Journal of Pediatric Surgery (2005) 40, 1957–1963

www.elsevier.com/locate/jpedsurg

Continuous noninvasive monitoring of cardiac performance

and tissue perfusion in pediatric trauma patientsB,BB

Matthew Martina,*, Carlos Browna, David Bayarda, Demetrios Demetriadesa,Ali Salima, Ryan Gertza, Kenneth Azarowb, Charles C.J. Woa, William Shoemakera

aDivision of Trauma and Surgical Critical Care, Los Angeles County Hospital + USC Medical Center,

Los Angeles, CA 90033, USAbDepartment of Surgery, Madigan Army Medical Center, Fort Lewis, WA 98431, USA

0022-3468/$ – see front matter D 2005

doi:10.1016/j.jpedsurg.2005.08.017

Presented at the 38th Annual MeetinB This study was supported in partBB The award DAMD 17-01-2-0070

necessarily reflect the position or the po

T Corresponding author. Tel.: +1 32

E-mail address: docmartin2@yahoo.

Index words:Trauma;

Thoracic bioimpedance;

Transcutaneous oxygen;

Noninvasive monitoring;

Survival prediction

Abstract

Purpose: The aim of this study was to assess the accuracy of a continuous survival probability

prediction using noninvasive measures of cardiac performance and tissue perfusion in severely injured

pediatric patients.

Methods: Review of all patients entered into a prospective noninvasive monitoring protocol. Cardiac

index (CI) was measured using a thoracic bioimpedance device and tissue perfusion was assessed

by transcutaneous carbon dioxide (Ptcco2) tension and oxygen tension indexed to the fraction of inspired

oxygen (Ptco2/Fio2). Survival probability (SP) was continuously calculated using a stochastic

analysis program.

Results:Therewere 45 patients with a total of 953 data sets. Themean agewas 11 years (range, 1-16 years)

with a mean Injury Severity Score of 24 (F16). There was no difference between survivors (n = 32) and

nonsurvivors (n = 13) at study entry for heart rate, blood pressure, CI, or pulse oximetry (all P N .05).

However, survivors demonstrated higher Ptcco2 (45 vs 35), higher Ptco2/Fio2 (236 vs 156), and higher

predicted SP (89% vs 62%) compared with nonsurvivors at study entry and throughout the monitoring

period (all P b .01). For the entire data set, the strongest independent predictors of survival were Ptco2/

Fio2 and SP. The area under the receiver operating characteristic curve for mortality prediction was

0.83 for SP and 0.71 for Ptco2/Fio2, compared with 0.6 for heart rate, 0.51 for blood pressure, and 0.53

for CI. Similar hemodynamic patterns were observed for all injury patterns with the exception of those

with severe brain injury.

Conclusions:Thoracic bioimpedance and transcutaneous monitoring give critical real-time hemodynamic

and tissue perfusion data that can provide early identification of pathologic flow patterns and accurately

predict survival.

D 2005 Elsevier Inc. All rights reserved.

Elsevier Inc. All rights reserved.

g of the Pacific Association of Pediatric Surgeons, May 22-26, 2005, Vancouver, Canada.

by grants RR-11526, GM-65619, and DOD BAA99-1 from the National Institutes of Health, Bethesda, Md.

is by the US Army Medical Research Acquisition Activity, Fort Detrick, Md. The content of the information does not

licy of the Government, and no official endorsement should be inferred.

3 226 8112; fax: +1 323 226 8116.

com (M. Martin).

M. Martin et al.1958

Restoration of optimal hemodynamics and end-organ

perfusion are key components in the resuscitation and

management of any critically ill or injured patient. Normal-

ization of routine parameters such as heart rate (HR), blood

pressure, pulse oximetry, and urine output are inadequate end

points in these complex patients, who may have impending

or ongoing shock in the face of bstableQ vital signs [1]. More

detailed and invasive monitoring techniques such as pulmo-

nary artery catheterization and arterial blood gas analysis are

frequently used to provide prognostic information and to

assess cardiorespiratory performance, as well as oxygen

delivery and utilization. However, this information is often

not obtained until later in the disease course, when the

patient is brought to an intensive care unit (ICU). Although

there is great debate about the utility of invasive monitoring

[2,3], several trials have demonstrated improved outcomes

when these techniques are used to guide therapy early in the

disease course [4-9].

