Noninvasive carbon dioxide monitoring during neurosurgical procedures in adults: end-tidal versus transcutaneous techniques.

ABSTRACT

Background. We prospectively compared transcutaneous (TC) versus end-tidal (ET) carbon dioxide monitoring during neurosurgical procedures in adults.

Methods. After calibration and an equilibration time for the TC-[CO.sub.2] monitor, arterial blood gas (ABG) values were obtained as clinically indicated. The [Paco.sub.2] values were compared with the values recorded by the noninvasive monitors (TC and ET).

Results. The ET-[CO.sub.2] to [Paco.sub.2] difference was 6.1 [+ or -] 5.6 mm Hg, and the TC-[CO.sub.2] to [Paco.sub.2] difference was 3.7 [+ or -] 2.9 mm Hg. The difference between the [Paco.sub.2] and ET-[CO.sub.2] was 3 mm Hg or less in 17 of 57 values, while the difference between the [Paco.sub.2] and TC-[CO.sub.2] was 3 mm Hg or less in 35 of 57 values. Linear regression analysis of ET-[CO.sub.2] versus [Paco.sub.2] revealed a slope of 0.381 [+ or -] 0.007. Linear regression analysis of TC-[CO.sub.2] versus [Paco.sub.2] revealed a slope of 1.17 [+ or -] 0.008.

Conclusion. Transcutaneous [CO.sub.2] monitoring provides a more accurate estimate of [Paco.sub.2] than ET-[CO.sub.2] monitoring during neurosurgical procedures.

**********

MEASUREMENT of the partial pressure of carbon dioxide ([Paco.sub.2]) in the arterial blood is used to assess the adequacy of ventilation during both spontaneous and controlled ventilation. Because of its effects on cerebral blood flow and intracranial pressure, the intraoperative control of [Paco.sub.2] can be of great significance during neurosurgical procedures. Although the standard for monitoring [Paco.sub.2] remains arterial blood gas (ABG) monitoring, ABGs provide only an intermittent sample of what can be a dynamic and rapidly changing value. To overcome such problems, noninvasive monitors are used to provide a continuous estimate of [Paco.sub.2] Routine monitoring in the operating rooms includes end-tidal [CO.sub.2] ([ET-CO.sub.2]) measurement; however, several patient-related issues such as ventilation-perfusion mismatch may influence the correlation of the ET-[CO.sub.2] with [Paco.sub.2]. (1,2) Grenier et al (3) showed a significant influence of patient positioning on the correlation between ET and arterial [CO.sub.2] as well as a significant variation or instability of the gradient over time, prompting them to question the utility of ET-[CO.sub.2] monitoring during neurosurgical procedures.

Transcutaneous (TC) [CO.sub.2] devices provide another option for the continuous noninvasive estimation of [Paco.sub.2]. Although these devices have been shown to be accurate in infants and children, (4,5) limited data discuss intraoperative TC-[CO.sub.2] monitoring in the adult population. The current study prospectively compares the accuracy of ET and TC monitoring of [CO.sub.2] during neurosurgical procedures in adults.

METHODS

The study was approved by the Institutional Review Board and the Committee for the Protection of Human Subjects of the University of Missouri. The patient population included adult patients (18 years of age or older) about to have neurosurgical procedures in which intra-arterial access was deemed necessary by the attending anesthesiologist.

End-tidal [CO.sub.2] was measured using infrared spectroscopy with a sidestream aspirator at a flow rate of 150 mL/min (Nellcor N-2500 Capnograph, Nellcor Inc, Hayward, Calif). The digital readout on the ET-[CO.sub.2] monitor is based on an algorithm that evaluates two successive waveforms and the valley between them. The ET-[CO.sub.2] reported by the monitor is the maximum value achieved on the first waveform. Transcutaneous [CO.sub.2] was measured using a standard TC-[CO.sub.2]/[O.sub.2] device (Radiometer, Copenhagen, Denmark). The protocol for TC-[CO.sub.2] monitoring included calibration, placement, and maintenance of the monitor by all of us. Before placement, the electrode was cleaned, a new membrane was applied, and calibration was performed according to the manufacturer's recommendations. The working temperature of the electrode was set at 45[degrees]C. Management of ABGs was done as clinically indicated during the surgical procedure. When ABG values were obtained, the ET-[CO.sub.2] and TC-[CO.sub.2] were noted and recorded on the data sheet. The ABGs were measured at 37[degrees]C. One to three sample sets (arterial blood gas, ET-[CO.sub.2], and TC-[CO.sub.2]) were collected from each patient.

