A critical appraisal of heterogeneity in Obsessive-Compulsive Disorder using symptom-based clustering analysis

Obsessive-compulsive disorder (OCD) encompasses a broad range of symptoms and is commonly considered a heterogeneous condition. Attempts were made to define discrete OCD subtypes using a range of symptom-based methods including factor and cluster analyses. The present study aims to find the most appropriate clustering model based on Yale-Brown obsessive-compulsive scale (YBOCS) checklist explaining OCD heterogeneity. Five different clustering algorithms (FCM, K-means, Ward, Ward+K-means and Complete) applied on YBOCS symptoms of 216 patients with OCD. Data studied as four different sets including item-level raw data, item-based factor scores, category-level raw data and category-based factor scores and clustering results for 2 to 6 cluster solutions evaluated by four clustering indices (Davies-Bouldin, Calinski-Harabasz, Silhouettes and Dunn indices). Two-cluster solution was detected as the most appropriate model for item and category-based clustering analyses of YBOCS checklist symptoms. Patients in each cluster were characterized based on their clinical and demographic properties and results showed that they had similar patterns of symptoms but in different severities. Heterogenity of OCD based on the YBOCS-symptoms has been challenged as OCD patients were classified based on their symptom severity not their symptom patterns. More investigations need to find appropriate measures explaining OCD heterogeneity with clinical importance.