CARDIOPULMONARY EXERCISE TESTING has considerable

1686 Arterial Blood Gases During Exercise: Validity of Transcutaneous Measurements Carole Planès, MD, PhD, Michel Leroy, MD, Evelyne Foray, IDE, Bernadette Raffestin, MD, PhD ABSTRACT. Planès C, Leroy M, Foray E, Raffestin B. Arterial blood gases during exercise: validity of transcutaneous measurements. Arch Phys Med Rehabil 2001;82:1686-91. Objective: To investigate the validity of transcutaneous measurements of blood gas tensions for the assessment of partial arterial pressure of oxygen (PaO 2 ) and carbon dioxide (PaCO 2 ) during treadmill exercise. Design: Experimental, self-controlled against a reference standard. Setting: Lung function laboratory. Patients: Eighty-one patients with various lung diseases. Interventions: At rest and at symptom-limited peak exercise, puncture of the radial artery with concurrent transcutaneous measures of blood gases. Main Outcomes Measures: Arterial blood samples were analyzed with a radiometer to measure PaO 2 and PaCO 2.A microgas apparatus was used to measure gas tensions transcutaneously. Values obtained transcutaneously (TcPO 2, TcPCO 2 ) were compared with those obtained by blood sample. TcPO 2 was adjusted as close as possible to the PaO 2 obtained in the same conditions, with the correction factor of the apparatus. Values obtained transcutaneously were compared with those obtained by blood sample to establish the sensitivity and specificity of the noninvasive method. Results: Mean differences standard deviation between transcutaneous and arterial tension at peak exercise were 0.4 7.0mmHg and 2.1 3.3mmHg for PaO 2 and PaCO 2, respectively. The transcutaneous device enabled us to predict a decrease in PaO 2 ( 2mmHg) from rest to exercise with a sensitivity of 92.1% and a specificity of 90% and an increase in PaCO 2 with a sensitivity of 88% and a specificity of 58.9%. Conclusions: Although transcutaneous measurement are sufficiently sensitive and specific to detect patients whose PaO 2 decreases during exercise, its precision is not sufficient for gas exchange calculations. Key Words: Blood gas monitoring, transcutaneous; Exercise test; Lung diseases, interstitial; Rehabilitation. 2001 by the American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and Rehabilitation From the Département de Physiologie, Université Paris 5, Hôpital Ambroise Paré Assistance Publique-Hôpitaux de Paris, Bolougne, France. Accepted in revised form December 27, 2000. Presented as an abstract at the American Thoracic Society s international conference, April 23-28, 1999, San Diego, CA. No commercial party having a direct financial interest in the results of the research supporting this article has or will confer a benefit upon the author(s) or upon any organization with which the author(s) is/are associated. Reprint requests to Carole Planès, MD, PhD, Service d Exploration Fonctionnelle, Hôpital Ambroise Paré, 9 Av Charles de Gaulle, 92104 Boulogne, France, e-mail: bernadette.raffestin@apr.ap-hop-paris.fr. 0003-9993/01/8212-6490$35.00/0 doi:10.1053/apmr.2001.26248 CARDIOPULMONARY EXERCISE TESTING has considerable clinical value in determining the primary cause for dyspnea on exertion. 1,2 This test is useful in the evaluation and follow-up of patients with various pulmonary disorders 3 and has been proposed in the preoperative evaluation of subjects considered for thoracotomy and other major surgical procedures. 4 By assessing arterial blood gas tension (PaO 2 ) during exercise, clinicians can better diagnose and characterize the pulmonary gas exchange response to exercise and its contribution to exercise limitation. 5 A decrease in PaO 2 during exercise may occur in patients with mild interstitial lung disease, even if they have normal diffusing capacity for carbon monoxide at rest. 6-9 Determinations of PaO 2 and arterial carbon dioxide pressure (PaCO 2 ) are also needed to calculate alveolar-arterial difference, 10 which is important in interstitial lung disease for appreciating disease severity and for follow-up evaluation of patients response to therapy. To sample arterial blood, one must either place an arterial line or perform an arterial puncture at peak exercise. Other alternatives such as pulse oxymetry, which allows oxyhemoglobin desaturation to be detected, have been proposed. 