Further development of YBOCS dimensions in the OCD Collaborative Genetics Study: Symptoms vs. categories
Introduction
Obsessive-compulsive disorder (OCD) is a chronic illness associated with substantial distress and functional impairment. Twin studies suggest OCD is heritable, with concordance rates higher in monozygotic (80–87%) than dizygotic twins (47–50%) (Carey and Gottesman, 1981). Family studies find relatives of cases affected substantially more often than relatives of controls (Pauls et al., 1995, Nestadt et al., 2000). Obsessions and compulsions are remarkably diverse and are expressed with great variability across patients with OCD (Rasmussen and Eisen, 1988). Phenotypic heterogeneity significantly complicates research on OCD pathophysiology and etiology, including the search for vulnerability genes.
Factor-analytic approaches have been used to derive useful phenotypic dimensions in other heterogeneous conditions, including schizophrenia (Liddle et al., 1992, Andreasen et al., 1995) and bipolar disorder (Cassidy et al., 1998). Similarly, in the last decade, at least 10 factor analyses (total N > 2000) have delineated clinically meaningful OCD symptom dimensions (see Mataix-Cols et al., 2005). While fairly similar in content, the number of factors reported ranges from three to six.
A major weakness of existing factor-analytically-derived OCD dimensions is that they depend on the a priori structure of the field's primary symptom measure, the Yale Brown Obsessive-Compulsive Scale Symptom Checklist (YBOCS-SC) (Goodman et al., 1989). This comprehensive list classifies obsessions and compulsions into 13 categories, based on the symptoms' overt similarities. Most factorial studies to date, including a recent report from this collaborative group (Hasler et al., 2007), have used composite scores on these a priori categories rather than individual symptoms (Mataix-Cols et al., 2005). However, category-level factor analyses are biased since they assume the validity of these symptom groupings (Summerfeldt et al., 1999), restrict the number of items available for analysis, and limit the symptom dimensions that can emerge (Denys et al., 2004). It is not surprising that an item-level confirmatory factor analysis (Summerfeldt et al., 1999) failed to empirically support the YBOCS-SC categories. Moreover, the published three-factor (Baer, 1994) and four-factor (Leckman et al., 1997) models, both derived from these categories, were not confirmed at the item-level. While adequate fit was found for the four-factor model at the second-order (or category) level, parameter estimates showed within-factor heterogeneity and overlap between factors (Summerfeldt et al., 1999). Despite these significant weaknesses, category-level factor analyses have been applied to the YBOCS-SC for practical reasons (e.g., fewer variables to manage; smaller samples required for power than in item-level analyses).
A comprehensive model of OCD symptom structure requires exploratory factor analysis of individual YBOCS-SC items to identify symptom clusters empirically. Three such item-level analyses have been reported (Hantouche and Lancrenon, 1996, Feinstein et al., 2003, Denys et al., 2004). All were limited by low subjects-to-items ratios (4–9 subjects per item) and the use of principal components analysis. When used with dichotomous (present versus absent), non-normally distributed data, as are obtained with the YBOCS-SC, traditional factor analysis procedures, like principal components analysis, can be misleading since commonly endorsed items cluster separately from less commonly endorsed items, even when all measure the same latent variable, yielding factors that reflect similarity in item prevalence rather than strength of association (Nunnally and Bernstein, 1994). An appropriate statistical alternative is dichotomous factor analysis (DFA) (Christoffersen, 1975), which explicitly accounts for the dichotomous, non-normal distribution of these data. Only one study (Cullen et al., 2007) has applied DFA to YBOCS-SC data but that attempt was at the category-level.
