Anesth Analg 2003;97:1414-1420
© 2003 International Anesthesia Research Society
TECHNOLOGY, COMPUTING, AND SIMULATION
Novel, Adjustable, Clinical Bymixer Measures Mixed Expired Gas Concentrations in Anesthesia Circle Circuit
Abraham Rosenbaum, MD*,
, and
Peter H. Breen, MD FRCPC*
*Department of Anesthesiology, University of California, Irvine, California; and
Department of Anesthesiology, The Technion-Israel Institute of Technology, Haifa, Israel
Address correspondence and reprint requests to Peter H. Breen, MD, FRCPC, Department of Anesthesiology, UCI Medical Center, Bldg. 53, Rm. 227, 101 The City Drive S., Orange, CA 92868. Address e-mail to pbreen{at}uci.edu
 |
Abstract
|
|---|
We have introduced a novel, parallel design into a new clinical bymixer (patent pending), named for the bypass of a constant fraction of total flow through a mixing chamber. Over a wide range of tidal volumes (300–1200 mL), frequency (6–20 breaths/min), and PCO2 (6–50 mm Hg), the bymixer provided accurate measurement of mixed expired gas fractions in the ventilation circuit compared with an expired gas collection in a metabolic lung bench setup (average slope, 1.00; average y intercept, -0.01; average coefficient of determination, R2 = 0.9988). Simple changes in mixing chamber volume provided adjustable bymixer response times. The fast bymixer response (time constant, 6.4 s) should allow measurements to be updated every 20 s (where 95% response occurs by three time constants). The new clinical bymixer is constructed from standard anesthesia circuit components, attaches easily to the anesthesia machine inspired outlet and expired inlet ports, is simple to clean and sterilize, and has no reservoir to trap condensed water vapor from expired gas. The new clinical bymixer may facilitate indirect calorimetry (CO2 elimination, 
CO2, and oxygen uptake, CO2) during anesthesia and the noninvasive detection of metabolic upset (e.g., onset of anaerobic metabolism) and critical events (e.g., pulmonary embolism).
IMPLICATIONS: A new clinical bymixer (inline mixing chamber) provides a fast response and accurate measurements of mixed expired gas fractions in the anesthesia circle circuit. A novel parallel design facilitates adjustable response, easy cleaning, and construction from standard airway circuit components. The new clinical bymixer may facilitate widespread introduction of indirect calorimetry during anesthesia.
|
Introduction
|
|---|
Measurement of mixed expired gas concentrations is an essential component of the methodology to measure CO2 elimination (CO2) and pulmonary oxygen uptake (O2) at the airway opening (1). In the normal condition in which CO2 is absent from inspired gas, CO2 is given by
where E is the expired ventilation and F
CO2 is the mixed expired CO2 fraction.
Conversely, O2 is the difference between inspired and expired oxygen volumes, as given by
where I denotes inspiration. Because expired volume is increased by increased temperature and added water vapor (increased humidity), volumes must be corrected to standard temperature and pressure, dry (STPD) conditions, or the error in O2 can approach 50% as FiO2 increases to unity (1,2). Because accurate differences between I and E are difficult to measure, the Haldane transformation (1,2) is usually used, invoking conservation of the inert gas nitrogen (I · FiN2 = E · FN2). By substitution into Equation 2, O2 can be expressed as a function of only E, where
|
Regardless of whether temperature and humidity differences between inspiration and expiration are managed by the Haldane transformation (Equation 3) or by separate measurements of airway temperature and relative humidity (RH) (Equation 2) (3), the determination of CO2 and O2 requires measurements of mixed expired and inspired gas fractions (4). The classic method to obtain mixed expired gas fractions is to collect exhaled gas over a number of breaths in a collection chamber connected to the expiratory outlet of the ventilator (5,6). However, expired gas collection cannot be conducted in the anesthesia semiopen or closed anesthesia circle ventilating circuit because expired gas passes through a CO2 absorber to become the next inspiration (1,2,7). Instead, to measure mixed expired gas fractions in the circle circuit, we (8) and others (4) have used an inline bypass mixing chamber (bymixer, Fig. 1). The bymixer is named for the bypass of a constant fraction of total flow through a mixing chamber. However, the response time of that bymixer is long and fixed, the mixing chamber is difficult to fabricate, clean, and sterilize, and the device is bulky.
