M18. Lack of Generalization From a High-Dose, Well-Powered Randomized Controlled Trial of Working Memory-Focused Training for Schizophrenia

Abstract

Abstract Background: Cognitive deficits contribute to the functional disability associated with schizophrenia. Cognitive training has shown promise as a method of intervention; however, there is considerable variability in the implementation of this approach. The aim of this study was to test the efficacy of a high dose of cognitive training that targeted working memory-related functions. Methods: A randomized, double blind, active placebo-controlled, clinical trial was conducted with 80 outpatients with schizophrenia (mean age 46.44 years, 25% female). Patients were randomized to either working memory-based cognitive training or a computer skills training course that taught computer applications. In both conditions, participants received an average of 3 hours of training weekly for 16 weeks. Cognitive and functional outcomes were assessed with the MATRICS Consensus Cognitive Battery, N-Back performance, 2 measures of functional capacity (UPSA and SSPA) and a measure of community functioning, the Social Functioning Scale. Results: An intent-to-treat analysis found that patients who received cognitive training demonstrated significantly greater change on a trained task (Word N-Back), F(78) = 21.69, P < .0001, and a novel version of a trained task (Picture N-Back) as compared to those in the comparison condition, F(78) = 13.59, P = .002. However, only very modest support was found for generalization of training gains. A trend for an interaction was found on the MCCB Attention Domain score, F(78) = 2.56, P = .12. Participants who received cognitive training demonstrated significantly improved performance, t(39) = 3.79, P = .001, while those in computer skills did not, t(39) = 1.07, P = .37. Conclusion: A well-powered, high-dose, working memory focused, computer-based, cognitive training protocol produced only a small effect in patients with schizophrenia. Results indicate the importance of measuring generalization from training tasks in cognitive remediation studies. Computer-based training was not an effective method of producing change in cognition in patients with schizophrenia.