International Journal of Academic Medicine

: 2019  |  Volume : 5  |  Issue : 1  |  Page : 51--56

End-tidal carbon dioxide on emergency department arrival predicts trauma patient need for transfusion, vasopressors, and operative hemorrhage control in the first 24 hours

Rebecca Jeanmonod, John Tran, Dhanalakshmi Thiyagarajan, Bryan Wilson, Jason Black, Saira Agarwala, Donald Jeanmonod 
 Department of Emergency Medicine, St. Luke's University Health Network, Pennsylvania, USA

Correspondence Address:
Dr. Rebecca Jeanmonod
St. Luke's University Health Network, 801 Ostrum St, Bethlehem, Pennsylvania 18015


Introduction: Predicting early need for trauma resuscitation is an important goal of trauma systems. Aim/Hypothesis: We sought to compare the predictive value of end-tidal carbon dioxide (EtCO2) compared to hemoglobin, shock index, systolic blood pressure, heart rate, and respiratory rate in determining transfusion requirement, need for operative hemorrhage control, or pressor use in the first 24 h after trauma alert activation. Materials and Methods: A convenience sample of trauma patients at a Level 1 community trauma center were enrolled. EtCO2 was measured via nasal cannula upon arrival, in addition to initial hemoglobin levels and vital signs. Once stable and competent, patients or families were consented. EtCO2 average over three measurements 3 min apart was used as the EtCO2 value. The electronic medical record was reviewed by a trained research associate who was not involved in the care of the patient to determine the need for transfusion, pressor use, or operative hemorrhage control within the first 24 h of hospitalization. Results: Fifty patients were enrolled, with a median age of 52 years. Seven patients required transfusion, pressor use, and/or surgical hemorrhage control. Vital signs between groups were statistically no different. The mean EtCO2 in patients requiring transfusion was 26.8 (19.5–34.1) versus 34.1 (31.8–36.3) in those who did not (P = 0.027). A cutoff value of EtCO2 ≤33 mmHg was 100% sensitive and 62.8% specific, with an area under the curve of 0.889. Conclusion: EtCO2 has a high sensitivity in predicting the need for intervention in trauma patients. Additional research is needed to determine further utility of this value in the triage and treatment of trauma patients. The following core competency statement: Patient care and Systems-based practice.

How to cite this article:
Jeanmonod R, Tran J, Thiyagarajan D, Wilson B, Black J, Agarwala S, Jeanmonod D. End-tidal carbon dioxide on emergency department arrival predicts trauma patient need for transfusion, vasopressors, and operative hemorrhage control in the first 24 hours.Int J Acad Med 2019;5:51-56

How to cite this URL:
Jeanmonod R, Tran J, Thiyagarajan D, Wilson B, Black J, Agarwala S, Jeanmonod D. End-tidal carbon dioxide on emergency department arrival predicts trauma patient need for transfusion, vasopressors, and operative hemorrhage control in the first 24 hours. Int J Acad Med [serial online] 2019 [cited 2019 May 21 ];5:51-56
Available from:

Full Text


Although it is declining in incidence, exsanguination remains a significant cause of morbidity and early death in trauma.[1],[2] Acute decrease in blood volume leads to reduced cardiac output, tissue hypoperfusion, and shock, which is sometimes reversible with early resuscitation and transfusion.[3] Eight percent of trauma patients require blood transfusion, with 62% of transfusions occurring in the first 24 h.[4] In spite of the frequency with which transfusions are required in the trauma patients, thresholds for transfusion are variable, and heart rate and blood pressure, which are commonly used as screening criteria, are not reliable enough to consistently identify early or impending hemorrhagic shock.[5] Serum lactate measurements are useful to predict shock and mortality in both sepsis and trauma, but because of time needed to draw blood and run the analyzer, the use is limited in real time.[5],[6],[7]