The benefits of invasive monitoring must be balanced

against the limitations, which include vascular or cardiac

injury, infection, thromboembolism, pneumothorax, and the

need for specialized equipment and personnel that are often

only available in the ICU setting [3,10]. In the pediatric

patient, vascular access may be particularly difficult because

of the small caliber and increased reactivity of target blood

vessels and the limited availability of appropriately sized

catheters and monitoring equipment [11]. In addition, the

utility and interpretation of the data provided by invasive

monitors, such as pulmonary artery catheters, in the pediatric

population have been questioned and continue to be sources

of controversy [12].

Continuous noninvasive monitoring of routine physiologic

variables such as HR, blood pressure, and pulse oximetry has

become standard practice for both adult and pediatric ICUs. In

addition to these routine parameters, more modern noninva-

sive monitoring technology has been developed that can

accurately and reliably assess ventilation, cardiac perfor-

mance, arterial and tissue oxygenation, and end-organ

perfusion without the risks associated with invasive monitor

placement [13-18]. These data could potentially be used to

stratify disease or injury severity, make therapeutic decisions,

and assess the response to interventions and to provide

continuous prognostic information. Recently, Bayard et al

[19-21] developed a mathematical method that used a large

database of noninvasively monitored hemodynamic variables

to provide online real-time displays, outcome prediction, and

therapeutic decision support for newly admitted acute

emergency patients. Although the hemodynamic profiles

and survival probability analysis provided by this noninvasive

monitoring system have been well characterized in adults

[22,23], there are no reports describing the applicability of this

system to the pediatric trauma patient. The purpose of this

report is to describe the hemodynamic patterns and assess the

accuracy of a continuous survival probability prediction using

noninvasive measures of cardiac performance and tissue

perfusion in severely injured pediatric patients.

1. Materials and methods

1.1. Clinical series

We retrospectively identified 45 noninvasively monitored

pediatric trauma patients (age b16 years) with major blunt or

penetrating injuries and significant risk of mortality or

morbidity within 1 hour of emergency department (ED)

admission. Noninvasive hemodynamic monitoring was

begun in the ED or immediately upon arrival to the ICU

and the patients were followed to the radiology suite when

indicated, to the operating room (OR), and then to the ICU.

Patients were selected for study entry at the discretion of

the investigators based on the injury severity and availability

of monitoring equipment. The time of monitoring, time

of operations, times of ICU admission, and hospital dis-

charge or death were recorded relative to time elapsed after

ED admission.

In addition, the following data were included in the data-

base: age, sex, presence of sepsis, Glasgow coma score

(GCS), Injury Severity Score (ISS), the primary bodily in-

juries, covariates, hemodynamic patterns by invasive and

noninvasive methods, organ failures, other complications,

hospital days, ICU days, and hospital outcome. The proposed

computerized program is an information system that is not

directly connected to the patient as a closed loop system. It is

comparable to monitoring the vital signs and the provision of

online real-time displays of data and calculations to inform

the attending without interfering with the attending’s respon-

sibilities or capacities to render patient care. The institutional

review board approved the protocol.

1.2. Noninvasive hemodynamic monitoring

Hemodynamic values were evaluated by continuous

online real-time display of noninvasive monitoring of

cardiac, respiratory, and tissue perfusion functions. The data

were downloaded every 30 seconds, averaged over 5-minute

intervals, and entered into the database. When consistent

hemodynamic patterns were demonstrated, the data were

averaged over 15-minute periods for presentation. The

noninvasive hemodynamic monitoring was continued until

the patients were stable or died.

1.2.1. Cardiac output and cardiac index

A thoracic bioelectric impedance device (IQ, model 101,

Noninvasive Medical Technologies LLC, Auburn Hills,

Mich) was applied upon study entry. The noninvasive

disposable prewired hydrogen electrodes were positioned

on the skin, and three electrocardiogram leads were placed on

the precordium and each shoulder as has been previously

described [22,24,25]. A 100-kHz, 4-mA alternating current

was passed through the patient’s thorax by the outer pairs of

electrodes and the voltage was sensed by the inner pairs of

electrodes that captured the baseline impedance (Zo), the first

derivative of the impedance waveform (dZ/dt), and the

electrocardiogram. Previous studies have documented satis-

Continuous noninvasive monitoring in pediatric trauma patients 1959

factory correlations between thermodilution and bioimpe-

dance cardiac output values for trauma patients in the ED,

OR, and ICU conditions [22,23].

