| JOURNAL HOME | CME HOME | THIS MONTH | PAST ISSUES | ETOC | COLLECTIONS |
| AUTHORS | REVIEWERS | EDITORIAL BOARD | FEEDBACK | RSS | HELP |
| ||||||||||||||
| ||||||||||||||
|
|
|
|
||||||||||||
|
||||||||||||||
||
*Department of Child Health; Division of Pediatric Critical Care/Pediatric Anesthesiology; Department of Surgery; Division of Pediatric Surgery; and ||Department of Anesthesiology, University of Missouri, Columbia, Missouri
Address correspondence and reprint requests to John W. Berkenbosch, MD, Assistant Professor, Pediatric Critical Care, Department of Child Health, The University of Missouri, One Hospital Drive, Columbia, MO 65212. Address e-mail to berkenboschj{at}health.missouri.edu
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionReferences |
|---|
Implications: The authors report on the accuracy of noninvasive, transcutaneous CO2 monitoring during mechanical ventilation in children 4 yr or older. Application of this technique should be useful by decreasing the need for repeated, costly, and sometimes painful arterial blood gas analysis, and the continuity of assessment should facilitate proactive, rather than reactive, ventilator manipulations.
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Two types of noninvasive CO2 monitor are available for clinical use. End-tidal CO2 (ETCO2) monitoring accurately estimates ventilation in patients with normal pulmonary function (2,3). However, alterations in ventilation/perfusion matching secondary to increased dead space or shunt fraction limit the accuracy of ETCO2 monitoring in patients with abnormal pulmonary function (4,5). Areas with high ventilation/perfusion ratios (alveolar dead space) have a tendency to dilute the CO2 from areas with normal ventilation/perfusion ratios, resulting in an underestimation of PaCO2. Among critically ill children, increased intrapulmonary shunting from pulmonary parenchymal disease is relatively common. Admixture of this blood into the arterial circulation contributes to an increased ETCO2 - PaCO2 difference (6). These problems suggest that ETCO2 monitoring may be of limited usefulness within the pediatric critical care setting.
Transcutaneous CO2 (TCCO2) monitoring may bypass some of the problems inherent in ETCO2 monitoring. The accuracy of TCCO2 monitoring has been demonstrated in neonates (7) because of their thin, poorly keratinized skin, which has fewer diffusion barriers to capillary gases (8). We have demonstrated the accuracy of TCCO2 monitoring in neonates and children after cardiothoracic surgery (9) and have demonstrated the superiority of TCCO2 versus ETCO2 monitoring in infants and toddlers with respiratory failure (10). There is little information comparing these two techniques in older pediatric patients. As part of our continuing investigation, we prospectively compared TCCO2 and ETCO2 monitoring in older pediatric patients with respiratory failure, hypothesizing that, as in our previous investigations, TCCO2 monitoring would provide a more accurate estimation of PaCO2 than ETCO2 monitoring.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
ETCO2 was continuously measured by infrared spectroscopy with a sidestream aspirator at a flow rate of 150 mL/min (Capnocheck Plus; Sims BCI International, Waukesha, WI). CO2 calculation is based on an algorithm that evaluates two successive wave forms and the valley between them, with the recorded CO2 being the maximum value achieved on the first waveform. TCCO2 was continuously measured by using a TCM3 TCCO2/O2 device (Radiometer, Copenhagen, Denmark). Monitor calibration, placement, and maintenance were performed by the respiratory therapy staff. The electrode membrane was cleaned and calibrated before each use. In accordance with manufacturer’s recommendations, the working temperature of the electrode was maintained at 43.5°C, and the monitoring site was changed every 4 h to avoid thermal injury. If burns were noted, the electrode site was changed every 3 h. The electrode was recalibrated before placement at a new site. When clinically indicated, arterial blood gas measurements were performed, and the simultaneous values for ETCO2 and TCCO2 were recorded. The ETCO2 and TCCO2 values recorded were the average reading during a 15-s time interval. To avoid biasing of the data, a maximum of five sample sets (PaCO2, ETCO2, and TCCO2) were recorded for each patient.
