More Related Content
Similar to HRV in trauma patients during prehospital transport (20)
HRV in trauma patients during prehospital transport
- 1. ORIGINAL ARTICLE
Heart Rate Variability as a Triage Tool in Patients With Trauma
During Prehospital Helicopter Transport
David R. King, MD, Michael P. Ogilvie, MD, MBA, Bruno M. T. Pereira, MD, FCCM, Yuchiao Chang, PhD,
Ronald J. Manning, RN, MSPH, Jeffrey A. Conner, RN, BSN, Carl I. Schulman, MD, MSPH, FACS,
Mark G. McKenney, MD, MBA, FACS, and Kenneth G. Proctor, PhD
Background: Prehospital triage of patients with trauma is routinely chal-
lenging, but more so in mass casualty situations and military operations. The
purpose of this study was to prospectively test whether heart rate variability
(HRV) could be used as a triage tool during helicopter transport of civilian
patients with trauma.
Methods: After institutional review board approval and waiver of informed
consent, 75 patients with trauma requiring prehospital helicopter transport to
our level I center (from December 2007 to November 2008) were prospec-
tively instrumented with a 2-Channel SEER Light recorder (GE Healthcare,
Milwaukee, WI). HRV was analyzed with a Mars Holter monitor system and
proprietary software. SDNN (standard deviation [SD] of the normal-to-
normal R-R interval), as an index of HRV, was correlated with prehospital
trauma triage criteria, base deficit, seriousness of injury, operative interven-
tions, outcome, and other data extracted from the patients’ medical records.
There were no interventions or medical decisions based on HRV. Data were
excluded only if there was measurement artifact or technical problems with
the recordings.
Results: The demographics were mean age 47 years, 63% men, 88% blunt,
25% traumatic brain injury, 9% mortality. Prehospital SDNN predicted
patients with base excess ՅϪ6, those defined as seriously injured and
benefiting from trauma center care, as well as patients requiring a life-saving
procedure in the operating room. No other available data, including prehos-
pital en-route vital signs, predicted any of these. The sensitivity, specificity,
positive predictive value, and negative predictive value were 80%, 75%,
33%, 96%, respectively, with and an overall accuracy of 76% for predicting
a life-saving intervention in the operating room.
Conclusions: This is the first demonstration that prehospital HRV (specifi-
cally SDNN) predicts base excess and operating room life-saving opportu-
nities. HRV triages and discriminates severely injured patients better than
routine trauma criteria or en-route prehospital vital signs. HRV may be a
useful civilian or military triage tool to avoid unnecessary helicopter evac-
uation for minimally injured patients. A prospective, randomized trial in a
larger patient population is indicated.
Key Words: Combat casualty care, Base deficit, Vital sign monitor.
(J Trauma. 2009;67: 436–440)
INTRODUCTION
Appropriate triage of traumatized patients remains a
challenge. Not all injured patients will require, or even benefit
from, care at a level I center and activation of a dedicated
trauma team. Rapidly deciding which patients will truly
benefit from this more costly trauma care delivery system is
difficult in many field situations. Additionally, unnecessary
helicopter transport of minimally injured patients is particu-
larly wasteful, cost consuming, and offers no additional care
to the patient. The overall global goal of trauma triage and
helicopter transport is to match the most aggressive and
available medical resources with the needs of the most
seriously injured patients (appropriate triage) while minimiz-
ing unintentional exclusion of a patient who would benefit
from a level I center, trauma team activation, and helicopter
transport (undertriage).
Currently, most prehospital trauma triage systems are
based on little data and generally reflect expert panel consen-
sus, a level III recommendation.1,2 Recent examination of
current trauma triage criteria demonstrate gross overtriage of
patients resulting in poorly allocated medical resources.3
Some subjective trauma triage criteria fail to reliably discrim-
inate patients benefiting from trauma center care (e.g. para-
medic judgment or high suspicion of injury by prehospital
providers), whereas others can predict operating room (OR)
use and emergent intensive care unit admission (e.g. hypo-
tensive and gunshot as wounding mechanism).4,5 Clearly,
there is an urgent need for more objective trauma triage
methods.