1.2.2. Pulse oximetry

Routine pulse oximetry (Nellcor, Pleasanton, Calif) was

used to assess continuously arterial oxygen saturation

(Sapo2). Values were observed continuously and recorded

along with cardiac index (CI) measurements. Sudden

changes in these values were confirmed by standard blood

gas analysis.

1.2.3. Transcutaneous oxygen and carbon dioxide

tensions

Conventional transcutaneous oxygen tension measure-

ments (Ptco2) were indexed to the fraction of inspired

oxygen concentration (Ptco2/Fio2) and continuously mon-

itored throughout the observation period. Ptco2 technology

uses the Clark polarographic oxygen electrode routinely used

in standard blood gas measurements [26-30]. Ptco2 tensions

were measured on an upper extremity skin surface heated to

448C to increase diffusion of oxygen across the stratum

corneum [27-30]. The Severinghaus electrode was used to

continually monitor transcutaneous CO2 tension [28]. A

limitation of the transcutaneous methods is that the thermal

environment should be reasonably constant. Marked changes

in room temperature from drafts or open windows should be

avoided and the electrode changed to a nearby site and

recalibrated at 4-hour intervals. Calibration of the system was

performed at the initiation of monitoring and checked every

several hours. The accuracy of the transcutaneous variables

was assessed by comparison with arterial blood gas analysis

and end-tidal carbon dioxide measurement in select patients.

1.3. Mathematical details: the search and

display program

Bayard et al [20] developed a search and display (sto-

chastic analysis) program to determine individual patients’

survival probabilities (SP) from a database of patients with

similar clinical-hemodynamic bstates,Q defined by the prima-

ry diagnosis, covariates, and hemodynamic variables. By

bsimilarQ is meant a group of patients with the same diagnosis

and covariates and with similar hemodynamic patterns to the

newly admitted patient under study. These are referred to as

bnearest neighbors.Q Mathematically, the stochastic analysis

was motivated by methods of machine learning [31,32]

and methods of dynamic programming for stochastic con-

trol [20,21].

1.3.1. Control input definition

The bcontrol inputQ (mode of therapy) will be chosen

from a finite set of control inputs that can be applied to

the system.

1.3.2. System dynamics

Both the clinical covariates and process noise help to

explain the variability of patient responses of the database.

The covariates help to distinguish gross differences in

responses of patients with major differences in the nature

of their disorders. Process noise helps to explain small

differences between patients with the same diagnosis and

covariates but different responses to the same therapy. It is a

measure of unmodeled dynamics, or intraindividual vari-

ability, from other sources of variability in the system [19].

1.4. Probability of survival

A patient’s survival probability (SP) for a given state x is

denoted by S(x), which is calculated by first extracting the

40 or more nearest neighbor states of patients having the

same diagnosis and covariates as well as hemodynamic

values that are closest to the given patients’ values. The SP

is then calculated as the fraction of these nearest neighbors

that survived. The prospectively determined predicted

outcome of each patient was validated by the patient’s

actual hospital outcome determined at the end of the study.

1.5. Statistical analyses

The survivors’ and nonsurvivors’ deficits of mean arterial

pressure (MAP), cardiac output, Sapo2, and transcutaneous

O2 were calculated for the periods of monitoring. For

categorical variables, differences in proportions between

survivors and nonsurvivors were tested using the v2 test or

the 2-tailed Fisher’s Exact test. For continuous variables, the

equality of the means between survivors and nonsurvivors

was tested by the 2-sample t test or the Wilcoxon’s 2-sample

test. The ability of the survival probability and the other

noninvasive hemodynamic variables to predict hospital

mortality was analyzed by generation of a receiver operating

characteristic (ROC) curve for each variable and by

comparison of the respective areas under the ROC curve

(AUC). All statistical analysis was done with SPSS 12.0 for

Windows (SPSS Inc, Chicago, Ill) and statistical significance

was set at P b .05.

