Previous reviews suggest there is minimal evidence for an association between duration of untreated psychosis (DUP) and neurocognition. This is based on tallied findings of studies with small samples and neurocognition viewed as a single construct. We aimed to conduct a systematic review and meta-analysis examining the association between DUP and individual neurocognitive domains and tests in first-episode psychosis (FEP).

MOOSE and PRISMA guidelines were followed. Forty-three studies involving 4647 FEP patients were included. For studies providing correlations between DUP and neurocognition, 12 separate meta-analyses were performed based on neurocognitive domains/indices. The influence of demographic/clinical variables was tested using weighted linear meta-regression analyses.

The relationship between DUP and most neurocognitive domains/indices was not significant. Longer DUP was associated with a larger cognitive deterioration index, i.e. current minus premorbid intellectual functioning (N = 4; mean ES −0.213, 95% confidence interval (CI) (−0.344 to −0.074), p = 0.003). Findings were homogeneous, with no evidence of publication bias or significant influence from moderators. For studies providing mean and standard deviations for neurocognitive measures and DUP, 20 meta-regressions were performed on individual neurocognitive tests. One significant finding emerged showing that longer DUP was associated with fewer Wisconsin Card Sorting Test-perseverative errors (mean ES −0.031, 95% CI (−0.048 to −0.013), p < 0.001). Exploratory meta-regressions in studies with mean DUP <360 days showed longer DUP was significantly associated with poorer performance on Trail Making Test A and B and higher Full-Scale IQ.

There may not be a generalised association between DUP and neurocognition, however, specific cognitive functions may be associated with longer DUP or delayed help-seeking.