Heart rate variability (HRV) may represent an addi-
tional objective trauma triage tool.6–8 The purpose of this
study was to test the hypothesis that HRV could be used as an
objective trauma triage tool in the prehospital environment.
PATIENTS AND METHODS
After institutional review board approval and waiver of
informed consent, 95 patients with trauma requiring prehos-
pital helicopter transport to our level I center (from December
Submitted for publication December 29, 2008.
Accepted for publication May 1, 2009.
Copyright © 2009 by Lippincott Williams & Wilkins
From the Daughtry Family Department of Surgery (D.R.K., M.P.O., B.M.T.P.,
R.J.M., J.A.C., C.I.S., M.G.M.K., K.G.P.), Division of Trauma and Surgical
Critical Care, University of Miami Miller School of Medicine, Miami,
Florida; Department of Surgery (D.R.K., Y.C.), Division of Trauma, Emer-
gency Surgery, and Surgical Critical Care, Massachusetts General Hospital,
Harvard Medical School, Boston, Massachusetts; and GE Healthcare Diag-
nostic Cardiology (J.A.C.), Marquette, Wisconsin.
Presented at the 22nd Annual Meeting of the Eastern Association for the Surgery
of Trauma, January 13–17, 2008, Lake Buena Vista, Florida.
Supported in part by the Office of Naval Research grant N140610670.
Address for reprints: David R. King, MD, Division of Trauma, Emergency
Surgery, and Surgical Critical Care, Massachusetts General Hospital and
Harvard Medical School, 165 Cambridge Street, Suite 810, Boston, MA
02141; email: Dking3@partners.org.
DOI: 10.1097/TA.0b013e3181ad67de
436 The Journal of TRAUMA®
Injury, Infection, and Critical Care • Volume 67, Number 3, September 2009
- 2. 2007 to November 2008) were prospectively instrumented
with a 2-Channel SEER Light recorder (GE Healthcare,
Milwaukee, WI). HRV was analyzed with a Mars Holter
monitor system (GE Healthcare) and proprietary software.
SDNN (standard deviation [SD] of the normal-to-normal R-R
interval) is a determination of HRV derived from the time
domain of a standard electrocardiogram, primarily deter-
mined by measuring the randomness of the exact occurrence
of when one R wave follows a preceding R wave. SDNN, as
an index of HRV, was correlated with prehospital en-route
vital signs, injury severity score, base deficit, current
trauma triage criteria, outcome, and other data extracted
from the patients’ medical records. Data were excluded
only if there was measurement artifact or technical prob-
lems with the recordings (including recordings of an ab-
breviated length). There were no interventions or medical
decisions based on HRV.
To define and identify “seriously injured” patients,
three blinded practicing trauma surgeons at our level I center
reviewed each patient chart and final diagnoses to determine
whether, in retrospect, each patient benefited from care at a
level I center and trauma team activation. Any death was
considered seriously injured even if the outcome would not
have been altered by medical care. No information about
triage classification or prehospital condition was provided.
Each surgeon was then asked to determine whether each
patient was seriously injured in a binary decision pattern. A
patient was classified as seriously injured when two of the
three blinded trauma surgeons classified the patient similarly.
A similar dichotomization was done to determine whether a
patient who underwent surgery had a “life-saving” operation.
Several prehospital triage measures were evaluated as
predictors of interest, including enroute SDNN, heart rate
(HR), systolic blood pressure (SBP), Glasgow Coma Scale
(GCS), and the paramedic subjective judgment of “high
suspicion” for severe injury. Wilcoxon rank sum tests were
used to compare continuous predictors, whereas 2
tests were
used to compare dichotomized predictors between those with
and without serious injury. Test characteristics (specificity
and positive/negative predictive values) of continuous pre-
dictors were reported based on threshold values chosen for a
minimum of 80% of sensitivity. Multiple logistic regression
models were used to examine the marginal effect of each
predictor. To avoid overfitting the models, only predictors
with p Ͻ 0.15 were included in the final models. Predictive
risks were calculated from the models with and without
adding SDNN. Next, the effect of adding SDNN to the triage
criteria was investigated by examining the change in pre-
dicted risk stratified by outcome status. Subgroup analysis
was conducted among those without the comorbid conditions
and circumstances that may alter the reliability of SDNN.
