13a

Epidemiological, Factor-Analytic, and Cognitive Factors in the Position of Obsessive-Compulsive Personality Disorder among the Cluster C Personality Disorders: Commentary on Cluster C Anxious-Fearful Personality Pathology and Avoidance

Arnoud Arntz

The chapter by Sanislow and Hector gives an excellent overview of the history of the Cluster C personality disorders (PDs). Currently, three PDs are assumed to be part of this cluster. However, one of them, Obsessive-Compulsive PD (OCPD), seems to deviate from the others in a range of areas. In support of this assertion, I will discuss research varying from epidemiological to information processing studies. Following Sanislow and Hector’s suggestion to go beyond variants of the DSM structure by using transdiagnostic constructs, this commentary addresses three key points including: (1) contributions of epidemiology research with a focus on differences between obsessive-compulsive personality disorder (OCPD) and other PDs; (2) factor-analytic evidence for the DSM cluster structure of PDs; and (3) the role of cognitive processes and structures.

Epidemiology: The Strange Position of OCPD Compared to Other PDs

Epidemiological research has documented peculiar characteristics of OCPD. First, whereas in the general population most if not all PDs are associated with indices of poorer social-economic functioning, this appears to be not the case for OCPD. On the contrary, in the general population OCPD is associated with higher educational level, higher income, socioeconomic status, and is not associated with unfortunate living situations (Torgersen, 2012). Moreover, in contrast to the data supporting the point made by Sanislow and Hector that OCPD is characterized by severe interpersonal problems such as a lack of empathy, there is no evidence at all that OCPD is associated with poorer marital functioning, more divorce, or more living without a partner (Torgersen, 2012). Third, a similar picture emerges across studies with respect to quality of life and level of functioning, with no association between OCPD and these variables in the general population. Fourth, whereas prevalences in the general population (in Western countries) are about 2.5 percent for Avoidant PD (AVPD), 1 percent for Dependent PD (DEPD) and 2 percent for OCPD, the prevalences in clinical samples show a very different picture: about 25 percent for AVPD and 15 percent for DEPD, but only 10 percent for OCPD. Thus, whereas the clinical vs. general prevalence ratios are 10:1 and 15:1 for AVPD and DEPD, the ratio is only 5:1 for OCPD (data from Torgersen, 2012). Thus, unexpectedly small numbers of people with OCPD seek help in mental healthcare, again indicating the relatively good functioning of many people with OCPD.

These findings question whether OCPD as defined by the DSM criteria is a disorder at all. Indeed, many of the OCPD traits as currently defined, and associated characteristics, match well with the current values of Western societies, such as putting ratio above emotion or social considerations, perfectionism, productivity, control, hard working, and taking high responsibility (e.g., Pfohl & Blum, 1995). This suggests that there is not so much deviation of OCPD traits from cultural norms. Interestingly, one usually sees OCPD patients after they decompensated because of working too hard, e.g., with an initial clinical presentation of depression or burnout. Alternatively, epidemiological studies among the general population might have overestimated the prevalence of OCPD, e.g., by not validly assessing whether the traits cause dysfunction –although this argument would hold for all PDs.

Following from questions about the role and level of functional impairment in OCPD, we next move to discuss the ways in which OCPD also has a status aparte from the other PDs at the level of cluster structure.

Factor-Analytic Evidence for the DSM Cluster Structure of PDs

How does OCPD fit within the cluster structure that the DSM hypothesizes? To answer this, a second order principal component analysis (PCA) was done on a SCID-II trait score data set of n = 2165 patients and non-patients (34.3 percent men; M age = 33.04, range 16–66, SD = 11.05). This data set is an extension of one that was previously used for taxometric analyses (Arntz et al., 2009). In that study good evidence was found for a factor structure following the DSM-IV classification of PDs, and for a dimensional rather than a categorical nature of the PD traits of six PDs (including the three Cluster C PDs).

