Manual of Emergency Airway Management, 3rd Edition

39.Pulse Oximetry and Capnography/Capnometry

Michael F. Murphy

Baruch Krauss

Pulse Oximetry

Pulse oximetry provides a noninvasive and continuous means of rapidly determining arterial oxygen saturation and its changes. The devices are easy to use and interpret, pose no risk to the patient, and are relatively inexpensive. However, reliable interpretation of the information provided by these devices requires an appreciation of their limitations in certain situations.

Principles of Measurement

There are two kinds of oximetry used in clinical practice: transmission and reflectance. Both transmission and reflectance oximeters consist of light sources, typically red and infrared (IR), and a photodetector (“photodiodes”). The transmission pulse oximeter is the most common type of oximetry employed in clinical practice. It positions light-emitting diodes on one side of a tissue bed (e.g., finger or ear lobe) and a photo detector on the opposite side of the tissue bed. Reflectance oximeters position the emission source and the detector side by side or at least in close proximity to each other. Transmission oximeters have to deal with other absorbers in the light path, including skin, soft tissue, and venous and capillary blood, whereas reflectance oximeters do not.

The principle of “oximetry” is based on oxyhemoglobin (OxyHb) and deoxyhemoglobin having different absorption spectra at the two commonly used sensor wavelengths of 660 nm (red) and 940 nm (IR). At the wavelength of red light (660 nm), reduced hemoglobin absorbs about ten times as much light as OxyHb.

Pulse oximeters assess the pulsatile variation of red and IR light transmitted through (transmission) or absorbed by (reflectance) a tissue bed. These factors are divided into a pulsatile (AC) component due to the pulsatile flow of arterial blood and a nonpulsatile (DC) component of the tissue bed that includes venous blood, capillary blood, and nonpulsatile arterial blood. Data averaged over several arterial pulse cycles are then presented as saturation (S_pO₂). Studies have shown an excellent correlation between arterial hemoglobin (Hb) oxygen saturation and pulse oximeter saturation.

Indications

Pulse oximetry is particularly useful in the emergency department (ED) evaluation of patients with acute cardiopulmonary disorders such as chest trauma, bronchiolitis, asthma, heart failure, and chronic obstructive pulmonary disease. It is a standard monitoring parameter for patients undergoing sedation and in patients with a decreased level of consciousness, such as intoxication, overdose, and head injury. Its ability to decrease the frequency with which arterial blood gases are done has also been demonstrated, particularly when used in conjunction with end-tidal carbon dioxide (ETCO₂) determination. Continuous monitoring may indicate the insidious development of shock as vasoconstriction develops. Continuous oximetry is mandatory in patients requiring definitive airway management.

Limitations and Precautions

Limitations to the accuracy of pulse oximetry exist with severe vasoconstriction (e.g., shock, hypothermia), excessive movement, synthetic fingernails and nail polish, severe anemia, or the presence of abnormal hemoglobins. Reflectance oximetry has been demonstrated to reflect oxygen saturations more accurately in the setting of hypothermia and vasoconstriction. Carboxyhemoglobin (COHb) and methemoglobin (MetHb) contribute to light absorption and cause errors in the pulse oximetry readings. The pulse oximeter sees COHb as though it was mostly OxyHb and gives a falsely high reading. MetHb produces a large pulsatile absorbance signal at both the red and IR wavelengths. This effect forces the absorbance ratio toward unity, which corresponds to a S_pO₂ of 85%. Thus, in the presence of high levels of MetHb, the S_pO₂ is erroneously low when the arterial saturation is above 85% and erroneously high when the arterial saturation is below 85%. In dark-skinned patients, erroneously high readings (about 3%–5%) and a higher incidence of failure to detect signal have been reported.

Pulse oximetry has been shown to be of limited accuracy and reliability during cardiopulmonary resuscitation (CPR) and may be misleading. Nevertheless, its use is indicated during resuscitation because information useful to patient management may be gleaned.

In general, transmission oximetry signals are weaker from ears than from fingers, except in the face of hypotension or peripheral vasoconstriction, but ear responses are faster. Nasal bridge probes have been reported to read falsely high in some circumstances. When compared to reflectance sensors, there are fewer potential sensor sites, the response time can be lengthy, and its performance is more adversely affected by ambient light, motion, and poor perfusion.

