Reliability and Factor Structure of the Short Problem Behaviors Assessment for Huntington’s Disease (PBA-s) in the TRACK-HD and REGISTRY studies

Abstract

The authors report the inter-rater reliability and factor structure of the Short Problem Behaviors Assessment (PBA-s), a semistructured interview to measure severity and frequency of behavioral problems in Huntington's disease. Video recordings of 410 PBA-s interviews were rescored by an independent rater, and Cohen's kappa calculated to assess inter-rater reliability. The mean kappa was 0.74 for severity and 0.76 for frequency scores, whereas weighted kappa (allowing scores to differ by 1 point) was 0.94 for severity and 0.92 for frequency scores. The results of factor analysis were consistent with previous studies using other measures. The authors conclude that the PBA-s is a reliable measure.

Exploring the Validity of the Short Version of the Problem Behaviours Assessment (PBA-s) for Huntington’s disease: A Rasch Analysis

“…The PBA-s is the recommended outcome measure for behavioural symptoms in HD, as it has been recognised as a reliable and valid tool using traditional psychometric techniques [10,15]. However, applying Rasch analysis has revealed a number of limitations when using total PBA-s scores.…”

“…However some studies have multiplied the severity and frequency scores for each item and subsequently added these individual item scores to create subscale scores [14,39]. For example, symptom clusters derived from previous factor analyses have led to the creation of composite affect, irritability and apathy subscales [1,14,15]. Broader multiplicative behaviour scores have been created using up to seven PBA-s items [39].…”

“…The PBA-s has been shown to have substantial agreement in scores given by different interviewers, indicating a good interrater reliability [15].…”

“…Although factor analysis in previous literature has identified subscales within the PBAs [15], the total PBA-s severity score created in this study was deemed to be measuring a unidimensional construct representing the overall severity of behavioural symptoms. The responses for depressed mood were related to the responses for suicidal ideation and the same dependency was found between irritability and aggressive behaviour.…”

“…CTT also uses test-level statistics such as Cronbach's alpha to assess reliability and factor analysis to determine the dimensionality of a measurement [12,13]. Factor analyses performed on the short and long version of the PBA have reported consistent findings, providing evidence to support its reliability from a traditional psychometric perspective [1,14,15].…”