RESULTS
A total of 95 patients were enrolled in the study.
Twenty were excluded; the reasons for exclusion were short
recording time of Ͻ200 QRS complexes (n ϭ 12), technical
problem with the recording produced by missing leads, ex-
treme artifact, or noise such that the recording could not be
interpreted meaningfully (n ϭ 2), or incomplete data from the
medical record or trauma registry (n ϭ 6). Data analysis was
conducted on 75 patients with complete HRV and trauma
registry data. Patient characteristics of the cohort are shown
in Table 1.
All patients met established standard prehospital
trauma triage criteria (Table 2). The majority of patients (n ϭ
46) were triaged to helicopter transport and trauma team
activation based on subjective paramedic judgment (i.e. “high
suspicion” for severe injury). The remainder met trauma
triage criteria based on GCS Ͻ13 (n ϭ 15), penetrating injury
to the head/neck/torso (n ϭ 9), two or more long bone
fractures (n ϭ 4), no radial pulse and HR greater than 120
bpm (n ϭ 1), and SBP Ͻ90 mm Hg (n ϭ 1). SDNN was
significantly correlated with HR (r ϭ Ϫ0.24, p ϭ 0.047), but
not with the paramedic subjective judgment, GCS, or blood
pressure.
Table 3 shows the relationship between each predictor
and three outcome variables (base excess [BE] ՅϪ6, seri-
ously injured condition, or life-saving interventions in the
operating room). SDNN was significantly lower among those
patients who had a BE ՅϪ6 (mean 17 msec versus 47 msec,
TABLE 1. Patient Characteristics
N ؍ 75
Age
Mean 47
SD 20
Sex
Female, N (%) 28 (37.3)
Male, N (%) 47 (62.7)
ISS
Mean 15
SD 16
Median 11
SDNN prehospital
Mean 42
SD 31
Median 34
HR prehospital
Mean 92
SD 21
Median 90
SBP prehospital
Mean 138
SD 28
Median 138
GCS prehospital 3–13, N (%) 16 (21.3)
GCS prehospital 14, N (%) 17 (22.7)
GCS prehospital 15, N (%) 40 (53.3)
CAD, N (%) 15 (20.0)
DBM, N (%) 6 (8.0)
BE ՅϪ6, N (%) 12 (16.7)
Seriously injured, N (%) 36 (48.0)
OR life saving opportunity, N (%) 10 (13.3)
ISS, Injury Severity Score; SBP, systolic blood pressure; CAD, coronary artery
disease; DBM, diabetes mellitus; BE, base excess; OR, operating room.
The Journal of TRAUMA®
Injury, Infection, and Critical Care • Volume 67, Number 3, September 2009 HRV in Patients With Trauma
© 2009 Lippincott Williams & Wilkins 437
- 3. p Ͻ 0.001), were classified as seriously injured (mean 28
msec versus 55 msec, p Ͻ 0.001), or underwent a life-saving
intervention in the OR (mean 24 msec versus 45 msec, p ϭ
0.016). Of the other routine prehospital triage measures, only
GCS was a significant predictor for BE ՅϪ6 (mean 13.7
versus 14.7, p ϭ 0.04) and seriously injured condition (mean
12.6 versus 13.9, p ϭ 0.037). HR, SBP, and high suspicion of
injury were not significant predictors of any of the three
outcomes.