The absolute difference between arterial and noninvasive (ETCO2 or TCCO2) CO2 values was calculated. Negative numbers were not used because this would have artificially lowered the mathematical mean of these differences. The ETCO2 to PaCO2 and TCCO2 to PaCO2 differences were compared with a two-tailed Wilcoxon’s signed rank test for pairs. By using both positive and negative values, a Bland-Altman analysis was performed (11). Bias, the mean difference between values, and precision, the SD of the bias, were calculated for ETCO2 to PaCO2 and TCCO2 to PaCO2 differences. The frequency of ETCO2 or TCCO2 values being 
3 or 5 mm Hg from PaCO2 was compared with a 
2 test with Yates correction by using a contingency table. Linear regression analysis was used to compare ETCO2 and TCCO2 values with measured PaCO2. A P value of <0.05 was considered significant.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
The ETCO2 to PaCO2 difference was 6.4 ± 6.3 mm Hg, whereas the TCCO2 to PaCO2 difference was 2.6 ± 2.0 mm Hg (P < 0.0001). In 58 of 82 samples, the TC measurement was more accurate, whereas in 16 of 82 samples, the ET measurement was more accurate. In 8 of 82 samples, ET and TC measurements were equally accurate. Bland-Altman analysis revealed a bias of -5.68 with a precision of ±6.93 when comparing ETCO2 with PaCO2 and a bias of 0.02 with a precision of ±3.27 when comparing TCCO2 and PaCO2 values (P < 0.00001) (Figs. 1 and 2).
|
|
3 mm Hg and by 5 mm Hg. ETCO2 values deviated from PaCO2 by 3 in 34 of 82 samples and by 5 in 47 of 82 samples, whereas TCCO2 values deviated from PaCO2 3 in 65 of 82 and by 5 in 76 of 82 samples (P < 0.00001). When both methods were used, the ET/TC to PaCO2 difference was >3 mm Hg in 3 of 82 samples and >5 in 1 of 82 samples. Regression analysis of ETCO2 and PaCO2 values revealed a correlation coefficient of 0.5418 and an r value of 0.8745. Regression analysis of TCCO2 and PaCO2 values revealed a correlation coefficient of 1.0160 and an r value of 0.9693 (Figs. 3 and 4).
|
|
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
In contrast, the adult literature suggests that TCCO2 monitoring is somewhat less accurate, with more limitations on clinically relevant applications (7,13). Although this suggests that TCCO2 monitoring should become less accurate in the largest patients in this study, our subgroup analysis of patients more than 40 kg did not find this to be the case. TCCO2 monitoring was equally accurate in these patients compared with the entire group and provided a more accurate estimate of PaCO2 than ETCO2 monitoring. However, these findings should be interpreted with some caution given the small sample size of this subgroup. Additionally, because the average size of patients in this subgroup is smaller than would be found in an adult study, direct comparison of this subgroup with adults is not valid.
With two possible exceptions (intracranial hypertension and pulmonary vascular disease), an estimated CO2 value that is within 3–5 mm Hg of PaCO2 should be acceptable for making clinical decisions when the arterial value is unavailable. This study shows that the majority of TCCO2 measurements were within this range, although almost half of the time ETCO2 values exceeded it. Of further interest, in only 1 of 82 measurements did both TC and ET values deviate from PaCO2 beyond this range, suggesting that one or the other form of noninvasive CO2 monitoring could be appropriate for the majority of pediatric intensive care unit patients.
The inaccuracies of ETCO2 monitoring have been well documented. Specifically, ETCO2 monitoring tends to underestimate PaCO2 levels. This relates primarily to failure to achieve, and thereby truly document, alveolar ventilation (6). Sivan et al. (14) have reported that these discrepancies start to occur below a PaO2/PAO2 ratio of 0.3. Although we did not specifically look at this ratio, we did note that the largest ETCO2 to PaCO2 differences occurred in patients having significant parenchymal lung disease or those with status asthmaticus. Removal of these patients from the analysis significantly improved the accuracy of ETCO2 monitoring (precision -3.6, bias ±3.7). In contrast, this systematic pattern of CO2 underestimation did not appear with TCCO2 monitoring. This is not surprising, however, because the CO2 measured during TCCO2 monitoring is arterialized blood and therefore is not dependent on ventilation-perfusion matching, as ETCO2 monitoring is.
