(1520) Evaluation of Coronary Perfusion Pressure Calculation Methods During Cardiopulmonary Resuscitation

Introduction: Effective CPR relies on achieving adequate myocardial perfusion, approximated by the coronary perfusion pressure (CPP), defined as the diastolic aortic-right atrial pressure gradient. Multiple discrepant CPP calculation methods may impact the translation of research-based CPP goals to clinical care. The degree of intermethod agreement and performance as predictors of ROSC during CPR are unknown. Further, existing methods do not account for the contribution of diastolic time to coronary perfusion. This study aimed to evaluate intermethod agreement, define a time-dependent CPP metric, and assess ability to discriminate ROSC during CPR.

Methods: Waveforms from porcine models of pediatric CPR were retrospectively analyzed. Four established CPP calculation methods, differentiated by the moment of pressure sampling, were applied to each compression, at end-diastole (CPPend), mid-diastole (CPPmid), minimum of diastole (CPPmin), and averaged through diastole (CPPmean). To account for perfusion duration, a new metric, diastolic coronary perfusion index (dCPI), was defined, integrating CPP through diastole normalized by compression cycle time. Agreement was assessed by comparing measurements from each method and by Cohen’s kappa statistic when dichotomizing CPP by a clinical target of 20 mmHg. Logistic regression and receiver operator characteristic (ROC) curves evaluated each as predictors of ROSC and identified optimal discriminator thresholds.

Results: The cohort of 204 subjects received a total of 198,960 compressions. Referenced to CPPend, CPPmid, CPPmin, and CPPmean resulted in median differences of 0.94 (IQR -0.28, 2.14), -2.57 (-7.26, -0.51), and -0.05 (-1.10, 1.16) mmHg with kappa 0.88, 0.76, and 0.91, respectively. The areas under the ROC curves (AUC) were 0.777, 0.792, 0.787, and 0.788. Optimal thresholds to discriminate ROSC were 15.3, 15.8, 12.3, and 14.7 mmHg, respectively. dCPI outperformed the remaining methods with AUC 0.799.

Conclusions: While overall intermethod agreement was strong, optimal thresholds for discriminating outcomes varied significantly. As a result, clinical targets may not reliably translate across methods and methodological standardization is necessary to ensure alignment between research findings and clinical targets.