Optimizing the prediction of sepsis-associated encephalopathy with cerebral circulation time utilizing a nomogram: a...

Sepsis-associated encephalopathy (SAE) is prevalent in intensive care unit (ICU) environments but lacks established treatment protocols, necessitating prompt diagnostic methods for early intervention. Traditional symptom-based diagnostics are non-specific and confounded by sedatives, while emerging biomarkers like neuron-specific enolase (NSE) and S100 calcium-binding protein B (S100B) have limited specificity. Transcranial Doppler (TCD) indicators, although is particularly relevant for SAE, requires high operator expertise, limiting its clinical utility.

This pilot study aims to utilize cerebral circulation time (CCT) assessed via contrast-enhanced ultrasound (CEUS) as an innovative approach to investigate the accuracy of SAE prediction. Further, these CCT measurements are integrated into a nomogram to optimize the predictive performance.

This study employed a prospective, observational design, enrolling 67 ICU patients diagnosed with sepsis within the initial 24 h. Receiver operating characteristic (ROC) curve analyses were conducted to assess the predictive accuracy of potential markers including NSE, S100B, TCD parameters, and CCT for SAE. A nomogram was constructed via multivariate Logistic Regression to further explore the combined predictive potential of these variables. The model's predictive performance was evaluated through discrimination, calibration, and decision curve analysis (DCA).

SAE manifested at a median of 2 days post-admission in 32 of 67 patients (47.8%), with the remaining 35 sepsis patients constituting the non-SAE group. ROC curves revealed substantial predictive utility for CCT, pulsatility index (PI), and S100B, with CCT emerging as the most efficacious predictor, evidenced by an area under the curve (AUC) of 0.846. Multivariate Logistic Regression identified these markers as independent predictors for SAE, leading to the construction of a nomogram with excellent discrimination, substantiated by an AUC of 0.924 through bootstrap resampling. The model exhibited satisfactory concordance between observed and predicted probabilities, and DCA confirmed its clinical utility for the prompt identification of SAE.

This study highlighted the enhanced predictive value of CCT in SAE detection within ICU settings. A novel nomogram incorporating CCT, PI, and S100B demonstrated robust discrimination, calibration, and clinical utility, solidifying it as a valuable tool for early SAE intervention.