Examining the variability of neurocognitive functioning in individuals at clinical high risk for psychosis: a meta-analysis.

This study aims to meta-analytically characterize the presence and magnitude of within-group variability across neurocognitive functioning in young people at Clinical High-Risk for psychosis (CHR-P) and comparison groups. Multistep, PRISMA/MOOSE-compliant systematic review (PROSPERO-CRD42020192826) of the Web of Science database, Cochrane Central Register of Reviews and Ovid/PsycINFO and trial registries up to July 1, 2020. The risk of bias was assessed using a modified version of the NOS for cohort and cross-sectional studies. Original studies reporting neurocognitive functioning in individuals at CHR-P compared to healthy controls (HC) or first-episode psychosis (FEP) patients were included. The primary outcome was the random-effect meta-analytic variability ratios (VR). Secondary outcomes included the coefficient of variation ratios (CVR). Seventy-eight studies were included, relating to 5162 CHR-P individuals, 2865 HC and 486 FEP. The CHR-P group demonstrated higher variability compared to HC (in descending order of magnitude) in visual memory (VR: 1.41, 95% CI 1.02-1.94), executive functioning (VR: 1.31, 95% CI 1.18-1.45), verbal learning (VR: 1.29, 95% CI 1.15-1.45), premorbid IQ (VR: 1.27, 95% CI 1.09-1.49), processing speed (VR: 1.26, 95% CI 1.07-1.48), visual learning (VR: 1.20, 95% CI 1.07-1.34), and reasoning and problem solving (VR: 1.17, 95% CI 1.03-1.34). In the CVR analyses the variability in CHR-P population remains in the previous neurocognitive domains and emerged in attention/vigilance, working memory, social cognition, and visuospatial ability. The CHR-P group transitioning to psychosis showed greater VR in executive functioning compared to those not developing psychosis and compared to FEP groups. Clinical high risk for psychosis subjects shows increased variability in neurocognitive performance compared to HC. The main limitation of this study is the validity of the VR and CVR as an index of variability which has received debate. This finding should be explored by further individual-participant data research and support precision medicine approaches.