2. Results

There were 45 pediatric trauma patients identified who

were entered in the noninvasive monitoring protocol from

September 1996 through January 2004. The patients ranged

in age from 1 to 16 years, with a mean age of 11. There were

13 hospital deaths, for an overall mortality rate of 29%.

Table 1 shows the clinical features of the survivors and

nonsurvivors in this series. Nonsurvivors had a higher

frequency of head injuries, higher ISS and APACHE II

scores, and significantly lower GCSs compared with survi-

vors. Eight (18%) patients had severe intracranial hemor-

rhage and edema that progressed to brain death.

Table 2 shows the noninvasive hemodynamic monitoring

data and the calculated survival probability collected in the

first hour after study admission. Early in the resuscitation

period, there is no significant difference between survivors

Fig. 1 Calculated survival probability plotted against time

from study admission for survivors (black line) and nonsurvivors

(white line).

Table 1 Comparison of survivors and nonsurvivors

Variable Survived

(n = 32)

Died

(n = 13)

P

Age (y) 11.7 F 4.9 11.7 F 5.2 .998

Sex

Male 26 (81) 10 (77) .70

Female 6 (19) 3 (23) .70

Mechanism

Blunt 19 (59) 6 (46) .52

Penetrating 13 (41) 7 (54) .52

Injury site

Head 8 (25) 7 (54) .09

Chest 9 (28) 2 (15) .47

Abdomen 10 (31) 2 (15) .46

Extremity 11 (34) 0 (0) .02

Emergent operation 12 (38) 4 (31) .74

Sepsis 7 (22) 2 (15) .98

ARDS 1 (3) 4 (31) .02

Systolic blood

pressure (mm Hg)

112 F 32.7 139 F 53 .10

Estimated

blood loss (mL)

1445 F 1784 3475 F 3220 .14

ISS 19.9 F 13.5 33.4 F 18.4 .012

GCS 11.7 F 4.6 5.0 F 4.1 b.001

Acute Physiology

Score

19.1 F 10 32 F 2 b.001

APACHE II 22.1 F 11.5 34.1 F 3.3 b.001

ICU stay (d) 13.5 F 26.2 5.9 F 4.9 .18

Hospital stay (d) 19.9 F 24 5.5 F 4.7 .005

Values are expressed as mean F SD or number (%). ARDS indicates

acute respiratory distress syndrome; APACHE II, Acute Physiology and

Chronic Health Evaluation score.

M. Martin et al.1960

and nonsurvivors in the standard parameters of HR, blood

pressure, pulse oximetry, or bioimpedance CI. However,

there were significant differences noted in the transcutaneous

oxygen measure (Ptco2/Fio2) and the calculated survival

probability was 86% for survivors vs 62% for nonsurvivors

(P b .001).

Table 2 Predicted survival probability and hemodynamics in

the first hour after admission

Variable Survived

(n = 32)

Died

(n = 13)

P

Survival probability (%) 86 F 9.9 62 F 22 .001

CI (L/min per m2) 3.4 F 2.5 3.5 F 2.1 .92

MAP (mm Hg) 87 F 21.5 74 F 20.6 .085

HR (beats per min) 108 F 30 123 F 39.4 .25

Sao2 (%) 99 F 1.7 97.7 F 3.8 .27

Ptcco2 (mm Hg) 44.9 F 14.9 35.8 F 12.2 .06

Ptco2/Fio2 235 F 106 155 F 133 .04

Sao2 indicates arterial oxygen saturation; Ptcco2, transcutaneous

carbon dioxide tension; Ptco2/Fio2, transcutaneous oxygen tension

indexed to fraction of inspired oxygen.

The calculated survival probability from the noninva-

sive monitoring data reliably separated survivors from

nonsurvivors at all time-points during the study period,

with improvement in the predictive ability over time as

more data were accumulated (Fig. 1). The mean calculated

survival probability for the entire data set was 85% (F10%)

for survivors compared with 64% (F19%) for nonsurvivors.

The transcutaneous oxygen index (Ptco2/Fio2) was also

significantly higher in survivors than nonsurvivors (251 vs

177, P b .001). In comparison, there was minimal difference

between the two groups noted in the standard parameters

of MAP (82 vs 86), bioimpedance CI (3.6 vs 3.4), HR

(112 vs 118), or pulse oximetry (98% vs 99%) during the

study period.