View larger version (33K):
[in this window]
[in a new window]
|
Figure 1. Classical bymixer design bypasses a constant fraction of main flow into the mixing chamber. Gas is sampled, through a secondary chamber, to measure flow-averaged gas concentration. Reprinted with permission from the Annals of Biomedical Engineering 1997;25:164–71 (8).
|
|
To solve these problems, we have developed a new clinical bymixer from common anesthesia circuit components (Fig. 2). Instead of diverting gas flow into a separate, large mixing chamber, the new clinical bymixer incorporates a novel parallel tubing design (patent pending). A constant fraction of total is diverted through the mixing chamber (tubing), whose volume can be adjusted by varying the length of corrugated tubing (collapsible/expandable pediatric anesthesia circuit tubing). The resistor controls the fraction of bypass to total flow. As gas passes through the mixing chamber, it mixes longitudinally in the tubing. Flow-averaged mixed gas is sampled at the port for analysis by a sidestream sampling monitor.
This study describes the development and construction of the new clinical bymixer. The following questions were tested and answered: Is a constant fraction of main gas flow diverted through the mixing chamber (mandatory for mixed bypass gas samples to accurately represent total gas flow)? Does the longitudinal design of the tubular mixing chamber provide adequate mixing (no significant breath-to-breath variation of mixed gas fractions)? What is the fastest accurate response of the new clinical bymixer when the mixing chamber volume is decreased (shortest length of mixing tubing)? Does the continuous sidestream sampling flow rate affect the measurement of the mixed gas fraction?
To test the performance of the new clinical bymixer during cyclical changes in gas fractions under actual expiratory flow conditions, the bymixer was interposed in the exhalation limb of a ventilator attached to a CO2-producing mechanical lung (Fig. 3). Measurement of bymixer mixed expired PCO2 was compared with the value in a gas collection from the exhaust port of the open circuit ventilator. The use of the metabolic lung simulator was mandatory for the execution of this study to provide a wide and controlled range of tidal volume (VT), respiratory frequency (f), and mixed expired PCO2.
View larger version (23K):
[in this window]
[in a new window]
|
Figure 3. Metabolic lung bench setup. A roller pump (5 L/min) circulated gas through a circular circuit between the mechanical lung simulator and the metabolic chamber. CO2 was infused into the metabolic chamber (200 mL/min). A fan and baffles inside the metabolic chamber ensured mixing of CO2. The open system ventilator provided inspired gas to the lung simulator. Expired gas passed through the bymixer and was also collected in a bag connected to the exhaled gas exhaust port of the ventilator. The ventilator is depicted during the inspiratory phase of respiration.
|
|
|
Methods
|
|---|
Design and Construction of the New Clinical Bymixer (Patent Pending)
The bymixer divided the incoming total gas flow into two parallel channels: main flow and bypass flow (Fig. 2). The main flow channel was a 24-cm length of standard 3/4-in. polyvinyl chloride (PVC) pipe (22-mm inner diameter [ID]). The bypass flow/mixing chamber channel was a length of expandable/collapsible pediatric anesthesia circuit tubing (15-mm ID, Expandoflex; Cleveland Tubing Inc., Cleveland, TN). The adjustable tubing was connected, in series, to a sampling port adapter (Datex-Engstrom Division, Instrumentarium Corp., Helsinki, Finland), a flow resistor, and a 12-cm length of standard 3/4-in. PVC tubing. The flow resistor was constructed by drilling a 4-mm-diameter hole in a plastic cap (NAS-820-10; Niagra Plastics, Erie, PA) placed inside a connector (Multi Adapter; Hudson RCI, Temecula, CA; 15-mm ID; 22-mm outer diameter). In this study, the adjustable tubing lengths were 50, 65.5, and 121 cm, which generated mixing chamber volumes (measured up to the sampling port) of 100, 150, and 200 mL, respectively. The volume of the bypass channel from the sampling port to the downstream Y-connector was 53 mL. The main flow and bypass flow channels were con-nected at each end by identical Y-connectors (supplied with standard anesthesia circle circuits). Volumes of channel components were determined by water displacement.
