RESUMO
OBJECTIVES: To determine the psychometric properties of an observational scale of anxiety. DESIGN: A cross-sectional and longitudinal survey with stroke survivor-carer dyads. METHODS: Eighty-nine dyads recruited in community stroke groups completed: a demographic questionnaire; the Behavioural Outcomes of Anxiety scale (BOA), survivor-rated (survivor BOA) and carer-rated (carer BOA) versions; the anxiety scale of the Hospital Anxiety and Depression Scales (HADS-A), also in carer and survivor versions. Twenty-seven survivors and carers repeated the BOA after 1 week. RESULTS: Correlations between the carer BOA and the survivor HADS-A (r = .55, p < .001) and the survivor BOA (r = .73, p < .001) demonstrated construct validity. Cronbach's alpha for the carer BOA was .81; item statistics did not identify any items for exclusion. The test-retest coefficient at 1 week was 0.83. Receiver operating characteristic analysis against the survivor HADS-A and BOA produced areas under the curve of 0.75 and 0.88, respectively. At a cut-off score of 13/14 sensitivity and specificity against the HADS-A were 0.77 and 0.58, respectively, and 0.86 and 0.68 against the survivor BOA. The impact of stroke on memory was associated with elevated anxiety. Scores for both BOA versions were independent of demographic variables. CONCLUSIONS: The carer BOA has acceptable psychometric properties and is independent of survivor demographic variables such as age. It identifies self-reported cases with acceptable sensitivity and specificity. It has potential for use with persons unable to self-report anxiety. Further validation is recommended, but its continuing use is supported.
