BACKGROUND: Cognitive and executive deficits lead to worsening of quality of life and are a risk factor for developing dementia in people with Parkinson's disease (PD) with psychosis (PDP). However, which key cognitive domains are differentially affected in PDP compared with those without (PDnP), remains unclear. Here, we examined this using a Bayesian meta-analytical approach.

METHODS: Searches were conducted on PubMed, Web of Science, SCOPUS, Medline and PsycINFO. Hedges' g effect-size estimates were extracted from eligible studies as a measure of standard mean differences between PDP and PDnP participants. Meta-analyses were conducted separately for each cognitive domain and subdomain, we examined the effect of age, PD medications, PD duration and severity, depression and psychosis severity for all major domains with meta-regressions.

RESULTS: Effect-size estimates suggest worse performance on all major domains (k=105 studies) in PDP compared with PDnP participants, with global cognition (k=103 studies, g=-0.57), processing speed (k=29 studies, g=-0.58), executive functions (k=33, g=-0.56), episodic memory (k=30 studies, g=-0.58) and perception (k=34 studies, g=-0.55) as the most likely affected domains. Age, depression and PD duration had moderating effects on task-related performance across most of the major nine domains.

CONCLUSIONS: We report extensive deficits across nine domains as well as subdomains in PD psychosis, with global cognition, processing speed and executive functions as the most likely impaired. The presence of depression may influence task-related performance in PDP, alongside age and PD duration, but not dose of dopamine replacement treatments.

Original languageEnglish
Article numberjnnp-2022-331028
JournalJournal of neurology, neurosurgery, and psychiatry
Publication statusE-pub ahead of print - 19 Jul 2023


Dive into the research topics of 'Cognitive and executive impairments in Parkinson's disease psychosis: a Bayesian meta-analysis'. Together they form a unique fingerprint.

Cite this