Confirmatory factor analysis of the DSM-IV criteria for borderline personality disorder: Findings from the Collaborative Longitudinal Personality Disorders Study

Charles A. Sanislow, Yale University School of Medicine
Carlos M. Grilo, Yale University School of Medicine
Leslie C. Morey, Texas A & M University - College Station
Donna S. Bender, New York State Psychiatric Institute and Columbia University College of Physicians and Surgeons
Andrew E. Skodol, New York State Psychiatric Institute and Columbia University College of Physicians and Surgeons
John G. Gunderson, McLean Hospital and Harvard Medical School
M. Tracie Shea, Veterans Affairs Medical Center and Brown University Medical School
Robert L. Stout, Decision Sciences Institute and Brown University Medical School
Mary C. Zanarini, McLean Hospital and Harvard Medical School
Thomas H. McGlashan, Yale University School of Medicine

Abstract

Objective: This study tested the factor structure of the DSM-IV criteria for borderline personality disorder by using confirmatory methods for the analysis of covariance structures in a large group from a multisite study.

Method: A total of 668 primarily treatment-seeking subjects were reliably assessed for personality disorders by using the Diagnostic Interview for DSM-IV Personality Disorders. Associations among criteria for borderline personality disorder were examined. A confirmatory factor analysis was performed to test diagnosis as a unitary construct and to test an earlier-reported three-factor model comprising disturbed relatedness, behavioral dysregulation, and affective dysregulation. The three-factor model was subsequently tested by using a subset (N=498) of the study group that was reassessed 2 years later by independent evaluators who were blind to original diagnoses.

Results: Internal consistency of the borderline personality disorder diagnosis was adequate for both baseline and follow-up assessments. A confirmatory factor analysis model testing the borderline personality disorder diagnosis as a unitary construct provided good fit, and the three-factor model offered a significantly better fit for the baseline assessment. The three-factor model was replicated with the 2-year follow-up data.

Conclusions: The diagnostic criteria for borderline personality disorder appear to reflect a statistically coherent construct. Three homogeneous components were supported empirically, lending conceptual clarity to different classes of the criteria for borderline personality disorder.