Low initial in-hospital end-tidal carbon dioxide predicts poor patient outcomes and is a useful trauma bay adjunct

a b s t r a c t

Introduction: appropriate triage of the trauma patient is critical. Low end-tidal carbon dioxide is associ- ated with mortality and hemorrhagic shock in trauma, but the relationship between low ETCO2 and important clinical variables is not known. This study investigates the association of initial in-hospital ETCO2 and patient out- comes, as well as the utility of ETCO2 as a predictive aid for blood transfusion.

Methods: Adult patients who presented to a Level One trauma center from 2019 to 2020 were eligible. Trauma bay ETCO2 measured by side-stream capnography was prospectively obtained for all Trauma activations at time of initial evaluation. Using the Liu method of cut point estimation, patients were stratified as having low (<=29.5 mmHg) or normal ETCO2 (>29.5 mmHg). multivariable regression was used to estimate the association of low ETCO2 with patient outcomes.

Results: A total of 955 patients underwent initial in-hospital ETCO2 measurement. Median time from arrival to ETCO2 measurement was 4 min. Among admitted patients (N = 493), 48.9% had low ETCO2. Compared to pa- tients with normal ETCO2, those with low ETCO2 were older (median age 53 vs 46, p = 0.01) and more likely to have the highest trauma activation (27.4% vs 19.8%, p = 0.048). There was no difference in head injury. After adjustment, patients with low ETCO2 had greater odds of blood transfusion (OR 4.65, 95%CI 2.0-10.7), mor- tality (OR 5.10, 95%CI 1.1-24.9), inferior disposition (OR 1.64, 95%CI 1.1-2.6), and complications (OR 3.35, 95%CI 1.5-7.4). ETCO2 was more predictive of early blood transfusion than Shock Index (area under ROC = 67.6% vs 58.2%).

Conclusions: Low trauma bay ETCO2 remains significantly associated with inferior clinical outcomes after adjust- ment. In comparison to other triage tools, low EtCO2 values may be more predictive of the need for blood trans- fusion. Further studies are needed to evaluate the role of ETCO2 as a decision making tool for early trauma management.

(C) 2022

  1. Introduction

Efforts to assess systemic perfusion have dominated the fields of crit- ical care and trauma resuscitation for over half a century. It is well estab- lished that trauma morbidity and the global state of perfusion are inversely related, echoing the association between ventilatory output and perfusion status [1,2]. Real-time appraisal of perfusion status re- mains imprecise, as traditional parameters such as heart rate and blood pressure are notoriously poor indicators of shock when used in isolation

* Corresponding author at: WakeMed Health and Hospitals, 3024 New Bern Avenue, Raleigh, NC 27610, United States.

E-mail addresses: [email protected], [email protected] (J.N. Portelli Tremont), [email protected] (R.A. Caldas), [email protected] (N. Cook), [email protected] (P.O. Udekwu), [email protected] (S.M. Moore).

[3]. One method commonly utilized by clinicians to extrapolate hemody- namic function is through quantitative evaluation of exhaled carbon dioxide via End-tidal capnography (ETCO2). Both side-stream and main-stream capnography depict the quantifiable and continuously re- corded concentration of end-expiratory carbon dioxide per respiratory cycle [4]. Oscillations in exhaled CO2 outside of normal parameters (35-45 mmHg) and deviations in capnogram morphology may be utilized to appraise hemodynamic and metabolic fluctuations [5-7]. In addition, end-tidal capnography may be employed as an adjunctive means from which to monitor the bedside progression of underlying tis- sue malperfusion, wherein rising serum lactate concentrations are accompanied by a decrease in ETCO2 [7-9].

Continuous ETCO2 monitoring is commonly used to evaluate chest Compression quality and advanced airway placement in cardiopulmo- nary arrest [6,10-12]. The triage utility of end-tidal capnography is not

0735-6757/(C) 2022

as well known since most studies focus on main-stream ETCO2 in the setting of intubation or cardiopulmonary arrest [11,13]. Past analyses corroborate the diagnostic value of ETCO2 in detecting hemorrhagic shock [6-8]. However, the bulk of literature on end-tidal capnography in critical care and major trauma has not differentiated between those values obtained immediately upon arrival and those obtained hours af- terwards. Most studies have focused on intubated patients, who by def- inition are already considered critically ill. Recently, we have shown that non-intubated trauma patients with a low pre-hospital ETCO2 are at in- creased risk for morbidity and mortality, which suggests that non- invasive ETCO2 measurements very early in trauma resuscitation may prove useful in identifying occult shock [14].

Resuscitation and patient decision making are fundamentally con- tingent on Hemodynamic state and clinical presentation. Current clinical practice may rely heavily on injury mechanism or other anatomic markers to predict which patients are the sickest. Over-reliance on mechanistic and anatomic criteria has been shown to increase mis- triage, in part due to reduced sensitivity of these measures [15-17]. One potential solution is to increase emphasis on physiologic parame- ters to aid in recognition of shock states and guide decision making. Most currently used parameters, however, are either limited by a lack of sensitivity (e.g., heart rate, blood pressure) or are not available in real-time (e.g., Base deficit, lactate) [3,18].