Fig. 2 shows the ROC curves for prediction of hospital

mortality by the study variables collected over the first

24 hours, with the curve for the calculated survival

probability clearly outperforming the other variables of

HR, MAP, and bioimpedance CI. Table 3 shows a

comparison of the respective AUCs for the study variables

at the time of admission, the last data set collected, and the

mean values during the entire study period. The calculated

survival probability and the Ptco2/Fio2 demonstrated

superior predictive ability from the time of study admis-

sion and throughout the study period (both AUC N0.7).

The other variables performed poorly for mortality predic-

tion, with the exception of the final MAP with an AUC

of 0.83.

Similar hemodynamic and perfusion patterns were ob-

served in all injury patterns with the exception of severe head

injury progressing to brainstem herniation. Six patients in this

series with severe head injury were monitored from

admission through the period surrounding brain death. The

standard hemodynamic variables demonstrated evidence of

increasing instability, with tachycardia, hypotension, and

increasing cardiac output leading up to the point of

herniation. However, the tissue perfusion variables showed

evidence of a massive increase in peripheral perfusion around

the time of brainstem herniation and brain death, with a sharp

Fig. 2 Receiver operating characteristic curves for survival prediction (SPmoni), bioimpedance cardiac index (CIbi), HR, and MAP.

Continuous noninvasive monitoring in pediatric trauma patients 1961

rise in the Ptco2/Fio2 ratio and decline in the Ptcco2

consistent with sympathetic collapse and vasodilatation.

3. Discussion

The critically injured pediatric patient presents a daunting

challenge to even the most experienced physicians. Major

diagnostic and therapeutic decisions must often be made

based on incomplete or imperfect data or are frequently

delayed until the patient is moved to a physical area of the

hospital (ICU) where appropriate invasive monitoring can be

performed. This can lead to significant delays in recognizing

states of bcompensatedQ shock and in initiating appropriate

and targeted resuscitation resulting in prolonged hypoperfu-

sion, organ failure, and death [33-39]. Modern noninvasive

monitoring systems have been proposed as a partial solution,

allowing early monitoring outside the ICU setting and

providing a mobility that allows extension of intensive care

to remote areas of the hospital [15].

Table 3 Area under the ROC curve for mortality prediction

Time SP Ptco2/Fio2 MAP HR CI Spo2

Initial values 0.84 0.80 0.63 0.36 0.43 0.55

Final values 0.85 0.83 0.83 0.43 0.61 0.23

All values 0.83 0.71 0.51 0.60 0.53 0.47

SP indicates survival prediction; Ptco2/Fio2, transcutaneous oxygen

index; Spo2, pulse oximetry.

The value of the hemodynamic and flow-related variables

obtained from invasive monitoring such as pulmonary ar-

tery catheters is a source of ongoing debate. Several large,

prospective randomized trials in adults have demonstrated no

benefit, and possibly increased morbidity, associated with

routine pulmonary artery catheterization [36,40-42]. How-

ever, early or preoperative hemodynamic optimization has

been reported to improve outcome in peripheral vascular

surgery [4], trauma [5], cardiogenic shock [6,7], and sepsis

[8,9]. There are much less data available on the use of

invasive monitoring in pediatric population. Several retro-

spective reports found pulmonary artery catheters to be

beneficial in guiding the management of children undergoing

cardiac surgery [11] and in pediatric burn patients [43]. These

studies support the concept that invasive monitoring and

aggressive resuscitation must be initiated very early in the

disease course to achieve improved outcomes. Noninvasive

monitoring systems that can provide similar data regarding

cardiac performance and oxygen delivery/perfusion may

provide an attractive alternative or even replacement for the

more cumbersome and less user-friendly invasive systems.

