Share this article on:

Reliability of a new device to assess the oxygen consumption of human respiratory muscles


Medicine & Science in Sports & Exercise: July 1999 - Volume 31 - Issue 7 - pp 1076-1082
Special Communications: Technical Note

Reliability of a new device to assess the oxygen consumption of human respiratory muscles. Med. Sci. Sports Exerc., Vol. 31, No. 7, pp. 1076-1082, 1999.

Purpose: This study tests the reliability of a new device for assessing the oxygen consumption of the respiratory muscles (V̇O2 resp.).

Methods: Fourteen healthy male volunteers participated in the study. The device consists of an expandable external ventilatory dead space created with pieces of plastic tubing and a spirometer filled with 100% oxygen. It also incorporates a carbon dioxide absorber. Total V̇O2 (V̇O2 tot.) was recorded from the spirometric closed circuit and ventilation (V̇E), from the spirometer tracing. For each subject the test procedure was carried out in duplicate (T1 and T2) after an overnight fast. The dead space was increased at a constant rate of 260 mL every 90 s, and V̇O2 tot. and V̇E increased progressively. Because log V̇O2 tot. was linearly related to V̇E, we calculated the slope value (log V̇O2-V̇E) and the Y-intercept (V̇E = 0) of the semilog regression representing, respectively, V̇O2 resp. and metabolic V̇O2 (V̇O2 met.).

Results: When compared with values in the literature, these values did not differ from those recorded in subjects of a similar age group. The V̇O2 resp. and V̇O2 met. calculated in T1 and T2 were not different (V̇O2 resp. = 0.0066 ± 0.0005 for T1 vs 0.0067 ± 0.0005 log mL·min−1/L·min−1 for T2 and V̇O2 met. = 269.3 ± 28.6 for T1 vs 281.9 ± 24.1 mL·min−1 for T2). The coefficients of variation were: 25% at T1 and 23% at T2 for V̇O2 resp. and 34% at T1 and 29% at T2 for V̇O2 met. Moreover, significant correlations (r = 0.96, P < 0.001 for V̇O2 resp., r = 0.95, P < 0.001 for V̇O2 met.), high coefficients of determination (r2 = 0.92 for V̇O2 resp., r2 = 0.90 for V̇O2 met.) and negligible SEE (0.0005 for V̇O2 resp., 0.2 mL·min−1 for V̇O2 met.) were found between the two tests. When we plotted the mean values of V̇O2 resp. and V̇O2 met. measured at T1 and T2 against their respective differences, more than 95% of the slight differences ranged between the limits defined by mean value ± 2 SD, reflecting the small discrepancy between duplicate measurements.

Conclusion: The results confirm that the test performed with this device is useful and reliable for assessing the V̇O2 resp. in healthy subjects.

TEF-Adaptations Physiologiques et Comportementales, Faculté de Médecine (EA 2088), Université de Picardie Jules Verne, 80036 Amiens, FRANCE; Laboratoire de Physiologie des Interactions (EA 701), Service d'Explorations Fonctionnelles Respiratoires, Hôpital Arnaud de Villeneuve, 34295 Montpellier, FRANCE; Service d'Explorations Fonctionnelles Cardio-pulmonaires Pédiatriques, Hôpital Nord, 80000 Amiens, FRANCE; Faculté des Sciences du Sport, Université de Picardie Jules Verne, 80025 Amiens, FRANCE

Submitted for publication July 1997.

Accepted for publication December 1997.

The authors gratefully thank the medical and technical staff of the service "Explorations Fonctionnelles Cardio-pulmonaires Pédiatriques," particularly Ms. and Mr. Goin and the subjects for their generous cooperation with our project. We also thank Ms. Stott Carmeni, Mr. Telliez, Mr. Bouferrache, and Mr. Bosle for valuable technical assistance.

Address for correspondence: Prof. S. Ahmaidi, Ph.D., Faculté des Sciences du Sport, Allée P. Grousset, 80025, Amiens Cedex 1, France. E-mail:

The metabolic requirement of the respiratory muscles is an important determinant of exercise capacity. To understand better the reduced exercise capacity resulting from pulmonary disease, it would be helpful to record the total work done by the respiratory muscles, determine the energy cost to perform this work, and evaluate the efficiency of these muscles in diseased and healthy states. The evaluation of respiratory muscle performance becomes crucial when this muscle function is impaired, particularly for the development and the benefits assessment of adapted exercise training programs. Most studies (5,9,10) have been conducted in subjects with chronic obstructive pulmonary disease (COPD). The authors showed that V̇O2 resp. increased more in COPD patients than in healthy subjects with increased minute ventilation. Recently Shindoh et al. (17) demonstrated that with COPD, V̇O2 resp. was greater compared with that of age-matched normal subjects at a given ventilation level. In healthy subjects it is well known that pulmonary resistance increases (4) while chest wall compliance decreases with age (15). Such functional changes with aging increase respiratory muscle work and the oxygen cost of breathing. In a recent study Takishima et al. (18) showed a positive relationship between V̇O2 resp. and the age of healthy subjects.