Because carbon dioxide and sodium bicarbonate are the major buffering system in the blood, it stands to reason that, in cases of hemorrhage when patients become acidotic, the value of carbon dioxide (CO2) will change immediately and will correlate with acidosis. End-tidal CO2 (EtCO2) can be measured using a nasal cannula, noninvasively at the bedside, providing the care team with real-time values. There is evidence that EtCO2 levels can provide some prognostic values to the management of cardiac arrest and correlate to a decreased cardiac output and, subsequently, a decreased cardiac index.[8],[9],[10] There is also limited evidence that EtCO2 levels are associated with in-hospital mortality in suspected sepsis.[11] In trauma patients, EtCO2 measured in the prehospital setting as well as on admission has been shown to correlate with mortality,[9],[12],[13] although, in a single study, it did not show adequate sensitivity to rule out admission to intensive care unit, transfusion, surgical intervention, acute blood loss anemia, or acute clinically significant findings on computed tomography imaging.[14] A pilot study demonstrated that end-tidal CO2 on admission might predict the need for massive transfusion in trauma patients, and a single study has assessed EtCO2 in penetrating trauma as a predictor for the need for operative hemorrhage control.[5],[15]

As the existing evidence is limited and conflicted, we sought to build on existing literature and assess whether EtCO2 measured in nonintubated trauma patients on arrival to the emergency department (ED) predicted (1) transfusion requirement in the first 24 h of hospitalization, (2) need for operative hemorrhage control, or (3) pressor use. We compared the predictive value of EtCO2 as compared to other noninvasive means of patient assessment, including shock index, systolic blood pressure, heart rate, and respiratory rate.

 Materials and Methods

Study design

This is an exploratory prospective cohort study of awake adult patients presenting as trauma alerts to a level 1 trauma center. A convenience sample of nonintubated trauma patients were identified by research associates or medical providers. The data for EtCO2 and vital signs were collected real time during the patients' evaluation, and patients or their surrogates provided written consent for participation in the study after the patient was stabilized, sober, and competent. Patient's charts were subsequently reviewed for the results of laboratory studies and interventions, including surgical hemorrhage control or transfusion requirement in the first 24 h after admission. The study protocol did not alter patient management. The provider recording EtCO2 measurements was not involved in decisions regarding patient care. The research protocol was reviewed and approved by the institutional review board, and all investigators completed mandatory research training and certification.

Study setting and population

The study was performed at a community Level 1 trauma center that has a volume of over 2000 trauma alerts each year. The study facility is in the third largest metropolitan area in Pennsylvania. The hospital hosts an emergency medicine residency, a general surgery residency, and a trauma/critical care fellowship.

Any patient who was designated as a trauma alert was eligible for the study if he or she wasage 18 years or older, English speaking, and breathing spontaneously without the need for immediate definitive airway protection with endotracheal intubation. Patients were required to be able to maintain adequate oxygen saturation (≥90%) with either no supplemental oxygen or nasal cannula oxygen alone. Patients were excluded if they were younger than 18 years old or non-English speaking. Patients requiring non-rebreather oxygen or intubation or patients presenting with an endotracheal tube or supraglottic airway already in place were excluded. Patients were not excluded on the basis of trauma type (penetrating versus blunt) or trauma acuity (at the study institution, “Level A” [with trauma team, emergency medicine, anesthesia, respiratory therapy, blood products, and an operating room immediately available] versus “Level B” [with trauma team and emergency medicine response only]). Patients were enrolled in the study by trained research personnel, and all patients provided written informed consent.

Study protocol and measurements

Patients were enrolled as a convenience sample during times when study investigators were available. Upon arrival to the trauma bay, eligible patients were placed on an EtCO2 nasal cannula detector (CapnoFlex LF CO2 Module Manufactured for GE Medical Systems Information Technologies Inc., 8200 W. Tower Ave, Milwaukee, USA) by the trauma airway team. EtCO2 measurements were gathered during the primary and secondary survey and recorded by the study personnel. EtCO2 measurement was recorded on a standardized data collection sheet every 3 min with a minimum of two recordings per patient. Demographic information was recorded after the initial point of care.

Once patients were determined to be stable, sober, and competent, they were approached to provide written informed consent for enrollment in the study. For some patients, this occurred after trauma evaluation while the patient was still in the ED or trauma bay. For other patients, this occurred during the course of their inpatient hospitalizations. No patients were consented after they had been discharged from the hospital. Patients who declined to participate had their EtCO2 data destroyed, and no further data were collected on these patients. Consenting patients were followed through review of their electronic medical records to determine the need for transfusion within the first 24 h of hospitalization, operative control of hemorrhage, and pressor use. Laboratory studies recorded including lactate, serum bicarbonate, anion gap, and hemoglobin. Patients' vital signs were also recorded.