If SDNN was used alone as a trauma triage test, the
overall performance (area under the receiver operating curve)
was 0.86 for predicting a BE ՅϪ6, 0.80 for predicting serious
injury, and 0.74 for predicting a life-saving intervention in the
OR, compared with 0.47 to 0.68 from using HR, SBP, or GCS
alone. When SDNN was dichotomized at values 24 msec, it has
a sensitivity of 80%, specificity of 75%, positive predictive value
of 33%, negative predictive value of 96%, and an overall
accuracy of 76% for predicting a life-saving intervention in the
OR. Similar test characteristics were observed for predicting
patients with a BE ՅϪ6 and patients determined to be seriously
injured. No other prehospital triage tests were performed as well.
The analysis was repeated excluding patients with known heart
disease or diabetes (conditions know to alter HRV), and the
results were very similar.
In this study population, 48% of patients were (retro-
spectively) triaged appropriately to helicopter transport and
trauma team activation, demonstrating a 52% over-triage rate.
If SDNN had been used as the sole trauma triage criteria (39
msec cutoff), then 46 patients (61%) would have met trauma
alert criteria: 29 of the 36 seriously injured patients (80%)
would have been appropriately captured, giving an over-
triage rate of 37% (17 of 46). Seven seriously injured patients
would have been miss-triaged by this test alone.
If SDNN had been used as the sole trauma triage
criteria with a different threshold (55 msec cutoff), then 58
patients (77%) would have met trauma alert criteria: 34 of 36
seriously injured patients (94%) would have been appropri-
ately captured, giving an over-triage rate of 41% (24 of 58).
Two seriously injured patients would have been miss-triaged
by this test alone.
In the multiple logistic regression models, the marginal
effect of SDNN was examined controlling for the other four
(HR, SBP, GCS, and subjective high suspicion of injury) crite-
ria. SDNN was the only significant predictor in the models for
all three outcomes. The adjusted odds ratio was 11.7 (95% CI,
2.1–65.4) for predicting a life-saving intervention in the OR
among those with SDNN Յ24 msec; was 5.8 (95% CI, 1.9–
17.1) for predicting a serious injury among those with SDNN
Յ39 msec, and was 14.8 (95% CI, 2.7–82.8) for predicting a BE
ՅϪ6 among those with SDNN Յ26 msec.
When comparing the models with and without SDNN,
adding SDNN to the model increased the predicted risk for
78% of the patients with a life-saving intervention in the
operating room and reduced the predicted risk for 52% of the
patients without a life-saving intervention in the OR; in-
creased the predicted risk for 89% of the patients identified
as seriously injured and reduced the predicted risk for 69%
of the patients not identified as seriously injured; and
increased the predicted risk for 82% of the patients with
BE ՅϪ6 and reduced the predicted risk for 76% of the
patients with BE ϾϪ6.
TABLE 2. Prehospital Trauma Triage Criteria
Category 1 Category 2
Age Age Ͼ55 yr
Airway Active airway support beyond supplemental O2 Respiratory rate Ͼ30 beats/min
Consciousness BMR 4, or paralysis, or suspicion of spinal cord injury, or loss of
sensation, or GCS Ͻ13
BMR 5
Circulation No radial pulse and sustained heart rate Ͼ120 beats/min, or SBP
Ͻ90 mm Hg
Sustained heart rate Ͼ120 beats/min
Fracture 2 or more long bone fractures Any long bone fracture sustained in a MVC or fall Ͼ10 feet
Cutaneous 2nd or 3rd degree burns to 15% TBSA, or amputation at or
proximal to wrist or ankle, or penetrating injury to head, neck,
or torso
Major degloving injury, or major flap avulsion Ͼ5 inches, or
GSW to the extremity
Mechanism Ejection from motor vehicle, or steering wheel deformity
resulting from driver impact
Other High Index of Suspicion
Any Category 1 (or 2 Category 2) meets trauma alert criteria.
BMR, best motor response of the GCS; SBP, systolic blood pressure; MVC, motor vehicle crash; TBSA, total body surface area; GSW, gunshot wound.