Three techniques for assessing V̇O2 resp. have been described in the literature: 1) voluntary hyperventilation, in which V̇O2 resp. is measured at several breathing rates (1,3); 2) flow resistive loading with a water manometer (5) and narrow tubing (8,11,16); and 3) externally added ventilatory dead space volume between the subject and a spirometer (6,14). Since these techniques require measurements at several discrete points and are time consuming, they are unsuitable for measuring V̇O2 resp. in large numbers of subjects. More recently, Takishima et al. (18) designed a device to record V̇O2 resp. by modifying the method of Cherniack (6). V̇O2 resp. can thus be recorded over a wide range of ventilation and requires only 15 min per subject for measurements. But the device is technically complex and difficult to use, and assessment is further limited by the fact that it requires a sophisticated data acquisition system. The device consists of a low-resistance plastic tube around a wooden disk, which is mounted on a flat acrylic disk as an anchor for a steel wire. The steel wire is wound by a motor so that the acrylic resin disk is rotated around the iron axis, which continuously expands the dead space volume at a constant rate. Such a device is not easy to use in routine practice and does not lend itself to testing large groups.

The goal of this study was to present a new and simpler device to assess V̇O2 resp. in healthy humans by increasing the external ventilatory dead space volume and then to analyze its reliability.

Back to Top | Article Outline


Subjects. Fourteen healthy male volunteers (age, 27.3 ± 5.5 yr, range 25-35; height, 179.5 ± 6 cm; body mass, 77.5 ± 0.9 kg; body surface area, 1.96 ± 0.09 m2; forced vital capacity, 5.37 ± 0.6 L; and forced expiratory volume in 1 s = 4.55 ± 0.5 L·s−1) participated in the study. The subjects were sedentary and none were smokers or on medication of any sort. One week before the experiment they underwent a physical examination, including a medical history and resting electrocardiogram. All were free from complaints referable to the cardiac, pulmonary, or neuromuscular system, and all had normal physical examinations. A preliminary assessment of the oxygen consumption of the respiratory muscles was conducted to adapt the subjects to the procedures, equipment, and laboratory environment. The nature, procedures, and possible risks were explained to each subject and consent was obtained before the experiment. The experimental procedures were approved by the local ethics committee.

Apparatus and respiratory muscle oxygen consumption measurement. The experimental device had two parts. The first concerned the system for increasing ventilatory dead space volume and the second concerned the measurement of oxygen consumption (Fig. 1). The first part included a circuit made of corrugated tubing, a three-way Hans-Rudolph valve, a Douglas bag, and a mass spectrometer gas analyzer (Perkin-Elmer MG-1100, Marquette, IL). The volume of this section was 300 mL, and the increases in ventilatory dead space volume were made possible with additional pieces of vinyl polychlorure plastic tubing. The tubes had internal and external diameters of 3.5 and 3.9 cm, respectively, and were fitted at both ends with three-way taps. The individual tubes were connected via these taps. Tube length was adjusted so that the increase in each dead space section did not exceed twice the theoretical physiologic dead space (approximately 2 mL·kg−1 body mass) estimated for a minimal standard body mass of 65 kg. The system thus allowed an increase in volume by steps at a constant rate of 260 mL every 90 s with the use of the taps. The minimal and maximal dead space volumes ranged between 1.1 and 2.9 L. The total dead space including the fixed portion ranged from 1.1 to 4.0 L. The resistance of the system varied from 0.01 to 0.04 cm H2O·L−1·s−1 at flow volumes of 10 and 60 L·min−1, respectively, in the minimal configuration up to 0.09-0.33 cm H2O·L−1·s−1 at flow volumes of 10 and 60 L·min−1 in the maximal configuration. This part of the device was connected to the spirometric oxygen consumption measuring system. The total oxygen consumption (V̇O2) was measured by the volume decrease of a 9-L Gould-Godard spirometer filled with 100% oxygen and connected to the expanding dead space by a pair of flexible tubes. A soda lime box was located in the expiratory airflow tube of the spirometer to absorb carbon dioxide. The spirometer was refilled with oxygen as necessary. The absence of air leakage through the system was verified before each measurement. The absence of outward air leakage through the system was tested by putting a 400-g weight on the spirometer bell and from spirometer tracings after calibrated volumes of air were introduced at various flow rates. Lung volume was monitored by inductance plethysmography (Respitrace, Ardsley, NY) with abdominal and rib cage belts according to the procedure of Konno and Mead (13). Calibration was performed with isovolume maneuvers at functional residual capacity (FRC), with the subject off the mouthpiece. Abdominal and rib cage displacements were displayed on a oscilloscope visible to the subject. Each subject was instructed to use the isovolume FRC line as a target for end expiration during the subsequent tests.