These challenges highlight the need for a sensitive and real-time

physiologic measure that can be used to augment existing triage tools to better patient outcomes and improve trauma service allocation. In this one-year study, we evaluate the association between initial in- hospital ETCO2 of a non-intubated trauma patient cohort and transfu- sion requirement, as well as other well-established clinical outcomes. Our secondary goal is to extrapolate the plausible utility of ETCO2 as a predictive aid for early trauma management.

  1. Methods
    1. Study design

This is a single-center retrospective observational cohort study of adult trauma patients >=16 years old who presented to a Level One trauma center with initial trauma bay ETCO2 obtained by side-stream nasal canula capnography. Patients who had a measured trauma bay ETCO2 and either presented as a trauma activation or were entered into the trauma registry between 2019 and 2020 were eligible for inclu- sion. Measurement of ETCO2 was obtained using GE CARESCAPE B650 monitors (GE Healthcare, Helsinki, Finland) with a standardized pre- ventive maintenance schedule for calibration. Trauma bay ETCO2 was prospectively obtained for all trauma activations at the time of initial in-hospital evaluation at the discretion of the attending provider. Pa- tients who were intubated prior to ETCO2 measurement and patients with erroneous ETCO2 (0- and 1-mmHg) values were excluded. There were no other exclusion criteria.

Demographic variables of interest included gender, patient age, and self-identified race. Clinical variables of interest included comorbid con- ditions associated with possible ETCO2 derangements (composite of current smoker, coronary artery disease, congestive heart failure, chronic obstructive pulmonary disease), mechanism of injury, injury se- verity score (ISS), Emergency Department (ED) disposition, ED vital signs, ED Glasgow coma scale (GCS), hospital discharge disposition, level of trauma activation, blood product administration, mortality, as well as ETCO2 measurement and time of ETCO2 measurement. These variables were obtained from the institutional trauma registry and the electronic medical record.

The primary outcome of interest was need for blood product transfu- sion within 24 h of hospital presentation. Secondary outcomes included total hospital length of stay , all-cause mortality and included deaths occurring in the trauma bay, intensive care unit admission, any hospital complication (composite of pneumonia, sepsis, wound

infection, alcohol withdrawal, pulmonary embolus, deep vein thrombo- sis, respiratory failure, urinary tract infection), need for mechanical ven- tilation (i.e., intubation), and discharge disposition. Complications were diagnosed by the attending physician during the hospital stay or at dis- charge and were documented in the trauma registry.

    1. Dichotomization of ETCO2

In order to compare clinical outcomes for varying ETCO2, patients were dichotomized into either a low or normal/high ETCO2 group by the Liu method of cut point estimation, which calculates the optimal cut point to predict the probability of an outcome by maximizing the product of the sensitivity and specificity [19]. Using this method,

29.5 mmHg was determined as the optimal threshold for ETCO2 to pre-

dict need for blood transfusion in this cohort. Clinically, this value mir- rors ETCO2 thresholds for malperfusion and is also within 1 mmHg of the optimal threshold value that was similarly determined for a sepa- rate cohort of pre-hospital patients at our institution [14].

    1. Statistical analyses

Patient demographics and clinical characteristics were examined by ETCO2 status (low ETCO2 vs normal/high ETCO2). Means, medians, standard deviations, and IQR ranges were calculated for continuous variables, and percentages were calculated for categorical variables. Bivariate analysis of demographics and clinical characteristics were compared using Chi-square tests for categorical data and Wilcoxon rank sum tests for continuous variables, where appropriate. Missing data were evaluated, and no variable met the a priori threshold of

>20% for exclusion from analysis. There were no missing data for our outcomes of interest.

Unadjusted and multivariable logistic and linear regression were used to estimate the association between ETCO2 status and blood trans- fusion, hospital admission, need for mechanical ventilation, need for op- eration, ICU stay, inpatient complications, inferior disposition (includes discharge to skilled nursing facility, long term care facility, rehabilita- tion, and hospice), and total hospital LOS for admitted patients. Collin- earity was assessed with tests of correlation for continuous variables. Hospital admission analysis was adjusted for patient age, head injury, and GCS. All other models were adjusted for patient age, GCS, head in- jury, and comorbidities using a change in effect estimate approach and clinical knowledge-based a priori determination [20]. We also exam- ined potential modification of low ETCO2 on patient outcomes by head injury (patients without head injury vs. patients with head injury). The purpose of this analysis was to determine whether the effect of low ETCO2 on patient outcomes differed in patients with and without head injury. Head injury status was determined by AIS codes obtained from the trauma registry. Mantel-Haenszel tests of homogeneity were used to compare effects across head injury status.

