Development and internal validation of an oxygen saturation-based prehospital cardiac risk assessment index (P-CRAI) for predicting 30-day mortality in acute myocardial infarction: a multicentre registry study from the Mie ACS Registry, Japan

Objectives Early risk stratification of patients with acute myocardial infarction (AMI) in the prehospital setting remains challenging. The aim of this study was to develop and internally validate a Prehospital Cardiac Risk Assessment Index (P-CRAI) based on peripheral oxygen saturation (SpO₂) for predicting 30-day mortality in patients with AMI. Design Retrospective analysis of a prospective multicentre registry. Setting 17 cardiovascular centres participating in the Mie ACS Registry in Japan. Participants A total of 3584 consecutive patients with AMI who were transported directly by ambulance between 2013 and 2023 were included. Primary outcome measure The primary outcome was all-cause mortality within 30-day. Candidate predictors were age, sex, heart rate, systolic blood pressure and peripheral SpO₂ at first medical contact. P-CRAI was developed by adding SpO₂ and sex to the Prehospital-Simple Risk Index (pre-SRI). Model performance was assessed by discrimination, calibration, overall prediction accuracy and decision curve analysis, with bootstrap internal validation. Results The P-CRAI was calculated as follows: (heart rate×[age/10]²)/systolic blood pressure+1.1(100−SpO₂)−1.5 (male sex). The model demonstrated good discrimination for predicting 30-day mortality (area under the curve (AUC) 0.830, 95% CI 0.817 to 0.842). Discrimination was significantly improved compared with the pre-SRI (AUC 0.805, p<0.001) and was comparable to that of the Global Registry of Acute Coronary Events score (AUC 0.846, p=0.28). Calibration analysis showed an excellent agreement between the predicted and observed mortality rates, with a mean absolute error of 0.018. Harrell’s C-index was 0.821 and the Brier score was 0.050. Decision curve analysis showed greater net benefit of P-CRAI than the pre-SRI across clinically relevant threshold probabilities. Kaplan-Meier analysis demonstrated significant graded separation of 30-day survival across increasing P-CRAI quartiles (log-rank p<0.001). P-CRAI remained independently associated with 30-day mortality after multivariable analysis (adjusted HR 6.37, 95% CI 4.35 to 9.32; p<0.001). Conclusions P-CRAI is a simple prehospital model using variables available at first medical contact that showed good discrimination for 30-day mortality in AMI and may support triage and early transfer of high-risk patients to specialised cardiovascular centres. External validation is required before clinical implementation. Trial registration number UMIN000036020.