Bayes AJ, Parker GB. Differentiating borderline personality disorder (BPD) from bipolar disorder: diagnostic efficiency of DSM BPD criteria.Objective: We sought to determine the differential diagnostic efficiency of all DSM-IV borderline personality disorder (BPD) criteria by studying a sample of those with BPD and a contrast group with a bipolar disorder (BP). Method: Participants were clinically assessed and assigned diagnoses based on DSM criteriawith prevalence rates and diagnostic efficiency values calculated. Results: Fifty-three participants were assigned a BPD diagnosis, 83 a BP diagnosis, with comorbid participants excluded. The mean number of DSM BPD criteria assigned was 6.6 (SD = 1.0) in the BPD group and 1.9 (SD = 1.3) in the BP group. The most prevalent criterion in the BPD group was 'affective instability' (AI) (92.5%), with 'inappropriate anger' least endorsed (49%). The highest specificity criterion was 'abandonment fears', which displayed the greatest positive predictive value (PPV) = 0.9, and with AI offering the lowest specificity. 'Unstable relationships' had the highest overall negative predictive value (NPV) = 0.91. The highest percentage accuracy of classification was provided by 'identity disturbance' and 'abandonment fears' criteria, both 85%. Conclusion: The transdiagnostic nature of 'affective instability' means it is less useful for diagnostic decisions, whereas 'abandonment fears' and 'identity disturbance' offer superior diagnostic efficiency in distinguishing BPD from BP.
Significant outcomes• Prevalence estimates of a DSM criterion alone are less useful in establishing a diagnosis if it occurs with a reasonable prevalence in other conditions.• The prevalence of 'affective instability' was high in both conditions, being sensitive to BPD but not specific in differentiating from BP.• Study results challenge the DSM polythetic model, which gives equal weighting to criteria, with diagnostic efficiency indices pointing to specific items being superior (i.e. 'identity disturbance' and 'abandonment fears') in making a diagnosis of BPD.
Limitations• Results may not generalise to differing age groups, as prevalence of certain BPD symptoms changes across the lifespan.• Prevalence estimates of criteria likely reflect the study's specific recruitment portals' base rates and may not be generalisable.• The study captures diagnostic efficiency in distinguishing BPD from BP, but not other potential differential diagnoses.