Cognition at illness onset as a predictor of later functional outcome in early psychosis: Systematic review and methodological critique

Background: Cognitive deficits occur early in the course of psychosis, are mostly stable, and have been identified as potential functional prognostic markers. Previous reviews of chronic schizophrenia have concluded that specific cognitive deficits are strongly associated with poorer functional outcomes. However, results of schizophrenia studies may be influenced by the effects of long-term illness or treatment or be biased toward individuals with poorer outcomes and may not be relevant to early psychosis (EP). This review aimed to systematically examine the evidence regarding general and social cognitive predictors of later functional outcome in EP and critique the methodology of the studies reviewed. A final aim was to conduct a meta-analysis of the studies reviewed, but methodological reasons precluded this. Method: A comprehensive search of PsycINFO and MEDLINE databases identified 15 relevant articles and 7 further articles following a reference list search, totaling 22 included articles. Results: Most studies found at least one cognitive domain predicted functional outcome, but examination of separate cognitive domains revealed there were more null than significant associations between cognition and functional outcome across every cognitive domain. No study examined social cognition as a predictor of outcome. The frequency with which different cognitive domains predicted outcome varied depending on study methodology and this was most noticeable when studies with short-term follow-up were compared with longer-term follow-up studies. Conclusions: Due to the methodological variability and limitations of the studies reviewed, firm conclusions regarding the relationship between cognition and functional outcomes in EP cannot be made. Tentatively, cognition may be prognostic in EP, especially for longer-term outcomes. However, further research that addresses the methodological issues identified, including examination of social cognition and other non-cognitive predictors, is needed. (C) 2010 Elsevier B.V. All rights reserved.