The absolute difference between the noninvasive monitor (ET-[CO.sub.2]/TC-[CO.sub.2]) and the [Paco.sub.2] was calculated. No negative numbers were used since we believed this would artificially lower the mathematical mean of the differences in the values between the noninvasive reading of [CO.sub.2] and the [Paco.sub.2]. For example, if the noninvasive reading deviated 7 mm Hg from the actual reading, 7 mm Hg was used, not -7 or +7. The ET-[CO.sub.2] to [Paco.sub.2] and TC-[CO.sub.2] to [Paco.sub.2] differences (absolute differences) were analyzed using a two-tailed Wilcoxon signed rank test for pairs. The correlation between the noninvasive monitor (ET-[CO.sub.2] or TC-[CO.sub.2]) and the [Paco.sub.2] was determined using a linear regression analysis. Additionally, using the raw data (positive and negative numbers), a Bland-Altman analysis was done, (6) comparing the ET-[CO.sub.2]/TC-[CO.sub.2] and the [Paco.sub.2]. The bias, defined as the mean difference between values, and precision, defined as the stand ard deviation of the bias, were determined for TC-[CO.sub.2] versus [Paco.sub.2] and ET-[CO.sub.2] versus [Paco.sub.2]. A chi-square analysis with Yates' correction and a two-way contingency table was used to compare the number of ET-[CO.sub.2] versus TC-[CO.sub.2] values whose absolute difference deviated 3 mm Hg or less and 5 mm Hg or less from the actual [PaCO.sub.2] value. All data are expressed as the mean [+ or -] SD, and P< .05 was considered significant.

RESULTS

The cohort for the study included 30 adults ranging in age from 18 to 76 years (43.7 [+ or -] 15.8 years) and in weight from 49 to 130 kg (81.1 [+ or -] 20.3 kg). There were 12 men and 18 women. In all cases, the surgical procedure included a craniotomy for various indications. No patient was receiving an inotropic agent or adrenergic agonist. None of the patients had significant tissue edema or manifested signs of a low cardiac output state.

Fifty-seven sample sets ([PaCO.sub.2], ET-[CO.sub.2], and TC-[CO.sub.2]) were obtained from the 30 patients. The actual [PaCO.sub.2] ranged from 26 to 62 mm Hg. The ET-[CO.sub.2] to [PaCO.sub.2] difference was 6.1 [+ or -] 5.6 mm Hg, and the TC-[CO.sub.2] to [PaCO.sub.2] difference was 3.7 [+ or -] 2.9 mm Hg (P = .0049). The difference between the [PaCO.sub.2] and ET-[CO.sub.2] was 3 mm Hg or less in 17 of 57 values, and the difference between the [PaCO.sub.2] and TC-[CO.sub.2] was 3 mm Hg or less in 35 of 57 values (P = .00251, odds ratio [OR] of 3.942). The difference between the [PaCO.sub.2] and ET-[CO.sub.2] was 5 mm Hg or less in 31 of 57 values, while the difference between the [PaCO.sub.2] and TC-[CO.sub.2] was 5 mm Hg or less in 47 of 57 values (P = .00139, OR 3.743). If both monitors were considered simultaneously, one of the two noninvasive monitors was 3 mm Hg or less from the actual [PaCO.sub.2] in 44 of 57 measurement points. In 55 of 57 sets, one of the noninvasive monitors was 5 mm Hg or less f rom the actual [PaCO.sub.2] value.

Linear regression analysis of ET-[CO.sub.2] versus [PaCO.sub.2] revealed a slope of 0.381 [+ or -] 0.007, r value = 0.5531, and [r.sup.2] = 0.3025. Linear regression analysis of TC-[CO.sub.2] versus [PaCO.sub.2] revealed a slope of 1.17 [+ or -] 0.008, r value = 0.8824, and [r.sup.2] = 0.7786. Bland-Altman analysis of ET-[CO.sub.2] versus [PaCO.sub.2] revealed a bias of -6.1 mm Hg and a precision of [+ or -] 5.6 mm Hg (Fig 1). Bland-Altman analysis of TC-[CO.sub.2] versus [PaCO.sub.2] revealed a bias of 2.1 mm Hg and a precision of [+ or -] 4.2 mm Hg (Fig 2).

