Abstract 233: Integrating Heart Rate Variability for the Prediction of Cardiac Arrest in Critically Ill Patients Presenting With Acute Cardiovascular Conditions
Background: Current triage scoring systems depend on traditional vital signs and have limitations in predicting morbidity and mortality. We aimed to explore the utility of heart rate variability (HRV) for predicting cardiac arrest for patients presenting with acute cardiovascular conditions.
Methods and Results: ECG data was obtained from 425 cardiovascular patients who were monitored with LIFEPAK 12 defibrillator monitors. After extracting the data, filtering for noise reduction and isolating non-sinus beats, 16 HRV parameters were computed. These include time domain (RMSSD, nn50, etc), frequency domain (VLF, LF, HF, etc) and geometric parameters (TINN, triangular index). Patient outcome and vital signs were obtained from Emergency Department and hospital records. 32·5% of patients had Acute Myocardial Infarction, 36·0% had angina and 17·9% had cardiac dysrhythmias. 2·8% subsequently developed cardiac arrest within 72 hours and mortality rate was 2·6%. Overall, HRV parameters performed better than ‘traditional’ vital signs like pulse rate or blood pressure. When patients' age, vital signs and HRV parameters were combined in logistic regression models, these parameters predicted cardiac arrest within 72 hours with sensitivity of 66·7% (95% confidence interval [CI] 5·4%−88·7%), specificity of 98·1% (95% CI 96·1%−99·1%), positive predictive value of 50·0% (95%CI 25·5%−74·5%), negative predictive value of 99·0% (95%CI 97·3%−99·7%) and area under ROC curve (AUC) of 0·96 (95%CI 0·90–1·00).
Conclusion: We found the combination of vital signs, HRV parameters and age gave better prediction of outcomes than vital signs alone. Integrating HRV with vital signs and age shows potential as a clinical triage tool.
- © 2010 by American Heart Association, Inc.