The noninvasive monitoring system in this series con-

sisted of a thoracic bioimpedance device to measure stroke

volume and cardiac output and transcutaneous assessment of

oxygen and carbon dioxide tension. The noninvasively

determined cardiac output has been found to correlate well

with thermodilution measurement across a wide variety of

patient populations [44]. The transcutaneous oxygen and

carbon dioxide monitor indirectly measures the arteriole

blood gas tension levels [16]. With tissue hypoperfusion, the

M. Martin et al.1962

transcutaneous oxygen and carbon dioxide levels may

diverge from their arterial counterparts and provide early

evidence of inadequate flow to the monitored tissue bed

[17,18]. These noninvasive monitoring techniques have been

previously found to provide accurate and clinically relevant

data in adult emergency patients [22] and severely injured

adult trauma patients [23,45]. In addition, when coupled with

a mathematical stochastic analysis program, the data

obtained from the noninvasive monitoring platform can

provide an accurate and continuous real-time measure of the

individual patient’s probability of survival [23,46].

The approach presented here is similar to that of

experienced clinicians who can recall similar patients who

responded to a specific therapy. In essence, the program

attempts to emulate the processes of good clinical judgment

by searching a database for patients with identical diagnoses

and covariates and who also have very similar hemodynamic

patterns. The program then uses these nearest neighbors as

surrogates for the present study patient. The patient’s

outcome may be estimated from the outcomes of the nearest

neighbors that are known from the database. The predicted

outcome of each patient on ED admission was validated by

the patient’s actual outcome determined at hospital discharge

several days or weeks later.

This technology was used to provide continuous and real-

time hemodynamic assessments of severely injured children

beginning very early in their hospital course and through the

early resuscitation period. The noninvasive data also

provided a framework for a continuous accurate mathemat-

ical prediction of survival probability. From the time of study

entry to death or termination of monitoring, the calculated

survival probability accurately identified survivors and

nonsurvivors, outperforming all of the individual measured

parameters. We also found the transcutaneous oxygen index

(Ptco2/Fio2) to be a useful early marker of hypoperfusion

and mortality, and a ratio of less than 200 should prompt

immediate evaluation and intervention. In the severely head-

injured patient, the monitoring data may provide an early

warning for impending herniation and brain death, allowing

opportunities for intervention to prevent herniation or

salvage the patient for potential organ donation.

The availability of continuous monitoring and feedback

also provides a unique opportunity for real-time assessment

of the need for therapeutic interventions and provides a

means to immediately assess the patient’s physiologic

response to treatment. We now have the ability to couple

this noninvasive monitoring system to a computer-based

therapeutic decision support package. This system can

analyze the patient’s hemodynamic profile and identify the

treatment options that have provided optimal outcomes

among patients with similar profiles from the database

[46]. This system will need further testing and validation to

determine its applicability to the pediatric trauma population.

This study does have several limitations. It is a

retrospective review of a highly select group of severely

injured children and is not a consecutive or all-encompassing

sample. The sample size is relatively small as only the most

severely injured patients were included, which limits

statistical analysis and broad conclusions. Although the

thoracic bioimpedance device has been well validated in

adults, there is limited validation of its accuracy in pediatric

patients. In addition, the calculated survival probability relies

on identifying similar patients from a large clinical database

that contains mostly older children and adults for compar-

ison. This may limit its ability to identify true nearest

neighbors for younger children until more pediatric patients

are included in the database. However, the calculated

survival probability obtained in this series performed well

for even the youngest patients.

Noninvasive monitoring technologies can now supple-

ment or entirely replace the volume and quality of data

provided by more invasive systems and have the benefits of

lower complication profiles, greater mobility, and wider

applicability [15,47]. In addition to providing critical

physiologic data, noninvasive monitoring can provide

accurate prognostic information and may be used to guide

and tailor appropriate and often lifesaving interventions.

These advances will help reinforce the concept of critical

care as a system of principles and practices that can be

applied anywhere in the hospital rather than only in a fixed

and isolated location. Further rigorous study of the utility,

costs, and benefits of these new and emerging technologies,

particularly in the pediatric population, will be required to

fully integrate them into today’s ICU environment.

References

[1] Scalea TM, Maltz S, Yelon J, et al. Resuscitation of multiple trauma

and head injury role of crystalloid fluids and inotropes. Crit Care Med

1994;22:1610-5.

[2] Bernard GR, Sopko F, Cerra F, et al. Pulmonary artery catheterization

and clinical outcomes: National Heart, Lung, and Blood Institute and

Food and Drug Administration Workshop Report. Consensus State-

ment. JAMA 2000;283:2568-72.