The utility of ETCO2 was then compared to the validated Shock Index as a predictor of the need for any blood transfusion and large vol- ume transfusion, defined as >=4 units, within the first 24 h. The SI is cal- culated by dividing heart rate by systolic blood pressure, with a value >=1 predicting the need for massive transfusion [21-23]. Therefore, we used

>=1 a priori as the optimal SI threshold. Sensitivity, specificity, and area under the receiver operating curve (ROC) were calculated for ETCO2 and SI.

Statistical significance was defined as p < 0.05 except as noted. Using the conservative assumption that 5% of patients with normal ETCO2 and 10% of patients with low ETCO2 will require a blood transfusion, relative sample size of 1:1, and an alpha of 0.05, this study has a minimum power of 0.8 to refute the null hypothesis for our primary outcome. All analyses were performed using Stata v17 (StataCorp 2021, College Sta- tion, TX). Study procedures were reviewed and approved by the hospi- tal Institutional Review Board.

  1. Results

Patients who were intubated prior to ETCO2 measurement (N = 9) and patients with ETCO2 values of 0- and 1-mmHg (N = 43) were ex- cluded. Final analyses included 955 patients. Of all patients with re- corded initial trauma bay ETCO2 measurements, median time from arrival to non-invasive side-stream ETCO2 measurement was 4 min (IQR 3-6 min). 493 (51.6%) of these patients were admitted to the hos- pital, and 241 (48.9%) had low ETCO2. Patients in the low ETCO2 cohort had a median ETCO2 of 22 mmHg (IQR 19-26), and patients in the nor- mal/high ETCO2 cohort had a median ETCO2 of 35 mmHg (IQR 33-38).

Compared to patients with normal/high ETCO2, patients with low ETCO2 were significantly older (53 vs 45.5, p = 0.011). There were no differences in gender or race distributions among low and normal/ high ETCO2 groups. With regard to injury characteristics, there were no differences in head injury prevalence. Vital signs and injury mecha- nisms were also statistically similar across ETCO2 groups (p > 0.05). Although overall activation levels were equivalent between the groups (p = 0.140), those with low ETCO2 were more likely to have the highest trauma activation (27.4% vs 19.8%, p = 0.048). Patient and injury char- acteristics are detailed in Table 1.

    1. Blood transfusion, mortality, and other clinical outcomes

The incidence of clinical outcomes among the low ETCO2 and nor- mal/high ETCO2 groups are listed in Table 2. Table 3 reports the adjusted

Table 1 Patient demographics and injury characteristics for admitted patients, stratified by ETCO2 status.


Normal/High ETCO2


N = 241

N = 252

Age, median (IQR)

53 (32-72)

45.5 (27-68)


Male, n (%)

156 (64.7)

161 (63.9)


Black, n (%)

55 (22.8)

77 (30.6)


Comorbidities presenta, n (%)

63 (26.1)

66 (26.2)


Primary Insurance, n (%)


Commercial/Private 76 (34.2) 95 (38.9)

Public/Government 99 (44.6) 96 (39.3)

Self-Pay 19 (8.6) 24 (9.8)

Other b 28 (12.6) 29 (11.9)

Table 2

Incidence of patient complications and outcomes after trauma admission, stratified by ETCO2 status

Low ETCO2 Normal/High ETCO2

N = 241 N = 252

Blood transfusion, n (%) 29 (12.0) 9 (3.6)

Total LOS, median (IQR) 4 (2-10) 4 (2-7)

Mechanical ventilation, n (%) 32 (13.3) 15 (6.0)

ICU stay, n (%) 99 (41.1) 85 (33.7)

Inpatient complications a, n (%) 28 (11.6) 9 (3.6) Discharge disposition, n (%)

Routine b 146 (60.6) 186 (73.8)

Continued care c 74 (30.7) 53 (21.0)

Transfer/Other d 9 (3.7) 11 (4.4)

Died 12 (5.0) 2 (0.8)

Abbreviations: ETCO2, end tidal carbon dioxide; IQR, interquartile range; ICU, intensive care unit; LOS, length of stay

a Includes respiratory failure, pneumonia, alcohol withdrawal syndrome, delay in treatment, unplanned intubation, unplanned admission to ICU, AKI, pneumothorax, ARDS, PE, stroke/CVA, delirium, cardiac arrest with CPR, SSI.

b Includes home with self-care and home with services.

c Includes rehabilitation, skilled nursing facility, hospice, and long-term care.

d Includes transfer to Burn center, trauma center, mental health facility, left against medical advice, and correctional facility.

effect estimates, as well as unadjusted odds ratios, of low ETCO2 on clinical outcomes. Blood transfusion was needed in 12.0% of patients with low ETCO2 compared to 3.6% with normal/high ETCO2. After ad- justment, patients with low ETCO2 were 4.65 times more likely to re- quire transfusion than patients with normal/high ETCO2 (95% CI 2.01-10.72). 5.0% of patients died in the low ETCO2 group compared to 0.8% in the normal high ETCO2 group. Odds of mortality were 5.10 times greater in patients with low ETCO2 after adjustment (95% CI 1.05-24.90).

