M25. Development and Validation of the Brief Assessment of Validation in Schizophrenia (BAC-App)

Background: The Brief Assessment of Cognition in Schizophrenia (BACS) is a pen-and-paper cognitive assessment tool that has been used in hundreds of research studies and clinical trials, and has normative data available for generating age- and gender-corrected standardized scores. A tablet-based version of the BACS called the BAC App has been developed to allow standardized presentation of task instructions and stimuli, audio-recording of subject responses, and automatized scoring and data management. Development of the BAC App was aimed at reducing rater burden and variability. Our validation study compared performance on the traditional BACS and the BAC App in patients with schizophrenia and healthy controls. Test equivalency was assessed, and the applicability of paper-based normative data was evaluated. An ongoing follow-up study examines BAC App performance in large-scale census-matched normative sample. Methods: Participants in the validation study included 48 patients (23 female) with schizophrenia and 50 healthy controls (25 female) recruited from 3 academic sites including the University of California-San Diego, the University of Miami—Miller School of Medicine, and the University of South Carolina. All participants were assessed with the standard pen-and-paper BACS and the BAC App. Ongoing normative evaluation of the BAC App will include a community sample of 650 individuals matched to the US census on age, education, race and gender. Results: In the validation study, distributions of standardized composite scores for the tablet-based BAC App and the pen-and-paper BACS were indistinguishable in both schizophrenia patients and healthy controls. Between-methods mean differences were not statistically significant. The discrimination between patients and controls was similarly robust with the BAC App ( d = 1.34) and the BACS ( d = 1.24). The between-methods correlations for individual measures in patients were r > .70 except Token Motor ( r = .43) and Tower of London ( r = .61). In patients, performance between the test methods was not significantly different on any test except the Token Motor Test. When data from the Token Motor Test were removed, the between-methods correlation of composite scores improved to r = .88 ( df = 48; P