[3] Coulter TD, Wiedemann HP. Complications of hemodynamic

monitoring. Clin Chest Med 1999;20:249 -67.

[4] Berlauk JF, Abrams JH, Gilmour IJ, et al. Preoperative optimization of

cardiovascular hemodynamics improves outcome in peripheral vas-

cular surgery. Ann Surg 1991;214:289-97.

[5] Moore FA, Haemel JB, Moore EE, et al. Incommensurate oxygen

consumption in response to maximal oxygen availability predicts

postinjury oxygen failure. J Trauma 1992;33:58-67.

[6] Creamer JE, Edwards JD, Nightingale P. Hemodynamic and oxygen

transport variables in cardiogenic shock secondary to acute myocar-

dial infarction. Am J Cardiol 1990;65:1297-300.

[7] Rady MY, Edwards JD, Rivers EP, et al. Measurement of oxygen

consumption after uncomplicated acute myocardial infarction. Chest

1993;103:886 -95.

[8] Kern JW, Shoemaker WC. Meta-analysis of hemodynamic optimiza-

tion in high-risk patients. Crit Care Med 2002;30:1686-92.

[9] Rivers E, Nguyen B, Havstad S, et al. Early goal-directed therapy in the

treatment of severe sepsis and septic shock. N Engl J Med

2002;345:1368-77.

[10] Becker K. Resolved: a pulmonary artery catheter should be used in the

management of the critically ill patient. Con J Cardiothorac Vasc

Anesth 1998;12:13 -6.

Continuous noninvasive monitoring in pediatric trauma patients 1963

[11] Damen J, Wever JE. The use of balloon-tipped catheters in children

undergoing cardiac surgery. Intensive Care Med 1987;13:266 -72.

[12] Berger S. Pulmonary artery catheters, children, and the twenty-first

century. Pediatr Crit Care Med 2001;2:286-7.

[13] Shoemaker WC, Wo CCJ, Bishop MH, et al. Multicenter trial of a new

thoracic electrical bioimpedance device for cardiac output estimation.

Crit Care Med 1994;22:1907-12.

[14] Shoemaker WC, Belzberg H, Wo CCJ, et al. Multicenter study of

noninvasive monitoring of acutely ill emergency patients. Chest

1998;114:1643-52.

[15] DeNicola LK, Kissoon N, Abram HS, et al. Noninvasive monitoring

in the pediatric intensive care unit. Pediatr Crit Care 2001;48:573 -88.

[16] Lubbers DW. Theoretical basis of the transcutaneous blood gas

measurements. Crit Care Med 1981;9:721-33.

[17] Janssens JP, Perrin E, Bennani I, et al. Is continuous transcutaneous

monitoring of PCO2 (tcPCO2) over 8h reliable in adults? Respir Med

2001;95:331 -5.

[18] Rosner V, Hannhart B, Chabot F, et al. Validity of transcutaneous

oxygen/carbon dioxide pressure measurement in the monitoring of

mechanical ventilation in stable chronic respiratory failure. Eur Respir J

1999;13:1044-7.

[19] Bayard DS, Botnen A, Shoemaker WC, et al. Stochastic analysis of

therapeutic modalities using a database of patient responses. IEEE

Symp Computer-Based Med Syst (CBMS) 2001;11:439-44.

[20] Bayard DS. A forward method for optimal stochastic nonlinear and

adaptive control. IEEE Trans Automat Contr 1991;36:1046-53.

[21] Bayard DS. Reduced complexity dynamic programming based on

policy iteration. J Math Anal Appl 1992;170:75 -103.

[22] Tatevossian RG, Shoemaker WC, Wo CC, et al. Noninvasive

hemodynamic monitoring for early warning of adult respiratory

distress syndrome in trauma patients. J Crit Care 2000;15:151 -9.

[23] Shoemaker WC, Wo CCJ, Chan L, et al. Outcome prediction of

emergency patients by noninvasive hemodynamic monitoring. Chest

2001;120:528 -37.

[24] Wang X, Sun H, Adamson D, et al. An impedance cardiography

system: a new design. Ann Biomed Eng 1989;17:535 -56.

[25] Wang X, Van De Water JM, Sun H, et al. Hemodynamic monitoring

by impedance cardiography with an improved signal processing

technique. Proc IEEE Eng Med Biol 1993;15:699-700.