Determination of Time Constant (Bymixer Response)
The anesthesia monitor (Capnomac Ultima; Datex Medical Instruments, Instrumentarium Corp.) sampling line was connected to the bymixer sampling port (200 mL/min), and the pneumotachometer adapter was attached to the inlet of the bymixer. FO2 (paramagnetic) and bymixer total flow were continuously captured (100 Hz) by an analog-to-digital acquisition PC card (DAQcard 700; National Instruments, Austin, TX) installed in a notebook computer (Inspiron 3800; Dell Computer Corp., Austin, TX). The digital data-acquisition system was driven by a custom program (Delphi Pascal; Borland International, Scotts Valley, CA) written by our computer support specialist (David Chien) and one author (PHB). The bymixer was flushed with air to provide a baseline FO2 of 21%. At Time 0, oxygen flow of 4, 8, or 12 L/min was abruptly connected to the bymixer input. The time constant (
) was the interval from Time 0 until FO2 increased to 63% of its maximal value (8). The was corrected for the FO2 transport delay (2.95 s) down the sidestream sampling system (9).
The bymixer response was also tested in the exhalation limb during mechanical ventilation of the CO2-producing metabolic lung simulator (Fig. 3; see below). Steady-state ventilation was established at minute ventilation of 4, 8, or 12 L/min (the respiratory f was 10 breaths/min, and the inspiration/expiration time ratio was 1:2). The bymixer (100-mL mixing chamber) was separately flushed with air. During the inspiratory phase, the bymixer was abruptly interposed in the exhalation limb of the ventilation circuit (Time 0). Because bymixer data during ventilation were periodic and available only during expiration, we measured the time for bymixer PCO2 (infrared analysis) to reach 95% of its maximum value. This time for 95% response was corrected for the PCO2 transport delay (1.76 s) down the sidestream sampling system (9).
Validation of the Accuracy of the New Clinical Bymixer
To test the bymixer, we used a modification of the metabolic lung simulator bench setup (8,10) (Fig. 3). The commercial lung simulator (Dual Adult TTL, Model 1600; MI Instruments, Inc., Grand Rapids, MI) generated a physiologic ventilation wave form by combining airway resistance elements to a bellows (residual volume, 920 mL), whose compliance can be adjusted by springs (10). The mechanical lung was connected by a circular circuit to a metabolic chamber (airtight 18.6 L pail). CO2 was continuously infused (200 mL/min) by calibrated rotameter into the metabolic chamber (11). A fan and a baffle system inside the chamber ensured a homogeneous gas mixture. An occlusion roller pump (15-mm ID tubing; Precision Blood Pump; COBE Perfusion System, Lakewood, CO) generated constant gas flow (5 L/min) between the metabolic chamber and the mechanical lung. The mechanical lung was ventilated with 30% oxygen (Servo Ventilator 900C; Siemens, Stockholm, Sweden). The bymixer was interpolated in the expiratory limb of the open circuit (no rebreathing).
For each length of mixing chamber expandable tubing, the bymixer was tested during different ventilatory patterns, encompassing combinations of VT (300–1200 mL) and respiratory f (6–20 breaths/min). The inspiratory/expiratory ratio was 1:2. Gas was continuously sampled from the bymixer by the sidestream capnometer (bymixer PCO2). Before measurements began at each ventilator setting, steady-state was confirmed by stable values of PetCO2 and bymixer PCO2. A measurement sequence consisted of continuous digital acquisition of bymixer PCO2 and simultaneous collection of expired gas in a 15-L gas-impermeable collection bag (Hans Rudolph, Kansas City, MO) connected to the ventilator exhaust port. Measurements were conducted for 3 min (higher minute ventilation) to 5 min (lower minute ventilation). After the measurement sequence, the expired gas collection was mixed by shaking and agitating small balls inside the bag. Gas collection of PCO2 was measured by attaching the sidestream sampling line to a stopcock on the collection bag. Before each measurement sequence, the collection bag was emptied by vacuum to prevent gas dilution error. After attaining steady-state and just before gas collection began, the deadspace of the bag was flushed with exhaled gas from the ventilator exhaust port. Time-averaged bymixer PCO2 was compared with the value measured in the simultaneous expired gas collection.
Effect of Tidal Volume and Respiratory Frequency on Oscillations of Bymixer FCO2
Using the above validation experimental setup and measurement sequences in the bymixer (150-mL mixing chamber volume), we conducted two additional protocols that measured oscillations of bymixer FCO2. First, respiratory f was held constant (10 breaths/min), and VT was varied from 300 to 1200 mL. Second, VT was held constant (900 mL), and respiratory f was varied from 6 to 20 breaths/min.
Effect of Intermittent (Instead of Continuous) Sampling from the Bymixer Port
We conducted an additional validation protocol of the bymixer (150-mL mixing chamber volume) to test the effect of intermittent (instead of continuous) sampling from the bymixer port. Gas was intermittently sampled from the bymixer sampling port, by manipulation of a three-way stopcock, for short periods (approximately 3 s). Several intermittent samples from the bymixer were averaged for comparison with the expired gas collection (3–5 min) at each ventilator setting of VT and f.
Data Analysis
Bymixer bypass flow (BYPASS) was calculated by
where VBYPASS was the volume of the mixing chamber (measured up to the sampling port) and was the measured time constant of the bymixer (4,8). Then,
where TOTAL was the total gas flow entering the bymixer.
In the validation of average bymixer PCO2 versus the value measured in the expired gas collection (metabolic lung simulator),
where dt is the digital sampling interval (1/100 Hz) and t0 and tend were the beginning and end sampling times (s) of bymixer PCO2.
Bymixer PCO2 values were compared with expired gas collection PCO2 values by least-squares linear regression (slope, y intercept, and coefficient of determination [R2]) and by the limits of agreement (LOA) technique described by Bland and Altman (12,13). Differences between groups were analyzed by Student’s t-test or by analysis of variance. Computer programs were used for data analysis (Excel; Microsoft Corp., Redmond, WA), statistical testing (SigmaStat; SPSS, Chicago, IL), and graphical presentation (SigmaPlot 8.0; SPSS).
|
Results
|
|---|
Figure 4 displays the excellent linear regression correlation between bymixer PCO2 and the value measured in the expired gas collection over a wide range of PCO2 (6–50 mm Hg) for the bymixer with mixing chamber volumes set to 100, 150, and 200 mL. There was no significant difference in bymixer PCO2 accuracy among the mixing chamber volumes (analysis of variance of the PCO2 differences between the bymixer and expired gas collection measurements). For the bymixer set to a mixing chamber volume of 150 mL, there was no significant difference in PCO2 accuracy between continuous and intermittent sampling from the bymixer port.
The Bland-Altman analysis (12) derived the LOA as 0.07 ± 0.93 mm Hg (Fig. 5, top). Measurements for mixing chamber volumes of 100, 150, and 200 mL were combined. Inspection of the graph revealed that the PCO2 difference between the bymixer measurement and the simultaneous value measured in the expired gas collection increased as the measurement increased along the x axis. To correct for this effect, the lower panel of Figure 5 plotted the PCO2 ratio (bymixer/bag) versus the average of the two values (13). The calculation of LOA (1.00 ± 0.03) demonstrated that 95% of the bymixer PCO2 measurements were within 3% of the expired gas collection value.
View larger version (23K):
[in this window]
[in a new window]
|
Figure 5. Bland-Altman analysis. Top, The difference between the bymixer PCO2 and the value measured in the mixed collection of expired gas (exhaust gas collection bag) was plotted against the average of the two values. Bottom, The ratio of the bymixer PCO2 to gas collection PCO2 was plotted against the average of the two values. Dotted lines denote the mean ± 1.96 SD, which encompass 95% of the measurement sequences. Each plotted point represents a steady-state ventilation sequence of a CO2-producing lung simulator over a range of tidal volume (300–1200 mL) and respiratory frequency (6–20 breaths/min). Measurements for mixing chamber volumes of 100, 150, and 200 mL are combined.
|
|
Figure 6 displays the breath-by-breath oscillations in PCO2 measured during continuous aspiration from the bymixer into the sidestream sampling gas analyzer. Oscillations in PCO2 were larger with the smaller bymixer mixing chamber volume. Table 1 shows that the average PCO2 oscillations increased from 0.1 to 0.7 mm Hg as the bymixer mixing chamber volume decreased from 200 to 100 mL. The plot of PCO2 oscillation (mm Hg) versus VT (mL) (constant f) generated a significant direct relationship (slope = 0.0016; y intercept = -0.69; R2 = 0.92). The plot of PCO2 oscillation (mm Hg) versus f (breaths/min) (constant VT) resulted in a significant inverse relationship (slope = -0.062; y intercept = 1.22; R2 = 0.91). Thus, PCO2 oscillations increased in magnitude as VT increased and f decreased.
View larger version (31K):
[in this window]
[in a new window]
|
Figure 6. Effect of mixing chamber volume on bymixer PCO2 during continuous sampling from the bymixer port (200 mL/min) (Fig. 2) by the sidestream gas analyzer. Data were digitally acquired at 100 Hz. The airway opening flow was processed by moving the average filter over seven data points to remove signal noise (14). For clarity, every 20th data point was plotted for bymixer PCO2. Relative to the flow signal, PCO2 was advanced in time by transport delay (the time to aspirate gas through the sampling line). Transport delay was measured previously in a bench setup (9). Respiratory frequency was 12 breaths/min, and tidal volume was 600 mL. With the mixing chamber volume of 100 mL (top), oscillations in bymixer PCO2 were approximately 1.3 mm Hg. The larger mixing chamber of 200 mL (bottom) generated only tiny oscillations in bymixer PCO2 of approximately 0.2 mm Hg.
|
|
The ratio of bypass flow to total flow was similar (1:9) for the three mixing chamber volumes (Table 1). The of the bymixer response to a change in input gas concentration (at 8 L/min) ranged from 6.4 to 14.1 s for the smallest (100 mL) to largest (200 mL) mixing chamber volumes. Tripling of the predicted 95% response. During minute ventilation of the metabolic lung simulator at 4, 8, and 12 L/min, the times for 95% response of the bymixer (100-mL volume) were 19.0, 12.6, and 6.6 s, respectively—significantly less than the values of 3 (Table 1).
|
Discussion
|
|---|
The new clinical bymixer incorporates a radically new design, compared with the classic bymixer (4,8). Instead of diverting a portion of main flow into a surrounding reservoir, as in the classic bymixer (Fig. 1), the new clinical bymixer diverts a fraction of main flow through a parallel, longitudinal, and adjustable mixing chamber (Fig. 2). The flow resistor (variable orifice) provided an easy control of fraction of bypass flow. To provide an accurate mixed average gas fraction of total flow, the ratio of bypass flow to total flow must remain constant, and gas must adequately mix by the time it reaches the sampling port. Figure 4 demonstrates the excellent correlation of bymixer PCO2 compared with the simultaneous value measured in the expired gas collection over a wide range of VT, f, and PCO2. The Bland-Altman LOA analysis (12,13) (Fig. 5) demonstrates excellent bymixer measurement accuracy, where 95% of the bymixer mea-surements were within 3% of the simultaneous value measured in the mixed expired gas collection. If present, the small bymixer PCO2 oscillations (Fig. 6, Table 1) were time-averaged and did not degrade bymixer performance for mixing chamber volumes of 100, 150, and 200 mL.
For these mixing chamber volumes, the ratio of bypass flow to total flow was similar (1:9) because the major impedance to gas flow was the flow resistor. The increased length of the large-bore mixing chamber tubing did not materially add to bypass flow resistance. Thus, the dynamic response of the bymixer can be improved by decreasing the volume of the mixing chamber tubing (decreased length). Table 1 suggests that the bymixer dynamic response (at 8 L/min) could be improved, beyond (less than) the 9.3-second of the bymixer with a 100-mL mixing chamber, by further decreasing the mixing chamber volume. However, at some point, time averaging of increasing FCO2 oscillations would significantly depart from the flow-averaged value and degrade bymixer accuracy. Interestingly, compared with constant gas flows (Table 1), the bymixer demonstrated a much faster response during mechanical ventilation of the metabolic lung simulator, presumably because the periodic peak expiratory flows enhanced gas mixing in the bymixer.
There was no difference in bymixer accuracy between continuous and intermittent aspiration at the sampling port. Accordingly, the downstream volume (measured from the sampling port) of the bypass channel was sufficiently large that sidestream sampling (200 mL/min) did not spuriously sample gas from the main flow outlet during inspiration (when gas flow through the bymixer was zero).
The small bymixer FCO2 oscillations, when present, represented slight incomplete mixing in bypass flow. FCO2 oscillations decreased with smaller VT because the ratio of VT to mixing chamber volume decreased. FCO2 oscillations decreased with higher f (at constant VT) because increased overall gas flow (and velocity) improved gas mixing. The corrugations of the mixing chamber tubing presumably added to gas mixing. The presence of bymixer PCO2 oscillations was not significant, because simple time averaging of the oscillations resulted in excellent bymixer accuracy (Figs. 4 and 5).
In summary, the novel, parallel design of the new clinical bymixer should provide accurate measurement of mixed expired gas fractions in the anesthesia circle circuit. Simple changes in mixing chamber volume allow adjustable bymixer response time. The fast bymixer response ( = 6.4 seconds) should permit measurements to be updated every 20 seconds (where 95% response occurs by 3). The new clinical bymixer is constructed from standard anesthesia circuit components, attaches easily to the anesthesia machine inspired outlet and expired inlet ports, is simple to clean and sterilize, and has no reservoir that can trap condensed water vapor from expired gas. The new clinical bymixer should help introduce, we believe, indirect calorimetry (O2 and CO2) during anesthesia and the noninvasive detection of metabolic upset (e.g., onset of anaerobic metabolism) and critical events (e.g., onset of pulmonary embolism) (1,7,10,14).
|
Acknowledgments
|
|---|
Supported by National Heart Lung and Blood Grant R01 HL-42637 (PHB, principal investigator). Supported in part by the Department of Anesthesiology, University of California-Irvine, and by National Center for Research Resources Grant M01 RR00827.
The authors acknowledge Duoyou Wang, MD, PhD, for preliminary studies of the clinical bymixer. The authors thank David Chien, BSc, computer support specialist, for assistance in bymixer design/testing and digital data acquisition program development and Chris Kirby, BSc, research associate, for assistance in validation studies. The authors also thank Jeffrey C. Milliken, MD, W. Lane Parker, CCP, and Berend J. Ages, CCP, Division of Cardio-Thoracic Surgery, for providing and helping with the precision occlusion roller pump.
|
References
|
|---|
- Breen PH, Isserles SA, Taitelman UZ. Non-steady state monitoring by respiratory gas exchange. J Clin Monit 2000; 16: 351–60.
- Breen PH. Importance of temperature and humidity in the measurement of pulmonary oxygen uptake per breath during anesthesia. Ann Biomed Eng 2000; 28: 1159–64.[Medline]
- Breen PH. Fast response humidity and temperature sensor device. US Patent 6,014,890. January 18, 2000.
- Sanjo Y, Ikeda K. A small bypass mixing chamber for monitoring metabolic rate and anesthetic uptake: the bymixer. J Clin Monit 1987; 3: 235–43.[ISI][Medline]
- Jones WJ. An oxygen replenishment technique for continuous monitoring of oxygen uptake rate. J Assoc Adv Med Instrum 1971; 5: 285–9.[Medline]
- Carter J, Jeukendrup AE. Validity and reliability of three commercially available breath-by-breath respiratory systems. Eur J Appl Physiol 2002; 86: 435–41.[ISI][Medline]
- Breen PH. Carbon dioxide kinetics during anesthesia: pathophysiology and monitoring. Anesthesiol Clin North Am 1998; 16: 259–93.
- Breen PH, Serina ER. Bymixer provides on-line calibration of measurement of CO2 volume exhaled per breath. Ann Biomed Eng 1997; 25: 164–71.[Medline]
- Breen PH, Mazumdar B, Skinner SC. Capnometer transport delay: measurement and clinical implications. Anesth Analg 1994; 78: 584–6.[Abstract/Free Full Text]
- Breen PH, Serina ER, Barker SJ. Measurement of pulmonary CO2 elimination must exclude inspired CO2 measured at the capnometer sampling site. J Clin Monit 1996; 12: 231–6.[Medline]
- Damask MC, Weissman C, Askanazi J, et al. A systematic method for validation of gas exchange measurements. Anesthesiology 1982; 57: 213–8.[Medline]
- Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 1: 307–10.[ISI][Medline]
- Altman DG, Bland JM. Commentary on quantifying agreement between two methods of measurement. Clin Chem 2002; 48: 801–2.[Free Full Text]
- Breen PH, Isserles SA, Harrison BA, Roizen MF. Simple, computer measurement of pulmonary VCO2 per breath. J Appl Physiol 1992; 72: 2029–35.[Abstract/Free Full Text]
Accepted for publication June 3, 2003.
Top
Abstract
Introduction
