Heart rate variability predicts 30-day all-cause mortality in intensive care units

Abstract

Background: Autonomic nervous function, as quantified by heart rate variability (HRV), has shown promise in predicting clinically important outcomes in the critical care setting; however, there is debate concerning its utility. HRV analysis was assessed as a practical tool for outcome prediction in two South African hospitals and compared with Acute Physiology and Chronic Health Evaluation II (APACHE II) scoring. Method: In a dual centre, prospective, observational cohort study of patients admitted to the intensive care units (ICU) of two hospitals in KwaZulu-Natal, South Africa frequency domain HRV parameters were explored as predictors of: all-cause mortality at 30 days after admission; ICU stay duration; the need for invasive ventilation; the need for inotrope/vasopressor therapy; and the need for renal replacement therapy. The predictive ability of HRV parameters against the APACHE II score for the study outcomes was also compared. Results: A total of 55 patients were included in the study. Very low frequency power (VLF) was shown to predict 30-day mortality in ICU (odds ratio 0.6; 95% confidence interval 0.396-0.911). When compared with APACHE II, VLF remained a significant predictor of outcome, suggesting that it adds a unique component of prediction. No HRV parameters were predictive for the other secondary outcomes. Conclusion: This study found that VLF independently predicted all-cause mortality at 30 days after ICU admission. VLF provided additional predictive ability above that of the APACHE II score. As suggested by this exploratory analysis larger multi-centre studies seem warranted.