With regard to secondary outcomes, patients with low ETCO2 were more likely to require mechanical ventilation (OR 2.50, 95% CI 1.22-5.15), need an emergent operation (OR 1.73, 95% CI 1.08-2.77), suffer inpatient complications (OR 3.35, 95% CI 1.52-7.38), and be discharged to a skilled nursing facility, long-term care facility, rehabili- tation, or hospice than those with normal/high ETCO2 (OR 1.64, 95% CI 1.05-2.56; see Table 3). In addition, low ETCO2 was significantly

Table 3

Association between low ETCO2 status and patient outcomes, unadjusted and adjusted models.

Unadjusted Model Adjusted Model a

Head injury, n (%)

Number of injuries, median (IQR)

63 (26.1)

5 (2-8)

60 (23.8)

4 (2-6)



OR (95% CI)


OR (95% CI)


Mechanism of injury, n (%)


Hospital admission b

1.72 (1.32, 2.23)


1.37 (1.03, 1.83)


Vehicle c

125 (51.9)

137 (54.4)

Blood transfusion

3.69 (1.71, 7.98)


4.65 (2.01, 10.72)



85 (35.3)

78 (31.0)

Mechanical ventilation

2.42 (1.27, 4.59)


2.50 (1.22, 5.15)



11 (4.6)

9 (3.6)


1.63 (1.03, 2.57)


1.73 (1.08, 2.77)



10 (4.2)

19 (7.5)

ICU stay

1.37 (0.95, 1.98)


1.31 (0.88, 1.94)


Other d

10 (4.2)

9 (3.6)


3.53 (1.63, 7.66)


3.35 (1.52, 7.38)


Blunt injury, n (%)

221 (91.7)

226 (89.7)



SBP on arrival, median (IQR)

135 (119-155)

138 (125-154)


Inferior disposition d

1.77 (1.18, 2.68)


1.64 (1.05, 2.56)


HR on arrival, median (IQR)

91 (78-106)

90 (78.5-101)


inpatient mortality

6.55 (1.45, 29.58)


5.10 (1.05, 24.90)


GCS on arrival, median (IQR)

ISS, median (IQR)

15 (14-15)

10 (5-17)

15 (15-15)

9 (5-14)



CIE (95% CI)


CIE (95% CI)


Trauma Activation Level, n (%)


Total LOS

1.85 (0.36, 3.34)


1.59 (0.12, 3.07)


Level 1

66 (27.4)

50 (19.8)

Level 2

172 (71.4)

199 (79.0)

No activation

3 (1.2)

3 (1.2)

Abbreviations: ETCO2, end-tidal carbon dioxide; IQR, interquartile range; SBP, systolic blood pressure; HR, heart rate; GCS, Glasgow Coma Scale; ISS, injury severity score; ED, emergency department. p < 0.05 is considered significant.

a Includes current smoker, coronary artery disease, congestive heart failure, chronic

obstructive pulmonary disease.

b Includes military, automobile and other.

c Includes motor vehicle collision, motorcycle crash, pedestrian versus vehicle, ATV, moped, and bicycle mechanisms.

d Includes machine and animal mechanisms.

Abbreviations: ETCO2, end tidal carbon dioxide; OR, odds ratio; CI, confidence interval; CIE, change in estimate; ICU, intensive care unit; LOS, length of stay. p < 0.05 is considered significant.

a Hospital admission adjusted for patient age, head injury, and GCS. All other models

adjusted for patient age, GCS, head injury, and comorbidities.

b Among all patients, n = 955.

c Includes respiratory failure, pneumonia, alcohol withdrawal syndrome, delay in treatment, unplanned intubation, unplanned admission to ICU, AKI, pneumothorax, ARDS, PE, stroke/CVA, delirium, cardiac arrest with CPR, SSI.

d Includes discharge to skilled nursing facility, long term care facility, rehabilitation, and hospice.

associated with a 1.59-day increase in hospital LOS (95% CI 0.12-3.07). After adjustment, there were no differences in ICU stay.

When stratified by the presence of head injury, we saw no difference in the unadjusted association of low ETCO2 on the odds of blood trans- fusion (p = 0.527), mechanical ventilation (p = 0.281), ICU stay (p = 0.165), operation (p = 0.762), inferior disposition (p = 0.271), and mortality (p = 0.966) among patients with and without head injury, Table 4. However, low ETCO2 decreased the likelihood of hospital ad- mission in patients with head injuries, compared to those without (OR 0.35, 95% CI 0.06-1.50 vs OR 1.71, 95% CI 1.27-2.30, p = 0.024).

