Societal representations of DSM-IV-TR personality disorder criteria: an examination of DSM and dimensional assessment of personality pathology structures

Abstract

This research provides an examination of DSM personality disorder criteria using the DSM-IV-TR Axis II and Dimensional Assessment of Personality Pathology (DAPP) frameworks.

Data
The British Psychiatric Morbidity of Adults Living in Private Households, 2000 (BPMS) survey, an epidemiological study covering England, Scotland and Wales (n=8580) was obtained from the UK Data Archives at the University of Essex. The personality disorder section of the interview was measured by self-report methods using the SCID-II (Structured Clinical Interview for DSM Axis II Disorders) screening questionnaire.

Methods
The statistical software programme MplusTM version 5.21 was used to carry out statistical analyses. Item response theory (IRT) methods produced discriminatory and severity characteristics of the DSM personality disorder criteria. Confirmatory factor analysis (CFA) was utilized to examine the latent factors of the DSM personality disorders and clusters, and DAPP primary traits and domains. Hybrid modelling that combines CFA with latent class analysis techniques (LCFA) identified sub-populations within both the DSM and DAPP structures. Binomial logistic regression analyses of these sub-populations provided examination of group characteristics.

Results
IRT item characteristic parameters indicated a number of DSM personality disorder are not performing well in terms of the underlying 11 disorders and domains. CFA indicated good model fit for the structures on inclusion of a number of extra specified relationships between some criteria. LCFA identified sub-populations that differ in terms of severity and likelihoods of criteria endorsements.

Conclusions
IRT and LCFA both confirm that a number of criteria used for the assessment of personality disorder have poor discriminant validity, and these should be reviewed as indicators in new guidelines. Analyses at subpopulation levels provide support for dimensional trait-type representations of personality disorder, and have provided useful profiles for hidden groups that have not been previously identified and should be of interest for clinicians, policy-makers and working groups for future editions of psychiatric guidelines.

Date of Award	Jun 2011
Original language	English
Sponsors	Department of Employment and Learning
Supervisor	Brendan Bunting (Supervisor) & Siobhan O'Neill (Supervisor)