Development and validation of a comprehensive early risk prediction model for patients with undifferentiated acute chest pain

Aims

Existing risk scores for undifferentiated chest pain focus on excluding coronary events and do not represent a comprehensive risk assessment if an alternate serious diagnosis is present. This study aimed to develop and validate an all-inclusive risk prediction model among patients with undifferentiated chest pain.

Methods

We developed and validated a multivariable logistic regression model for a composite measure of early all-inclusive risk (defined as hospital admission excluding a discharge diagnosis of non-specific pain, 30-day all-cause mortality, or 30-day myocardial infarction [MI]) among adults assessed by emergency medical services (EMS) for non-traumatic chest pain using a large population-based cohort (January 2015 to June 2019). The cohort was randomly divided into development (146,507 patients [70%]) and validation (62,788 patients [30%]) cohorts.

Results

The composite outcome occurred in 28.4%, comprising hospital admission in 27.7%, mortality within 30-days in 1.8%, and MI within 30-days in 0.4%. The Early Chest pain Admission, MI, and Mortality (ECAMM) risk model was developed, demonstrating good discrimination in the development (C-statistic 0.775, 95% CI 0.772-0.777) and validation cohorts (C-statistic 0.765, 95% CI 0.761-0.769) with excellent calibration. Discriminatory performance for the composite outcome and individual components was higher than existing scores commonly used in undifferentiated chest pain risk stratification.

Conclusions

The ECAMM risk score model can be used as an all-inclusive risk stratification assessment of patients with non-traumatic chest pain without the limitation of a single diagnostic outcome. This model could be clinically useful to help guide decisions surrounding the need for non-coronary investigations and safety of early discharge.