszyc, J, Dupuis, A, Thayapararajah, SW, Schachar, R. Response inhibition and psychopathology: a meta-analysis of go/no-go task performance. J Abnorm Psychol. 2014; 123(2): 429–439.Google Scholar
32. Lei, H, Zhu, X, Fan, J, et al. Is impaired response inhibition independent of symptom dimensions in obsessive-compulsive disorder? Evidence from ERPs. Sci Rep. 2015; 5: 10413 Available at: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4438428/.Google Scholar
33. Bannon, S, Gonsalvez, CJ, Croft, RJ, Boyce, PM. Response inhibition deficits in obsessive-compulsive disorder. Psychiatry Res. 2002; 110(2): 165–174.CrossRefGoogle ScholarPubMed
34. Chamberlain, SR, Fineberg, NA, Blackwell, AD, Robbins, TW, Sahakian, BJ. Motor inhibition and cognitive flexibility in obsessive-compulsive disorder and trichotillomania. Am J Psychiatry. 2006; 163(7): 1282–1284.CrossRefGoogle ScholarPubMed
35. Figee, M, van den Munckhof, P, Schuurman, R, Denys, D. Neuroimaging deep brain stimulation in psychiatric disorders. In Denys D et al eds. Deep Brain Stimulation: A New Frontier in Psychiatry. New York: Springer Science; 2012: 225–236.CrossRefGoogle Scholar
36. Basar, K, Sesia, T, Groenewegen, H, Steinbusch, HWM, Visser-Vandewalle, V, Temel, Y. Nucleus accumbens and impulsivity. Prog Neurobiol. 2010; 92(4): 533–557.CrossRefGoogle ScholarPubMed
37. Knutson, B, Adams, CM, Fong, GW, Hommer, D. Anticipation of increasing monetary reward selectively recruits nucleus accumbens. J Neurosci. 2001; 21(16): 1–5; Available at: http://www.jneurosci.org/content/jneuro/21/16/RC159.full.pdf.Google Scholar
38. Foa, EB, Huppert, JD, Leiberg, S, et al. The Obsessive-Compulsive Inventory: development and validation of a short version. Psychol Assess. 2002; 14(4): 485–495.Google Scholar
39. Carleton, RN, Norton, MA, Asmundson, GJ. Fearing the unknown: a short version of the Intolerance of Uncertainty Scale. J Anxiety Disord. 2007; 21(1): 105–117; Epub ahead of print May 2, 2006.Google Scholar
40. Obsessive-Compulsive Cognitions Working Group. Psychometric validation of the obsessive belief questionnaire and interpretation of intrusions inventory, part 2: factor analyses and testing of a brief version. Behav Res Ther. 2005; 43(11): 1527–1542.Google Scholar
41. Tolin, DF, Abramowitz, JS, Brigidi, BD, Foa, EB. Intolerance of uncertainty in obsessive-compulsive disorder. J Anxiety Disord. 2003; 17(2): 233–242.Google Scholar
42. Gentes, EL, Ruscio, AM. A meta-analysis of the relation of intolerance of uncertainty to symptoms of generalized anxiety disorder, major depressive disorder, and obsessive-compulsive disorder. Clin Psychol Rev. 2011; 31(6): 923–933.Google Scholar
43. Cyders, MA, Smith, GT, Spillane, NS, Fischer, S, Annus, AM, Peterson, C. Integration of impulsivity and positive mood to predict risky behavior: development and validation of a measure of positive urgency. Psychol Assess. 2007; 19(1): 107–118.Google Scholar
44. Whiteside, SP, Lynam, DR, Miller, JD, Reynolds, SK. Validation of the UPPS impulsive behaviour scale: a four-factor model of impulsivity. Eur J Pers. 2005; 19(7): 559–574.CrossRefGoogle Scholar
45. Beck, AT, Steer, RA, Brown, GK. Manual for the Beck Depression Inventory–II. San Antonio, TX: Psychological Corporation; 1996.Google Scholar
46. Hastie, T, Tibshirani, R, Friedman, J. The Elements of Statistical Learning: Data Mining, Inference, and Prediction Springer Series in Statistics New York: Springer Science; 2009. Available at: http://statweb.stanford.edu/~tibs/ElemStatLearn/printings/ESLII_print10.pdf.Google Scholar
47. Cuthbert, BN, Kozak, MJ. Constructing constructs for psychopathology: the NIMH research domain criteria. J Abnorm Psychol. 2013; 122(3): 928–937.Google Scholar
48. Zhang, T, Ramakrishnan, R, Livny, M. BIRCH: an efficient data clustering method for very large databases. In: Proceedings of the 1996 ACM SIGMOD International Conference on Management of Data. New York: Association for Computing Machinery; 1996: 103–114. Available at: https://www.cs.sfu.ca/CourseCentral/459/han/papers/zhang96.pdf.Google Scholar
49. Fineberg, NA, Potenza, MN, Chamberlain, SR, et al. Probing compulsive and impulsive behaviors, from animal models to endophenotypes: a narrative review. Neuropsychopharmacology. 2010; 35(3): 591–604; Epub ahead of print Nov 25, 2009. Available at: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3055606/pdf/npp2009185a.pdf.Google Scholar
50. Abramovitch, A, McKay, D. Behavioral impulsivity in obsessive-compulsive disorder. J Behav Addict. 2016; 5(3): 395–387; Epub ahead of print May 9. Available at: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5264405/.Google Scholar
51. Stein, DJ, Hollander, E, Simeon, D, Cohen, L. Impulsivity scores in patients with obsessive-compulsive disorder. J Nerv Ment Dis. 1994; 182(4): 240–241.CrossRefGoogle ScholarPubMed
52. Visser, HA, van Oppen, P, van Megen, HJ, Eikelenboom, M, van Balkom, AJ. Obsessive-compulsive disorder; chronic versus non-chronic symptoms. J Affect Disord. 2014; 152–154(1): 169–174; Epub ahead of print Sep 13, 2013.Google Scholar
53. Sandyk, R. Naloxone abolishes obsessive-compulsive behavior in Tourette's syndrome. Int J Neurosci. 1987; 35(1–2): 93–94.Google Scholar
54. Berlin, HA, Koran, LM, Jenike, MA, et al. Double-blind, placebo-controlled trial of topiramate augmentation in treatment-resistant obsessive-compulsive disorder. J Clin Psychiatry. 2011; 72(5): 716–721; Epub ahead of print Aug 10, 2010.Google Scholar
55. Amiaz, R, Fostick, L, Gershon, A, Zohar, J. Naltrexone augmentation in OCD: a double-blind placebo-controlled cross-over study. Eur Neuropsychopharmacol. 2008; 18(6): 455–461; Epub ahead of print Mar 18.Google Scholar
56. Insel, TR, Pickar, D. Naloxone administration in obsessive-compulsive disorder: report of two cases. Am J Psychiatry. 1983; 140(9): 1219–1220.Google Scholar
57. Grant, JE, Kim, SW. Brain circuitry of compulsivity and impulsivity. CNS Spectr . 2014; 19(1): 21–27; Epub ahead of print May 10, 2013.Google Scholar
Figure 0

