Brief psychiatric rating scale – expanded version : construct validity using Rasch model analysis

Abstract

BACKGROUND : The Brief Psychiatric Rating Scale – Expanded version (BPRS-E) is a 24-item clinician-administered scale whereby severity of psychopathology is rated using seven scoring categories for each item. Although useful in research and clinical settings, the construct validity has not been tested in South Africa. AIM : Examine the construct validity of the BPRS-E using Rasch model analysis. SETTING : Community psychiatric clinics in the Sedibeng District of Gauteng province, South Africa. METHODS : A retrospective record review was conducted of adult psychiatric patients in whom the BPRS-E was used in routine clinical assessment by trained psychiatric nurses and doctors. Clinical records with completed BPRS-Es were purposively sampled from three community psychiatric clinics in the Sedibeng District. Data were entered into RUMM2030® software, and construct validity was analysed using the Rasch model, a probabilistic model that assesses item fit, response category functioning, and unidimensionality. RESULTS : Clinical records of 192 patients (93 males and 99 females; aged between 18 and 79 years) were reviewed. Total BPRS-E scores ranged from 24 to 93, with a score of 39 or less in 52% of records (n = 100). Rasch analysis revealed good item fit and unidimensionality for the 24 BPRS-E items but disorganised threshold curves and inconsistent differential item functioning for the severity scoring categories. CONCLUSION : This study supports the construct validity of the BPRS-E items when used clinically in a South African community psychiatric setting. However, severity scoring using the BPRS-E scoring categories in this setting requires further investigation. CONTRIBUTION : This study provides evidence that the BPRS-E is valid in a community psychiatric setting in South Africa.