11 However, because of the oxyhemoglobin dissociation curve s shape, one may find a significant decrease in PaO 2 with no concomitant decrease in oxygen saturation. Moreover, movement artifacts recorded during exercise often produce erroneous results. PaCO 2 may be assessed indirectly in normal subjects by measuring end-tidal PCO 2 ; however, this method is incorrect when perfusion does not increase appropriately in well-ventilated lung units, causing an arterial end, tidal difference in PCO 2. 9,10,12 Transcutaneous oxygen tension (TcPO 2 ) and carbon dioxide tension (TcPCO 2 ) measured with skin probes equipped with combined Clark and Severinghaus electrodes should reflect partial pressures in arterial blood after arterialization of the skin by warming. 13 The few researchers 14,15 who have reported on the validity of assessing arterial blood gases transcutaneously during exercise concluded that transcutaneous measurements are valid provided the transcutaneous device is calibrated under resting conditions against values obtained by arterial blood analysis. However, because these studies 14,15 had small samples, multiple comparisons of the same exercise test were repeated within subjects but between-subject variability was not assessed. The purpose of the present study was to determine the validity of transcutaneous measurements to assess arterial blood gases at symptom-limited exercise in patients with various lung diseases. Our main hypothesis was that patients with a gas exchange abnormality such as decreased PaO 2 or increased PaCO 2 could be identified by noninvasive, transcutaneous monitoring of gas tensions during exercise. METHODS Subjects We studied 81 consecutive patients who had been referred to the lung function laboratory. Patients characteristics are listed

TRANSCUTANEOUS GASES DURING EXERCISE, Planès 1687 Table 1: Characteristics of the 81 Patients Age (yr) 55.4 16.3 (18 81) Gender (M/F) 45/36 Height (cm) 167 8.87 (147 193) Weight (kg) 68.9 13.5 (45 101) Hemoglobin (g/dl) 14.2 1.47 (11 17.4) NOTE. Values are mean SD (range) or n. Abbreviations: M, male; F, female. in table 1. Arterial blood gas analysis at rest and during exercise was part of the investigation for evaluation of unexplained dyspnea on exertion (n 24) or follow-up of lung disease. Pathologic conditions were chronic obstructive lung disease (n 14), sarcoidosis (n 15), idiopathic pulmonary fibrosis (n 14), interstitial lung disease associated with collagen vascular disorder (n 7) or secondary to occupational exposure (n 4), histiocytosis X (n 1), and other (n 2). No patient had a recent history of acute respiratory failure, heart failure, or angina pectoris. All were in a stable clinical state and none had anemia. Protocol Arterial blood was obtained at rest in the sitting position by puncture of the radial artery with a heparinized arterial blood sampler. a The blood was immediately analyzed in duplicate for blood gas determination with a Radiometer ABL520 b that had its accuracy verified with tonometered blood at least twice a week. Transcutaneous blood gas was measured with a commercially available device. c According to the manufacturer s data, the 90% response time is below 25 seconds for the oxygen electrode and below 60 seconds for the carbon dixode electrode. At 1-second intervals, the device updated PO 2 and PCO 2 values, displaying them by light-emitting diode. After calibrating the monitor according to the manufacturer s instructions, the investigator placed its transcutaneous sensor (set at 45 C) on the skin of the patient s volar side of the arm. After a 20-minute equilibration, during which time the patient remained sitting, the investigator adjusted the patient s TcPO 2 value with a correction factor that placed it as close as possible to the PaO 2 obtained in the same conditions. The patient then exercized on a treadmill d with an increase of the load by approximately 30 watts every 3 minutes until he/she reached the level of symptom-limited exercise. Brachial artery pressure was measured every 3 minutes with a sphygmomanometer cuff. Transcutaneous blood gas and electrocardiogram were continuously monitored throughout the exercise period. No test had to be stopped because of decreased systemic artery pressure, anginal pain, ST-segment depression on electrocardiogram, or arrhythmias. On reaching symptom-limited exercise, just before stopping, the patient again provided arterial blood, which was obtained by puncture of the radial artery. The blood was immediately analyzed for comparison with gas tensions obtained concurrently by the transcutaneous method. Data Analysis Agreement between transcutaneous and arterial blood gas tensions at maximal exercise was evaluated according to Bland and Altman. 