A (re)-evaluation of the symptom structure of borderline personality disorder.

OBJECTIVE Despite evidence of significant symptom heterogeneity and excessive diagnostic comorbidity, many contend that borderline personality disorder (BPD) is unidimensional, an assumption that rests primarily on results from factor analytic investigations of BPD symptom criteria. We note several limitations in the literature and argue that the symptom structure of BPD can be best clarified by using both factor analytic techniques and examining the BPD symptom dimensions in relation to external criteria (that is, personality traits). Our goals were to: examine if the symptoms of BPD are best conceptualized as unidimensional or multidimensional, and determine the extent to which personality traits account for any symptom dimensions that underlie BPD. METHOD All published structural models of the BPD symptom criteria were identified and tested for statistical fit using confirmatory factor analysis in a sample of 373 patients who had completed the Structured Clinical Interview for the Diagnostic and Statistical Manual of Mental Disorders-Axis II Personality Questionnaire BPD scale. Dimensions from the best fitting model were examined in relation to traits from the Five-Factor Model (FFM) and the Personality Psychopathology Five (PSY-5) using correlational and regression analyses. RESULTS Sanislow's 3-factor model, containing affect dysregulation, behavioural dysregulation, and disturbed relations symptom dimensions, provided the best fit; the unidimensional model produced the worst. The symptom dimensions of the 3-factor model were differentiable from one another and had unique associations with the FFM and PSY-5 personality traits. CONCLUSION BPD is a multidimensional construct.