Despite prior findings of OCD's heritability, there have been few attempts to evaluate the familiality of OCD symptom dimensions. The first two such studies used the four-factor model developed by Leckman et al. (1997). Alsobrook et al. (1999) reported that the recurrence risk of OCD among first-degree relatives of probands with high scores on either the obsessions/checking or symmetry/ordering factors was much higher than the recurrence risk among relatives of probands with low scores on these factors. A segregation analysis by Leckman et al. (2003) yielded significant correlations between siblings and mother–child pairs for the same two factors in a sample of Tourette Syndrome affected sibpairs and relatives. Cullen et al. (2007) examined affected sibling similarity on four DFA-derived category-based factors in the Johns Hopkins OCD Family Study. Significant intrafamilial sib–sib correlations were found for the symmetry/ordering and hoarding factors despite limited power. Finally, in a category-based principal components analysis of affected sibling pairs from the OCD Collaborative Genetics Study (OCGS), Hasler et al. (2007) derived the same factor structure as in Leckman et al. (1997). All four factors showed statistically significant sib–sib correlations, with hoarding and obsessions/checking demonstrating the strongest familiality. Further and more consistent evidence of the familiality of symptom dimensions is needed to support the use of factor scores as quantitative traits in genetic linkage studies (Baer, 1994, Miguel et al., 2005).
The present study extends the work of Hasler et al. (2007) by re-analyzing YBOCS-SC data in the OCGS and applying the first-ever exploratory DFA of individual OCD symptoms. Results of this item-level analysis were compared to a refined category-level solution (see Section 2.3) from the same sample. We also examined affected sibling similarity on the derived symptom dimensions via intraclass sibling correlations as in Korszun et al. (2004). We expected the highest familial correlation for the hoarding factor based on clinical impressions, prior analyses by this group in the OCGS (Hasler et al., 2007), and findings from the Johns Hopkins OCD Family Study (Samuels et al., 2002, Cullen et al., 2007), a precursor to the OCGS.
Section snippets
Subjects
Data were collected between 2001 and 2004 as part of the OCGS (Samuels et al., 2006). The OCGS is a federally-funded collaboration among six sites in the United States: Johns Hopkins University (JHU) School of Medicine [lead site], Butler Hospital/Brown University, Columbia University, Massachusetts General Hospital/Harvard Medical School, National Institute of Mental Health (NIMH), and University of California—Los Angeles. The study targeted families with OCD-affected sibling pairs, and
Dichotomous factor analysis of YBOCS-SC items
Table 1 displays frequencies of the 44 individual YBOCS-SC items (lifetime). The most frequently endorsed items were concern with dirt or germs, checking locks or appliances, ordering/arranging compulsions, and rereading or rewriting. Inspection of the scree plot (see Fig. 1) generated by DFA indicated four to seven possible factors. These factor solutions were compared for optimal interpretability, pattern of loadings, and model fit. While the RMSEA for the four-factor model (0.052) did not
Discussion
The aim of the present study was to empirically derive a multidimensional model of OCD symptom structure by applying exploratory DFA to individual YBOCS-SC items, extending previous work by this collaborative group using YBOCS-SC a priori categories (Hasler et al., 2007). The new item-based model was then compared with a factor solution based on a traditional principal components analysis of YBOCS-SC categories, adjusted according to a method developed by Pinto et al. (2007). Data were
Acknowledgments
This research was supported by NIMH grants R01 MH50214, K23 MH064543, and K23 MH066284. The authors are grateful to Karen Bandeen-Roche, PhD, and Paul T. Costa, Jr., PhD, for their input on factor-analytic methods. The authors also thank the many families participating in the study; the Obsessive-Compulsive Foundation; and clinicians and coordinators at each site: Butler Hospital/Brown University (Maria Mancebo, PhD, Richard Marsland, RN, Shirley Yen, PhD); Columbia University (Renee Goodwin,
References (55)
- et al.
Obsessive–compulsive disorder subgroups: a symptom-based clustering approach
Behaviour Research and Therapy
(1999) - et al.
Obsessive–compulsive disorder subtypes: an attempted replication and extension of a symptom-based taxonomy
Behaviour Research and Therapy
(2004) - et al.
Use of factor analysis to detect potential phenotypes in obsessive–compulsive disorder
Psychiatry Research
(2004) - et al.
Familiality of factor analysis-derived YBOCS dimensions in OCD-affected sibling pairs from the OCD Collaborative Genetics Study
Biological Psychiatry
(2007) - et al.