16 We used a paired t test to evaluate whether a systematic difference existed between transcutaneous and arterial measurements. To examine the validity of the transcutaneous method in following the trend of gas exchange, we compared the amplitudes of the changes in transcutaneous and arterial tensions from rest to exercise. Differences of gas tension from rest sitting (R) to symptom-limited exercise (E) were plotted for transcutaneous versus arterial blood measurements expressed as (PaO 2 /PaCO 2E R ) and (TcPO 2 /TcPCO 2E R ). We then calculated the sensitivity and specificity of transcutaneous gas tension measurements to predict a decrease in PaO 2 or an increase in PaCO 2 with exercise. The sensitivity of TcPO 2 to detect a decrease in PaO 2 was calculated as the percentage of subjects with an exercise-induced decrease in PaO 2 who also had a decrease in TcPO 2 and the specificity as the percentage of patients with no change or an increase in PaO 2 who had no decrease in TcPO 2. Similar analysis was performed for patients with an exercise-induced increase in PaCO 2. Results were expressed as mean standard deviation (SD). RESULTS According to reported normal values for age, 17 PaO 2 at rest (PaO 2R ) was low in 34 of the 81 patients, being below 60mmHg in 7 subjects. Only 8 patients had PaCO 2R values above 44mmHg. None of the patients had ph below 7.36 at rest and H concentration rose from 37.7 2.6nmol/L to 40.4 4.1nmol/L at peak exercise, which was reached in 11.3 6.7 minutes. Heart rate reached 78.2% 14.2% of maximal predicted heart rate. There was a rather large scatter of differences between oxygen or carbon dioxide tensions at peak exercise for the noninvasive and invasive methods. However, no systematic difference existed between TcPO 2E and PaO 2E, the difference (TcPO 2E PaO 2E ) being 0.4 7.0mmHg (not significant, NS). In contrast, TcPCO 2E overestimated PaCO 2E, the difference (TcPCO 2E PaCO 2E ) being 2.1 3.3mmHg (p.001; fig 1). The differences between methods did not appear related to the average value of PO 2 or PCO 2 or to the intensity of exercise. The plot of TcPO 2E R versus PaO 2E R is given in figure 2. Among the 55 patients with a decrease ( 1mmHg) in PaO 2 from rest to maximal exercise, 49 had a decrease ( 1mmHg) in TcPO 2. Four of the 26 patients with no change or increase in PaO 2 had a decrease in TcPO 2. Therefore, measurement of TcPO 2 predicted a decrease in PaO 2 with a sensitivity of 89.1% and a specificity of 84.6%. Because a decrease in PaO 2 below 2mmHg is within the limits of precision of blood gas analyzers, 18 the sensitivity and specificity of the transcutaneous device to detect a decrease in PaO 2 larger than 2mmHg was also computed for varying cutoff values of the decrease in TcPO 2 (table 2). Sensitivity was greater when only the decreases in PaO 2 larger than 2mmHg were taken into account. Moreover, as the threshold drop in TcPO 2 increased above 2mmHg, specificity reached at least 90% at the expense of a small decrease in sensitivity. In 2 patients whose TcPO 2 increased despite a decrease in PaO 2 from rest to exercise (10 and 15mmHg, respectively), PaO 2 at peak exercise was nevertheless equal to or above 85mmHg. In the 49 patients with an exercise-induced decrease in both PaO 2 and TcPO 2, the difference between the amplitudes of changes in transcutaneous and arterial PO 2 ( TcPO 2E R PaO 2E R ) was 2 6.55mmHg (p.05). The significant underestimation of the amplitude of changes in PaO 2 by TcPO 2 measures was not related to the average decrease in oxygen pressure (fig 3). Seven patients those with interstitial lung disease and a fall in PaO 2 from rest to exercise of at least 7mmHg repeated the exercise tests within 6 to 24 months after the initial tests. In each of these patients, the concomitant decrease in TcPO 2 and PaO 2 with exercise was consistently found during each of the other tests (1 7 per patient). When TcPO 2 under- or overesti-

