ABSTRACT
BACKGROUND: Postoperative cognitive dysfunction (POCD) is an adverse outcome that impacts patients' quality of life. Its diagnosis relies on formal cognitive testing performed before and after surgery. The substantial heterogeneity in methodology limits comparability and meta-analysis of studies. This systematic review critically appraises the methodology of studies on POCD published since the 1995 Consensus Statement and aims to provide guidance to future authors by providing recommendations that may improve comparability between future studies. METHODS: This systematic review of literature published between 1995 and 2019 included studies that used baseline cognitive testing and a structured cognitive test battery, and had a minimal follow-up of 1 month. For cohorts with multiple publications, data from the primary publication were supplemented with available data from later follow-up studies. RESULTS: A total of 274 unique studies were included in the analysis. In the included studies, 259 different cognitive tests were used. Studies varied considerably in timing of assessment, follow-up duration, definition of POCD, and use of control groups. Of the 274 included studies, 70 reported POCD as a dichotomous outcome at 1 to <3 months, with a pooled incidence of 2998/10 335 patients (29.0%). CONCLUSIONS: We found an overwhelming heterogeneity in methodology used to study POCD since the publication of the 1995 Consensus Statement. Future authors could improve study quality and comparability through optimal timing of assessment, the use of commonly used cognitive tests including the Consensus Statement 'core battery', application of appropriate cut-offs and diagnostic rules, and detailed reporting of the methods used. PROSPERO REGISTRY NUMBER: CRD42016039293.