Reflectance oximeters have emitting and sensing diodes on a single surface. Therefore, they can be placed in anatomical locations that do not require two surfaces in close proximity with a vascular bed in between. In fact, “oximetric Swan-Ganz” catheters employ this technology to monitor venous oxygen saturation. Placement on the forehead implies that there is better correlation with “core” arterial oxygen saturation. The big disadvantage of reflectance sensors lies in their propensity to be subject to contaminating sources of tissue (e.g., arteries, pigmentation).

In general, the reflectance sensor is more sophisticated in design, requires more user skill (to correctly apply the sensor and interpret the results, including waveform morphology), can provide more parameters, and has greater versatility than transillumination methods.

The pulse oximeter measures oxygen saturation, not the partial pressure of oxygen. This means that it is possible for the partial pressure of oxygen to fall substantially before the oxygen saturation starts to fall. The explanation lies in the sigmoid shape of the OxyHb dissociation curve. If a healthy adult patient is given 100% oxygen to breathe for a few minutes and then ventilation ceases for any reason, several minutes may elapse before the oxygen saturation starts to fall (see Chapter 3). The pulse oximeter in these circumstances therefore warns of a potentially fatal complication several minutes after it has begun. This is sometimes referred to as monitor lag.

Finally, adequate oxygen saturation does not indicate that the patient is ventilating adequately. This is particularly important in patients with decreased levels of consciousness, such as during procedural sedation. Routine carbon dioxide (CO₂) monitoring during procedural sedation is a better means of assessing the adequacy of minute ventilation (see the next section).

End-Tidal Carbon Dioxide Monitoring

The concentration of CO₂ in exhaled breath is intrinsically linked to tissue metabolism, pulmonary (central) circulation, and ventilation. Capnography is the noninvasive graphic record of instantaneous CO₂concentrations in the respired gases during a respiratory cycle represented as a waveform or capnogram. Although the concentrations of CO₂can be displayed continuously through the respiratory cycle, it is conventional that only the maximum CO₂ concentration at the end of each tidal exhalation, the ETCO₂, is displayed. Capnometry is the quantitative measurement of ETCO₂ on a visual display without a waveform. Colorimetric detectors use color scales to estimate ranges of ETCO₂, but are not sufficiently accurate to give quantitative measurements. Their use is therefore limited to confirmation of endotracheal tube (ETT) placement in the trachea and its continuous location in the trachea.

ETCO₂ monitoring is becoming increasingly common in the ED and prehospital setting. A 1998 survey of emergency medicine training programs in the United States found that more than 80% use CO₂ monitoring, with colorimetric devices being the most frequently used at that time.

Principles of Measurement

Four spectrographic methods are used to measure ETCO₂ concentration in expired gases: IR, mass, Raman, and photoacoustic spectrography with most stand-alone monitors using the IR technique. CO₂selectively absorbs IR light with a wavelength of 4.26 µ. The concentration of CO₂ in a gas can be determined by passing filtered IR light through the gas and comparing the amount of light that is absorbed to a reference light beam passing through a CO₂-free chamber.

CO₂ monitors are configured as either sidestream or mainstream, depending on the location of the photoelectric detector or sensor. Sidestream monitors, more likely to be encountered in emergency medical services and the ED, aspirate a sample of gas through a small catheter into a measuring chamber located inside the monitor. They are lightweight, can be used in intubated and nonintubated patients, and have nasal-oral airway interfaces (cannula) that simultaneously sample CO₂ and deliver low-flow oxygen. This allows for preoxygenation and continuous oxygen delivery during procedural sedation and analgesia. High-flow sidestream systems (150 cc/minute) have several disadvantages, including plugging by secretions, 2- or 3-second delays in response time, and air leaks, which can dilute the sample. Newer generation low-flow systems (50 cc/minute) do not have these problems. Mainstream monitors, useful only in intubated patients, are bulky and heavy, with the sensor positioned at the hub of the ETT, and because they must be heated to prevent condensation, may burn patients.

Colorimetric CO₂ detectors use pH-sensitive filter paper impregnated with metacresol purple, which changes color from purple (<4 mm Hg CO₂) to tan (4–15 mm Hg CO₂) to yellow (>20 mm Hg CO₂), depending on the concentration of CO₂, although there is some variability in absolute numbers based on the brand of device. The indicator, housed in a plastic casing, is inserted between the ETT and the ventilator bag, and responds quickly enough to detect changes on a breath-by-breath basis. They are inexpensive and easy to use, and should be available in every ED for confirmation of ETT placement if quantitative methods are not available.