1688 TRANSCUTANEOUS GASES DURING EXERCISE, Planès Fig 1. Differences between blood gas tensions measured transcutaneously and by arterial blood sampling (A) oxygen and (B) carbon dioxide at symptom-limited exercise vs average of pressures obtained with the 2 methods. No systematic difference existed between TcPO 2E and PaO 2E, but TcPCO 2E significantly overestimated PaCO 2E (p <.001, paired t test). Legend:, standard deviation;, mean. mated PaO 2 at exercise in the first test, this under- or overestimation was consistently found in all other tests performed on the same subject, but varied in magnitude. Figure 4 shows a typical example of a patient who underwent 7 successive tests over a period of 2 years. Overestimation of PaO 2 at exercise varied in amplitude from 1 test to another but was consistent with all the tests. After adjusting TcPO 2 with the correction factor of the apparatus as close as possible to PaO 2 obtained in the same conditions (rest, sitting position), TcPCO 2R did not differ significantly from PaCO 2R, the difference (TcPCO 2R PaCO 2R ) being 0.4 3.11mmHg. The plot of TcPCO 2E R versus PaCO 2E R is given in figure 5. Among the 25 patients with an increase in PaCO 2 from rest to maximal exercise, 22 had an increase in TcPCO 2 and 33 of the 56 patients with no change or a decrease in PaCO 2 had a decrease or no change in TcPCO 2. Therefore, transcutaneous measurement, TcPCO 2, can predict an increase in PaCO 2 with a sensitivity of 88% and a specificity of 58.9%. Taking into account the precision limits of blood gas analyzers, 18 the sensitivity and specificity of the transcutaneous device to detect an increase in PaCO 2 larger than 2mmHg was also computed for the varying cutoff values of the increase in TcPCO 2 (table 3). Sensitivity was greater when only the increases in PaCO 2 larger than 2mmHg were taken into account. Table 2: Predictive Accuracy of a Decrease in PaO 2 From Rest to Exercise by a Decrease in TcPO 2 of Various Amplitudes Fig 2. Magnitudes of O 2 change from rest sitting to symptomlimited exercise, ( TcPO 2E R vs PaO 2E R ). Because resting values were subtracted from values obtained at exercise, values of patients with a decrease in both PaO 2 and TcPO 2 at exercise are negative and therefore in the left lower quadrant. Legend:, identity line;, regression line. PaO 2E R 1mmHg PaO 2E R 2mmHg TcPO 2E R 1mmHg Sensitivity (%) 89.1 (49/55) 94.1 (48/51) Specificity (%) 84.6 (22/26) 83.3 (25/30) TcPO 2E R 2mmHg Sensitivity (%) 85.4 (47/55) 92.1 (47/51) Specificity (%) 92.3 (24/26) 90.0 (27/30) NOTE. Parentheses indicate numbers of positive over true positive patients (sensitivity) and numbers of negative over true negative patients (specificity) according to the cutoff values of the decrease in TcPO 2.

TRANSCUTANEOUS GASES DURING EXERCISE, Planès 1689 Fig 3. Differences between the amplitudes of exercise-induced changes in PO 2 values obtained transcutaneously and by blood sampling ( TcPO 2E R PaO 2E R ) versus average of changes with the 2 methods. Only values of patients with a decrease in both TcPO 2E and PaO 2E are presented. For clarity, PO 2 values during exercise were here subtracted from those at rest (PO 2E R ). This approach yielded positive values. Underestimation of changes in arterial PO 2 by transcutaneous measures was statistically significant (p <.05, paired t test). Legend:, standard deviation;, mean. Moreover, as the threshold increase in TcPCO 2 increased above 2mmHg, specificity rose but sensitivity decreased greatly. In the 22 patients whose PaCO 2 and TcPCO 2 values both increased from rest to exercise, the difference in the amplitudes of the increase in transcutaneous and arterial PCO 2 ( TcPCO 2E R PaCO 2E R ) was.18 4.9mmHg (NS) and did not appear to be related to the average increase in CO 2 tension (fig 6). Fig 5. Magnitudes of CO 2 change from rest sitting to symptomlimited exercise ( TcPCO 2E R vs PaCO 2E R ). Because resting values were subtracted from values obtained at exercise, values of patients with an increase in both PaCO 2 and TcPCO 2 at exercise are positive and therefore in the right upper quadrant. Legend:, regression line;, identity line. DISCUSSION The present study shows that after