DISCUSSION

In adult patients receiving general anesthesia for neurosurgical procedures, transcutaneous [CO.sub.2] monitoring provided a more accurate estimate of the [PaCO.sub.2] than end-tidal [CO.sub.2] monitoring. Although previous studies have evaluated the accuracy of these noninvasive monitors in various settings, there is a paucity of reports directly comparing these techniques during intra-operative anesthetic care in adults. In the current study, the two noninvasive monitors were placed on the same patient, thereby allowing us to directly compare the reliability of the two techniques in adults during neurosurgical procedures. When compared with ET-[CO.sub.2] monitoring, TC-[CO.sub.2] monitoring showed a lower mean absolute difference from [PaCO.sub.2], a greater number of values that deviated 3 mm Hg or less and 5 mm Hg or less from the actual [PaCO.sub.2], a slope of closer to 1 using linear regression analysis, and a more accurate bias and precision using Bland-Altman analysis.

Since its inception, TC-[CO.sub.2] has received the greatest use in the neonatal population. Previous studies have also validated the accuracy of TC-[CO.sub.2] monitoring in the pediatric intensive care unit population with respiratory failure of various causes. (47) In these studies, TC-[CO.sub.2] monitoring proved especially efficacious in settings where ventilation-perfusion disturbances can be expected to interfere with the gradient between the end-tidal and arterial [CO.sub.2]. (4,5)

To date, reports of TC-[CO.sub.2] in the adult population remain relatively limited. Previous authors have reported their experiences with TC-[CO.sub.2] monitoring in adults in the following clinical scenarios: (1) noninvasive mechanical ventilation, (8) (2) sleep studies, (9) (3) mechanical ventilation with endotracheal intubation in the ICU, (10) (4) fiberoptic bronchoscopy, (11) and (5) during the immediate postoperative period. (12) When reviewing the literature concerning TC-[CO.sub.2] monitoring, the type and manufacturer of the monitor and whether a "calibration factor" is used by the monitor can affect the accuracy of the monitoring and thereby the results of the study. Additionally, regardless of the monitor used, the tissue [CO.sub.2] and therefore the TC-[CO.sub.2] are invariably higher than the Pa[CO.sub.2] whereas the ET-[CO.sub.2] is invariably lower than the Pa[CO.sub.2].

The TC-[CO.sub.2] monitor by Radiometer, the one used in the current study, corrects for the alterations in [CO.sub.2] caused by heating the skin to 45[degrees]C. With heating of the skin, the uncorrected TC-[CO.sub.2] is significantly above the Pa[CO.sub.2] (measured at 37[degrees]C) because of increased [CO.sub.2] production from the elevated local tissue metabolism and the altered solubility of [CO.sub.2] at higher temperatures. (12) More information concerning these issues and the calibration factor used in the TC-[CO.sub.2] monitor is available in the Radiometer User's Handbook (Radiometer, Copenhagen, Denmark). The importance of these factors is illustrated by the study of Tremper et al. (10) Without correcting the TC-[CO.sub.2] value for temperature differences, they compared the transcutaneous and arterial values of [CO.sub.2] in 435 data sets in 44 patients in a mixed OR/ICU setting. (10) Although they noted a wide difference between the TC and the Pa[CO.sub.2] value (23 [+ or -] 11 mm Hg), linear re gression analysis yielded a correlation coefficient of 0.80. They also noted that with a low cardiac output state (cardiac index less than 1.5 L/min/[m.sup.2]), there was a further increase in the TC to arterial [CO.sub.2] gradient and the TC-[CO.sub.2] trended inversely with the cardiac index and not the Pa[CO.sub.2]. A similar effect of cardiovascular performance and the gradient between arterial and TC carbon dioxide has been reported by other investigators. (13-16) No patient in our current study manifested clinical signs or symptoms of cardiovascular dysfunction or a low cardiac output state.