    1. ETCO2 compared to SI to predict blood transfusion

ETCO2 <= 29.5 mmHg had a sensitivity of 76.3% and a specificity of 58.9% for predicting the need for any blood transfusion. SI >= 1 had a sen- sitivity of 21.1% and a specificity of 95.4% for predicting the need for any blood transfusion. The area under the ROC for ETCO2 <= 29.5 mmHg was 67.6% compared to 58.2% for SI >= 1. For patients with large volume blood transfusion, ETCO2 <= 29.5 mmHg had a sensitivity of 77.8%, a specificity of 57.8%, and an area under the ROC of 51.4%. The sensitivity, specificity, and area under the ROC for SI >= 1 in predicting large volume transfusion was 33.3%, 95.0%, and 57.7% respectively.

  1. Discussion

In our study, approximately 50% of admitted trauma patients had an initial non-invasive ETCO2 in the trauma bay of <29.5 mmHg, with a median ETCO2 in the low group of 22 mmHg, and this was associated with increased odds of blood transfusion, mortality, and adverse clinical outcomes, such as need for mechanical ventilation, inpatient complica- tions, and inferior disposition. Our findings extend the existing pre- hospital and intraoperative ETCO2 literature by examining the effect of prospectively obtained initial trauma bay ETCO2 on a wide range of clinical outcomes in a diverse non-intubated trauma population. Fur- thermore, most studies to date focus on mortality and blood transfusion as primary endpoints, but our data examine a broad range of clinical outcomes that allow us to extrapolate the efficacy of ETCO2 as a real- time trauma decision making tool.

Compared to our normal/high ETCO2 group, those patients with low ETCO2 were significantly older, which may predispose them to greater comorbidities, such as smoking status, COPD, and congestive heart fail- ure, which in turn may affect ETCO2 measurements. However, this was not shown to be the case such that the presence of comorbidities was evenly distributed between the low and normal/high ETCO2 groups. Nevertheless, we accounted for the potential ecological association of comorbidities, age, and ETCO2 by including these variables as covariates in our adjusted regression models.

Similarly, it is known that the presence of traumatic brain injury can affect ETCO2 measurements since it can drastically alter Respiratory function and is a leading cause of early mortality in trauma patients [24]. It was therefore hypothesized that the presence of head injury may confound the relationship between ETCO2 and trauma outcomes. Indeed, initial trauma bay GCS was significantly lower in the low ETCO2 group, although there was no difference in the presence of head injury among patients with low and normal/high ETCO2. To ad- dress this, we conducted a stratified analysis by presence of head injury, which was defined by AIS injury codes obtained from the trauma regis- try. Traditionally, a p-value <0.10 is considered potentially significant in an effect measure modification analysis, and we found that the presence of head injury did not meaningfully change the effect estimates of low ETCO2 on patient outcomes. Notably, odds of death were approximately equal among non-head injured and head-injured patients, which sug- gests that traumatic brain injury (in isolation) does not adequately ex- plain the greater risk of mortality with low ETCO2 measurements. It should be noted, however, that our stratified estimates are unadjusted, and outcomes are relatively rare, so it is possible that with a larger sam- ple size and more prevalent outcomes, some comparisons would be- come significant. Interestingly, in our stratified analysis, patients with low ETCO2 and head injury were less likely to be admitted to the hospi- tal compared to patients without head injury. The rationale for this find- ing is likely multifactorial, but it is possible that patients with low ETCO2 and head injury were more likely to die in the ED.

Previous studies examining low pre-hospital and intraoperative ETCO2 have reported greater odds of mortality in these patients com- pared to those with normal/high ETCO2 [13,14,25-27]. Indeed, our study extends these findings to the period immediately after patient ar- rival to the emergency department. After adjustment for age, comorbid- ities, GCS, and head injury status, patients with low ETCO2 in this critical time period had more than five times the odds of death compared to pa- tients with normal ETCO2. These results suggest the association be- tween low ETCO2 and mortality after trauma is robust and consistent regardless of the patient population or method of measurement. Inter- pretation of pre-hospital ETCO2 results can be challenging due to the lack of standardized pre-hospital indications for ETCO2 measurements in most EMS systems and is especially susceptible to bias if there is a tendency to measure ETCO2 only on higher risk patients [16]. Similarly, intraoperative ETCO2 measurement utilizes mainstream capnography in intubated patients, and comparisons to our in-hospital data pre- sented here is limited since our ETCO2 data was obtained using side- stream capnography in a non-intubated cohort [28]. Future studies may overcome these limitations by using standardized pre-hospital ETCO2 protocols, and by comparing pre-hospital, in-hospital, and intra- operative ETCO2 measurements to clinical outcomes in the same pa- tients using a repeated-measures or within-subjects study design.

Table 4

Unadjusted effect of low ETCO2 on patient outcomes, stratified by presence of head injury.

