Many highly trained endurance athletes exhibit exercise-induced arterial hypoxemia (EIH) during heavy exercise at sea level (4,9,10,16,25). From an exercise performance standpoint, this phenomenon is of interest as the decreased arterial oxyhemoglobin saturation (HbSaO2) associated with EIH may reduce O2 delivery to the working muscles, possibly limiting V˙O2max (9,25). Pulmonary gas exchange limitations, indicated by an excessively widened alveolar-arterial O2 pressure difference (A-aDO2), have been identified as a primary determinant of the reduced HbSaO2 and arterial PO2 (PaO2) in EIH subjects (10,36). However, many athletes with EIH also demonstrate a reduced hyperventilatory response during exercise, resulting in an alveolar O2 partial pressure (PAO2) that is inadequate to maintain PaO2 near resting levels as exercise workload increases (10,16).
By increasing PAO2 during exercise in EIH athletes, it may be possible to determine the nature of the ventilatory role in the formation of the hypoxemia or determine whether hypoxemia could be prevented or mitigated. PAO2 has been directly increased in EIH athletes with a mildly hyperoxic inspirate, reducing the hypoxemia and improving V˙O2max (10,25). PAO2 may also be increased during exercise by increasing alveolar ventilation that is reportedly lower in EIH athletes compared with normoxic athletes (10,16). During submaximal exercise, stimulating ventilation (V˙E) in trained athletes has been successfully accomplished with a mildly hypercapnic inspirate (18), a mildly hypoxic inspirate (3,4), increased dead space (21), and with pharmacologically augmented peripheral chemosensitivity (15,22). However, during maximal exercise, many highly trained endurance athletes reach an expiratory flow limit due to mechanical constraints of the chest wall and, subsequently, are unable to further increase V˙E despite the presence of potent chemical stimuli (3,12,18). Increasing alveolar ventilation may also increase the work of breathing, which has been shown to negatively affect skeletal muscle blood flow and exercise performance (17). Therefore, it is unknown if ventilation during heavy or maximal exercise could be augmented in EIH athletes or if an increase in maximal V˙E could improve end-exercise HbSaO2 or V˙O2max.
In examining potential ventilatory stimulants, caffeine has been historically cited for its stimulating effects on respiration (20,28). Clinically, xanthine compounds have been used as safe and effective ventilatory stimulants for the treatment of chronic obstructive pulmonary disease and neonatal apnea (5,19,20,30). Caffeine in moderate doses has been demonstrated to significantly raise the resting hypoxic and hypercapnic ventilatory responses (HVR and HCVR, respectively) (15,24), which have been reported to contribute to the inadequate hyperventilatory response associated with EIH (16). Caffeine also has a long history of central nervous system stimulation (20,23), and evidence exists of a strong central effect of caffeine on ventilation (28,33). During moderate exercise, caffeine dosages in the clinical range have been reported to cause increased alveolar ventilation (1) and an increased ventilatory equivalent for CO2 (V˙E/V˙CO2) (15). However, data on ventilatory changes during maximal exercise with caffeine are lacking.
Therefore, the purpose of this investigation was to further examine the role of ventilation and ventilatory chemoresponsiveness on the formation of EIH in highly trained endurance athletes. We specifically examined if the analeptic effects of caffeine would augment the ventilatory response to exercise in EIH athletes, perhaps via changes in resting ventilatory chemosensitivity. If the inadequate hyperventilation observed in the EIH population is due in part to low ventilatory responsiveness, then pharmacological augmentation may improve exercise HbSaO2 and V˙O2max.
Seventeen college-aged males volunteered to participate in the study, each meeting a stringent set of criteria: a history of extensive endurance training, no previous history of exercise-induced asthma, caffeine-naive status (defined as daily intakes of <25 mg (14)), and no previous history of sensitivity or discomfort associated with caffeine ingestion. Subjects gave written informed consent before testing, and all protocols and procedures used in testing were approved by the Institutional Review Board of Indiana University. Initial testing involved a maximal graded exercise test on a treadmill to screen for the presence of EIH. The criterion measure for further inclusion in the study was HbSaO2 less than 92% for the final minute of exercise (3,16,25). Of the 17 initial subjects, 8 subjects did not meet this additional criterion and were not included in further portions of the study. One of the remaining nine subjects became injured through training unrelated to data collection and could not complete the additional testing sessions. The eight remaining subjects participated in two additional experimental sessions, each separated by 5 to 16 d. Individual subject characteristics are found in Table 1.