Experimental protocol. The spirometry and flow-volume analysis were conducted before the experiment. The maximum expiratory flow volume curves were measured with a digital spirometer (Datalink Pulmochort, Montpellier, France) to record the forced vital capacity and the maximal expiratory volume per second (FEV1). Two tests (T1 and T2) measuring the oxygen consumption of the respiratory muscles were then performed by each subject with a 3-d interval. The measurements were performed in the morning after an overnight fast before breakfast and after the subject had been resting, sitting in a comfortable chair, for 30 min.

The speed of the unwinding paper of the spirometer (1 mm·s−1) was verified before and after each recording. The subject assumed a sitting position, and the rib cage and abdomen Respitrace belts were fitted. The subject could not observe the dead space expansion. At first, the three-way valve between the mouth and the Douglas bag was opened to allow breathing of 100% oxygen for 3 min. This procedure allows the elimination of nitrogen (N2) from the lungs. The oxygen concentration in the expandable circuit was verified before each experiment by a mass spectrometer gas analyzer. The subjects were asked to maintain their control FRC (i.e., the FRC off the mouthpiece) by matching their rib cage and abdominal displacement signals at end expiration to the isovolume FRC control line on the oscilloscope. The subjects were required to maintain the control FRC level on the oscilloscope. The three-way valve was then turned, connecting the subject to the expandable dead space volume and oxygen measurement device. The expandable dead space was kept at minimal volume for 3 min and then was increased by 260 mL every 90 s. The 3-min period allowed the subject to reach a nearly steady state at the initial dead space load of 1.1 L. Each subject breathed continuously with this increasing dead space volume until he could tolerate it no longer. This point was signaled by the subject who raised his hand, and the procedure was accordingly stopped. Heart rate was recorded for each dead space load with a cardioscope.

Data analysis. The total oxygen consumption (V̇O2 tot. mL·min−1 STPD) was calculated, for each dead space load, as the slope of the 90-s increase in end-expiratory volume recorded with the spirometer. We assumed that metabolic V̇O2 (V̇O2 met.) was constant during V̇O2 measurement and we defined V̇O2 tot. as the addition of V̇O2 met. and V̇O2 resp. (V̇O2 tot. = V̇O2 met. + V̇O2 resp.). Therefore, the increment of V̇O2 tot. reflects that of V̇O2 resp., which is closely related to pulmonary ventilation. The paired points of V̇O2 tot. and V̇E were obtained every 90 s and plotted on the semilog chart. The individual logarithm of V̇O2 tot. was linearly related to ventilation (V̇E). The individual logarithm of V̇O2 tot. was therefore regressed by least squares against V̇E (log V̇O2 tot. = aV̇E + b). The slope represents the V̇O2 increase in the respiratory muscles. The Y-intercept of the semilog regression line represents the metabolic V̇O2 of muscles other than respiratory ones. The V̇O2 resp. was thus calculated by the slope of the semilog regression (log V̇O2-V̇E) line and the V̇O2 met., by the Y-intercept (V̇E = 0).

Statistics. The data are expressed by mean ± SEM. During the two tests, the relationship between log V̇O2 tot. and V̇E was studied by linear regression analysis. Comparisons with literature data were made with a Student unpaired t-test. The values of V̇O2 resp. and V̇O2 met. obtained from the duplicate measurements were compared by the Student paired-t test. Degrees of relation between the two tests for the V̇O2 resp. and V̇O2 met. assessments were determined from correlation coefficient analysis. To test the agreement between the two measurements, the individual differences of V̇O2 resp. and of V̇O2 met. measured at T1 and T2 were plotted against their respective means. The agreement was calculated from the biases, estimated by the differences between the means and the SD of the differences according to Bland and Altman's method (2). The statistical significance was set at P < 0.05.

