Development and Psychometric Properties of the Dimensional Obsessive-Compulsive Scale (DOCS) Jonathan S. Abramowitz, PhD University of North Carolina at Chapel Hill
Collaborators Brett Deacon Bunmi Olatunji Michael Wheaton Noah Berman Diane Losardo Kiara Timpano Patrick McGrath Bradley Riemann Thomas Adams Throstur Bjorgvinsson Eric Storch Lisa Hale
Assessment of OCD OCD is heterogeneous Challenges in assessing OCD symptoms – Measure a wide range of possible symptoms – Do so efficiency (as few items as possible) Approaches – Assess only the quintessential obs. and comps. PI/PI-R, OCI/OCI-R, MOCI – Assess the severity of the patient’s main symptoms YBOCS/D-YBOCS
Limitations of Existing OCD Measures Symptom severity is confounded with breadth Severity is confounded with popularity of sx One dimensional assessment of severity Obsessions assessed separately from rituals Avoidance? Hoarding?
We Need New OCD Measures Assess the severity of empirically supported symptom dimensions – Contamination, responsibility for harm/mistakes, Unacceptable thoughts, symmetry/incompleteness Multiple severity parameters Assess avoidance Assess severity independent of type or range of symptoms Easy to administer
DOCS Four sections (one for each symptom dimension) – Contamination, responsibility, unacceptable thoughts, symmetry/incompleteness Each section contains – Examples of obsessions, compulsions, avoidance – Five severity questions (time, avoidance, distress, interference, control) rated 0 to 4 20 items in all Approximately 5-10 mins. to read and complete
Study Aims Factor structure Reliability and Validity Diagnostic accuracy Treatment sensitivity
Method Participants (data collected at 8 sites) – 315 adults with OCD – 198 adults with other anxiety disorders (OAD) – 1,044 unselected undergraduate students Measures GroupDOCSOCI-RYBOCSBDIBAIDASSSIAS OCD X X X X X X OAD X X X X X Students X X X X
Results: Exploratory Factor Analysis with Half of the Student Sample
Results: Confirmatory Factor Analyses Goodness-of-fit parameters indicated that the student and clinical data fit the four-factor structure very well. Goodness-of-fit index GroupRMSEASRMRTLICFI Students OCD + OAD
Correlations among factors showing weak to moderate relationships
Reliability Cronbach’s alpha and item-total correlation Test-retest
Convergent and Discriminant Validity: DOCS Total Score
Convergent and Discriminant Validity: DOCS Factors/Subscales
Known-Groups Validity
Diagnostic Accuracy: ROC Analyses DOCS total score discriminates: – OCD patients from nonclinicals (AUC =.86) – OCD patients from OAD patients (AUC =.77) Cutoff scores – A score of 21 correctly classified 70% of OCD patients and 70% of OAD patients – A score of 18 correctly classified 78% of OCD patients and 78% of students
Diagnostic Accuracy: DOCS vs. OCI-R OCD vs. Nonclinical OCD vs. OAD Diff in AUC =.06; Z = 4.68, p <.01 Diff in AUC =.08; Z = 3.57, p <.01
Sensitivity to Treatment
Conclusions The DOCS is a conceptually & psychometrically sound measure of OC symptoms in patients and nonpatients – Addresses many of the limitations of existing measures Uses – Baseline assessment of symptom dimensions – Treatment response (use “main” symptom dimension) – Measure of psychopathology for research Future work – Test-retest in a clinical (OCD) group – Child version – Translations