All raw data from hospitalization were entered into a standardized data collection sheet by the trained study personnel and transferred into a standardized Microsoft Excel 2007 spreadsheet (Microsoft Corporation, Redmond, WA) by a single nonphysician study associate who was not involved in patient enrollment or consent. EtCO2 data were analyzed using both the mean value and using the lowest value obtained.

Data analysis

EtCO2 values were normally distributed. A two-tailed t-test was used to compare EtCO2 values between patients requiring transfusion/operative intervention in the first 24 h after admission and those who did not. Blood pressure, heart rate, and hemoglobin were also compared in these two groups. Pearson's correlation (R) and correlation coefficient (r2) were calculated to determine any correlations between EtCO2 values and outcomes. In addition to correlations, we used a cutoff of EtCO2 =35 mmHg to calculate sensitivity and specificity of EtCO2 as a screening test for transfusion need based on a prior study.[15] We explored other cutoff values to optimize the performance of EtCO2 as a transfusion predictor. Demographic data and summary statistics were analyzed using descriptive statistics and Chi-square. All data were analyzed using MedCalc (© 1993–2013, Ostend, Belgium) and (© Richard Lowry 1998–2018).


A total of 50 patients were enrolled with a median age of 52 years. Ages ranged from 18 to 88 years. Seven patients required transfusion, surgical hemorrhage control, pressor use, or some combination of these. The median age of patients undergoing intervention/stabilization was 64.5 years (interquartile range [IQR]: 30–84) whereas those not undergoing intervention/stabilization was 50.5 years (IQR: 33–56) (P = 0.07). Ninety-four percent of patients suffered from blunt trauma, with 60% related to motor vehicle crashes. Further demographic information is listed in [Table 1].{Table 1}

Vital signs (including respiratory rate, heart rate, systolic blood pressure, and shock index) on arrival to the trauma bay were not significantly different in patients requiring transfusion versus those who did not [Table 2]. EtCO2, however, was significantly lower in patients requiring transfusion, operative hemorrhage control, or pressors as compared to those who did not [Table 3].{Table 2}{Table 3}

Using the prior published EtCO2 level of 35 mmHg as a cutoff threshold, EtCO2 had 100% sensitivity for predicting need for transfusion, surgery, or pressors (confidence interval [CI]: 72%–100%). However, the specificity was only 56% (CI: 40%–71%), given that our mean EtCO2 in patients not requiring intervention was 34. In this small cohort, lowering the threshold to 33 did not change the sensitivity of EtCO2, but improved the specificity to 63% (CI: 47%–77%). Receiver operator curve for sensitivity and specificity of EtCO2 is shown in [Figure 1]. Area under the curve = 0.89.{Figure 1}


EtCO2 monitors measure the amount of exhaled carbon dioxide as a partial pressure (normal range, 35–45 mmHg). EtCO2 values provide insight into the body's ability to move air (ventilation), CO2 transportation around the body (perfusion), and CO2 production at the cellular level (metabolism). These values are distinct from oxygen saturation, measured by arterial blood gas. There are two forms of CO2 measurement: numerical or capnometry and waveform or capnography.[16] EtCO2 capnometry is a measurement of the amount of CO2 in the airway at a point in a time, specifically at the end of exhalation, while waveform capnography shows the progression of CO2 over the course of the breathing cycle. The shape of the waveform during normal and pathological respiratory states has numerous clinical applications related to CO2 production, transport, ventilation, and ventilation-to-perfusion ratio, but these are beyond the scope of this study. EtCO2 does not rely on interpretation of waveforms for its use, and this technology has become the gold standard in prehospital endotracheal tube confirmation as well as for sedation monitoring in emergency settings.[16],[17] EtCO2 monitoring relies on infrared radiation to detect CO2 concentration in a sample line based on the wavelength of radiation absorbed by CO2 and has been shown to be a reliable and accurate measure that correlates well with PaCO2.[18] Carbon dioxide–carbonic acid–bicarbonate is the primary buffer system in blood, which makes carbon dioxide a marker for changes in acid/base status. Because it is noninvasive, rapid, and reliable, the clinical use of EtCO2 is actively evolving for prognostic applications in sepsis, trauma, and cardiac arrest.