Back to Top | Article Outline


The relationships between the mean values of V̇O2 and V̇E, measured at each dead space load and expressed, respectively, in mL·min−1 and L·min−1, are presented in Figure 2.

Heart rate values increased with increasing dead space loading (T1: HR rest = 80 ± 10, HR exhaustion = 115 ± 4 beats·min−1 vs T1: HR rest = 87 ± 8, HR exhaustion = 110 ± 6 beats·min−1). However, the difference between HR values recorded during the duplicate measurements was not significant.

Comparison with conventional methods. The mean peak V̇E value obtained reached 60 ± 2.9 L·min−1. This value is close to those reported by Cherniack (6) and Fritts et al. (12) and lower than those of Bradley and Keith (3). The experimental value of V̇O2 resp. (0.0066 log mL·min−1/L·min−1) was intermediate between those noted by Bartlett et al. (1), Campbell et al. (5), Cournand et al. (9), and Milic-Emili and Petit (14) (0.0042, 0.0011, 0.0054, and 0.0011 log mL·min−1/L·min−1, respectively) and those of Fritts et al. (12) (0.0083 log mL·min−1/L·min−1), whereas they did not differ from those of Cherniack (6) and Takishima et al. (18) (0.0071 and 0.0072 log mL·min−1/L·min−1, respectively).

Reliability of the new device. Table 1 shows the values of the T1 and T2 trials. No differences were observed for V̇O2 resp. and V̇O2 met. The intersubject coefficients of variation ranged between 23 and 34%. Significant correlation coefficients between V̇O2 resp. at T1 and T2 (Fig. 3) and V̇O2 met. at T1 and T2 (Fig. 4), and high coefficients of determination (r2) were found. The SEE were small. Moreover, the two measurements were in close agreement. When the mean values of V̇O2 resp. (Fig. 5) and V̇O2 met. (Fig. 6) measured at T1 and T2 were plotted against the corresponding differences for each individual, at least 95% of the small differences ranged between mean ± 2 SD, reflecting the small discrepancy between the duplicate measurements.

Back to Top | Article Outline


In this study a new device is presented and tested for reliability. The device was designed to assess the oxygen consumption of the respiratory muscles in healthy subjects by increasing external ventilatory dead space volume according to the method of Cherniack (6). The findings indicate that it gives reliable values of metabolic V̇O2 and the V̇O2 of respiratory muscles. The values of respiratory muscle oxygen consumption are comparable with those of age-matched subjects reported by other authors (6,18). Moreover, duplicate measurements demonstrated the high agreement for this functional parameter. As recommended by Takishima et al. (18), we used a method of expanded external dead space volume, which has many advantages. First, this technique gives a more detailed analysis of V̇O2 resp. than is possible with other methods since V̇O2 is continuously measured from resting to high levels of ventilation. In contrast, other methods use discrete measurements and do not cover a wide range of ventilation. The second advantage is that V̇O2 resp. can be measured more easily. We generally were able to record a V̇E-V̇O2 curve during an experimental session that lasted only 20 min, whereas with discrete measurements the test duration can be as long as 40 min, seriously impeding routine clinical use in large populations. Finally, this technique also has the advantage of not requiring voluntary hyperventilation.

Methodological considerations. All methods to measure the oxygen consumption of the respiratory muscles depend on the total body oxygen consumption recorded at various levels of ventilation. As the oxygen consumption of the respiratory muscles is small compared with that of the total body, any error in the measurement or change in the metabolic rate or activity of other muscles will lead to a false determination of the oxygen consumption of these muscles. In the present study, the variability in the nonrespiratory muscle oxygen consumption was reduced since all the subjects were in steady metabolic state. They were tested after an overnight fast without breakfast, and careful attention was paid to postural comfort (sitting position) so that the body muscles were as relaxed as possible.

Usually with a closed circuit spirometer there is no large instrumental error. The only main error concerns the determination of total V̇O2 from the slope of the spirograph recording, which may be difficult to draw because of increases in respiratory instability. This error is reduced by accustoming the subjects to the procedure and by the use of the Respitrace, which improves the control rate of voluntary depth of inspiration and ensures a ventilatory steady state. Indeed, hyperventilation has been shown to elicit a larger oxygen consumption than ventilation stimulated by hypercapnea. Moreover, in this study, at the end of the resting period and before connecting to the spirometer, the subjects breathed 100% O2 for 3 min to replace the nitrogen in the lungs to minimize apparent changes in metabolic oxygen consumption resulting from changes in body oxygen storage.