TABLE 3. p Value, Relationship Between Predictors and
Outcome
Variable Test BE, p
Serious
Injury, p
OR Life
Saving, p
SDNN Wilcoxon 0.0002 Ͻ0.0001 0.016
t Ͻ0.0001 Ͻ0.0001 0.002
HR Wilcoxon 0.49 0.65 0.57
t 0.35 1.00 0.66
SBP Wilcoxon 0.56 0.62 0.38
t 0.75 0.77 0.24
GCS Wilcoxon 0.040 0.037 0.17
t 0.11 0.095 0.20
High
suspicion
Chi square 0.34 1.00 0.31
BE, base excess; HR, heart rate; SBP, systolic blood pressure; high suspicion, high
suspicion of injury as determined by paramedics.
King et al. The Journal of TRAUMA®
Injury, Infection, and Critical Care • Volume 67, Number 3, September 2009
© 2009 Lippincott Williams & Wilkins438
- 4. DISCUSSION
This study demonstrates the usefulness of HRV (spe-
cifically SDNN) as a prehospital trauma triage tool. This
technology was more accurate in predicting patients who
would benefit from trauma center care and trauma team
activation than any other of our currently used trauma triage
criteria. This could offer significant advantages in trauma
resource allocation on the battlefield or during mass casualty
scenarios.
HRV is generally regarded as a nonspecific indicator of
health and reflects autonomic dysfunction. Loss of variability,
or regularization of HR, is considered a global index of poor
health.9 Some chronic disease, such as ischemic heart disease
and diabetes, are known to reduce HRV and therefore repre-
sent potential confounding comorbidities.10,11 Our initial
analysis, however, was conducted on all patients regardless of
comorbidites. We believe this strengthens the utility of this
technology. Our analysis of the subpopulation of patients
with confounding diseases (heart disease and diabetes) dem-
onstrated an effect of these comorbid conditions, but presence
of these conditions did not significantly alter the predictive
usefulness of HRV as a triage tool. To truly determine the
effects of comorbid conditions on HRV in the setting of
trauma, a study should be designed to examine HRV in this
specific population. Our study was not powered or designed
to discriminate difference between otherwise healthy trauma-
tized patients and those with known comorbid conditions.
This study has several limitations. First, for a variety of
logistic reasons, we only enrolled patients undergoing heli-
copter transport after meeting existing prehospital trauma
triage criteria. It is our general impression that sicker patients
are selected for helicopter evacuation, however we cannot
confirm this. Consequently, this study group may represent a
different patient population than patients transported by
ground. Second, our study population was comprised of
Ͻ100 patients. This investigation needs to be repeated in a
much larger population. Third, HRV is affected by multiple
disease states. It is possible that profound alterations in HRV
may occur is nontraumatized patients with certain chronic
diseases who later are involved in a traumatic incident. This
may result in over-triage of these chronically diseased pa-
tients as sick patients with trauma. Only a larger study of this
technology can demonstrate the extent and prevalence of this
occurrence. Fourth, a very large proportion of our patient
population was triaged as high suspicion of injury by para-
medics. It is possible that some of these patients with trauma
met alternative standard trauma triage criteria that were not
captured by our data collection process. It is generally be-
lieved that prehospital providers commonly use high suspi-
cion of injury as a “catch-all” for many patients with trauma
to avoid having to articulate the more complex trauma triage
criteria. It is possible that this may confound the results.