On arrival at the laboratory, each subject ingested a capsule containing caffeine (CAF, 8 mg·kg−1 body weight) or a gelatin placebo (PLA) in a double-blind, randomized, crossover fashion. This dosage of caffeine is within the clinical range (23,30) and is consistent with similar studies investigating the effects of caffeine on resting ventilatory responsiveness (15,24) and V˙O2max (13,14). A 60-min rest period followed capsule ingestion to insure adequate absorption of the caffeine (26). Each subject then performed, in order, an HVR test, an HCVR test, and a graded maximal exercise test. All testing was completed within 2.5 h of the capsule ingestion, which is within a previously reported time frame of peak plasma caffeine concentration (26).
Measures of resting ventilatory responsiveness.
The techniques used to determine HVR and HCVR are slight modifications of the techniques of Weil et al. (34) and Read (27), respectively, and have been detailed previously (12,16). For the HVR test, subjects rested quietly watching a video documentary while the inspired oxygen concentration was reduced in a stepwise fashion via the manual addition of 100% N2 to a balloon reservoir filled with room air. HbSaO2 and breath-by-breath V˙E were measured continuously, end-tidal CO2 was monitored in real time, and CO2 was manually added to the inspirate to maintain isocapnia. HVR testing was concluded when HbSaO2 fell to less than 70% or end-tidal PO2 (PetO2) was less than 40 mm Hg, with a typical test lasting approximately 10 to 15 min. HVR was expressed as the slope of the regression line relating V˙E to HbSaO2, with units of L·[min·%HbSaO2]−1. By convention, HVR regression slopes are displayed as positive numbers. Data points representing every breath during the test period were used in the analysis for each subject, and the relationship between V˙E and HbSaO2 was consistently linear for all subjects. The HCVR test was performed approximately 15 min after the conclusion of the HVR test. HCVR was determined by having the subject rebreathe from a 10-L bag filled with 100% O2. The HCVR test was terminated when end-tidal PCO2 (PetCO2) reached 55 mm Hg or the subject indicated discomfort. A typical test lasted approximately 7 to 12 min. HCVR was expressed as the slope of the regression line relating V˙E to PetCO2, with units of L·[min·mm Hg]−1. Data points representing every breath during the test period were used in the analysis for each subject, and the relationship between V˙e and PetCO2 was consistently linear for all subjects.
Graded exercise test protocol.
After 4 min of resting data collection, each subject began the maximal graded exercise test by walking on a motor-driven treadmill at 4.8 km·h−1 (3 mph) with 0% grade. After 2 min of walking, the speed of the treadmill was increased gradually to a speed, chosen by the subject, between 10.5 and 13.7 km·h−1 (6.5-8.5 mph). The speed of the treadmill remained constant throughout the test, and the same speed was used for both the PLA and CAF trials. The grade of the treadmill was increased 2% every 2 min until volitional fatigue. Criteria for assessment of V˙O2max included 1) a heart rate (HR) of ±10% of age-predicted maximum, 2) a respiratory exchange ratio (RER) of 1.10 or higher, and 3) a plateau (≤150-mL increase) in V˙O2 with an increase in workload. If two of the three criteria were met, the highest V˙O2 recorded was chosen as the subject's V˙O2max.
Ventilatory and metabolic variables were continuously measured and monitored during exercise using a computer-interfaced, open-flow, indirect calorimetry system. The expired side of a two-way, large-bore nonrebreathing valve (Hans Rudolph 2700, Kansas City, NO) was connected to a 5-L mixing chamber. Fractional concentrations of O2 and CO2 were determined from a continuous sample of dried expired gas at a rate of 300 mL·min−1 using an Applied Electrochemistry S-3A oxygen analyzer and CD-3A CO2 analyzer (Ametek, Thermox Instruments, Pittsburgh, PA). End-tidal partial pressures (PetO2, PetCO2) were measured using a second set of the same model of O2 and CO2 analyzers. End-tidals and HbSaO2 were continuously measured and monitored with a data acquisition control system (WorkBench PC 2.0; Strawberry Tree, Sunnyvale, CA) and averaged for each minute of exercise. The analyzers were calibrated with a gas of known concentration in the physiological range before and after each test. V˙E was determined by a turbine-based electronic flow meter (Model VMM-2: Sensormedics Anaheim, CA) situated on the inspired side.