The extent to which the technique is performed under steady-state conditions remains problematic since the dead space volume is progressively increased. To address this concern, we increased the rate of dead space volume by steps of 90-s duration, which provides a much longer period for measurement than the discrete points of measurement in the studies of Shindoh et al. (17) and Takishima et al. (18), in which the dead space volume was increased linearly. A second problem concerns the effect of FRC changes during dead space rebreathing. Collett et al. (8) reported that for large lung volumes, the oxygen cost of breathing was increased even when work rate and the pressure-time product were matched. They suggested that the decrease in the efficiency of the inspiratory muscles is related to a drop in their strength. In this study we minimized such effects by compelling the subjects to set their FRC at control values. Without such a target when ventilation is markedly increased, the contraction of abdominal muscles at end expiration usually occurs, thereby decreasing FRC. This mechanism probably operates to distribute the respiratory work load between inspiratory and expiratory muscles and is of paramount importance in the energy cost of breathing. Thus, compelling the subjects to set their control FRC during rebreathing hyperventilation may be unphysiological. But because the V̇O2 alteration occurring with lung volume changes may effect the recorded measurements, we chose to fix FRC despite its possibly unphysiological character.

The mean peak ventilation (60 L·min−1) was lower than that achieved during voluntary hyperventilation (1,3) but was close to the values reported by Cherniack (6) and Fritts et al. (12). Rebreathing probes are different from voluntary hyperventilation and comparisons should be made with caution. Moreover, the V̇O2 resp.-V̇E relationship is generally curvilinear, and the analysis of the oxygen cost of breathing at high voluntary ventilation is difficult. In the range of pulmonary ventilation associated with rebreathing, the logarithm of V̇O2 was linearly related to V̇E. It is well known that values of respiratory muscle oxygen consumption for handling added work loads will depend on the resistance of the device. In our study, the resistance determined at its minimal and maximal configurations, with low and high constant flow volumes, was lower than the values found in other studies. Coast and Krause (7) showed that an increased work of breathing results in increased cardiac output. This finding may be relevant for the estimation of the amount of oxygen consumed by the respiratory muscles. The results of our study indicate that heart rate values were increased with increases in dead space breathing, reflecting an augmentation in cardiac output. This would cause an increase in total oxygen consumption during the dead space loading and thus artifactually increase the calculated oxygen consumption of the respiratory muscles. The reported V̇O2 values therefore probably include an increase in myocardial oxygen consumption, resulting in an overestimation of the oxygen consumption of the respiratory muscles. However, since the measurements were done in similar conditions and no difference in heart rate was shown between the two tests, this consideration could not influence the result in terms of reliability obtained for the repeated tests.

Reliability of the device and comparison with conventional methods. A large discrepancy in the oxygen consumption of the respiratory muscles was found between studies (5,6,8,17,18). In the present study, the log V̇O2-V̇E for subjects 25-35 yr old was 0.0066 ± 0.005 log mL·min−1/L·min−1. This value is similar to those reported by other studies for this age group (6,18). However, our value is higher than the values of Bartlett et al. (1), Campbell et al. (5), Cournand et al. (9), and Milic-Emili and Petit (14), whereas it is lower than that found by Fritts et al. (12). Experimental design differences contribute to this variability. Indeed, some authors used a rebreathing circuit adding external flow resistive loading and estimated the additional O2 consumption associated with the additional mechanical work (5). Other investigators calculated the V̇O2 resp. from a gas-sampling method (12), while Bradley and Leith (3) measured V̇O2 resp. with the sustained hyperpnea method. Some investigators tested not only young healthy subjects but also elderly subjects (18) and patients with respiratory muscle impairment, including chronic obstructive pulmonary disease (17) or obesity and/or emphysema (5,6,12). A wide intra-experimental session variability also exists. In our study we found large inter-individual variability in V̇O2 resp. (0.0044 to 0.0090 log mL·min−1/L·min−1), which may result from differences in muscle fiber composition, the mechanical properties of the lung and/or chest wall, or the response to increased ventilation. Undoubtedly, V̇O2 resp. values are age-, anthropometric-, and spirometric value dependent. To our knowledge most of the healthy subjects explored in the previous studies were younger than 45 yr, so the effect of aging would be minimal. However, Fritts et al. (12) included somewhat older subjects, aged 47 to 56 yr, who had higher values of the oxygen cost of breathing than the younger subjects. This is consistent with our results and might explain part of the data variability. In the present study, duplicate values of V̇O2 resp. and V̇O2 met. did not differ and the coefficients of variation were less than 35%. Furthermore, significant correlations were found, as well as high coefficients of determination, and the SEE was negligible. Moreover, the two tests were in close agreement. The differences within the range of mean ± 2 SD were small, and more than 95% of the differences were ranged between these limits, showing the small discrepancy between the duplicate measurements and the reliability of the technique.