For the present PCA, 1–2–3 trait scores were summed per PD to get a dimensional score per PD, and a first PCA was run to inspect the scree plot to decide on how many components to extract. Next, a PCA was run with this number of components followed with oblimin rotation, thus allowing for correlation between components. For the first analysis, the Depressive and the Negativistic (= Passive Aggressive) PDs were excluded, as not being fully acknowledged in the DSM-IV.

There were three components with eigenvalue > 1 and the scree plot indicated three components. These explained 58 percent of the variance. The factor loadings after oblimin rotations are given in Table 13.a.1. As can be seen, the first component reflects Cluster A, the second Cluster C, and the third Cluster B. There are three additional cross-loadings, though they are all ≤ .35: the Paranoid scale also loads on components 2 and 3, thus shares variance with clusters B and C; and the Borderline scale loads on component 2, i.e., shares variance with Cluster C. Nevertheless, grosso modo the clusters are quite convincingly recovered in the data. Table 13.a.2 presents the correlations between the components, which are very modest, giving further evidence for relatively independent clusters.

Table 13.a.1Factor loadings of SCID-II trait scores per PD on components after oblimin rotation

Note: Loadings < .25 are not presented. Loadings > .40 printed bold.

Table 13.a.2Component correlation matrix

When Depressive and Negativistic PD scales were added, results were very similar: again three components appeared, corresponding with the three PD-clusters. Depressive PD loaded solely on the Cluster C factor (.75), Negativistic PD on both Cluster B (.51) and C (.32).

The three clusters hypothesized by the DSM model of PDs were well recovered in the data as second order factors. The three cross-loadings were rather small in size, and fit with clinical observations and research (e.g., Arntz, Weertman, & Salet, 2011) that Borderline PD shows phenomena that overlap with Cluster C, such high fear, guilt and shame feelings, feelings of inferiority, and rejection sensitivity, and that many patients with Paranoid PD are less “psychotic” than the other Cluster A PDs, and share features with Cluster B and Cluster C patients – for example, their treatment response to Cluster B and C based treatment models is quite good (e.g., Bamelis, Evers, Spinhoven, & Arntz, 2014).

Replicating previous studies discussed by Sanislow and Hector, the results also show that Negativistic (Passive-Aggressive) PD loaded on both Cluster B and C, notably even stronger on Cluster B. Thus, passive-aggressive traits might be more characteristic of Cluster B PD pathology than of Cluster C, perhaps related to the relatively strong presence of anger in the criteria (Sanislow and Hector, this volume). Another remarkable finding is that Obsessive-Compulsive PD (OCPD) does not load highly on the second order Cluster C factor. Unlike Paranoid PD, which also shows a limited loading on its hypothesized cluster, there are no cross-loadings that help us to explain. This suggests that OCPD differs from other PDs in Cluster C in ways that cannot be explained by characteristics that belong to Cluster A or B. Indeed, it has been questioned before whether a fourth general factor underlies OCPD (de Reus & Emmelkamp, 2012). I will now address how OCPD has its own characteristics from the point of view of cognitive models of PDs.

Cognitive Processes and Structures

Cognitive models of PDs hypothesize that schemas that steer information processing underlie the PD. At least part of schemas can be represented in beliefs that people may be aware of, which they then can evaluate as to how strongly they believe them. Indeed, studies indicated that different PDs are characterized by specific sets of beliefs as assessed with self-report, including all Cluster C PDs (Arntz, Dreessen, Schouten, & Weertman, 2004; Beck et al., 2001). An important information processing bias is interpretational bias. One study indeed found evidence for interpretational biases in Borderline, Avoidant, and Dependent PD, but not in OCPD, when assessing choice and believability of preformulated interpretations of ambiguous scenarios (Arntz et al., 2011). Moreover, the interpretations hypothesized to be specific for OCPD turned out to be very popular and highly believable among non-patients. This suggests that the way OCPD patients tend to view situations is not strongly deviating from what is normal in Western culture. Open responses to ambiguous scenarios were also investigated, and here OCPD showed increased responses expressing compulsiveness and worry, as well as reduced flexibility and acceptance, compared to non-patients. These categories point more to processing peculiarities in OCPD than to characteristic content of interpretations. Moreover, OCPD was not related to open responses related to avoidance and self-criticism, neither to guilt and fear of judgment; whereas the other Cluster C PDs showed elevated scores on these types of open responses. In sum, this interpretation study suggested that OCPD is not so much characterized by typical interpretational content, compared to non-patients, but rather by deviating cognitive styles of rigidity, compulsiveness, and worry.