HbSaO2 was estimated using an ear oximetry (Model 47201A; Hewlett Packard, Waltham, MA). The oximeter was calibrated using an internal protocol before each test, according to the manufacturer's instructions. This instrument has been reported to be accurate across a wide range of HbSaO2, with HbSaO2 values greater than 75% underestimated by less than 2% (31). In addition, a standard spectrophotometer cuvette was specially adapted in our laboratory to fit in the sample viewing slit of the oximeter. Freshly drawn whole blood was analyzed with a tonometer to produce a range of O2 saturations and was sampled in the cuvette by the ear oximeter. One-minute computer averages of HbSaO2 via the ear oximeter were referenced against the average of four samples analyzed by an OSM3 hemeoximeter (Radiometer, Copenhagen, Denmark). In the range of 60-95% HbSaO2, the two independent measures never differed by more than 1.9% and were tightly correlated (r = 0.99) (4). After data collection in this study, the ear oximeter was again checked versus blood samples analyzed with an ABL3000 blood gas analyzer (Radiometer, Copenhagen, Denmark) with similar results. For HbSaO2 ranging from 65% to 98%, the ear oximeter and blood gas analyzer were significantly correlated (r = 0.99), the average deviation across the range was less than 1.0%, and the maximum deviation at any point was 2.1% (12).
Differences in HVR and HCVR test values between treatments were analyzed using a paired t-test. Data from the graded exercise tests was analyzed using a 4 × 2 repeated-measures ANOVA, with workload at four levels (75%, 80%, 90%, and 100% of V˙O2max) as one independent variable and treatment (PLA and CAF) as the second independent variable. Tests of a priori simple main effects were used to determine differences between treatments at each workload. For all tests, the level of significance was established at P < 0.05.
Data during progressive exercise were analyzed at workloads corresponding to 75%, 80%, 90%, and 100% of V˙O2max for each condition (PLA and CAF) and are found in Table 2. V˙O2, V˙CO2, and RER were not different between PLA and CAF at any level of exercise.
CAF resulted in significant increases in V˙E (compared to PLA) at 75%, 80%, and 100% of V˙O2max (Fig. 1). A corresponding increase in V˙E/V˙O2 occurred at 75%, 80%, and 90% of V˙O2max with CAF, whereas V˙E/V˙CO2 was higher with CAF at all levels of exercise. PetO2 was significantly elevated during submaximal exercise with CAF (75%, 80%, and 90%), whereas PetCO2 was reduced at 75% and 80% of V˙O2max in CAF. HbSaO2 was significantly higher at all levels of submaximal exercise (75%, 80%, and 90%); however, neither PetO2, PetCO2, nor HbSaO2 was different between treatments at V˙O2max. This result occurred despite increases in V˙e and ventilatory equivalent for CO2 during maximal exercise with CAF. HR was significantly elevated with CAF only at V˙O2max (PLA 184.9 ± 2.4 bpm, CAF 191.6 ± 1.3 bpm [mean ± SE]).
Individual measures of peripheral chemoresponsiveness are shown in Figure 2. The HVR was determined at a PetCO2 of 41.3 ± 1.2 mm Hg and 39.7 ± 0.7 mm Hg for the PLA and CAF conditions respectively. PetCO2 did not change significantly from normoxic to hypoxic conditions in either PLA (42.0 ± 1.3 vs 40.9 ± 0.8 mm Hg) or CAF (40.2 ± 0.8 vs 39.3 ± 0.9 mm Hg). Ventilation at rest was significantly higher in CAF (8.9 ± 1.7 L·min−1) versus PLA (7.0 ± 1.2 L·min−1), and remained higher at any common HbSaO2. However, the gain in peripheral chemoresponsiveness was unchanged, as caffeine ingestion was not associated with a change in the resting HVR (PLA 0.30 ± 0.07 L·[min·%HbSaO2]−1, CAF 0.29 ± 0.09 L·[min·%HbSaO2]−1). Similarly, resting HCVR values in the CAF condition were not significantly different from PLA (PLA 1.79 ± 0.23 L·[min·mm Hg]−1, CAF 2.20 ± 0.40 L·[min·mm Hg]−1).
The primary findings of this study are that 1) caffeine ingestion of 8 mg·kg−1 body wt caused an increase in ventilation during both submaximal and maximal exercise, 2) the increase in ventilation resulted in an improvement in HbSaO2 only during submaximal exercise, 3) no change was observed in V˙O2max between treatment conditions, and 4) caffeine does not augment resting ventilatory responsiveness in athletes with EIH.
