HR SD5 predicts patient outcome as early as 12 hours: Utility of dense physiologic data captured in 1316 trauma patients

Abstract

Introduction. Our previous work demonstrated Dense Physiologic Data Capture ( DPDC) in the ICU, defined a new vital sign Cardiac Volatility Related Dysfunction ( CVRD) , and demonstrated CVRD predicts death during the hospital stay adjusting for age and injury severity score (ISS). We hypothesized that a more precise statistical definition of CVRD improves predictive power earlier in ICU stay, without adjusting for covariates. Methods. Approximately 120 million heart rate (HR) data points were prospectively collected and archived from 1316 trauma ICU patients, linked to TRACS outcome data, and de-identified. HR standard deviation was computed in each 5-min interval (HR SD5). HR SD5logistic regression (MATLAB) identified ranges predictive of death. The study group was randomly divided. CVRD (% time HR SD5 in predictive distribution ranges) models were developed on the first set ( N = 658) at 1, 2, 4, 6, 12, and 24 h and validated ( N = 658). Results. HR SD5 is bimodal, predicts death at low (0.3–0.6), and survival at high (2.1–2.2) ranges (Fig. 1). CVRD predicts trauma patient death as early as 12 h (Fig. 2). CVRD in a moving 1-h window is a simple graphic display technique (Fig. 3). Conclusions. DPDC allows calculation of Cardiac Volatility Related Dysfunction which (1) Independently predicts hospital death in trauma patients at 12 h (optimized at HR SD5 0.4–0.6); (2) Shows evolving differences by survival when viewed in a moving window; (3) May have implications for military and civilian triage.