In conclusion, a new device was designed to assess V̇O2 resp. by increasing external ventilatory dead space volume as recommended by Cherniack (6). In 14 healthy subjects, this technique was reliable and gave results comparable with those found in the literature. Furthermore, the method can be easily used and is suitable for evaluating large populations.

Back to Top | Article Outline


1. Bartlett, R. G., Jr. H. J. Bruback, and H. Specht. Oxygen cost of breathing. J. Appl. Physiol. 12:413-424, 1958.
2. Bland, J. M. and D. G. Altman. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 8:307-310, 1986.
3. Bradley, M. E. and D. E. Leith. Ventilatory muscle training and the oxygen cost of sustained hyperpnea. J. Appl. Physiol. 45:885-892, 1978.
4. Briscoe, W. A. and A. B. Dubois. The relationship between airway resistance, conductance, and lung volume in subjects of different age and body size. J. Clin. Invest. 37:1279-1285, 1958.
5. Campbell, E. J. M., E. K. Westlake, and R. M. Cherniack. Simple methods of estimating oxygen consumption and efficiency of the muscles of breathing. J. Appl. Physiol. 11:303-308, 1957.
6. Cherniack, R. M. The oxygen consumption and efficiency of the respiratory muscles in health and emphysema. J. Clin. Invest. 38:494-499, 1959.
7. Coast, J. R. and K. M. Krause. Relationship of oxygen consumption and cardiac output to work of breathing. Med. Sci. Sports Exerc. 3:335-340, 1993.
8. Collett, P. W. and L. A. Engel. Influence of lung volume on oxygen cost of resistive breathing. J. Appl. Physiol. 61:16-24, 1986.
9. Cournand, A., D. W. Richards, R. A. Bader, M. E. Bader, and A. P. Fishman. The oxygen cost of breathing. Trans. Assoc. Am. Physicians 67:172-173, 1954.
10. Donahoe, M., R. M. Rogers, D. O. Wilson, and B. E. Pennock. Oxygen consumption of the respiratory muscles in normal and malnourished patients with chronic obstructive pulmonary disease. Am. Rev. Resp. Dis. 140:385-391, 1989.
11. Field, S., S. Sanci, and A. Grassino. Respiratory muscle oxygen consumption estimated by the diaphragm pressure-time index. J. Appl. Physiol. 57:44-51, 1984.
12. Fritts, H. W., Jr., J. Filler, A. P. Fishman, and A. Cournand. The efficiency of ventilation during voluntary hyperpnea: studies in normal subjects and in dyspneic patients with either chronic pulmonary emphysema or obesity. J. Clin. Invest. 38:1339-1348, 1959.
13. Konno, K. and J. Mead. Measurement of separate volume changes of rib cage and abdomen during breathing. J. Appl. Physiol. 22:407-422, 1967.
14. Milic-Emili, G. and J. M. Petit. Mechanical efficiency of breathing. J. Appl. Physiol. 15:359-362, 1960.
15. Mittman, C., N. H. Edelman, A. H. Norris, and N. W. Shock. Relationship between chest wall and pulmonary compliance and age. J. Appl. Physiol. 20:1211-1216, 1965.
16. Nickersson, B. G. and T. G. Keens. Measuring ventilation muscle endurance in humans as sustainable inspiratory pressure. J. Appl. Physiol. 52:768-772, 1982.
17. Shindoh, C., W. Hida, Y. Kikuchi, et al. Oxygen consumption of respiratory muscles in patients with COPD. Chest 105:790-797, 1994.
18. Takishima, T., C. Shindoh, Y. Kikuchi, W. Hida, and H. Inoue. Aging affect on oxygen consumption of respiratory muscles in humans. J. Appl. Physiol. 69:14-20, 1990.


© 1999 Lippincott Williams & Wilkins, Inc.