In otherwise healthy patients, a close correlation exists between ETCO₂ (i.e., the alveolar CO₂ [PaCO₂]) and the arterial partial pressure (PaCO₂). The ETCO₂ is approximately 2 to 5 mm Hg less than the PaCO₂ because of the contribution of physiological dead space to the end-expiratory gases. Most conditions that affect ventilation-perfusion ratios can widen the Pa-ETCO₂ gradient, including pulmonary embolism, cardiac arrest, hypovolemia, obstructive lung disease, and the lateral decubitus position. Although ETCO₂ may not always accurately reflect the absolute PaCO₂ in critically ill patients, it is still valuable in detecting ventilatory trends and identifying sudden airway events (e.g., apnea, extubation, pulmonary embolism).

Phases of the Capnogram

Analysis of the waveform by the measurement of expired CO₂ concentration over time can yield valuable clinical information. A normal capnogram has four phases (Fig. 39-1). Phase A-B represents the CO₂-free portion of the respiratory cycle. Most often, this occurs at the end of the inspiratory phase, although it may represent apnea or a disconnection of the device from the patient. An elevation of this baseline above zero implies CO₂rebreathing, as might be the case with increased dead space in the circuit, hypoventilation (high-flow sidestream sensors only), or contamination of the sensor.

Phase B-C, the rapid upstroke of the curve, represents the transition from inspiration to expiration and the mixing of dead space and alveolar gas. Prolongation of phase B-C (Fig. 39-1) occurs with obstruction to expiratory gas flow (e.g., obstructive lung disease, bronchospasm, kinked ETT) or leaks in the breathing system.

Phase C-D, the alveolar plateau, represents the predominance of alveolar gas rich in CO₂ and tends to slope gently upward with the uneven emptying of alveoli. The maximum CO₂ concentration in each breath (i.e., ETCO₂) represents point D and is the number that appears on the monitor. The slope of this phase can be increased by the same obstructive factors that increase the slope of phase B-C, as well as a normal physiological variation in pregnancy. A dip in the plateau indicates a spontaneous respiratory effort during mechanical ventilation, as might occur in patients with hypoxia, hypercarbia, or emerging from anesthesia.

Phase D-E, the inspiratory downstroke, is a nearly vertical drop to baseline. This slope can be prolonged and blend in with the beginning of the expiratory phase in endotracheal cuff leaks. Abnormal respiratory patterns that are chaotic limit the usefulness of ETCO₂ monitoring because characteristic patterns are difficult to discern.

Clinical Uses of Capnography

Capnography can be used in the ED for many clinical scenarios in both intubated and nonintubated patients. These include confirmation of ETT placement in the trachea, continuous monitoring of tube position, providing qualitative and quantitative methods of assessing cardiac output, gauging effectiveness of CPR during cardiac arrest (Fig. 39-2), determining prognosis in CPR and in trauma, maintaining appropriate ETCO₂ levels in patients with suspected increases in intracranial pressure, aiding in the detection and diagnosis of pulmonary embolism (air and clot), assessing response to treatment in patients with acute respiratory distress, determining adequacy of ventilation in patients with altered mental status (including drug-induced alterations in consciousness during procedural sedation and analgesia; Fig. 39-3), assessment of ventilatory status of actively seizing patients, and detection of metabolic acidosis in diabetes and gastroenteritis.

Along with visualizing an ETT going through the vocal cords and seeing tracheal rings on bronchoscopy, CO₂ monitoring is another “gold standard” used to confirm intubation of the trachea (see Chapter 3). Misleading ETCO₂ readings can occur with erroneous esophageal intubation after bag-mask ventilation and ingestion of carbonated beverages or antacids. These tracings are abnormal in appearance and resolve after six breaths. ETCO₂ is also falsely elevated for about 5 minutes after injection of sodium bicarbonate. In nonarrest settings, the ETCO₂ approaches 100% sensitivity and specificity in confirming correct ETT placement; conversely, it is also useful in detecting accidental extubation.