However, this interpretation study can be criticized for at least two reasons. First, response options for healthy interpretations were missing in the closed format part, to prevent socially desirable answers. However, this might have led the non-patients to opt for the relatively least dysfunctional options, those hypothesized to be specific for OCPD. Second, interpretational processes may take place at an automatic level, and explicit assessment might therefore miss important biases. Indeed, a study into implicit cognitive biases in OCPD found evidence for biases that people are not necessarily aware of, with OCPD being associated with self-views of being responsible, conscientious, disciplined, etc., and views of other people as being irresponsible, lazy, undisciplined, etc. (Weertman, Arntz, de Jong, & Rinck, 2008). Interestingly, this tendency to view others in derogatory ways, and the self as superior in these areas is not what one would expect in a Cluster C PD. Is there further evidence that OCPD shows a cognitive profile that is atypical for people from the anxious-fearful cluster?

A study investigating profiles in schema modes of six PDs indeed indicates a very different constellation in OCPD than in the other Cluster C PDs (Bamelis et al., 2014). (Schema modes refer to a construct describing the momentary cognitive-emotional-behavioral state the person is in, resulting from the way a person copes with an activated schema.) For instance, whereas AVPD and DEPD show increased vulnerable child modes, reflecting an emotional vulnerable state that would be common in children when they are in emotional need, OCPD did not show any association with such modes (despite that theory assumes an association, and despite associations between OCPD and childhood emotional abuse, dysfunctional parenting experiences, etc. (e.g., de Reus & Emmelkamp, 2012; Lobbestael, Arntz, & Bernstein, 2010)). Moreover, whereas a punitive internalization of caregivers characterized AVPD and DEPD, it was a demanding internalization that characterized OCPD. A powerful use of an overcompensating style of coping with schema activation is expected to lead to successful keeping vulnerable child modes out of awareness. Indeed, whereas AVPD and DEPD were characterized by avoidant types of coping, OCPD was characterized by overcompensating. Interestingly, overcompensating to keep vulnerable schemas out of awareness is characteristic of Narcissistic, Histrionic, and Anti-Social PDs – that are usually viewed as quite distinct from Cluster C PDs.

In sum, studies of cognitive models of PDs have indicated quite a difference in cognitive processes and structures between OCPD and the other Cluster C PDs. Superficially, there seems to be little difference between OCPD and non-patients when one examines conscious interpretations, whereas there is a clear distinction between the other Cluster C PDs and non-patients. However, there appear to be distinctive styles in OCPD, characterized by lack of flexibility, problems with accepting problematic situations, compulsiveness and worry. When one examines implicit measures, OCPD is clearly different from non-patients, with superior views of the self and inferior views of others. Lastly, OCPD seems associated with keeping vulnerable feelings out of awareness by an overcompensating style of control and perfectionism, which is unique among Cluster C PDs.

Conclusions

Research into epidemiology, factor-analytic structures, and cognitive models all indicated that OCPD has a position that is different from the other Cluster C PDs. In the general population, there is little indication that OCPD is associated with (highly) problematic dysfunctioning. Given the general prevalence, the prevalence in clinical sample is, compared to other PDs, unexpectedly low.

