Towards a definitive symptom structure of obsessive-compulsive disorder: a factor and network analysis of 87 distinct symptoms in 1366 individuals

Abstract

Background: The symptoms of obsessive-compulsive disorder (OCD) are highly heterogeneous and it is unclear what is the optimal way to conceptualize this heterogeneity. This study aimed to establish a comprehensive symptom structure model of OCD across the lifespan using factor and network analytic techniques.

Methods: A large multinational cohort of well-characterized children, adolescents, and adults diagnosed with OCD (N = 1366) participated in the study. All completed the Dimensional Yale-Brown Obsessive-Compulsive Scale, which contains an expanded checklist of 87 distinct OCD symptoms. Exploratory and confirmatory factor analysis were used to outline empirically supported symptom dimensions, and interconnections among the resulting dimensions were established using network analysis. Associations between dimensions and sociodemographic and clinical variables were explored using structural equation modeling (SEM).

Results: Thirteen first-order symptom dimensions emerged that could be parsimoniously reduced to eight broad dimensions, which were valid across the lifespan: Disturbing Thoughts, Incompleteness, Contamination, Hoarding, Transformation, Body Focus, Superstition, and Loss/Separation. A general OCD factor could be included in the final factor model without a significant decline in model fit according to most fit indices. Network analysis showed that Incompleteness and Disturbing Thoughts were most central (i.e. had most unique interconnections with other dimensions). SEM showed that the eight broad dimensions were differentially related to sociodemographic and clinical variables.

Conclusions: Future research will need to establish if this expanded hierarchical and multidimensional model can help improve our understanding of the etiology, neurobiology and treatment of OCD.

Keywords: Obsessive–compulsive disorder; factor analysis; heterogeneity; network analysis; symptom dimensions.

Fig. 1.

Latent factor model representing empirically derived symptom dimensions of OCD. Notes. Item-level data (i.e. indicators) are not shown. Dashed lines indicate which parameter that was fixed in model identification. OCD, obsessive−compulsive disorder; NJR, not just right.

Fig. 2.

Network model and centrality for empirically derived symptom dimensions of obsessive−compulsive disorder. Notes. In the network, symptom dimensions are represented by nodes (circles) and the unique inter-relationship between each symptom dimension pair is depicted as an edge (line). Blue edges indicate positive interconnections. Red edges indicate negative interconnections. For the black and white version of this figure, solid edges indicate positive associations and dashed edges indicate negative associations. Wider and more saturated edges indicate stronger interconnections. Centrality (expected influence) is a numeric estimate for the positive interconnectedness of a specific node; higher values indicate a higher degree of overall interconnectedness. Z-standardized centrality values are presented.