To complete an up-to-date meta-analysis of studies investigating the correlation between insight and various measures of neurocognition in psychosis.

We completed a comprehensive literature search using Pubmed and ‘Web of Science’. After applying inclusion and exclusion criteria we isolated 34 papers published on this topic since 2004. This list was then added to those published prior to 2004 that had been found in an earlier meta-analysis performed by our group to give a complete collection of 69 papers. Metaanalysis of data was performed in the Comprehensive Meta-Analysis software using random effects models.

Overall our sample included 5127 patients with psychosis across 69 studies. We performed analyses for overall cognition, IQ, memory, executive function and specifically the Wisconsin Card Sorting Test (WCST). We found a small but highly significant correlation between insight and each measure of neurocognition in all patients with psychosis (mean weighted r = 0.13–0.16 across domains, p < 0.001). When restricted to those patients with diagnosis of schizophrenia (n = 1708) the strength of correlation in each domain increased slightly (mean weighted r = 0.14–0.18, p < 0.01). We found no evidence of significant publication bias.

In this comprehensive meta-analysis we found a small but robust correlation between insight and a variety of neurocognitive domains in patients with psychosis. We believe that this data supports the role of neurocognitive processes in some aspects of insight but believe that a more precise experimental approach is required to elucidate the exact mechanisms of this relationship.