Prediction Models for Postoperative Delirium in Patients With Stanford Type A Aortic Dissection: A Systematic Review and Meta-Analysis.

Background: Postoperative delirium (POD) is a prevalent neurological complication following Stanford type A aortic dissection (STAAD), significantly impacting patient prognosis and cognitive function. While the number of models predicting POD risk in STAAD patients has been steadily rising, their quality and clinical applicability, as well as their potential utility in future research, remain uncertain.

Aim: To systematically assess the performance and predictors of existing POD risk prediction models in STAAD patients.

Study design: A comprehensive systematic search was conducted across PubMed, Embase, Web of Science, Ovid, CINAHL, CNKI, Wanfang, VIP, and SinoMed databases up to February 2025. Two independent reviewers screened the articles and assessed study quality using the PROBAST tool. Data extraction was performed independently by two reviewers using standardised forms, followed by a meta-analysis of predictive model performance using STATA 18.0 software.

Results: A total of 605 studies were identified, of which 14 prediction models from 9 studies met the inclusion criteria. Seven studies were assessed as having a high risk of bias, and five showed high concerns regarding applicability. Meta-analysis yielded the pooled area under the curve of 0.87 (95% CI: 0.81-0.93), indicating moderate discriminatory ability. After analysing the sources of high heterogeneity (90.21%) and conducting sensitivity analyses, no significant changes were observed in the meta-analysis results, indicating high stability. Additionally, Egger's test and Begg's test revealed no evidence of small-sample bias. Existing POD prediction models have identified multiple high-frequency common predictors, including blood markers, acute kidney injury and male.

Conclusions: Although the included model demonstrated good discriminatory power, further refinements are necessary to enhance its applicability and reduce the risk of bias. Future research can utilise the common predictors identified in this study to develop predictive models and proactively investigate high-risk factors associated with POD. Furthermore, research should prioritise the development of innovative models utilising multicentre data sources and validate their effectiveness in reducing the incidence of POD in STAAD patients through clinical trials.

Relevance to clinical practice: A systematic review and meta-analysis of risk prediction models for POD in STAAD patients were conducted to assess the predictive performance of existing models and identify common high-risk factors. These findings will serve as a reference for precise clinical screening of high-risk patients and the development of targeted prevention strategies.