No Head Injury Head Injury Mantel-Haenszel Test of Homogeneity

N = 365 N = 123

OR (95% CI) OR (95% CI) p-value a

Hospital admission a 1.71 (1.27, 2.30) 0.35 (0.06, 1.50) 0.024

Blood transfusion 3.52 (1.39, 10.05) 7.38 (0.89, 337.84) 0.527

Mechanical ventilation 1.84 (0.79, 4.45) 4.00 (1.14, 17.61) 0.281

Operation 1.59 (0.92, 2.74) 1.90 (0.59, 6.72) 0.762

ICU stay 1.17 (0.73, 1.86) 2.13 (0.97, 4.69) 0.165

Inpatient complications b 1.85 (0.74, 4.94) – –

Inferior disposition c 1.50 (0.92, 2.45) 2.64 (1.01, 7.24) 0.271

Inpatient mortality 6.64 (0.79, 306.75) 6.21 (0.71, 290.36) 0.966

Abbreviations: ETCO2, end tidal carbon dioxide; OR, odds ratio; CI, confidence interval; ICU, intensive care unit. p < 0.05 is considered significant.

a Among all patients, n = 955.

b Includes respiratory failure, pneumonia, alcohol withdrawal syndrome, delay in treatment, unplanned intubation, unplanned admission to ICU, AKI, pneumothorax, ARDS, PE, stroke/ CVA, delirium, cardiac arrest with CPR, SSI. No patients with head injury had a complication listed.

c Includes discharge to skilled nursing facility, long term care facility, rehabilitation, and hospice.

ETCO2 has been suggested as a sensitive indicator of the need for blood transfusion among trauma and emergency department patients [26,29,30]. As a result, ETCO2 may play an important role in trauma tri- age and appropriate resource utilization that does not rely on pattern of injury or Imaging results. Compared to two commonly used physiologic indicators of occult shock, lactate and base deficit, ETCO2 can be ob- tained more quickly and is not contingent on Blood draws, and in our study, the median time from arrival to the trauma bay and ETCO2 mea- surement was 4 min. The SI is another widely used physiologic param- eter with potential value in detecting occult shock. As a screening tool, our results suggest that ETCO2 may be more predictive of the need for transfusion than SI, with improved sensitivity and greater area under the ROC. This is similar to previously published literature, which has shown that ETCO2 in the pre-hospital setting may outperform the SI for predicting blood transfusion [26,30]. ETCO2 was similarly more sen- sitive in predicting large volume transfusion than SI, although SI was more specific. However, it should be noted that previous studies have focused on the use of SI to predict massive transfusions, which is typi- cally defined as >10 units in 24 h, and the relationship of SI with smaller transfusion volumes is less defined [22]. Whether there is a graded re- sponse between degree of ETCO2 depression and volume of transfusion is of interest and yet to be determined. In practice, both ETCO2 and SI can be utilized synergistically, and incorporation of these more physio- logic patient measures may offer a more accurate assessment of trauma resource need than anatomic and mechanistic criteria, and may be par- ticularly helpful in avoiding under triage among patients who would otherwise qualify for lower trauma activation levels [31]. Indeed, in our study, five of six (83.3%) patients who died with the lowest tier trauma activation level had low ETCO2 (median 22.5 mmHg).

This study is not without limitations. First, although ETCO2 was

measured for all trauma patients at our institution as part of a prospec- tive research study, it was at the trauma attending’s discretion to opt out of ETCO2 collection, which may introduce some selection bias. Even though patients with low ETCO2 were significantly older, comorbidity prevalence and other demographic characteristics were similarly dis- tributed between low and normal/high ETCO2 groups. In addition, ISS, trauma activation level, presence of head injury, and initial trauma bay vitals were not significantly different among ETCO2 groups, which suggests that there was not a bias toward obtaining ETCO2 in only sicker or higher acuity patients. Second, we did not have access to administra- tion of pre-hospital medications, such as sedatives, which may affect ETCO2 measurements, although this would likely have occurred non- differentially, and thus affected both ETCO2 groups similarly. Third, it is known that side stream capnography is inferior to mainstream capnography, and is affected by a variety of patient factors, including hyperventilation, excess secretions, and mouth breathing; however, use of mainstream capnography limits the generalizability of ETCO2 in non-intubated patients and prevents its use as an easily obtainable prognostic indicator. Fourth, dichotomization of ETCO2 at one time point limits the nuance of data analysis in exchange for greater clinical utility. Examining trends in ETCO2 values would be helpful in discerning the effect of low ETCO2 on patient outcomes. Finally, it should be noted that cut point estimation relies on the dataset and not necessarily real- world physiology and may vary slightly among different patient popu- lations and protocols. However, the cut point of 29.5 mmHg, which was obtained using the Liu method of estimation, has moderate sensi- tivity, specificity, and is similar to previously published ETCO2 thresh- olds for malperfusion and to our previously calculated cut point using pre-hospital measurements [14,32].

