Associations Between Heart Rate Variability and Need for Lifesaving Intervention in a Large Helicopter EMS Service.

Background: Heart rate variability (HRV) measures give insight into the autonomic regulation of cardiac function in healthy and critically ill patients. The ease and predictive potential of HRV measures may be valuable in optimizing prehospital triage. In this retrospective study, we hypothesized that HRV measures, specifically sample entropy, measured early in emergency transport would predict the need for a prehospital lifesaving intervention (LSI) in a large, heterogenous cohort of critically ill patients.

Methods: We obtained patient records from a large helicopter critical care transport service. Continuous electrocardiogram (ECG) data were processed and screened for signal artifacts. Time, frequency, and complexity domain HRV measures were calculated and averaged. Multivariable logistic regression models and t-tests were constructed to establish associations between selected HRV measures and the need for a prehospital LSI, adjusting for demography and case characteristics including patient age, sex, scene run, and trauma/non-trauma. A suite of machine learning algorithms was applied to optimize prediction of outcome measures.

Results: A total of 4,521 cases were included for analysis. Of all patients, 68.8% of patients received prehospital LSI. Sample entropy, as well as other HRV measures, was associated with reception of prehospital LSI (OR 0.50 (95% CI [0.43,0.59])). Gradient boosting and random forest algorithms showed the best performance in predicting LSI (AUROC scores = 0.78 - 0.79).

Conclusions: Certain HRV measures are associated with prehospital LSI. Subsequent studies should focus on clinical utility and actionable thresholds for triage and initiation of LSIs.