The effect of sampling, diagnostic criteria and assessment procedures on the observed prevalence of DSM-III-R personality disorders among treated alcoholics

OBJECTIVE. In a recent review of empirical studies on the prevalence of DSM-III-R personality disorders among substance abusers, wide ranges of prevalence rates for overall Axis II, antisocial personality disorder (APD) and borderline personality disorder (BPD) were shown. Utilizing subsamples from within a single study population, the current report explores the effect of sampling, diagnostic criteria and assessment procedures on the observed prevalence of DSM-III-R personality disorders among treated alcoholics. Personality disorders were assessed with the Personality Diagnostic Questionnaire Revised (PDQR) at two times of measurement (Time 1 n = 459; Time 2 n = 90). In addition, APD was measured with the Composite International Diagnostic Interview (CIDI; n = 587). Finally, an unselected subgroup (n = 136) was interviewed with the International Personality Disorder Examination (IPDE). There were few differences between inpatients or outpatients or between males and females. Using the IPDE, the exclusion of substance-related pathology did not affect the prevalence estimate. However, the prevalence rates according to the PDQR varied greatly across age groups. In a representative subsample (n = 109), the prevalence rates also varied greatly across assessment methods (PDQR, 52% vs IPDE, 31%). The prevalence estimate of APD according to the CIDI was related to setting, gender, age group and the applied time-frame. These findings indicate the examined factors to be necessary qualifiers of prevalence estimates and, consequently, support the use of a multiple-criteria/multimethod assessment battery in research as well as in clinical wor