Symptom patterns in children and adolescents with obsessive–compulsive disorder (OCD)
Behaviour Research and Therapy
(2006) - et al.
Schedule for Affective Disorders and Schizophrenia-Lifetime Version modified for the study of anxiety disorders (SADS-LA): rationale and conceptual development
Journal of Psychiatric Research
(1986) - et al.
The structure of childhood obsessions and compulsions: dimensions in an outpatient sample
Behaviour Research and Therapy
(2006) - et al.
Taboo thoughts and doubt/checking: a refinement of the factor structure for obsessive compulsive disorder symptoms
Psychiatry Research
(2007) - et al.
Hoarding in obsessive compulsive disorder: results from a case-control study
Behaviour Research and Therapy
(2002) - et al.
Paroxetine treatment of compulsive hoarding
Journal of Psychiatric Research
(2007)
Principal components analysis of obsessive–compulsive disorder symptoms in children and adolescents
Biological Psychiatry
Symptom structure in obsessive–compulsive disorder: a confirmatory factor-analytic study
Behaviour Research and Therapy
Genomewide scan of hoarding in sib pairs in which both sibs have Gilles de la Tourette syndrome
American Journal of Human Genetics
Long-term follow-up and predictors of clinical outcome in obsessive–compulsive patients treated with serotonin reuptake inhibitors and behavioral therapy
Journal of Clinical Psychiatry
Segregation analysis of obsessive–compulsive disorder using symptom-based factor scores
American Journal of Medical Genetics
Symptoms of schizophrenia. Methods, meanings, and mechanisms
Archives of General Psychiatry
Factor analysis of symptom subtypes of obsessive compulsive disorder and their relation to personality and tic disorders
Journal of Clinical Psychiatry
Hoarding and treatment response in 38 nondepressed subjects with obsessive–compulsive disorder
Journal of Clinical Psychiatry
Alternative ways of assessing model fit
Sociological Methods and Research
Twin and family studies of anxiety, phobic, and obsessive disorders
A factor analysis of the signs and symptoms of mania
Archives of General Psychiatry
Factor analysis of dichotomous variables
Psychometrika
Factor analysis of the Yale-Brown Obsessive Compulsive Scale in a family study of obsessive–compulsive disorder
Depression and Anxiety
Evaluating the use of exploratory factor analysis in psychological research
Psychological Methods
Item-by-item factor analysis of the Yale-Brown Obsessive Compulsive Scale Symptom Checklist
Journal of Neuropsychiatry and Clinical Neurosciences
The Yale-Brown Obsessive Compulsive Scale: I. Deveopment, use, and reliability
Archives of General Psychiatry
Factor Analysis
Cited by (133)
A critical comparison of aggressive intrusive thoughts in obsessive compulsive disorder and aggressive scripts in offender populations
2024, Aggression and Violent BehaviorObsessive-compulsive symptom profiles vary by index trauma type in a trauma-exposed community sample
2023, Journal of Obsessive-Compulsive and Related DisordersThe relationship between obsessive-compulsive disorder symptom subtypes and social adjustment
2023, Journal of Obsessive-Compulsive and Related DisordersYear of Zoom in a Year of Doom: Lessons Learned Delivering ERP Remotely During the COVID-19 Pandemic
2023, Cognitive and Behavioral PracticeA network approach to understanding obsessions and compulsions
2023, Journal of Obsessive-Compulsive and Related DisordersValidation of the Chinese version of the dimensional obsessive-compulsive scale (DOCS) in Chinese clinical samples
2023, Journal of Obsessive-Compulsive and Related DisordersCitation Excerpt :For example, one may struggle with intense fears of contamination and the other may be plagued by intrusive violent images with religious rituals. As research grows, OCD is increasingly conceptualised to be multi-dimensional, with factor analyses showing compelling evidence on several symptom domains: contamination/cleaning, doubt/checking, taboo thoughts and symmetry/ordering (Pinto et al., 2008). Development of better instruments that could fully capture its complex phenomenology was advocated (Mataix-Cols et al., 2005).