Heart rate variability and its association with mortality in prehospital trauma patients

Abstract

Background: Accurate prehospital triage of trauma patients is difficult, especially in mass casualty situations. Accordingly, the U.S. Military has initiated a program directed toward improving noninvasive prehospital triage algorithms based on available physiologic data. The purpose of this study was to assess heart rate variability and its association with mortality in prehospital trauma patients.

Methods: Trauma patients without significant head injury requiring helicopter transport were identified from a retrospective research database. An equal number, unmatched sample of patients who lived were compared with those who died (n = 15 per group). All patients were transported to a single Level I urban trauma center. The primary independent variable was mortality. Patients with Abbreviated Injury Scale head scores >2 were excluded from the analysis, so that the effects seen were based on hemorrhagic shock. Age, sex, Glasgow Coma Scale score (GCS), blood pressure, pulse pressure, pulse, intubation rate, SpO2, mechanism of injury, transport time, and time of death after admission were recorded. R-waves from the first available 120 seconds of usable data were detected from normal electrocardiograms and heart rate variability was assessed.

Results: Patients who died demonstrated a lower GCS (7.9 +/- 1.4 versus 14.4 +/- 0.2; p = 0.0001) and higher intubation rate (53% of patients who died versus 0% patients who lived). Pulse rate, arterial pressure, and SpO2 were not distinguishable statistically between groups (p = 0.08), but pulse pressure was lower in patients who died (39 +/- 3 versus 50 +/- 2 mm Hg; p = 0.01). Compared with patients who lived, those who died had lower normalized low-frequency (LF) power (42 +/- 6 versus 62 +/- 4 LFnu; p = 0.009), higher high-frequency (HF) power (42 +/- 3 versus 32 +/- 3 HFnu; p = 0.04) and higher HF-to-LF ratio (144 +/- 30 versus 62 +/- 11nu; p = 0.01). With absolute HF/LF adjusted for GCS, the intergroup variance accounted for by HF/LF was reduced to 6% (p = 0.16).

Conclusions: Analysis of heart rate variability provides insight into adequacy of autonomic compensation to severe trauma. In our cohort of trauma patients, low pulse pressures coupled with relatively higher parasympathetic than sympathetic modulation characterized and separated patients who died versus patients who survived traumatic injuries when standard physiologic measurements are not different. These data do not suggest advantages of heart rate variability analysis over GCS scores, but suggest future possibilities for remote noninvasive triage of casualties when GCS scores are unattainable.