ETCO₂ can estimate PaCO₂ in hemodynamically stable patients with normal lung function. Although the normal Pa-ETCO₂ gradient is approximately 2 to 5 mm Hg, this may increase in patients with hemodynamic instability and pulmonary disorders, with the width of the gradient dependent on the severity of the lung disease or decrease in perfusion. Characteristic capnographic patterns associated with restrictive and obstructive lung disease are shown in Figure 39-4. Initially, it may be helpful to compare the PaCO₂ with the ETCO₂ to determine whether a difference is present and to quantify the difference, particularly in ventilated patients with obstructive lung disease (where the ventilator inspiratory cycle ordinarily interrupts the plateau phase before it becomes flat, underestimating the true end-tidal concentration of CO₂) or in unstable patients. With correction for this gradient, ETCO₂ trends can generally be used as a substitute for serial PaCO₂ measurements.

The airway, breathing, and circulation of critically ill or injured patients can be rapidly assessed using ETCO₂ values and the capnogram. The presence of a normal waveform denotes a patent airway and spontaneous breathing, and normal ETCO₂ levels indicate adequate perfusion. Therefore, capnography can be used to assess critically ill patients (including victims of chemical terrorism with nerve gas exposure) and actively seizing patients. Unlike pulse oximetry, capnography does not misinterpret motion artifact and provides reliable readings in low perfusion states.

Animal and human studies have shown that ETCO₂ is a noninvasive measurement that is highly correlated with cardiac output and is the earliest indicator of return of spontaneous circulation (ROSC) in CPR. ROSC is heralded by an almost immediate increase in ETCO₂from baseline. Multiple studies have shown that ETCO₂ has prognostic value in terms of mortality during CPR. No patient with a mean ETCO₂ level <10 mm Hg after 20 minutes of CPR survived, giving ETCO₂ measurement a high negative predictive value for failure of resuscitation. Despite these promising findings, capnography requires further prospective validation to confirm its utility as a prognostic tool in cardiac arrest.

The measurement of cardiac output ordinarily requires the placement of pulmonary arterial catheters and is rarely done in the ED, although noninvasive methods are continually being identified, including the use of ETCO₂. Using modified forms of the direct Fick equation, animal and human studies have shown excellent correlation between ETCO₂ measurements and cardiac output over a wide range of values. These measurements seem independent of changes in dead space (hypovolemia) or shunt (pulmonary edema) and may be adaptable to nonintubated patients.

Capnography is the only monitoring modality that is accurate and reliable in actively seizing patients. Capnographic data (respiratory rate, ETCO₂, and capnogram) can be used to distinguish among actively seizing patients with apnea (flatline waveform, no ETCO₂ readings, and no chest wall movement), with ineffective ventilation and low tidal volume breathing (small waveforms, low ETCO₂ values), and with effective ventilation (normal waveform, normal ETCO₂ values).

Capnography can also rapidly detect the common airway, respiratory, and central nervous system adverse events associated with the nerve agents in chemical terrorism, including apnea, upper airway obstruction, laryngospasm, bronchospasm, and respiratory failure.

Capnography provides dynamic monitoring of ventilatory status in patients with acute respiratory distress, including asthma, bronchiolitis, chronic obstructive pulmonary disease, congestive heart failure, croup, and cystic fibrosis. By measuring ETCO₂ and respiratory rate with each breath, capnography provides instantaneous feedback on the clinical status of the patient. Respiratory rate is measured directly from the airway by nasal-oral cannula, providing a more reliable reading than impedance respiratory monitoring. In upper airway obstruction and laryngospasm, impedance monitoring detects chest wall movement, interprets this as a valid breath, and displays a respiratory rate, even though the patient is not ventilating. In contrast, capnography will detect no ventilation and shows a flatline waveform.

ETCO₂trends can be rapidly assessed in tachypneic patients. A patient with a respiratory rate of 30 will generate 150 ETCO₂readings in 5 minutes. This provides sufficient information to determine the vector of the patient's ventilatory status: worsening despite treatment (increasing ETCO₂), stabilizing (stable ETCO₂), or improving (decreasing ETCO₂).

Bronchospasm in obstructive lung disease leads to upward slanting of the expiratory plateau of the capnogram (Fig. 39-4). Changes in ETCO₂ over time and the slope of this phase of the capnogram have been shown to correlate well with spirometric measurements (FEV₁ and PEFR). Capnography has the advantage of being independent of effort, gender, age, and height, and is a useful objective measure in asthmatic patients who are unwilling or unable to cooperate with spirometry (e.g., young children, ventilated patients, patients in acute respiratory distress).