an initial in vivo calibration at rest, the trend of PaO 2 change from rest to peak exercise may be assessed by transcutaneous measurements with a 92% sensitivity and a 90% specificity. Transcutaneous measurement of PCO 2 detected an increase in PaCO 2 with exercise with 92% sensitivity but only 56% specificity. Because accuracy and precision of the technique are greatly enhanced by in vivo calibration, TcPO 2 was adjusted as close as possible to PaO 2 obtained by a simple artery puncture in the resting condition. One limitation of transcutaneous assessment of arterial blood gas tensions is the skin blood flow dependence of transcutaneous gas tensions. As previously shown, 19 transcutaneous values deviate from their relationship with the arterial tensions when cardiac output and peripheral perfusion are reduced. The same limitation may apply to pulse oxymetry used during exercise to monitor patients with reduced peripheral perfusion. Of note, none of the patients in the present sample had overt heart failure. Another concern was that the slow response time of transcutaneous probes could affect val- Table 3: Predictive Accuracy of an Increase in PaCO 2 From Rest to Exercise by an Increase in TcPCO 2 of Various Amplitudes Fig 4. Typical example of O 2 tensions at rest and with symptomlimited exercise in a 68-year-old man with interstitial lung disease who underwent repeated exercise tests over 2 years to follow response to steroid treatment. Oxygen tensions measured at rest in the arterial blood (PaO 2R ) at symptom-limited peak exercise by the transcutaneous probe (TcPO 2E ) and arterial blood analysis (PaO 2E ) are given for each of 7 successive tests. PaCO 2E R 1mmHg PaCO 2E R 2mmHg TcPCO 2E R 1mmHg Sensitivity (%) 88.0 (22/25) 92.8 (13/14) Specificity (%) 58.9 (33/56) 56.7 (38/67) TcPCO 2E R 2mmHg Sensitivity (%) 56.0 (14/25) 64.3 (9/14) Specificity (%) 80.3 (45/56) 77.6 (52/67) NOTE. Parentheses indicate numbers of positive over true positive patients (sensitivity) and numbers of negative over true negative patients (specificity) according to the cutoff values of the increase in TcPCO 2.

1690 TRANSCUTANEOUS GASES DURING EXERCISE, Planès Fig 6. Differences in the magnitudes of exercise-induced change in PCO 2 values obtained transcutaneously and by blood sampling ( TcPO 2E R PaO 2E R ) versus average of changes with the 2 methods. Only values of patients with an exercise-induced increase in both TcPCO 2 and PaCO 2 are presented. No systematic significant difference existed between the measurement methods (paired t test). Legend:, standard deviation;, mean. ues during a rapidly progressive exercise protocol. We attempted to limit this problem by establishing an exercise protocol that had workload increases every 3 minutes instead of every 1 minute. However, because values of PaO 2 and PaCO 2 were not followed continuously during exercise, but only measured at peak exercise, we can only speculate about their stability during the last minutes of exercise. The fact that no patient discontinued exercise before the end of sampling rules out a rapid return of arterial blood gas tensions to their baseline values. Nevertheless, in the most severe patients, response time of the transcutaneous electrodes might be too long to follow accurately some of the changes of arterial blood gas tensions at the end of exercise. However, as shown in figure 2, the differences between the amplitudes of the decrease in transcutaneous and arterial oxygen tension does not appear to be larger for these subjects. It is therefore unlikely that this technical limitation significantly affected our results. Moreover, for a large range of PO 2 values (39 105mmHg), the transcutaneous method did not systematically or significantly under- or overestimate any patient s oxygen tension. The difference between TcPO 2 and PaO 2 values was not related to the magnitude of PO 2. These results contrast with a previous study 20 in which the transcutaneous probe underestimated O 2 pressure, its results increasing linearly with O 2 pressure obtained invasively, a discrepancy that may be explained by the lack of in vivo calibration in that study. Nevertheless, in the present study, the SD of the differences between the 2 methods of measuring PO 2 was 7mmHg, which means that the 95% confidence limit of agreement between the 2 methods was clinically significant ( 14.5 to 13.5mmHg). In a large range of PCO 2 (25 55mmHg), PaCO 2 was overestimated by transcutaneous measurements; the magnitude of the overestimation was not dependent on the PCO 2 value. The mean difference of CO 2 pressure was 2mmHg with 1 SD of the differences (3mmHg) smaller than for O 2 pressure. In studies comparing transcutaneous and arterial tensions, the latter are taken as the reference values. However, limited precision and accuracy of blood gas analyzers must be taken into account. 