In our convenience sample of patients, we found statistically significant differences in mean EtCO2 and hemoglobin between patients requiring blood transfusion versus those that did not. Indeed, hemoglobin level had a stronger association with transfusion than EtCO2 in this cohort. Finding a statistically significant correlation between anemia and blood product transfusion makes sense, as often anemia itself is an indication for transfusion, and practitioners were not blinded to hemoglobin measurements. Especially in medical management, there are established thresholds for transfusion based on hemoglobin level.[19] In the case of acute blood loss anemia in the setting of trauma, patients bleed whole blood without time for equilibration, and therefore, hemoglobin at the point of care may be initially misleading, causing delay to resuscitation. We did not assess timing of transfusion, operative intervention, or pressor use beyond a binary assessment of intervention before 24 h. Therefore, additional research is needed to determine if EtCO2 values decline faster than measured hemoglobin levels in actively bleeding patients over time.

A significant downside to hemoglobin measurement is that it requires phlebotomy as well as time for blood analysis (whether by traditional laboratory analyzer or i-stat analysis), which delays the ability to act upon that data point. EtCO2 is obtained instantaneously upon patient arrival, requiring only placement of a nasal cannula. Vital signs can also be obtained readily and in a noninvasive manner, but EtCO2 performed favorably as compared to all vital signs as well as shock index. Interestingly, there was a trend for higher respiratory rate in patients requiring transfusion versus those who did not (P = 0.09), which likely reflects their increased respiratory drive as their bodies attempt to eliminate CO2 and buffer their acidosis with ventilation. Although we did not study it here, EtCO2 is recorded along with vital signs in many prehospital settings. Therefore, EtCO2 has potential to be used as a physiologic triage criterion in the trauma patients (vs. traditional vital signs or mechanism of injury alone) to guide transportation decisions and care before arrival.

The CIs surrounding the intervention group in our study are wide due to the relatively low number of patients who required intervention. Our study was not designed to assess forany correlation between the amount of bleeding (as measured by hemoglobin decline or by total amount of blood transfused) and EtCO2 value, although this is an area for future study. Although we report out the EtCO2 values for all of our intervention endpoints, most patients met more than one endpoint. For instance, for the two patients who received pressors (neither for spinal shock), one was also transfused and the other required operative hemorrhage control. For this reason, our receiver operating characteristic represents the pooled data of all endpoints.

Our sample size also precludes making any conclusions on the use of EtCO2 on the basis of mechanism of trauma. Although we did not exclude patients with penetrating trauma, the patients enrolled were largely victims of blunt trauma. We expect findings would be similar in the setting of penetrating trauma as demonstrated in a prior study but cannot extrapolate on the basis of our data alone.[5] It is also unclear that what role comorbid disease or other injury might contribute to fluctuations in EtCO2. We excluded patients with chronic lung disease on the presupposition that their baseline CO2 values may be elevated from chronic retention, thus reducing EtCO2 utility. However, we did not assess foracute or chronic alcohol use, liver disease (which may cause a primary respiratory alkalosis), or congestive heart failure. Especially in an older patient population like ours, these factors may be important. We also did not have enough injured patients to study the impact of chest/lung trauma or head injury on EtCO2 levels. It would be helpful to better delineate how all these factors contribute to EtCO2 levels.


Our study is limited by its small sample size and convenience population enrolled from a single trauma center. There is inherent bias in all convenience samples, and our results may not be generalizable to all institutions. Overall, we had a small number of patients who actually needed intervention (14% of total). Our population was slightly older than many patients in trauma-based studies, and therefore, our data should be interpreted with caution in younger populations. In addition, our population were primarily victims of blunt trauma, and therefore, our data may not be applied to penetrating trauma victims.

Our study also only enrolled consenting, nonintubated patients. This necessarily eliminated enrollment of the sickest patients, many of whom are intubated before arrival, or are too ill to undergo consent. This precluded our study of the population in whom this technology may have its most significant application.

