Measuring insight through patient self-report: An in-depth analysis of the factor structure of the Birchwood Insight Scale

Little research has focused on item analysis and factor structure of the most commonly used measures of insight. We examined the factorial structure of the Birchwood Insight Scale (BIS), a brief, easy-to-administer, self-report measure. We studied the BIS in 327 first-episode psychosis patients, including a test sample (n=163) and a validation sample (n=164). We then used data from 100 patients with chronic serious mental illnesses as a second, external validation sample. Exploratory factor analysis was conducted with the test subsample, and confirmatory factor analyses with the two validation samples. Confirmatory factor analyses (in both the first-episode psychosis validation sample and the chronic serious mental illness sample) indicated that a single-factor solution, with seven items loading on a single factor-with item 1 ("Some of your symptoms are made by your mind") eliminated-was the best-fitting model. Seven of the eight original BIS items loading on a single factor fit the data well in these samples. Researchers using this efficient measure of patient-reported insight should assess the item distributions and factor structure of the BIS in their samples, and potentially consider eliminating item 1.