Effective TC monitoring is dependent on both technical and patient factors. Staff members must be carefully trained in the proper use of the equipment to avoid technical problems such as trapped air bubbles, improper placement, damaged membranes, or inappropriate calibration. These factors require more time and effort in comparison with ET monitoring. Additionally, care must be taken so that excessive thermal injury does not occur. Indeed, most of our patients did experience mild redness at the electrode site. However, if this was deemed to be more than usual, the electrode site was rotated more frequently than our standard of every four hours, which, in our population, was adequate to prevent any significant thermal injury. This may, however, further increase the costs related to the TC device.
Numerous investigators have reported that the quality of skin perfusion is an important factor in determining the accuracy of TC monitoring. This is related to the means by which TC analysis is performed. Application of heat to the surface of the skin increases blood flow into the arteriovenous anastomoses and the venous plexus. When this occurs, capillary blood is arterialized, and the oxygen and CO2 concentrations at the skin surface become reflective of the arterial tensions (8). In addition, heat disrupts the stratum corneum layer of the skin, which is relatively impermeable to gases, enabling arterial gases to diffuse more easily to the skin surface (8). Therefore, variations such as skin thickness, edema, tissue hypoperfusion, or the administration of vasoconstricting drugs can limit the accuracy of the measurements (9,14). Although we did not formally assess skin perfusion, it is of interest that TCCO2 accuracy did not appear to be significantly altered in five of our patients receiving moderate to large doses of vasoactive drugs (dopamine 5–10 µg · kg-1 · min-1 ± dobutamine 5–20 µg · kg-1 · min-1 ± phenylephrine 0.6 µg · kg-1 · min-1 ± epinephrine 0.25–0.6 µg · kg-1 · min-1). It is possible that the slightly higher electrode temperature that we used (43.5°C) relative to some other studies provided adequate skin vasodilation to compensate for this. Others have reported an increase in the TCCO2 to PaCO2 gradient at higher CO2 levels (15). In accordance with our previous studies (9,10), we did not find this to be the case, although only a limited number of CO2 values (10 of 82) were more than 50 mm Hg.
This study demonstrates that the TC method is more accurate than the ET method in estimating PaCO2 values over a wide range of PaCO2 in older pediatric patients receiving mechanical ventilation for respiratory failure and may be considered a useful adjunct for monitoring ventilation in this population. This would provide for continuous, rather than intermittent, assessment of ventilation and decrease the need for repeated, costly arterial blood gas measurement, which may be of particular benefit for the patient without indwelling arterial access. This finding does not exclude the use of ET monitoring, because it is reliable in patients with relatively normal lung function and requires less maintenance. Additionally, ETCO2 monitoring can be used for confirmation of endotracheal tube placement, which is particularly important in areas such as the transport environment. These two monitors may be used to complement each other, and the choice of monitor should be individualized for each patient to provide the most accurate means of estimating PaCO2. When one is inaccurate, the other may be tried, because at least one method was accurate within clinical relevance in the majority of cases. Because neither method can be expected to be accurate 100% of the time, blood gas correlation should be performed periodically or when noninvasive measurements do not appear to be consistent with clinical findings.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  J. Wilson, P. Russo, J. Russo, and J. D. Tobias Noninvasive Monitoring of Carbon Dioxide in Infants and Children With Congenital Heart Disease: End-Tidal Versus Transcutaneous Techniques J Intensive Care Med, September 1, 2005; 20(5): 291 - 295. [Abstract] [PDF]
|
|
|
|
 |
|
 O. Senn, C. F. Clarenbach, V. Kaplan, M. Maggiorini, and K. E. Bloch Monitoring Carbon Dioxide Tension and Arterial Oxygen Saturation by a Single Earlobe Sensor in Patients With Critical Illness or Sleep Apnea Chest, September 1, 2005; 128(3): 1291 - 1296. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Oshibuchi, S. Cho, T. Hara, S. Tomiyasu, T. Makita, and K. Sumikawa A Comparative Evaluation of Transcutaneous and End-Tidal Measurements of CO2 in Thoracic Anesthesia Anesth. Analg., September 1, 2003; 97(3): 776 - 779. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. M. Broka, J. W. Berkenbosch, J. Lam, R. S. Burd, and J. D. Tobias About the Origin of the PaCO2/ETCO2 ratio... Anesth. Analg., December 1, 2001; 93(6): 1625 - 1626. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