In conclusion, trauma bay ETCO2 measurements can be rapidly ob- tained using non-invasive side-stream capnography and can serve as an important prognostic indicator for a broad range of clinical outcomes, including the need for blood transfusion, mechanical ventilation, inpa- tient complications, and inferior discharge disposition. Furthermore, these data suggest that ETCO2 is a useful tool for detecting those pa- tients that would otherwise escape the more traditional activation

criteria and may allow earlier intervention among injured patients who may otherwise not receive timely Hemorrhage control or blood component therapy. We envision ETCO2 as an important trauma bay adjunct that allows providers to individualize and optimize early trauma management.



Author contributions

JPT contributed to the literature search, data analysis, data interpre- tation, writing, and critical revision, and final review. RC contributed to literature review, data collection, Data interpretation, and final review. NC contributed to study design, data interpretation, critical revision, and final review. POU and SMM contributed to the study design, data in- terpretation, critical revision, and final review.


Portions of this manuscript were presented at the 2021 American Association for the Surgery of Trauma (AAST) conference (poster pre- sentation).

CRediT authorship contribution statement

Jaclyn N. Portelli Tremont: Writing – review & editing, Writing – original draft, Resources, Methodology, Formal analysis, Conceptualiza- tion. Ricardo A. Caldas: Data curation, Writing – original draft, Writing – review & editing. Nicole Cook: Writing – review & editing, Project ad- ministration, Methodology, Data curation, Conceptualization. Pascal Osi Udekwu: Conceptualization, Methodology, Supervision, Writing – review & editing. Scott M. Moore: Writing – review & editing, Writing – original draft, Supervision, Methodology, Formal analysis, Conceptual- ization.

Declaration of Competing Interest

The authors report no proprietary or commercial interest in any product mentioned or concept discussed in this article. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.


  1. Hess JR, Ramos PJ, Sen NE, Cruz-Cody VG, Tuott EE, Louzon MJ, et al. Quality manage- ment of a massive transfusion protocol. Transfusion. 2018;58(2):480-4.
  2. Dzik WS, Ziman A, Cohn C, Pai M, Lozano M, Kaufman RM, et al. Biomedical excel- lence for safer transfusion collaborative. Survival after ultramassive transfusion: a review of 1360 cases. Transfusion. 2016;56(3):558-63.
  3. Frankel DAZ, Acosta JA, Anjaria DJ, Porcides RD, Wolf PL, Coimbra R, et al. Physiologic response to hemorraghic shock depends on rate and means of hemorrhage. J Surg Res. 2007;143:276-80.
  4. Balogh AL, Petak F, Fodor GH, Tolnai J, Csorba Z, Babik B. Capnogram slope and ven- tilation dead space parameters: comparison of mainstream and sidestream tech- niques. Br J Anaesth. 2016;117(1):109-17.
  5. Hunter CL, Silvestri S, Dean M, Falk JL, Papa L. End-tidal carbon dioxide is associated with mortality and lactate in patients with suspected sepsis. Am J Emerg Med. 2013; 31(1):64-71.
  6. Long B, Koyfman A, Vivirito MA. Capnography in the emergency department: a re- view of uses, waveforms, and limitations. J Emerg Med. 2017;53(6):829-42.
  7. Stone Jr ME, Kalata S, Liveris A, Adorno Z, Yellin S, Chao E, et al. end-tidal CO2 on ad- mission is associated with hemorrhagic shock and predicts the need for massive transfusion as defined by the critical administration threshold: a pilot study. Injury. 2017;48(1):51-7.
  8. Williams DJ, Guirgis FW, Morrissey TK, Wilkerson Wears RL, Kalynych C, Kerwin AJ, et al. End-tidal carbon dioxide and occult injury in trauma patients: ETCO2 does not rule out severe injury. Am J Emerg Med. 2016;34(11):2146-9.
  9. Caputo ND, Fraser RM, Paliga A, Matarlo J, Kanter M, Hosford K, et al. Nasal cannula end-tidal CO2 correlates with Serum lactate levels and odds of operative

intervention in penetrating trauma patients: a prospective cohort study. J Trauma Acute Care Surg. 2012;73(5):1202-7.