18 A previous study 18 of arterial blood gas analyzer performance showed that the systematic tendency to under- or overestimate PO 2 relative to values in tonometered human blood may reach 5%, and variation between values obtained with the same sample and the same analyzer may reach 3%, whereas identical samples of PO 2 analyzed on several brands of analyzers may differ by 10% or more. The main purpose of the present study was to examine whether the noninvasive, transcutaneous monitoring of gas tensions during exercise can help identify patients whose arterial blood gas values during exercise indicate a gas exchange abnormality. In this setting, a decrease in PaO 2 during exercise is abnormal. Transcutaneous monitoring enabled us to identify, with a sensitivity of 92.1% and a specificity of 90%, those patients whose decrease in PaO 2 at peak exercise was larger than 2mmHg. Noteworthy, 2 of the 4 patients whose significant decrease in PaO 2E R was unpredicted by a concomitant decrease in TcPO 2E R, did not have significantly impaired gas exchange at peak exercise. Indeed, these 2 patients maintained PaO 2 and the alveolar to arterial PO 2 difference (as calculated by alveolar equation) within normal limits. 21 The amplitude of the decrease in PaO 2E R was underestimated by the transcutaneous device. However, considering the accuracy and precision of blood gas analyzers, the mean difference of 2mmHg between values obtained by invasive versus noninvasive methods was not clinically significant. Moreover, the transcutaneous device s estimate of PaO 2 variations was not related to the amplitude of the variations. This finding agrees with that for O 2 pressures obtained during exercise. Whether obtained invasively or noninvasively, no relation existed between differences in TcPO 2E and PaO 2E and the magnitude of PO 2. In healthy subjects, arterial carbon dioxide tension remains within resting values below the anaerobic threshold and decreases above it as a reflection of respiratory compensation for metabolic acidosis. An increase in PaCO 2 during exercise is therefore abnormal. Although TcPCO 2 measurements could predict the occurrence of an abnormal increase in PaCO 2 with rather good sensitivity, specificity was poor and could be improved only at the expense of an unacceptable decrease in sensitivity. Of note, the patient with the largest increase in PaCO 2 from rest to exercise (14mmHg) had only a 1mmHg increase in TcPCO 2. Seven patients with interstitial lung diseases had repeated exercise tests with concomitant measurement of TcPO 2 and PaO 2 during follow-up. Although the limited number of patients as well as variations in the number of tests per patient preclude any statistical analysis, we found that in each patient, the over- or underestimation of TcPO 2E was consistent from 1 test to another with a trend for it to follow PaO 2 at symptomlimited exercise. However, further studies with a larger sample of patients will be needed to confirm that TcPO 2 monitoring during repeated exercise tests can be useful in the follow-up of patients with lung diseases. CONCLUSION Transcutaneously obtained values for arterial oxygen pressures can be used to predict a decrease in PaO 2 from rest to exercise with good sensitivity and specificity, provided the device has an initial in vivo calibration, the workload is gradually increased at 3-minute intervals, and the patient has no overt heart failure. However, with the transcutaneous device tested in the present study, we could not predict an increase in PaCO 2 with both satisfactory sensitivity and specificity. The estimates of PaO 2 and PaCO 2 by transcutaneous measurements were not sufficiently precise to calculate accurately alveolararterial difference.