Capnography can also detect the common adverse airway and respiratory events associated with procedural sedation and analgesia. Capnography is the earliest indicator of airway or respiratory compromise and will manifest an abnormally high or low ETCO₂ before pulse oximetry detects falling OxyHb saturation, especially in patients receiving supplemental oxygen. Both central and obstructive apnea can be almost instantaneously detected by capnography. Loss of the capnogram, in conjunction with no chest wall movement and no breath sounds on auscultation, confirms the diagnosis of central apnea. Obstructive apnea is characterized by loss of the capnogram, chest wall movement, and absent breath sounds. The absence of the capnogram in association with the presence or absence of chest wall movement distinguishes apnea from upper airway obstruction and laryngospasm. Response to airway alignment maneuvers can further distinguish upper airway obstruction from laryngospasm.

Capnography may be more sensitive than clinical assessment of ventilation in detection of apnea. In a recent study, 10 of 39 (26%) patients experienced 20-second periods of apnea during procedural sedation and analgesia. All ten episodes of apnea were detected by capnography, but not by the anesthesia providers.

Obtunded or unconscious patients, including those with alcohol intoxication, intentional or unintentional drug overdose, and postictal patients (especially those treated with benzodiazepines), may have impaired ventilation. Capnography can differentiate between postictal patients with effective ventilation and those with ineffective ventilation, as well as provide continuous monitoring of ventilatory trends over time to identify those patients at risk for respiratory depression and respiratory failure.

In addition to its established uses for assessment of ventilation and perfusion, capnography is a valuable tool for assessing metabolic status. Recent studies have shown that ETCO₂and serum bicarbonate (HCO₃) are linearly correlated in diabetes and in gastroenteritis, and ETCO₂ can be used as an indicator of metabolic acidosis in these patients (Figs. 39-5 and 39-6, respectively). As a patient becomes acidotic, HCO₃ decreases, and a compensatory respiratory alkalosis develops with an increase in minute ventilation and a resultant decrease in ETCO₂. The more acidotic the patient, the lower is the HCO₃, the higher the respiratory rate, and the lower the ETCO₂. Furthermore, ETCO₂ can be used to distinguish diabetics in ketoacidosis (metabolic acidosis, compensatory tachypnea, low ETCO₂) from those who are not (nonacidotic, normal respiratory rate, normal ETCO₂). In a study of diabetic children presenting to the ED, ETCO₂ <29 mm Hg identified 95% of the patients with ketoacidosis, with 83% sensitivity and 100% specificity. No ketoacidosis was detected in patients with ETCO₂ >36 mm Hg (Fig. 39-5).

A similar association between ETCO₂ and HCO₃ was demonstrated in children with gastroenteritis, with maximal sensitivity occurring at ETCO₂ ≤34 mm Hg (sensitivity 100%, specificity 60%) and optimal specificity without compromise of sensitivity occurring at ETCO₂ ≤31 mm Hg (sensitivity 76%, specificity 96%) (Fig. 39-6). As a potential triage tool for determining the need for oral versus intravenous rehydration, ETCO₂ could identify patients with a clinically significant acidosis, with an ETCO₂ ≤31 mm Hg giving a positive likelihood ratio of 20.4 in detecting HCO₃ ≤15 mmol/L and a ratio of 14.1 for HCO₃≤13 mmol/L. These ETCO₂ values are 14 or 20 times more likely to occur in an acidotic patient than in a patient with an HCO₃ >13 mmol/L or >15 mmol/L.

Sublingual Capnography

Studies in the mid-1990s showed that increases in esophageal and gastric PCO₂ (“gastric tonometry”) are associated with tissue dysoxia, most commonly found in low perfusion states, occurring early in the course of shock before more conventional measures of tissue oxygenation, such as heart rate, blood pressure, serum lactate, and arterial blood gases. Sublingual capnometry (P_SLCO₂) correlates with gastric tonometry and is a useful, noninvasive alternative to visceral PCO₂ monitoring.

Sublingual capnometry is relatively simple to monitor. A noninvasive microelectrode CO₂ probe is placed under the tongue. A handheld device similar in appearance to a digital thermometer provides a continuous P_SLCO₂ reading. The device consists of three major components: a disposable CO₂ sensor, a fiberoptic cable that connects the disposable sensor to a blood gas analyzer, and a blood gas monitoring instrument.

