End-tidal CO2 changes as predictors of significant blood pressure variations during general anesthesia

Abstract

Purpose: This study aimed to establish thresholds for end-tidal CO2 (EtCO2) changes that indicate hemodynamic alterations occurring between intermittent blood pressure measurements. Given that EtCO2 reflects cardiac output under conditions of stable systemic vascular resistance, it may serve as a continuous hemodynamic surrogate when noninvasive blood pressure monitoring is intermittent. Methods: This prospective observational study included participants (ASA I-IV) undergoing elective surgery with general anesthesia and controlled ventilation. EtCO2 and blood pressure were recorded every 3 min intraoperatively. Changes in EtCO2 (ΔEtCO2) between consecutive recordings were analyzed against corresponding percentage changes in systolic blood pressure (%SBP). Results: Analysis of 4,068 paired measurements from 120 patients showed 77.1% of events maintained hemodynamic stability (± 10% SBP change). For substantial blood pressure elevations (%SBP ≥ 20%), ΔEtCO2 ≥ 2 demonstrated sensitivity 95.9%, specificity 97.7%, and negative predictive value 99.8%. For severe blood pressure reductions (%SBP ≤ -20%), ΔEtCO2 ≤ -2 showed sensitivity 94.7%, specificity 97.3%, and negative predictive value 99.9%. Both thresholds exhibited strong positive likelihood ratios (42.5 for increases, 35.0 for decreases) and minimal negative likelihood ratios (0.04 for increases, 0.05 for decreases), with superior performance in predicting severe (%SBP ± 20%) versus moderate (%SBP ± 10%) blood pressure changes. Conclusions: ΔEtCO2 effectively predicted significant systolic blood pressure changes in anesthetized patients, with ± 2 mmHg thresholds showing excellent diagnostic performance for detecting clinically significant blood pressure variations. These promising findings suggest ΔEtCO2 could serve as a valuable component in hemodynamic monitoring, particularly when integrated with other clinical parameters. Future studies should explore artificial intelligence models incorporating multiple parameters alongside ΔEtCO2 for more accurate hemodynamic monitoring.