TRANSCUTANEOUS GASES DURING EXERCISE, Planès 1691 References 1. Weisman I, Zeballos R. Clinical evaluation of unexplained dyspnea. Cardiologia 1996;41:621-34. 2. Mohnssen S, Weisman I, Zeballos R, Hall A, Wasserman K, Tavel M. Heart or lung disease. Determining the primary cause for dyspnea on exertion. Chest 1998;113:1705-10. 3. Wasserman K, Whipp B. Exercise physiology in health and disease. Am Rev Respir Dis 1975;112:219-49. 4. Older P, Smith R, Courtney P, Hone R. Preoperative evaluation of cardiac failure and ischemia in elderly patients by cardiopulmonary exercise testing. Chest 1993;104:701-4. 5. Raffestin B, Escourrou P, Legrand A, Duroux P, Lockhart A. Circulatory transport of oxygen in patients with chronic airflow obstruction exercising maximally. Am Rev Respir Dis 1982;125: 426-31. 6. Crystal R, Roberts W, Hunninghake G, Gadek J, Fulmer J, Line B. Pulmonary sarcoidosis: a disease characterized and perpetuated by activated lung T-lymphocytes. Ann Intern Med 1981;94:73-94. 7. Gaensler E, Carrington C. Open biopsy for chronic diffuse infiltrative lung disease: clinical, roentgenographic, and physiological correlations in 502 patients. Ann Thorac Surg 1980;30:411-26. 8. Xaubet A, Agusti C, Luburich P, Roca J, Monton C, Ayuso MC, et al. Pulmonary function tests and CT scan in the management of idiopathic pulmonary fibrosis. Am J Respir Crit Care Med 1998; 158:431-6. 9. Sue D, Oren A, Hansen J, Wasserman K. Diffusing capacity for carbon monoxide as a predictor of gas exchange during exercise. N Engl J Med 1987;316:1301-6. 10. Jones N, McHardy G, Naimark A, Campbell E. Physiological dead space and alveolar-arterial gas pressure differences during exercise. Clin Sci 1966;31:19-29. 11. Escourrou P, Delaperche M, Visseaux A. Reliability of pulse oximetry during exercise in pulmonary patients. Chest 1990;97: 635-8. 12. Lewis D, Sietsema K, Casaburi R, Sue D. Inaccuracy of noninvasive estimates of VD/VT in clinical exercise testing. Chest 1994;104:701-4. 13. Clark J, Votteri B, Ariagno R, Cheung P, Eichhorn JH, Fallat RJ, et al. Noninvasive assessment of blood gases. Am Rev Respir Dis 1992;145:220-32. 14. Sridhar M, Carter R, Moran F, Banham S. Use of a combined oxygen and carbon dioxide transcutaneous electrode in the estimation of gas exchange during exercise. Thorax 1993;48:643-7. 15. Brudin L, Berg S, Ekberg P, Castenfors J. Is transcutaneous PO 2 monitoring during exercise a reliable alternative to arterial PO 2 measurements? Clin Physiol 1994;14:47-52. 16. Bland J, Altman D. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;i: 8476:307-10. 17. Crapo R, Jensen R, Hegewald M, Tashkin D. Arterial blood gas reference values for sea level and an altitude of 1400 meters. Am J Respir Crit Care Med 1999;160:1525-31. 18. Scuderi P, Macgregor D, Bowton D, Harris L, Anderson R, James R. Performance characteristics and interanalyzer variability of PO 2 measurements using tonometered human blood. Am Rev Respir Dis 1993;147:1354-9. 19. Tremper K, Shoemaker W. Transcutaneous oxygen monitoring of critically ill adults, with and without low flow shock. Crit Care Med 1981;9:706-9. 20. Rosner V, Hannhart B, Chabot F, Polu J. Validity of transcutaneous oxygen/carbon dioxide pressure measurement in the monitoring of mechanical ventilation in stable chronic respiratory failure. Eur Respir J 1999;13:1044-7. 21. Hansen J, Sue D, Wasserman K. Predicted values for clinical exercise testing. Am Rev Respir Dis 1984;129:S49-55. Suppliers a. AVL Medical Instruments AG, Stettermerstr 28, CH-8207 Schaffhausen, Switzerland. b. A/S, Emdrupveg 72, DK-2400, Copenhagen NV, Denmark. c. Kontron Microgas 7650; Linde Medical Sensors AG, Austr 25, CH4051 Basel, Switzerland. d. Marquette Medical System, GE Medical Systems, 3000 N Grandview Blvd, Waukesha, WI 53188.