The Problem Behaviours Questionnaire (PBQ) is designed to measure treatment outcome in patients with psychotic disorders. The interviewer fills out the questionnaire after an interview with the patient s key relative (i.e. family representative significant and close to the patient).
Aim –The aim of the study was to assess psychometric properties of the questionnaire.
Subjects – Three groups participated in the study: (1) 113 patients included in a program of evaluation of care provided by mobile community treatment teams; (2) 40 patients admitted to an inpatient department of the Institute of Psychiatry and Neurology, and (3) 32 representatives of the general population.
Method – The following properties of' the questionnaire were analysed: (1) reliability (internal consistency, using the Cronbach α; (2) validity (a) theoretical validity (including an analysis of intergroup differences, a matrix of intercorrelation between items, and factor analysis), (b) diagnostic criterion validity (i.e. correlation between the PBQ score and scores on a social functioning scale); (3) test-retest stability (or susceptibility to change over lime).
Results – The hitherto used version of the PBQ (14 items) was found to include three items dealing with behaviours that had never occurred in a vast majority (i.e. over 90%) of our patients. This resulted in an unsatisfactory differential validity of the questionnaire and besides could decrease its test-retest stability. Consequently, these 3 items were excluded from the questionnaire and its abbreviated version consisting of 11 items was submitted to validation. The Cronbach α reliability coefficient for the abbreviated version was .76; the level of problem behaviours turned out to be significantly correlated with the patients' social functioning scores, and the three groups differed significantly from each other in terms of their problem behaviour levels. Three factors were distinguished in factor analysis: aggression/bizarre behaviour (reliability coefficient = .77), withdrawal (reliability coefficient = .71), and self-destruction/alcohol abuse (reliability coefficient = .46).
Discussion – The study indicated that the former version of the PBQ (14 items) included three items evidently diminishing its psychometric properties. Validation of the abbreviated version (11 items) has evidenced that its reliability, diagnostic criterion validity, theoretical validity. And test-retest stability are satisfactory. The questionnaire may be used either as a single global scale, or its two sub-scales may be analysed separately – of aggression/bizarre behaviour (4 items) and withdrawal (3 items). The remaining four items do not constitute a subscale. Conclusions – The PBQ has rather good psychometric properties and may be used for treatment outcome evaluation.