This multicenter retrospective study developed an internally validated model using sequential end-tidal carbon dioxide measurements to predict ROSC in the next 2-minute cycle during adult out-of-hospital cardiac arrest resuscitation. Analysis included 1,651 event-level sequences from 1,480 patients, with ROSC occurring in 583 events. Event-level discrimination was good, with an AUROC of 0.876 (95% CI, 0.863-0.890). However, across 17,029 cycle-level predictions, AUROC fell to 0.685 (95% CI, 0.664-0.706), and the model systematically overestimated ROSC probability, indicating that recalibration is needed before cycle-to-cycle use.
Sequential EtCO2 measurements showed predictive value for ROSC, but event-level performance did not translate into reliable cycle-level probabilities. These findings do not establish readiness for real-time clinical use.
Study limitation: This retrospective model underwent internal cross-validation only; cycle-level probability overestimation limits its immediate clinical applicability.
AI-generated from the published abstract with automated source checks. For research discovery; consult the original paper before applying findings.