To date, information is limited concerning intraoperative TC-[CO.sub.2] monitoring. BhavaniShankar et al (17) reported the use of TC-[CO.sub.2] during intraoperative anesthetic care during laparoscopic cholecystectomy in a parturient. Their use of TC-[CO.sub.2] monitoring was prompted by reports of the potential problems with ET-[CO.sub.2] during laparoscopy. Since there was no indication for placement of an indwelling arterial catheter and since there was a limited increase in the TC-[CO.sub.2], no ABG values were obtained. The authors did note an increase in the TC-[CO.sub.2] from a baseline of 39.1 [+ or -] 0.1 to a maximum of 45.7 [+ or -] 0.1 mm Hg, as well as an increase in the TC to ET gradient from 6.6 [+ or -] 0.3 to 13.7 [+ or -] 0.2 mm Hg during peritoneal insufflation with [CO.sub.2].

Reid et al (18) compared ET and TC [CO.sub.2] monitoring during general anesthesia in 22 adults during 3 different levels of mechanical ventilation. A total of 66 data sets with the Pa[CO.sub.2] ranging from 28 to 62 mm Hg showed an ET to arterial gradient of 7.0 [+ or -] 3.1 mm Hg with a correlation coefficient of r = 0.89 and a TC to arterial gradient of 2.3 [+ or -] 2.4 mm Hg (P < .05 when compared with TC to arterial gradient) with r = 0.92. The differences were greatest at the higher Pa[CO.sub.2] values.

Phan et al (19) also compared ET and TC-[CO.sub.2] monitoring during general anesthesia in 24 adults. The correlation coefficient, bias, and precision for the ET to arterial comparison were 0.67, -7.8 mm Hg and + 6.1 mm Hg, respectively, while the correlation coefficient, bias, and precision for the TC to arterial comparison were 0.87, -1.6 mm Hg, and [+ or -] 4.3 mm Hg.

Our current data further support the accuracy and clinical utility of TC-[CO.sub.2] during general anesthesia, especially in situations where there may be a widened ET to arterial gradient. The success of TC-[CO.sub.2] can be affected by both monitor and patient factors. Our current protocol and technique for TC-[CO.sub.2] monitoring has been previously described. (4,5) When compared with ET-[CO.sub.2] monitoring, TC-[CO.sub.2] monitoring requires a much longer preparation time, including a 5-minute calibration period before placement and then an additional 10- to 15-minute equilibration period after placement on the patient to allow for an equilibration between the TC and arterial [CO.sub.2] values. Once the electrode is placed, recalibration and replacement at another site every 4 hours is recommended by the manufacturer, making this monitor more labor intensive than ET-[CO.sub.2] monitoring. Although we found that TC-[CO.sub.2] monitoring provided a more accurate reflection of [PaCO.sub.2] in most patients , several factors may affect this accuracy including technical variables such as trapped air bubbles, improper placement technique, damaged membranes, and inappropriate calibration techniques. In addition to technical problems, patient problems may affect the accuracy of TC-[CO.sub.2] monitoring. These may include variations in skin thickness, the presence of edema, tissue hypoperfusion, or the administration of vasoconstricting drugs. (5,13-16) Because no technique can be expected to be 100% reliable, periodic calibration with Paco2 values is recommended.

The two noninvasive [CO.sub.2] monitors (end-tidal and transcutaneous) should be used to complement rather than to exclude one another. In the current study, 44 of 57 times, one of the two noninvasive monitors was 3 mm Hg or less from the actual [PaCO.sub.2] value, while in 55 of 57 instances, one of the two noninvasive monitors was 5 mm Hg or less from the actual [PaCO.sub.2] value. Although TC-[CO.sub.2] monitoring may be a more accurate means of estimating [PaCO.sub.2], ET-[CO.sub.2] provides additional useful information, since it documents the intratracheal position of the endotracheal tube and serves as an additional safety monitor to alert one to a ventilator disconnection.

