Dr. Mataix-Cols and Colleagues Reply

To the Editor: We were pleased to read Dr. Calamari’s letter in relation to our recently published review. It highlights methodological and conceptual issues that are unlikely to be easily resolved. Rather than diametrically opposite techniques, factor analysis and categorical approaches, such as cluster analysis, are likely to be complementary because they constitute different ways of looking at the same phenomenon—the heterogeneity of OCD (1). Both have demonstrated their usefulness. For example, tic-related OCD and early-onset OCD both appear to be overlapping and valid subtypes (2). Our preference for factor analytical techniques to address the classic symptoms of OCD is twofold.

First, our model hypothesizes that obsessive-compulsive phenomena are normally distributed in the general population (3, 4) and are not limited to the traditional diagnostic boundaries of OCD, i.e., they may be present in many other neurological and psychiatric conditions. Conceptually, a dimensional approach seems to reflect this more accurately. Second, if one adopts a strictly categorical approach, patients need to be unequivocally allocated to only one subtype: a patient is either in cluster X or in cluster Y but not both. We doubt that nature is so exact regarding these symptoms. This is one of the main limitations of the DSM-IV multiaxial system and has been heavily criticized. Along with other theoreticians (5, 6), we propose that a dimensional approach can better deal with the problem of comorbidity or the coexistence of various symptom types in OCD. In short, we reiterate the idea that different methods of analysis are probably likely to yield complementary results. We are glad that Dr. Calamari concurs that considering the heterogeneity of OCD is the direction to take in this important area of research.