Influencing factors analysis and prediction model development for ROSC in adult non-traumatic OHCA: a retrospective study based on pre-hospital emergency settings

Background Cardiac arrest is a major global public health problem. This study aimed to explore factors influencing return of spontaneous circulation (ROSC) and develop a prediction model for out-of-hospital cardiac arrest (OHCA) patients.Methods We retrospectively collected electronic medical records of adult non-traumatic OHCA patients at Tianjin Emergency Medical Center from January 1, 2023 to December 31, 2024. The primary outcome was sustained ROSC up to emergency department arrival and handover to receiving-hospital staff. Multivariable mixed-effects logistic regression with clinician ID as random intercept adjusted clustering bias; predictors with p < 0.05 formed the nomogram. Patients were randomly split 7:3 for training and internal testing. Discrimination was assessed by ROC-AUC (acceptable threshold >0.7). We performed 1000 bootstrap resampling to correct C-index against overfitting. Calibration curves and decision curve analysis (DCA) further evaluated predictive consistency and net clinical benefit.Results Among 9,386 included cases, 678 (7.2%) achieved ROSC. Mixed-effects regression identified independent ROSC predictors: public location (aOR = 1.285, 95% CI:1.038–1.592), 06:00–12:00 onset (aOR = 1.308, 95% CI:1.018–1.682), prolonged emergency response time (aOR = 0.971, 95% CI:0.945–0.999), extended onset-to-ambulance intervals (5–10 min: aOR = 0.758;10–30 min: aOR = 0.578;>30 min:aOR = 0.498 vs ≤5 min), no OHCA occurred at emergency call (aOR = 1.540, 95% CI:1.048–2.262), male attending physicians (aOR = 1.255, 95% CI:1.077–1.464), longer physician practice years (aOR = 1.043, 95%CI: 1.022–1.065), intermediate physician title (aOR = 1.236, 95% CI:1.047–1.459), shockable initial rhythm (aOR = 2.850, 95% CI:2.390–3.397), endotracheal intubation (aOR = 2.294, 95% CI:1.936–2.718), epinephrine >5 mg (aOR = 0.765,95% CI:0.626–0.935). The nomogram AUC reached 0.775 (95% CI:0.756–0.794) in the training set and 0.775 (95% CI:0.745–0.805) in the test set; 1000 bootstrap internal validation gave an optimism-corrected C-index of 0.7679 with minimal optimism of 0.0071, showing negligible overfitting, good calibration and favorable net clinical benefit via DCA.Conclusion This study identified multiple factors significantly associated with ROSC in OHCA patients; the constructed model showed good predictive value.