[26] Tremper KK, Waxman K, Shoemaker WC. Effects of hypoxia and

shock on transcutaneous PO2 values in dogs. Crit Care Med 1979;7: 

526-31.

[27] Tremper KK, Shoemaker WC. Transcutaneous oxygen monitoring of

critically ill adults with and without low flow shock. Crit Care Med

1981;9:706 -9.

[28] Lubbers DW. Theoretical basis of transcutaneous blood gas measure-

ments. Crit Care Med 1981;9:721-30.

[29] Tremper KK, Huxtable RF. Dermal heat transport analysis for transcu-

taneous O2 measurements. Acta Anaesthesiol Scand 1978; 68:4 -9.

[30] Tremper KK, Waxman K, Bowman R, et al. Continuous transcu-

taneous oxygen monitoring during respiratory failure, cardiac decom-

pensation, cardiac arrest, and CPR. Crit Care Med 1980;8:377-81.

[31] Sutton RS, Barto AG. Reinforced learning. Cambridge7 MIT Press;

1998.

[32] Bertsekas DP, Tsitsiklis JN. Neuro-dynamic programing. Belmont

(Mass)7 Athena Scientific; 1996.

[33] Bishop MW, Shoemaker WC, Appel PL, et al. Relationship between

supranormal values, time delays and outcome in severely traumatized

patients. Crit Care Med 1993;21:56 -61.

[34] Shoemaker WC, Appel PL, Kram HB, et al. Prospective trial of

supranormal values of survivors as therapeutic goals in high risk

surgical patients. Chest 1988;94:1176-82.

[35] Boyd O, Grounds M, Bennett D. Preoperative increase of oxygen

delivery reduces mortality in high risk surgical patients. JAMA

1993;270:2699-705.

[36] Wilson J, Woods I, Faucett J, et al. Reducing the risk of major elective

surgery: randomized control trial of preoperative optimization of

oxygen delivery. BMJ 1999;318:1099 -103.

[37] Yu M, Levy MM, Smith P, et al. Effect of maximizing oxygen

delivery on mortality and mortality rates in critically ill patients: a

prospective randomized controlled study. Crit Care Med 1993;21: 

830 -8.

[38] Bishop MH, Shoemaker WC, Kram HB, et al. Prospective randomized

trial of survivor values of cardiac index, oxygen delivery, and oxygen

consumption as resuscitation endpoints in severe trauma. J Trauma

1995;38:780 -7.

[39] Boyd O, Bennett D. Enhancement of perioperative tissue perfusion

as a therapeutic strategy for major surgery. New Horiz 1996;4: 

453 -65.

[40] Conners Jr AF, Speroff T, Dawson NV, et al. The effectiveness of right

heart catheterization in the initial care of critically ill patients. JAMA

1996;276:889 -97.

[41] Guyatt G.Ontario Intensive care group. A randomized control trial of

right heart catheterization in critically ill patients. J Intensive Care Med

1991;6:91-5.

[42] Zion MM, Balkin J, Rosenmann D, et al. Use of pulmonary artery

catheters in patients with acute myocardial infarction: analysis of

experience in 5841 patients in the SPRINT registry. Chest 1990;98: 

1235-331.

[43] Reynolds EM, Ryan DP, Sheridan RL, et al. Left ventricular failure

complicating severe pediatric burn injuries. J Pediatr Surg 1995;30: 

264 -70.

[44] Brown CV, Martin MJ, Shoemaker WC, et al. The effect of obesity on

bioimpedance cardiac index. Am J Surg 2005;189:547-51.

[45] Velmahos GC, Demetriades D, Shoemaker WC, et al. Endpoints of

resuscitation of critically injured patients: normal or supranormal?

Ann Surg 2000;232:409 -18.

[46] ShoemakerWC, Bayard DS, BotnenA, et al.Mathematical program for

outcome prediction and therapeutic support for trauma beginning

within 1 hour of admission: a preliminary report. Crit Care Med

2005;33:1499-1506.

[47] Chaney JC, Derdak S. Minimally invasive hemodynamic monitoring

for the intensivist: current and emerging technology. Crit Care Med

2002;30:2338-45.