Key result
Preoperative risk prediction models show fair to good discrimination for postoperative delirium.
Why the study?
Various preoperative prediction models for postoperative delirium in cardiac surgery exist, but there is no consensus on which model is most suitable.
Systematic Review
Existing preoperative prediction models exhibit fair to good discrimination in identifying cardiac surgery patients at elevated risk for postoperative delirium, but extensive heterogeneity and limited validation currently challenge their routine clinical application.
May inform targeted delirium prevention in cardiac surgery; leaves open optimal model selection pending external validation and standardization.
BACKGROUND: Postoperative delirium is believed to be preventable in up to 40% of all cases. Researchers have proposed various preoperative risk prediction models for postoperative delirium in patients undergoing cardiac surgery, however, no consensus exists on which model is the most suitable. AIM: To identify and map existing preoperative risk prediction models, detecting cardiac surgery patients at elevated risk of developing postoperative delirium. DESIGN: This scoping review considered cohort and case-control studies eligible if they developed or validated preoperative prediction models for postoperative delirium, in adult patients admitted for cardiac surgery via sternotomy. DATA SOURCES: The primary search was conducted on May 6th, 2022, and a secondary search was conducted on September 18th, 2024. We searched MEDLINE, CINAHL, Embase, and PsycINFO where 2126 references were identified and 15 were included for full-text analysis. METHOD: This scoping review was conducted in line with the Systematic Reviews and Meta-Analyses extension for Scoping Reviews (the PRISMA-ScR) guideline. RESULTS: Twelve unique risk prediction models and three validation studies were included in this review, comprising between 77 and 45,744 participants. In total, 157 candidate prognostic variables were investigated of which 40 had a predictive value and thus, were included in the prediction models. The included models revealed an AUC from 0.68-0.93 in the derivation cohorts and 0.61-0.89 in the validation cohorts. CONCLUSIONS: Twelve unique prediction models and 3 validation studies were identified and mapped. Collectively, the models demonstrated an AUC ranging from 0.61-0.93, indicating a fair to good discrimination performance. PROTOCOL REGISTRATION: A protocol is registered at Open Science Framework (OSF) https://osf.io/wr93y/?view_only=d129c3bb6be04357bac35c2c41ba2a40.
No takes yet. Share an insight, caveat, or question.
Blakoe et al. (2025) conducted a systematic review in Postoperative delirium following cardiac surgery. Preoperative risk prediction models for postoperative delirium was evaluated on Predictive performance of delirium prediction models (Area Under the Curve). Twelve unique preoperative risk prediction models and three validation studies demonstrated fair to good discrimination performance for postoperative delirium, with AUCs ranging from 0.61 to 0.93.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: