Twelve male subjects, who were local competitive cyclists or triathletes, gave written informed consent to participate in this study. Their physical characteristics are contained in Table 1. The Human Ethics Committee of the Australian Institute of Sport approved all the test procedures.
The subjects attended the laboratory on two separate occasions. The first visit consisted of three 25-min bouts (5 min at 100, 150, 200, 250, and 300 W) on an air-braked cycle ergometer with an intervening 30-min rest between bouts (Fig. 1). One exercise bout was completed on each of the following systems: MedGraphics metabolic cart (CPX/D, St Paul, MN), the MedGraphics metabolic cart with altered software (CPX/DΔ), and a custom-built V̇O2 system at Flinders University (FU), Adelaide, South Australia (22). Details of the software changes are described in the Methods subsection titled “MedGraphics software and alterations.” The three tests were randomized and counterbalanced to minimize any order effect.
The second visit was used to obtain each subject’s V̇O2max on the cycle ergometer and utilized the FU V̇O2 system. The V̇O2max protocol comprised a 3-min warm-up at 100 W and 1 min at 150 W followed by 25-W increments every minute thereafter until the subject reached volitional exhaustion. All subjects received vigorous verbal encouragement during this test and were considered to have attained V̇O2max when they complied with the three following requirements: an increase in V̇O2 of less than 150 mL between successive workloads (27), a RER ≥ 1.15 (14), and an HRmax of at least 90% of that predicted (26).
The laboratory was well ventilated, and immediately before each test, the respective ranges for temperature, relative humidity, and barometric pressure were 17–21°C, 29–53%, and 746–760 mm Hg. Ambient temperature and relative humidity were measured using a calibrated thermohygrometer (Cole Parmer 37951-00, Vernon Hills, IL) while barometric pressure was measured using a mercury barometer (Casella, London, UK) that was verified by the South Australian Bureau of Meteorology.
A three-lead ECG (Electrondyne, ST-219, Sharon, MA), which was connected to the FU V̇O2 system, monitored HR during all tests.
Measure of oxygen consumption: FU V̇O2 system.
The FU V̇O2 system utilizes the Douglas bag principle (6) and has been described previously (22). Briefly, inspired gas volume and respiratory frequency were measured using a turbine volume transducer (Mark II ventilometer; P. K. Morgan, Chatham, Kent, UK) on the inspiratory side of an R2730 respiratory valve (Hans Rudolph, Kansas City, MO), while the expired gas was collected in one of three vertically mounted 200-L meteorological balloons (Kaysam Corp. of America, Paterson, NJ). The percentage of CO2 (%CO2) and O2 (%O2) in mixed dry expirate was measured continuously with CO2 and O2 analyzers (Applied Electrochemistry CD-3A and S-3A, respectively, AIE Technologies, Pittsburgh, PA). Water vapor was extracted from the expirate by drawing it through a column of CaSO4 (WA Hammond Drierite Company, Xenia, OH) before analysis. Temporal alignment of the inspired minute volume and the associated expired gas percentages was obtained via the customized valve and stepper motor arrangement of the three meteorological balloons that allowed simultaneous collection in one balloon, analysis of expirate from a second balloon while the third was evacuated.
The ventilometer was calibrated in accordance with the manufacturer’s specification using a 1.0-L syringe. Additionally, before and at the completion of all tests, the ventilometer was also calibrated using a sinusoidal artificial lung and Tissot chain-compensated gasometer (11) to establish its accuracy throughout the physiological range of tidal volumes and respiratory frequencies. The maximum error was ± 2.9% for tidal volumes of 1.5–4.0 L at flow rates of 22.5–200.0 L·min−1. This is within the accepted standard (±3%) for volume calibration according to the American Thoracic Society (1).
The O2 and CO2 analyzers were calibrated across the physiological range of measurement both immediately before and after each test with standard gas mixtures (BOC Gases Ltd., Australia) that had been authenticated by Lloyd-Haldane analyses. Furthermore, before and at the completion of all tests, the gas analyzers were assessed for drift over a 60-min interval. No drift was observed and the analyzers were accurate to ± 0.01%. Finally, the gas permeability of the meteorological balloons were checked before and after all tests by filling each balloon with expirate and then sampling the CO2 and O2 concentrations every min for 20 min. The changes in gas concentrations for the three bags were ≤ 0.01% for O2 and CO2, respectively.
FU V̇O2 system software.
The FU V̇O2 system’s program was written in Pascal and facilitates the on-line acquisition, reduction, and display of data (22). After entering the subject details, barometric pressure, relative humidity, and time interval, standard algorithms (29) were used to calculate V̇O2, carbon dioxide production (V̇CO2), lung ventilation under BTPS conditions (V̇E), tidal volume (Vt), respiratory frequency (fr), HR, and RER.
Oxygen consumption: MedGraphics system.
The CPX/D V̇O2 system is a breath-by-breath metabolic cart that uses a pneumotachograph (18) to estimate expired volume from a pressure differential and proprietary gas analyzers to measure gas fractions.
During each test, expired gas samples were drawn into the gas analyzers via a 2.13-m pink, Teflon sampling line (0.45-mm internal diameter) that originated from the pneumotachograph. The pneumotachograph was held in close proximity to the mouth via a rubber mouthpiece. Immediately before every test, the pneumotachograph was calibrated in accordance with the procedures outlined in the operator’s manual. A 4.0-L calibration syringe (SRL Medical Inc, M20, Dayton, OH) was used to deliver five strokes over flow ranges of ∼20–200 L·min−1. Both the pink sampling line and the pneumotachograph were also treated in accordance with the guidelines contained in the operator’s manual. Accordingly, the pink sampling line was replaced immediately before any of our subjects were tested and was not replaced thereafter because inspection immediately before each test verified that it was not crimped, twisted, or flattened and was free of saliva. The pneumotachograph was replaced after 10 tests; hence, three new pneumotachographs were used to test the 12 subjects.
Oxygen and carbon dioxide percentages were analyzed by zirconium and infrared analyzers (MedGraphics CPX/D, St Paul, MN), respectively. Before analysis, water vapor content was standardized by drawing the gas sample through an external drying cartridge and then internal Nafion® tubing (Perma Pure, Toms River, NJ). Before starting every test, the gas analyzers were calibrated using the CPX/D manual calibration routine and two precision grade gas mixtures (CO2/O2 mixture 1, 5/12%; mixture 2, 0/21%) supplied by Medical Graphics Corporation.
MedGraphics software and alterations.
The CPX/D used breath-by-breath data acquisition via Breeze3 software version 3.05. Raw data were collected breath-by-breath and the CPX/D software was used to report minute values for %O2, %CO2, V̇O2, V̇CO2, V̇E, Vt, fr, and RER.
The Breeze3 software was altered by editing the CAL.CFG (calibration configuration) file in the C:\breeze3¢fg directory. The CAL.CFG file was edited at lines 202 and 203 in the penultimate columns, where line 202 is for %CO2 and line 203 is for %O2. Proctor and Beck (19) note that the numbers in the final columns of lines 202 and 203 are the lag times. We derived empirically that the penultimate column of these lines apparently represents a correction factor for the lag time by examining multiple CPX/D CAL.CFG files. A survey regarding the use of the Medical Graphics CPX/D metabolic cart (via the Internet SportSci Listserve http://www.sportsci.org/forum/index.html) resulted in 22 respondents and 14 of these had −60.000 on both lines, 5 had 0.000 on both lines, and 3 respondents had combinations of 0.000 and −60.000, with the lag times ranging from 240 to 580 ms (Table 2). We therefore tested the CPX/D with lines 202 and 203 set to −60.000, and the CPX/DΔ with both lines set to 0.000.
The custom-built, air-braked cycle ergometer (B. L. Hayes, Adelaide Superdrome, Australia) was calibrated at 25-W increments using a dynamic calibration rig (30) before and after all testing. The mean difference in power between pre- and postcalibration results was 0.6% (range = −1.0 to 1.6%) for the span of 100–300 W.
Each dependent variable (V̇O2, V̇CO2, V̇E, Vt, fr, HR, RER, %O2, and %CO2) was analyzed by a three (system: CPX/D, CPX/DΔ, FU) by five (workload: 100, 150, 200, 250, 300 W) factorial design ANOVA (P ≤ 0.05) with repeated measures across both dimensions. These analyses were conducted using Statistica (Statsoft Inc., OK, version 6.0). Tukey post hoc tests were used to identify differences between cell means in the event of significant interactions or main effects. Values are reported as mean ± standard deviation, or mean and 95% confidence limits (95% CL).
Oxygen consumption, carbon dioxide production, and RER.
The CPX/D V̇O2 and V̇CO2 were significantly lower than the criterion FU system at 100–300 W (Fig. 2). The mean absolute and relative differences (and 95% CLs) for the V̇O2 and V̇CO2 of the CPX/D versus the criterion system at 100, 150, 200, 250, and 300 W are presented in Table 3.
Altering the software from −60.000 to 0.000 ms for CPX/DΔ elevated the V̇O2 and V̇CO2, but each still remained significantly lower than the FU system at 100–300 W for V̇O2 and at 200–300 W for V̇CO2 (Fig. 2). At 300 W, the differences and CLs were −0.22 L·min−1 (−0.31 to −0.12) for V̇O2 and −0.09 L·min−1 (−0.23 to 0.05) for V̇CO2.
RER was elevated (main effect for system, F(2,22) = 5.78, P < 0.01) for both the CPX/D (0.96 ± 0.07) and CPX/DΔ (0.96 ± 0.06) compared with the FU system (0.93 ± 0.06). Figure 2 illustrates the means for each submaximal workload.
Ventilation and gas fractions.
V̇E for both the CPX/D (67.2 ± 26.4 L·min−1) and CPX/DΔ (67.5 ± 26.9 L·min−1) was lower than for the FU (70.5 ± 27.1 L·min−1) system (F(2,22) = 14.5, P < 0.0001). The absolute and relative differences (and CLs) of the CPX/D versus the FU system are shown in Table 3, and at 300 W the difference was −3.4% (−7.7 to 0.8) for the CPX/DΔ. The Vt and fr for the CPX/D and CPX/DΔ were lower and higher, respectively, than for the FU system. Accordingly, the Vt of the CPX/D, CPX/DΔ and FU systems were 2.20 ± 0.53, 2.20 ± 0.53, and 2.45 ± 0.51 L·breath−1 (F(2,22) = 36.0, P < 0.0001), respectively, and the fr of the CPX/D, CPX/DΔ, and FU systems were 30 ± 7, 30 ± 7, and 28 ± 7 breaths·min−1 (F(2,22) = 9.8, P < 0.005), respectively.
The %O2 (16.24 ± 0.40%) and %CO2 (4.40 ± 0.29%) for the CPX/D were significantly higher and lower, respectively, than those for the FU system (16.15 ± 0.39 and 4.52 ± 0.30%). Changing the software from −60.000 to 0.000 ms reversed this result for the CPX/DΔ, but only the %O2 (16.04 ± 0.39%) was significantly different. Although higher than for the FU system, the %CO2 (4.62 ± 0.26%) of the CPX/DΔ failed to achieve conventional significance (P = 0.08). The overall main effects between the three systems were significant for both %O2 (F(2,22) = 19.8, P < 0.0001) and %CO2 (F(2,22) = 14.3, P < 0.0001), and mean values at each submaximal workload are shown in Figure 2.
Power, cadence, and heart rate agreement.
Power (F(2,22) = 1.17, P = 0.33), cadence (F(2,22) = 0.18, P = 0.84), and HR (F(2,22) = 1.18, P = 0.32) were not significantly different for the tests conducted on the three different systems throughout the five work stages. The largest mean differences at 100, 150, 200, 250, and 300 W were 0.6, 0.8, 1.5, 1.1, and 0.6 W, respectively.
The main finding of this study is that compared with a calibrated automated Douglas bag system, the Medical Graphics CPX/D metabolic cart yields V̇O2 and V̇CO2 values that are respectively ∼11% and ∼8% lower during submaximal exercise. Additionally, the trend in our submaximal data suggests that V̇O2max would be underestimated by a similar amount. Our results challenge previously published “validation” studies that have used the CPX/D as the benchmark to evaluate other computerized V̇O2 systems (12,17,24,28). It is also of concern that the CPX/D software appears to contain a lag time correction factor that varies between 0 and 60 ms on different systems throughout the world (Table 2).
Critique of criterion system.
Before reviewing our results with the CPX/D metabolic cart, it is salient to establish the accuracy of the criterion system, and this depends largely on the accuracy of the volume and gas fraction measurements. The calibration of the FU automated Douglas bag system is described in the Methods but some key issues are worth repeating. The FU system volume transducer and gas analyzers were thoroughly calibrated before every test. Additionally, before and after completing the test block on all 12 subjects, the turbine volume device was verified throughout the physiological range of exercise tidal volumes and respiratory frequencies against the first principles standard of a Tissot gasometer. Similarly, the gas analyzers were checked for stability over 60 min and the meteorological balloons were verified as not gas permeable over 20 min before and at the conclusion of the test block. These time intervals far exceed the 25 min of each five-stage submaximal test and the 1–2 min during which expirate was inside any meteorological balloon. Drying of expirate collected in the balloons was achieved via a column of the commercial dessicant, Drierite (CaSO4), and the moisture remaining in gases dried with this compound is 0.005 mg·L−1 at 25 to 30°C (10).
CPX/D versus criterion differences.
Without access to the Medical Graphics research and development files, it is impossible to identify the exact source of the discrepancy between the CPX/D and the FU automated Douglas bag system. Relative errors in V̇E usually translate directly to similar magnitude errors in V̇O2; for example, each 1% too low V̇E equates to a 1% too low V̇O2 (29). But the V̇E means for the CPX/D and CPX/DΔ were within 3.4–4.0% of those from the FU system. This is therefore only a partial explanation, and furthermore these differences are barely outside the ± 3% accuracy standard recommended by the American Thoracic Society (1). However, it was interesting to note that the pooled data of all five submaximal workloads for both the CPX/D and CPX/DΔ systems produced an fr that was 2 breaths·min−1 more but a Vt that was 0.25 L·breath−1 less than those from the FU system (28 breaths·min−1 and 2.45 L·breath−1).
The gas fraction differences also do not account for the dissimilarities between the FU and CPX/D metabolic cart. Using standard algorithms (29), one can recalculate the V̇O2 when V̇E is equated for all systems but the CPX/D and CPX/DΔ measured gas fractions are used. In this case, the relative V̇O2errors are −2% and +2% for the CPX/D and CPX/DΔ, respectively, for the pooled data of the five submaximal workloads presented under the “Ventilation and gas fractions” subsection of the Results. When both the mean V̇E and gas fractions measured by the CPX/D and CPX/DΔ are used, the recalculated mean V̇O2 differences are −6% and −2% compared with those measured by the FU system. These differences are approximately half of the actual discrepancies between the respective systems.
A further potential source of divergent results is the method used to dry the gas. Failure to dry gas will dilute the O2 and CO2 fractions and increase V̇O2 (29). But even an additional 30% of water vapor will only lower a true gas fraction of 16.24% by ∼0.10% with a resultant increase in V̇O2 of ∼3%. However, the gas fractions measured by the CPX/D were higher, not lower, than those measured by the FU system. An extra confounding factor is that if lag times are inappropriate then this will alter gas fractions independent of any changes due to gas drying problems.
Regardless of the source of error, others have been able to demonstrate agreement within a few percent for V̇O2 from breath-by-breath systems compared with both the Douglas bag method (4) and a mixing box system (13). The 11% lower submaximal V̇O2 values of the CPX/D compared with an automated Douglas bag system is disturbing. Furthermore, in the context of comparing results of athletes between different laboratories (8), the difference is unacceptably large because V̇O2max changes of just a few percent in response to a training stimulus such as altitude (25) are deemed important for international-level athletes.
CPX/D lag time differences.
A disconcerting finding was the variation in the apparent lag time correction for different CPX/D systems around the world. The notion that this may be a correction factor is supported by our observation that when set to 0.000 ms the %O2 and %CO2 were lower and higher, respectively, which is consistent with later sampling of the expirate; furthermore, this change halved the error in V̇O2 compared with the FU system. Thus, the average O2 delay of 415 ms (Table 2) would be used if this factor was set to zero, but the delay would be 355 ms (415–60) if the factor was set to −60.000 ms. All except one respondent did not know about this file, which suggests that the settings in these files (Table 2) were factory defaults. In fact, virtually all respondents to the Listserve survey had to be instructed how to locate the CAL.CFG file which is accessed via DOS commands via these paths; C:\breeze3¢fg (for CPX/D) or C:\breeze¢fg (for CardiO2). Particularly perplexing is the observation that one CPX/D in Australia and one in Latvia had an apparent correction factor of −60.000 ms for the O2 channel and 0.000 ms for the CO2 channel.
Overall, changing the apparent lag time correction factor from −60.000 to 0.000 ms halved the difference between the V̇O2 measured by CPX/D metabolic cart and the FU criterion system. A change of this magnitude is considerable when the data of Beaver and coworkers (4; their Fig. 4) reveals a 15% difference in V̇O2 for a −60 versus 0 ms comparison. However, direct comparison with the Medical Graphics metabolic cart is somewhat tenuous when the lag time for the Beaver et al. (4) system was 285 ms compared with ∼415 for the CPX/D. Likewise, the conclusion that errors of up 30% for V̇O2 can occur at high respiratory frequencies (70 breaths·min−1) should be tempered by the shorter lag time (mean of 180 ms) for a system using a mass spectrometer (19). Nevertheless, regardless of the magnitude, variations in the factory default settings are cause for concern.
In conclusion, compared with a criterion automated Douglas bag system, the CPX/D metabolic cart yields V̇O2 values that are ∼11% lower during submaximal cycle ergometry. The source of the discrepancy is unknown but it appears not to be related to volume measurement. A disturbing observation is that the factory defaults for the lag time uses different correction factors that vary by 60 ms, which according to others (4) may substantially alter V̇O2 particularly at higher respiratory frequencies (19). These findings therefore challenge the accuracy of data collected on any CPX/D system.
We wish to thank the Burnside War Memorial Hospital for use of their oxygen consumption system, Mr. Tom Stanef from the South Australian Sports Institute for cycle ergometer calibration, and Mr. Anthony Brooks for help with data collection.
This study was funded in part by the Laboratory Standards Assistance Scheme of the Australian Sports Commission.
1. American Thoracic Society. Standardization of spirometry: 1987 update. Statement of the American Thoracic Society. Am. Rev. Respir. Dis. 136: 1285–1298, 1987.
2. Bard, R. L., and J. M. Nicklas. New graphical method for evaluating gas exchange in congestive heart failure. Med. Sci. Sports Exerc. 32: 870–876, 2000.
3. Bassett, D. R., Jr., E. T. Howley, D. L. Thompson, et al. Validity of inspiratory and expiratory methods of measuring gas exchange with a computerized system. J. Appl. Physiol. 91: 218–224, 2001.
4. Beaver, W. L., N. Lamarra, and K. Wasserman. Breath-by-breath measurement of true alveolar gas exchange. J. Appl. Physiol. 51: 1662–1675, 1981.
5. Beaver, W. L., K. Wasserman, and B. J. Whipp. On-line computer analysis and breath-by-breath graphical display of exercise function tests. J. Appl. Physiol. 34: 128–132, 1973.
6. Douglas, C. G. A method for determining the total respiratory exchange in man. J. Physiol. 42: 17–18, 1911.
7. Gonzalez-Alonso, J., R. S. Richardson, and B. Saltin. Exercising skeletal muscle blood flow in humans responds to reduction in arterial oxyhaemoglobin, but not to altered free oxygen. J. Physiol. 530: 331–341, 2001.
8. Gore, C. Quality assurance in exercise physiology laboratories. In: Physiological Tests for Elite Athletes, C. J. Gore (Ed.). Champaign, IL: Human Kinetics, 2000, pp. 3–11.
9. Gore, C. J., P. G. Catcheside, S. N. French, J. M. Bennett, and J. Laforgia. Automated V̇O2max
calibrator for open-circuit indirect calorimetry systems. Med. Sci. Sports Exerc. 29: 1095–1103, 1997.
10. Hammond, W. A., and J. R. Withrow. Soluble anhydrite as a desiccating agent. Ind. Eng. Chem. 25: 653–659, 1933.
11. Hart, J. D., and R. T. Withers. The calibration of gas volume measuring devices at continuous and pulsatile flows. Aust. J. Sci. Med. Sport 28: 61–65, 1996.
12. Hausswirth, C., A. X. Bigard, and J. M. Le Chevalier. The Cosmed K4 telemetry system as an accurate device for oxygen uptake measurements during exercise. Int. J. Sports Med. 18: 449–453, 1997.
13. Hughson, R. L., D. R. Northey, H. C. Xing, B. H. Dietrich, and J. E. Cochrane. Alignment of ventilation and gas fraction for breath-by-breath respiratory gas exchange calculations in exercise. Comput. Biomed. Res. 24: 118–128, 1991.
14. Issekutz, B., Jr., N. C. Birkhead, and K. Rodahl. Use of respiratory quotients in assessment of aerobic work. J. Appl. Physiol. 17: 47–50, 1962.
15. Koppo, K., and J. Bouckaert. In humans the oxygen uptake slow component is reduced by prior exercise of high as well as low intensity. Eur. J. Appl. Physiol. 83: 559–565, 2000.
16. McLaughlin, J. E., G. A. King, E. T. Howley, D. R. Bassett, Jr., and B. E. Ainsworth. Validation of the COSMED K4 b2 portable metabolic system. Int. J. Sports Med. 22: 280–284, 2001.
17. Miles, D. S., M. H. Cox, and T. J. Verde. Four commonly utilized metabolic systems fail to produce similar results during submaximal and maximal exercise. Sports Med. Training Rehabil. 5: 189–198, 1994.
18. Porszasz, J., T. J. Barstow, and K. Wasserman. Evaluation of a symmetrically disposed Pitot tube flowmeter for measuring gas flow during exercise. J. Appl. Physiol. 77: 2659–2665, 1994.
19. Proctor, D. N., and K. C. Beck. Delay time adjustments to minimize errors in breath-by-breath of VO2
during exercise. J. Appl. Physiol. 81: 2495–2499, 1996.
20. Rietjens, G. J. W. M., H. Kuipers, A. D. Kester, and H. A. Keizer. Validation of a computerized metabolic measurement system (Oxycon-Pro®
) during low and high intensity exercise. Int. J. Sports Med. 22: 291–294, 2001.
21. Robach, P., M. Dechaux, S. Jarrot, et al. Operation Everest III: role of plasma volume expansion on VO2max
during prolonged high-altitude exposure. J. Appl. Physiol. 89: 29–37, 2000.
22. Sainsbury, D. A., C. J. Gore, R. T. Withers, and A. H. Ilsley. An on-line microcomputer program for the monitoring of physiological variables during rest and exercise. Comput. Biol. Med. 18: 17–24, 1988.
23. Saltin, B., H. Larsen, N. Terrados, et al. Aerobic exercise capacity at sea level and at altitude in Kenyan boys, junior and senior runners compared with Scandinavian runners. Scand. J. Med. Sci. Sports 5: 209–221, 1995.
24. Schulz, H., S. Helle, and H. Heck. The validity of the telemetric system CORTEX X1 in the ventilatory and gas exchange measurement during exercise. Int. J. Sports Med. 18: 454–457, 1997.
25. Stray-Gundersen, J., R. F. Chapman, and B. D. Levine. The “living high–training low” altitude training paradigm improves sea level performance in elite male and female runners. J. Appl. Physiol. 91: 1113–1120, 2001.
26. Tanaka, H., K. D. Monahan, and D. R. Seals. Age-predicted maximal heart rate revisited. J. Am. Coll. Cardiol. 37: 153–156, 2001.
27. Taylor, H. L., E. Buskirk, and A. Henschel. Maximal oxygen intake as an objective measure of cardio-respiratory performance. J. Appl. Physiol. 8: 73–80, 1955.
28. Wideman, L., N. M. Stoudemire, K. A. Pass, C. L. Mcginnes, G. A. Gaesser, and A. Weltman. Assessment of the Aerosport TEEM 100 portable metabolic measurement system. Med. Sci. Sports Exerc. 28: 509–515, 1996.
29. Withers, R. T., C. J. Gore, G. Gass, and A. Hahn. Determination of maximal oxygen consumption (VO2max
) or maximal aerobic power. In: Physiological Tests for Elite Athletes, C. J. Gore (Ed.). Champaign, IL: Human Kinetics, 2000, pp. 114–127.
30. Woods, G. F., L. Day, R. T. Withers, A. H. Ilsley, and B. F. Maxwell. The dynamic calibration of cycle ergometers. Int. J. Sports Med. 15: 168–171, 1994.