Understanding the Heterogeneity of OCD

Abstract

Back to table of contents Previous article Next article Letter to the EditorFull AccessUnderstanding the Heterogeneity of OCDJOHN E. CALAMARI, Ph.D., JOHN E. CALAMARISearch for more papers by this author, Ph.D., North Chicago, Ill.Published Online:1 Nov 2005https://doi.org/10.1176/appi.ajp.162.11.2193AboutSectionsPDF/EPUB ToolsAdd to favoritesDownload CitationsTrack Citations ShareShare onFacebookTwitterLinked InEmail To the Editor: I read with interest the excellent review of dimensional approaches to understanding obsessive-compulsive disorder (OCD) heterogeneity by David Mataix-Cols, Ph.D., and colleagues (1). I agree with the authors that OCD heterogeneity is an important issue and that failure to identify differences within the condition has significantly hindered advances in theory and treatment. My comments focus on the authors’ contention that a dimensional approach to understanding OCD heterogeneity is an inherently superior method.There have been three recent approaches to understanding OCD symptom heterogeneity. Some researchers have focused on patients’ dominant compulsive behavior to form symptom subgroups (e.g., washers versus checkers). This approach is limited and fails to capture most cases in which patients are seen with multiple classes of symptoms. In recent investigations, the diversity and complex patterns of symptoms seen in clinical presentations have been characterized with multivariate statistical analyses. Factor analysis has been used to identify the latent dimensions of several comprehensive OCD symptom measures. Alternatively, symptom measures have been subjected to cluster analysis to form symptom-based subgroups of individuals. In cluster analysis, individuals are assigned to groups created by maximizing between-group differences and minimizing within-group variability on a set of measures (2).Cluster analysis may offer several advantages over factor analysis in characterizing OCD heterogeneity, and this categorical approach is not limited in some of the ways Dr. Mataix-Cols et al. implied. In cluster analysis, individuals are unambiguously assigned to unique groups, whereas in factor analysis, each individual is assigned a score on all of the identified latent dimensions. Thus, the factor scores estimated for individuals may not connect the person to a specific dimension. As Dr. Mataix-Cols et al. pointed out, hoarding symptoms have emerged as a symptom dimension that predicts unresponsiveness to current pharmacotherapy and standard behavior therapy. Although there has been limited study, similar results have been reported with a cluster analysis approach in which the hoarding subgroup was less responsive to behavior treatment (3). The results of several recent cluster analyses (e.g., reference 4) suggest that complex symptom presentations can be captured with a cluster analysis approach and that resultant clusters are far from monosymptomatic.The relative merits of categorical and dimensional approaches to psychiatric classification have long been debated. The use of each of these approaches to understanding OCD heterogeneity warrants further investigation.