Studies currently underway to determine what absolute measures of P_SLCO₂ or trends in P_SLCO₂ will be useful in diagnosing early tissue dysoxia. One study of patients with penetrating torso trauma found that a P_SLCO₂ <45 mm Hg accurately predicted hemodynamic stability. With further study, this technology may prove useful in emergency medicine and prehospital care, where the early detection of tissue dysoxia is important and currently imprecise. Unfortunately, the discovery of contamination of the sublingual probe with Burkholderia cepacia has led to a voluntary recall of the product.

Conclusion

Capnography is a versatile noninvasive diagnostic monitoring modality that provides real-time information on the ventilatory, perfusion, and metabolic status of both intubated and nonintubated patients.

Evidence

1. What are some of the key issues that the emergency airway manager needs to be aware of when it comes to pulse oximetry in the ED? Monitor lag (the time taken for OxyHb saturation to reflect falling oxygen partial pressure) and response delay (the time taken to equilibrate because of intermittent sampling) are well described by Hill and Stoneham (1). Kellerman et al. (2) demonstrated that ready availability of pulse oximetry in the ED reduces arterial blood gas sampling. Reflectance oximetry has been demonstrated to more accurately reflect oxygen saturations in the setting of hypothermia and vasoconstriction (3,4). The limitations of pulse oximetry in reflecting the adequacy of oxygenation during CPR are well documented (5,6,7).

2. What are the key indications for ETCO₂ monitoring in the ED? The most common indications in the ED are to evaluate the adequacy of ventilation in intubated patients, for verification of ETT placement, for nonintubated patients with acute respiratory distress and during procedural sedation and analgesia (8,9). CO₂ monitoring is also commonly used to evaluate perfusion status during cardiac arrest and shock states, having long been known to be an indicator of pulmonary blood flow (10,11,12,13,14,15).

Disclosure

Baruch Krauss, MD, EdM, is a consultant for Oridion Medical, a capnography company, and holds two patents in the area of capnography.

References

1. Hill E, Stoneham MD. Practical applications of pulse oximetry. Update in anaesthesia 2000;11:1–2. Available at: http://www.nda.ox.ac.uk/wfsa/html/u11/u1104_01.htm.

2. Kellerman AL, Cofer CA, Joseph S, et al. Impact of portable oximetry on arterial blood gas test ordering in an urban emergency department. Ann Emerg Med 1991;20:130–134.

3. Bebout DE, Mannheimer PD, Wun CCW. Site-dependent differences in the time to detect changes in saturation during low perfusion [abstract]. Crit Care Med 2001;29:115a.

4. Keogh BF, Kopotic RJ. Recent findings in the use of reflectance oximetry: a critical review. Curr Opin Anesthesiol 2005;18:649–654.

5. Griffin M, Cooney C. Pulse oximetry during cardiopulmonary resuscitation. Anaesthesia 1995;50:1008.

6. Spittal MJ. Evaluation of pulse oximetry during cardiopulmonary resuscitation. Anaesthesia 1993;48:701–703.

7. Moorthy SS, Didorff SF, Schmidt SI. Erroneous pulse oximeter data during CPR. Anesth Analg 1990;70:339.

8. Krauss B, Hess DR. Capnography for procedural sedation and analgesia in the emergency department. Ann Emerg Med 2007;50:172–181.

9. Soto RG, Fu ES, Vila H, et al. Capnography accurately detects apnea during monitored anesthesia care. Anesth Analg 2004;99:379–382.

10. Levine RL, Wayne MA, Miller CC. End-tidal carbon dioxide and outcome of out-of-hospital cardiac arrest. N Engl J Med 1997;337:301.

11. Falk JL, Rackow EC, Weil MH. End-tidal carbon dioxide concentration during cardiopulmonary resuscitation. N Engl J Med 1988;318:607.

12. Garnett AR, Ornato JP, Gonzalez ER, et al. End-tidal carbon dioxide monitoring during cardiopulmonary resuscitation. JAMA 1987;257:512.

13. Deakin CD, Sado DM, Coats TJ, et al. Prehospital end-tidal carbon dioxide concentration and outcome in major trauma. J Trauma 2004;57:65.

14. Cantineau JP, Lambert Y, Merckx P, et al. End-tidal carbon dioxide during cardiopulmonary resuscitation in humans presenting mostly with asystole: a predictor of outcome. Crit Care Med 1996;24:791–796.

15. Ahrens T, Schallom L, Bettorf K, et al. End-tidal carbon dioxide measurements as a prognostic indicator of outcome in cardiac arrest. Am J Crit Care 2001;10:391–398.