[FIGURE 1 OMITTED]

[FIGURE 2 OMITTED]

References

(1.) Bhavani-Shankar K, Moseley H, Kumar AY et al: Capnometry and anesthesia. Can J Anaesth 1992; 39:617- 632

(2.) Pansard JL, Cholley B, Devilliers C, et al: Variation in arterial to end-tidal [CO.sub.2] tension differences during anesthesia in the "kidney rest" lateral decubitus position. Anesth Analg 1992; 75:506-510

(3.) Grenier B, Verchere E, Meslie A, et al: Capnography monitoring during neurosurgery: reliability in relation to various intraoperative positions. Aneth Analg 1999; 88:43-48

(4.) Tobias JD, Wilson WR Jr, Meyer DJ: Transcutaneous monitoring of carbon dioxide tension after cardiothoracic surgery in infants and children. Anesth Analg 1999; 88:531-534

(5.) Tobias JD, Meyer DJ: Noninvasive monitoring of carbon dioxide during respiratory failure in toddlers and infants: end-tidal versus transcutaneous carbon dioxide. Anesth Analg 1997; 85:55-58

(6.) Bland JM, Altman DG: Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 1:307-310

(7.) Sivan Y, Eldadah MK, Cheah TE, et al: Estimation of arterial carbon dioxide by end-tidal and transcutaneous [PCO.sub.2] measurements in ventilated children. pediatr Pulm 1992; 12:153-157

(8.) Janssens JP, Howarth-Frey C, chevrolet JC, et al: Transcutaneous [PCO.sub.2] to monitor nonivasive mechanical ventilation in adults. Chest 1998; 113:768-773

(9.) Sanders MH, Kera NE, costantino JP, et al: Accuracy of end-tidal and transcutaneous [PCO.sub.2] monitoring during sleep. Chest 1994; 106:472-483

(10.) Tremper KK, Shoemaker WC, Shippy CR, et al: Transcutaneous [PCO.sub.2] monitoring on adult patients in the ICU and the operating room. Crit Care Med 1981; 9:752-755

(11.) Evans EN, Ganeshalingam K, Ebden P: Changes in oxygen saturation and transcutaneous carbon dioxide and oxygen levels in patients undergoing fiberoptic bronchoscopy. Respir Med 1998; 92:739-742

(12.) Drummond KJ, Fearnside MR, Chee A: Transcutaneous carbon dioxide measurement after craniotomy in spontaneously breathing patients. Neurosurgery 1997; 41 :36 1-367

(13.) Rithalia SVS, Ng YN, Tinker J: Measurement of transcutaneous [PCO.sub.2] in critically ill patients. Resuscitation 1982; 10:13-18

(14.) Rithalia SVS, Clutton-Brock TH, Tinker J: Characteristics of transcutaneous carbon dioxide tension monitors in normal adults and critically ill patients. Intensive Care Med 1984; 10:149-153

(15.) Shoemaker WC: Physiologic and clinical significance of [P.sub.tc][O.sub.2] and [P.sub.tc][CO.sub.2]. Crit Care Med 1981; 9:689-690

(16.) Hasibeder w, Haisjackl M, Sparr H, et al: Factors influencing transcutaneous oxygen and carbon dioxide measurements in adult intensive care patients. Intensive Care Med 1991; 17:272-275

(17.) Bhavani-Shankar K, Steinbrook RA, Mushlin PS, et al: Transcutaneous [PCO.sub.2] monitoring during laparoscopic cholecystectomy in pregnancy. Can J Anaes 1998; 45:164-169

(18.) Reid CW, Martineau RJ, Miller DR, et al: A comparison of transcutaneous, end-tidal and arterial measurements of carbon dioxide during general anesthesia. Can J Anaesth 1992; 39:31-36

(19.) Phan CQ Tremper KK, Lee SE, et al: Noninvasive monitoring of carbon dioxide: a comparison of the partial pressure of transcutaneous and end-tidal carbon dioxide with the partial pressure of arterial carbon dioxide. J Clin Monitor 1987; 3:149-154

RELATED ARTICLE: KEY POINTS

* Because of its effects on cerebral blood flow and intracranial pressure, precise control of the partial pressure of carbon dioxide can be crucial during neurosurgical procedures.

* Recent clinical investigations have questioned the accuracy of end-tidal carbon dioxide monitoring during neurosurgical procedures in adults.

* Data are limited regarding the intraoperative use of transcutaneous carbon dioxide monitoring.

* With simultaneous use of both noninvasive techniques, transcutaneous monitoring provided a more accurate estimate of arterial carbon dioxide values.

From the Departments of Child Health and Anesthesiology, University of Missouri, Columbia.

Reprint requests to Joseph D. Tobias, MD, University of Missouri, Department of Child Health, M658 Health Sciences Center, One Hospital Dr, Columbia, MO 65212.