  1. Pantazopoulos C, Xanthos T, Pantazopoulos I, Papalois A, Kouskouni E, Iacovidou N. A review of carbon dioxide monitoring during adult cardiopulmonary resuscitation. Heart Lung Circ. 2015;24(11):1053-61.
  2. Link MS, Berkow LC, Kudenchuk PJ, Halperin HR, Hess EP, Motra VK, et al. Part 7: adult advanced cardiovascular life support: 2015 American Heart Association guide- lines update for cardiopulmonary resuscitation and emergency cardiovascular care. Circulation. 2015;132(18 Suppl 2):S444-64.
  3. Murphy RA, Bobrow BJ, Spaite DW, Hu C, McDannold R, Vadeboncoeur TF. Associa- tion between prehospital CPR quality and end-tidal carbon dioxide levels in out-of- hospital cardiac arrest. Prehosp Emerg Care. 2016;20(3):369-77.
  4. Deakin CD, Sado DM, Coats TJ, Davies G. Prehospital end-tidal carbon dioxide con- centration and outcome in major trauma. J Trauma Acute Care Surg. 2004;57(1): 65-8.
  5. Bryant MK, Portelli Tremont JN, Patel Z, Cook N, Udekwu PO, Reid T, et al. Low initial pre-hospital end-tidal carbon dioxide predicts inferior clinical outcomes in trauma patients. Injury. 2021.;S0020-1383(21) (00624-0).
  6. Newgard CD, Zive D, Holmes JF, Bulger EM, Staudenmayer K, Liao M, et al. WESTRN investigators. A multisite assessment of the American College of Surgeons Commit- tee on trauma field triage decision scheme for identifying seriously injured children and adults. J Am Coll Surg. 2011;213(6):709-21.
  7. van Rein EAJ, van der Sluijs R, Houwert RM, Gunning AC, Lichtveld RA, Leenen LPH, et al. Effectiveness of prehospital trauma triage systems in selecting severely injured patients: is comparative analysis possible? Am J Emerg Med. 2018;36(6):1060-9.
  8. Brown JB, Stassen NA, Bankey PE, Sangosanya AT, Cheng JD, Gestring ML. Mecha- nism of injury and special consideration criteria still matter: an evaluation of the Na- tional Trauma Triage Protocol. J Trauma Acute Care Surg. 2011;70(1):38-44.
  9. Liu X. Classification accuracy and cut point selection. Stat Med. 2012;31:2676-86.
  10. Maegele M, Brockamp T, Nienaber U, Probst C, Schoechl H, Gorlinger K, et al. Predic- tive models and algorithms for the need of transfusion including massive transfu- sion in severely injured patients. Transfus Med Hemother. 2012;39:85-97.
  11. McNamee R. Regression modelling and other methods to control confounding. Occup Env Med. 2005;62:500-6.
  12. Cannon CM, Braxton CC, Kling-Smith M, Mahnken JD, Carlton E, Moncure M. Utility of the shock index in predicting mortality in traumatically injured patients. J Trauma Acute Care Surg. 2009;67(6):1426-30.
  13. Schroll R, Swift D, Tatum D, Couch S, Heaney JB, Llado-Furrulla M, et al. Accuracy of shock index versus ABC score to predict need for massive transfusion in trauma pa- tients. Injury. 2018;49:15-9.
  14. Zhu CS, Cobb D, Jonas RB, Pokorny D, Rani M, Cotner-Pouncy T, et al. Shock index and Pulse pressure as triggers for massive transfusion. J Trauma Acute Care Surg. 2019; 87:S159-64.
  15. Sauaia A, Moore FA, Moore EE, Moser K, Brennan R, Read RA, et al. Epidemiology of Trauma deaths. J Trauma. 1995;38(2):185-93.
  16. Childress K, Arnold K, Hunter C, Ralls G, Papa L, Silvestri S. Pre-hospital end-tidal car- bon dioxide predicts mortality in trauma patients. Prehosp Emerg Care. 2018;22(2): 170-4.
  17. Campion EM, Cralley A, Robinson C, Sauaia A, Pieracci FM, Lawless RA, et al. Prehos- pital end-tidal carbon dioxide predicts massive transfusion and death following trauma. J Trauma Acute Care Surg. 2020;89(4):703-7.
  18. Demaree C, Simpson JT, Smith A, Guidry C, McGrew P, Schroll R, et al. Intraoperative end-tidal CO2 as a predictor of mortality in trauma patients receiving massive trans- fusion protocol. Am Surg. 2019;85(12):e617-9.
  19. Pekdemir M, Cinar O, Yilmaz S, Yaka E, Yuksel M. Disparity between main- stream and sidestream End-tidal carbon dioxide values and arterial carbon dioxide levels. Respir Care. 2013;58:1152-6.
  20. Wilson BR, Bruno J, Duckwitz M, Akers N, Jeanmonod D, Jeanmonod R. Prehospital end-tidal CO2 as an early marker for transfusion requirement in trauma patients. Am J Emerg Med. 2021;45:254-7.
  21. Bulger N, Harrington B, Krieger J, Latimer A, Arbabi S, Counts CR, et al. Prehospital end tidal carbon dioxide predicts hemorrhagic shock upon emergency department arrival. J Trauma Acute Care Surg. 2021;91:457-64.
  22. Day DL, Terada KEF, Vondrus P, Watabayashi R, Severino R, Inn H, et al. Correlation of nasal cannula end-tidal carbon dioxide concentration with need for critical re- sources for blunt trauma patients triaged to lower-tier trauma activation. J Trauma Nurs. 2020;27(2):88-95.
  23. Dunham CM, Chirichella TJ, Gruber BS, Ferrari JP, Martin JA, Luchs BA, et al. In emergently ventilated trauma patients, low end-tidal CO2 and low cardiac output are associated and correlate with hemodynamic instability, hemorrhage, abnormal pupils, and death. BMC Anesthesiol. 2013;13(1):20.

Leave a Reply

Your email address will not be published. Required fields are marked *