Although this study was conducted prospectively, anal-
ysis of the HRV data was not available in real time. After the
recordings were made, the software analysis of the recording
took anywhere between 1 and 10 days. This is a significant
limitation of the practical usefulness of this technology. The
mathematical manipulations and analysis of the recording are
not currently automated on a small, durable, portable, read-
able device, and this would have to occur in an integrated
fashion in real time for this to truly become a useful triage
tool. Additionally, the method by which other triage infor-
mation should be integrated into SDNN triage remains to be
determined. It is clear that SDNN triage should not stand
alone, and should be integrated with all other available
prehospital information to create a maximally predictive
model for identifying patients who stand to benefit the most
from trauma team activation.8,9
Changes in HRV are an accepted method of assessing
autonomic dysfunction in patients in several pathologic
states, with and without structural heart disease.12–14 Absence
of HRV predicts brain death15–18 and low HRV correlates
with increased mortality and morbidity after trauma,19–21
increased intracranial pressure and decreased cerebral perfu-
sion pressure.22,23 Recently, it was suggested that HRV is a
“new vital sign” that could be used as a trauma triage tool24,25
and that reduced HRV predicts mortality in the prehospital
setting26 or within the first 24 hr of trauma ICU stay.27,28
Recently, we developed a simple HRV-based algorithm with
improved specificity and efficiency for predicting outcome in
hospitalized patients with trauma.8,9 This study attempts to
extend our knowledge of HRV characterize the prehospital
usefulness of HRV and as a trauma triage tool.
In conclusion, we have demonstrated the usefulness of
HRV as a prehospital trauma triage tool during helicopter
transport to a level I trauma center. This technology seems to be
better than any single standard traditional trauma triage criteria.
The addition of this technology, in an integrated fashion to
improve overall prediction of severe injury, may improve med-
ical resource utilization, especially in austere environments. This
technology must be tested in a much larger population.
ACKNOWLEDGMENTS
We appreciate the efforts of Miami-Dade Emergency
Medical Services who instrumented all patients during helicop-
ter transport. We also recognize the efforts of Paul J. McMahon
who assisted with portions of the data collection. We also
appreciate the contributions of several student volunteers who
assisted with data collection, especially Hiamine Maas and
Rajesh Reddy. We are especially indebted to three colleagues
(George C. Velmahos, Hasan B. Alam, and Marc A. deMoya)
for their suggestions, comments, and insight.
REFERENCES
1. Tinkoff GH, O’Connor RE. Validation of new trauma triage rules for
trauma attending response to the emergency department. J Trauma.
2002;52:1153–1158, discussion 1158–1159.
2. Leach SR, Swor RA, Jackson RE, Fringer RC, Bonfiglio AX. Do
outcome measures for trauma triage agree? Prehosp Emerg Care.
2008;12:467–469.
3. Kann SH, Hougaard K, Christensen EF. Evaluation of pre-hospital
trauma triage criteria: a prospective study at a Danish level I trauma
centre. Acta Anaesthesiol Scand. 2007;51:1172–1177.
4. Henry MC, Hollander JE, Alicandro JM, Cassara G, O’Malley S, Thode
HC Jr. Incremental benefit of individual American College of Surgeons
trauma triage criteria. Acad Emerg Med. 1996;3:992–1000.
5. Tiamfook-Morgan TO, Kociszewski C, Browne C, Barclay D, Wedel
SK, Thomas SH. Helicopter scene response: regional variation in com-
The Journal of TRAUMA®
Injury, Infection, and Critical Care • Volume 67, Number 3, September 2009 HRV in Patients With Trauma
© 2009 Lippincott Williams & Wilkins 439
- 5. pliance with air medical triage guidelines. Prehosp Emerg Care. 2008;
12:443–450.
6. Cancio LC, Batchinsky AI, Salinas J, et al. Heart-rate complexity for
prediction of prehospital lifesaving interventions in trauma patients.
J Trauma. 2008;65:813–819.
7. Batchinsky AI, Cancio LC, Salinas J, et al. Prehospital loss of R-to-R
interval complexity is associated with mortality in trauma patients.
J Trauma. 2007;63:512–518.
8. Proctor KG, Atapattu SA, Duncan RC. Heart rate variability index in
trauma patients. J Trauma. 2007;63:33–43.
9. Norris PR, Anderson SM, Jenkins JM, Williams AE, Morris JA Jr. Heart
rate multiscale entropy at three hours predicts hospital mortality in 3,154
trauma patients. Shock. 2008;30:17–22.
10. Maule S, Rabbia F, Perni V, et al. Prolonged QT interval and reduced
heart rate variability in patients with uncomplicated essential hyperten-
sion. Hypertens Res. 2008;31:2003–2010.