We did not apply continuous EtCO2 monitoring to our patients and, therefore, are unable to state whether the measurement is helpful as a trend over time, either in the trauma bay or on an ongoing basis during admission. This would be an area for further study.


EtCO2 has a high sensitivity in predicting the need for transfusion, pressor use, and operative intervention in trauma patients. Additional research is needed to determine further utility of this value in the triage and treatment of trauma patients.

Ethical conduct of research

The study was reviewed and approved by the St. Luke's University Health Network IRB.

Financial support and sponsorship


Conflicts of interest

There are no conflicts of interest.


1Jochems D, Leenen LP, Hietbrink F, Houwert RM, van Wessem KJ. Increased reduction in exsanguination rates leaves brain injury as the only major cause of death in blunt trauma. Injury 2018;49:1661-7.
2Sauaia A, Moore FA, Moore EE, Moser KS, Brennan R, Read RA, et al. Epidemiology of trauma deaths: A reassessment. J Trauma 1995;38:185-93.
3Gutierrez G, Reines HD, Wulf-Gutierrez ME. Clinical review: Hemorrhagic shock. Crit Care 2004;8:373-81.
4Como JJ, Dutton RP, Scalea TM, Edelman BB, Hess JR. Blood transfusion rates in the care of acute trauma. Transfusion 2004;44:809-13.
5Caputo 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;73:1202-7.
6Hunter 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;31:64-71.
7Goyal M, Pines JM, Drumheller BC, Gaieski DF. Point-of-care testing at triage decreases time to lactate level in septic patients. J Emerg Med 2010;38:578-81.
8Paiva EF, Paxton JH, O'Neil BJ. The use of end-tidal carbon dioxide (ETCO2) measurement to guide management of cardiac arrest: A systematic review. Resuscitation 2018;123:1-7.
9Dunham 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;13:20.
10Lepilin MG, Vasilyev AV, Bildinov OA, Rostovtseva NA. End-tidal carbon dioxide as a noninvasive monitor of circulatory status during cardiopulmonary resuscitation: A preliminary clinical study. Crit Care Med 1987;15:958-9.
11Domsky M, Wilson RF, Heins J. Intraoperative end-tidal carbon dioxide values and derived calculations correlated with outcome: Prognosis and capnography. Crit Care Med 1995;23:1497-503.
12Holmes J, Peng J, Bair A. Abnormal end-tidal carbon dioxide levels on emergency department arrival in adult and pediatric intubated patients. Prehosp Emerg Care 2012;16:210-6.
13Childress K, Arnold K, Hunter C, Ralls G, Papa L, Silvestri S, et al. Prehospital end-tidal carbon dioxide predicts mortality in trauma patients. Prehosp Emerg Care 2018;22:170-4.
14Williams DJ, Guirgis FW, Morrissey TK, Wilkerson J, Wears RL, Kalynych C, et al. End-tidal carbon dioxide and occult injury in trauma patients: ETCO2 does not rule out severe injury. Am J Emerg Med 2016;34:2146-9.
15Stone ME Jr. Kalata S, Liveris A, Adorno Z, Yellin S, Chao E, et al. End-tidal CO2 on admission is associated with hemorrhagic shock and predicts the need for massive transfusion as defined by the critical administration threshold: A pilot study. Injury 2017;48:51-7.
16Long B, Koyfman A, Vivirito MA. Capnography in the emergency department: A Review of uses, waveforms, and limitations. J Emerg Med 2017;53:829-42.
17Donald MJ, Paterson B. End tidal carbon dioxide monitoring in prehospital and retrieval medicine: A review. Emerg Med J 2006;23:728-30.
18Zhang C, Wang M, Wang R, Wang W. Accuracy of end-tidal CO2 measurement through the nose and pharynx in nonintubated patients during digital subtraction cerebral angiography. J Neurosurg Anesthesiol 2013;25:191-6.
19Shander A, Javidroozi M, Naqvi S, Aregbeyen O, Caylan M, Demir S, et al. An update on mortality and morbidity in patients with very low postoperative hemoglobin levels who decline blood transfusion (CME). Transfusion 2014;54:2688-95.