Examining findings from factor-analytic studies, the association between OCPD and the second order Cluster C factor is rather weak. As a result, it has been proposed that a distinct latent trait not subsumed by the three clusters underlies OCPD. Considering the results of studies into cognitive models, we see again evidence for a relatively weak distinction between normality and OCPD when it comes to interpretations. However, a specific overcompensating style of dealing with vulnerable feelings might create a superficially similarity with “normal” Western beliefs and values. Nevertheless, the degree to which perfectionism, control, responsibility, conscientiousness and the like are used as compensating strategies is excessive and creates the risk of burnout and depression.

Building on the points raised, the somewhat narcissistic way in which OCPD is characterized by superior self- and derogatory other views is very different from what is commonly seen in the other Cluster C PDs. Thus, anxiety, fear, and avoidance seem to play a much less central role in OCPD than in AVPD and DEPD, whereas a more overcompensating style seems to be much more central.

However, other processes than discussed so far might contribute to a distinct position of OCPD. For instance, it has been proposed that OCPD develops from another compensating dynamic, using control and obsession for details to deal with cognitive disorganization caused by executive control deficits (Aycicegi-Dinn, Caldwell-Harris, & Dinn, 2009).

Clearly, further research into cognitive processes and structures will further advance our understanding of the differences and commonalities between OCPD and the other Cluster C PDs. That said, there might be a need for a refined formulation of OCPD traits so that a better distinction can be made between functional and dysfunctional forms of this personality dimension, thus allowing for a better understanding of its role as a Cluster C PD.

References

Arntz, A., Bernstein, D., Gielen, D., van Nieuwenhuyzen, M., Penders, K., Haslam, N., & Ruscio, J. (2009). Taxometric evidence for the dimensional structure of Cluster C, paranoid, and borderline personality disorders. Journal of Personality Disorders, 23(6), 606–628.

Arntz, A., Dreessen, L., Schouten, E., & Weertman, A. (2004). Beliefs in personality disorders: A test with the personality disorder belief questionnaire. Behaviour Research and Therapy, 42(10), 1215–1225.

Arntz, A., Weertman, A., & Salet, S. (2011). Interpretation bias in Cluster C and borderline personality disorders. Behaviour Research and Therapy, 49(8), 472–481.

Aycicegi-Dinn, A., Caldwell-Harris, C. L., & Dinn, W. M. (2009). Obsessive-compulsive personality traits: Compensatory response to executive function deficit? International Journal of Neuroscience, 119, 600–608.

Bamelis, L. L. M., Evers, S. M. A. A., Spinhoven, P., & Arntz, A. (2014). Results of a multicentered randomised controlled trial of the clinical effectiveness of schema therapy for personality disorders. American Journal of Psychiatry, 171, 305–322.

Beck, A. T., Butler, A. C., Brown, G. K., Dahlsgaard, K. K., Newman, C. F., & Beck, J. S. (2001). Dysfunctional beliefs discriminate personality disorders. Behaviour Research and Therapy, 39(10), 1213–1225.

de Reus, R. J. M. & Emmelkamp, P. M. G. (2012). Obsessive-compulsive personality disorder: A review of current empirical findings. Personality and Mental Health, 6(1), 1–21.

Lobbestael, J., Arntz, A., & Bernstein, D. P. (2010). Disentangling the relationship between different types of childhood maltreatment and personality disorders. Journal of Personality Disorders, 24(3), 285–295.

Pfohl, B., & Blum, N. (1995). Obsessive-compulsive personality disorder. In W. J. Livesly (Ed.), The DSM-IV Personality Disorders (pp. 261–276). New York: Guilford Press.

Torgersen, S. (2012). Epidemiology. In T. A. Widiger (Ed.), The Oxford Handbook of Personality Disorders (pp. 186–205). New York: Oxford University Press.

Weertman, A., Arntz, A., de Jong, P. J., & Rinck, M. (2008). Implicit self- and other-associations in obsessive-compulsive personality disorder traits. Cognition and Emotion, 22(7), 1253–1275.