11. Ko SH, Song KH, Park SA, et al. Cardiovascular autonomic dysfunction
predicts acute ischaemic stroke in patients with Type 2 diabetes mellitus:
a 7-year follow-up study. Diabet Med. 2008;25:1171–1177.
12. Buchman TG, Stein PK, Goldstein B. Heart rate variability in critical
illness and critical care. Curr Opin Crit Care. 2002;8:311–315.
13. Stein PK, Schmieg RE Jr, El-Fouly A, Domitrovich PP, Buchman TG.
Association between heart rate variability recorded on postoperative day
1 and length of stay in abdominal aortic surgery patients. Crit Care Med.
2001;29:1738–1743.
14. Godin PJ, Buchman TG. Uncoupling of biological oscillators: a com-
plementary hypothesis concerning the pathogenesis of multiple organ
dysfunction syndrome. Crit Care Med. 1996;24:1107–1116.
15. Goldstein B, Kempski MH, DeKing D, et al. Autonomic control
of heart rate after brain injury in children. Crit Care Med. 1996;24:
234–240.
16. Goldstein B, Toweill D, Lai S, Sonnenthal K, Kimberly B. Uncoupling
of the autonomic and cardiovascular systems in acute brain injury. Am J
Physiol. 1998;275:R1287–R1292.
17. Rapenne T, Moreau D, Lenfant F, et al. Could heart rate variability
predict outcome in patients with severe head injury? A pilot study.
J Neurosurg Anesthesiol. 2001;13:260–268.
18. Rapenne T, Moreau D, Lenfant F, Boggio V, Cottin Y, Freysz M. Could
heart rate variability analysis become an early predictor of imminent
brain death? A pilot study. Anesth Analg. 2000;91:329–336.
19. King ML, Lichtman SW, Seliger G, Ehert FA, Steinberg JS. Heart-rate
variability in chronic traumatic brain injury. Brain Inj. 1997;11:445–453.
20. Grogan EL, Norris PR, Speroff T, et al. Volatility: a new vital sign
identified using a novel bedside monitoring strategy. J Trauma. 2005;
58:7–12, discussion 12–14.
21. Grogan EL, Morris JA Jr, Norris PR, et al. Reduced heart rate volatility:
an early predictor of death in trauma patients. Ann Surg. 2004;240:547–
554, discussion 554–556.
22. Winchell RJ, Hoyt DB. Analysis of heart-rate variability: a noninvasive
predictor of death and poor outcome in patients with severe head injury.
J Trauma. 1997;43:927–933.
23. Biswas AK, Scott WA, Sommerauer JF, Luckett PM. Heart rate vari-
ability after acute traumatic brain injury in children. Crit Care Med.
2000;28:3907–3912.
24. Morris JA Jr, Norris PR. Role of reduced heart rate volatility in
predicting death in trauma patients. Adv Surg. 2005;39:77–96.
25. Norris PR, Morris JA Jr, Ozdas A, Grogan EL, Williams AE. Heart rate
variability predicts trauma patient outcome as early as 12 h: implications
for military and civilian triage. J Surg Res. 2005;129:122–128.
26. Cooke WH, Salinas J, Convertino VA, et al. Heart rate variability and its
association with mortality in prehospital trauma patients. J Trauma.2006;
60:363–370, discussion 370.
27. Norris PR, Ozdas A, Cao H, et al. Cardiac uncoupling and heart rate
variability stratify ICU patients by mortality: a study of 2088 trauma
patients. Ann Surg. 2006;243:804–812, discussion 812–814.
28. Morris JA Jr, Norris PR, Ozdas A, et al. Reduced heart rate variability:
an indicator of cardiac uncoupling and diminished physiologic reserve in
1,425 trauma patients. J Trauma. 2006;60:1165–1174.
King et al. The Journal of TRAUMA®
Injury, Infection, and Critical Care • Volume 67, Number 3, September 2009
© 2009 Lippincott Williams & Wilkins440