Risk prediction strategies using intraoperative physiological data in adults undergoing surgery: a systematic review study protocol

Abstract

Abstract Postoperative mortality accounts for 7.7% of all global deaths, while postoperative complications increase 1-year mortality by 60%. Risk prediction models for postoperative complications and mortality can facilitate tailored risk mitigation strategies. However, most models incorporate only preoperative patient-related factors as predictors and do not capture dynamic risks or intraoperative events. This systematic review seeks to evaluate the predictive capability of intraoperative physiology derived from routine anaesthetic monitoring and the feature extraction methods for these variables. This review will include both prospective and retrospective studies that incorporate intraoperative physiological measurements into the development, validation or updating of a statistical prediction model, to identify those at risk of major end-organ (cardiovascular, pulmonary, renal and neurological) complications and mortality up-to 90 days postoperatively. We will identify models developed in two settings: those undergoing cardiac surgery and heterogeneous adult patient cohorts undergoing non-cardiac surgery. The review will be reported according to the 2020 Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. This review will evaluate the available literature on intraoperative physiology as predictor variables, to improve feature extraction methods for risk prediction models development while understanding how to capitalize on the vast routinely collected intraoperative physiological data sets that are increasingly available. This review is registered on PROSPERO, registration number CRD42023474384.