Table 1 Pattern Matrix from principal component analysis showing variable loadings on the Compulsive and Impulsive components for a sample of OCD participants (N=182). The coefficients are displayed in descending loading order on the first component.

Figure 1

Figure 1 K-means clustering reveals four distinct groupings of OCD participants based on measures of compulsivity and impulsivity. Clustering yields a subgroup of OCD participants low on both compulsivity and impulsivity (Low/Low), a more classical OCD subgroup (High/Low), an impulsive subgroup (Low/High), and a subgroup high on both (High/High).

Figure 2

Figure 2 K-means clustering reveals four distinct groupings of HC and OCD participants based on measures of compulsivity and impulsivity. Clustering yields a subgroup of participants low on both compulsivity and impulsivity (Low/Low), a compulsive subgroup (High/Low), an impulsive subgroup (Low/High), and a subgroup high on both (High/High).

Figure 3

Table 2 Pattern Matrix from principal component analysis showing variable loadings on the Compulsive and Impulsive components for a sample of OCD and HC participants (N=217). The coefficients are displayed in descending loading order on the first component.

Figure 4

Figure 3 Proportions of HC and OCD subjects within each of the four subgroups from the k-means clustering solution presented in Figure 2. The majority of the HC participants are within the Low/Low cluster.

Supplementary material: File

Prochazkova supplementary material

Prochazkova supplementary material

Download Prochazkova supplementary material(File)
File 76.8 KB