Although there are several studies that have examined the metabolic and hormonal effects of caffeine on endurance exercise performance, few investigations have specifically measured the stimulatory effects of caffeine on ventilation during endurance exercise. In two previous studies that have specifically reported the ventilatory response to exercise with caffeine, one used constant load treadmill running at 50% of V˙O2max (1) and a second only reported ventilatory results for exercise at a workload "below the anaerobic threshold" in untrained subjects (15). In our subject cohort of highly trained distance runners who exhibit EIH during maximal exercise, we specifically wished to examine the effect of augmenting the ventilatory response to exercise at maximal and near-maximal workloads, where arterial oxyhemoglobin desaturation is most likely to occur.
Although caffeine was successful in augmenting ventilation across workloads, differences in V˙E, V˙E/V˙O2, V˙E/V˙CO2, and end-tidal measures between PLA and CAF seem to become narrower as workload increased from submaximal to maximal exercise (Fig. 1). This finding indicates that the effect of caffeine on exercise ventilation may be workload-dependent. Existing evidence suggests a decrease in the gain of the ventilatory response to exercise with increasing work rate, using inhaled CO2 (18), a hypoxic inspirate (3,4), or increased dead space (21) as stimuli. Our results seem to parallel the ventilatory response in those studies but with the new intervention of pharmacological ventilatory stimulation with caffeine. The reduced gain in the ventilatory response to exercise with increasing workload may be related to the approaching or frank achievement of expiratory flow limitation during heavy exercise, which could be a causal factor behind the reduced V˙E/V˙O2 generally found with EIH (3,11,18). Although seven of our eight athletes increased V˙E at V˙O2max with CAF and mean V˙E increased significantly by 6.5 L·min−1, the presence of ventilatory flow limitation to any extent in our subjects may have compromised the magnitude of a potential increase in V˙E with caffeine at maximal workloads.
Although ventilatory effects of caffeine were successful in increasing PetO2 and HbSaO2 at submaximal workloads, these variables were not significantly elevated at V˙O2max-despite a significant 4.2% (6.5 L·min−1) increase in V˙E. Considering that a similar 4.0% (5.0 L·min−1) change in V˙E at 90% of V˙O2max with caffeine resulted in significantly higher values for HbSaO2 and PetO2, the question remains: why do we not observe an improvement in HbSaO2 at V˙O2max? A failure of PetO2 to significantly increase with CAF during maximal exercise suggests that a large portion of the increase in V˙E at that workload may have gone to dead space ventilation (V˙d), with only a slight increase in alveolar ventilation (V˙a). If we were to use our measures of PetO2 as accurate estimates of PAO2, the alveolar gas equation [PAO2 = PIO2 − RT (V˙O2 / VA)] (8) can be used with the relationship V˙E = V˙d + V˙a (36) to obtain an estimate of Vd. We find that with CAF at V˙O2max, estimated V˙d significantly increased by 22% for PLA (PLA V˙d 17.1 ± 1.7 L·min−1, CAF V˙d 21.0 ± 1.9 L·min−1, P < 0.05). Therefore, of the 6.5-L·min−1 increase in V˙e at V˙O2max with CAF, a calculated 2.6 L·min−1 or only approximately 40% went toward increasing alveolar ventilation. At 90% of V˙O2max (where HbSaO2 was significantly increased by 1.2% with CAF), estimated V˙d was not significantly higher with CAF, and calculated increases in V˙a with CAF accounted for a much larger percentage of the increase in V˙E (3.4-L·min−1 increase in V˙d of a 5.0-L·min−1 increase in V˙E, or approximately 68%).
At 90% of V˙O2max, seven of eight subjects each had an individual increase in both V˙E and HbSaO2 with caffeine. However, at 100% of V˙O2max, although the same seven subjects had individual increases in V˙E with CAF, only three of those subjects demonstrated increases in HbSaO2. Figure 3 shows the minute-by-minute V˙E and HbSaO2 values during progressive exercise in both experimental conditions for Subject #2. Note that this individual had higher V˙E and HbSaO2 values with CAF during the same minute of progressive exercise at common workloads but was able to complete one additional minute of exercise in the caffeine trial. As HbSaO2 may be affected by pH changes independent of changes in PO2, it is possible that this subject and the others had differences in metabolic acidosis between trials, affecting HbSaO2 measures. It is also important to note that although the use of oximetry to provide a noninvasive measurement of HbSaO2 was logistically preferred over two arterial catheterizations on separate days, HbSaO2 is likely not sensitive or specific enough to determine whether PaO2 changed a few millimeters of mercury with the magnitude of increase in ventilation we observed with caffeine in EIH subjects. Now that the ventilatory response to heavy exercise with caffeine has been characterized in EIH subjects, direct measures of PaO2 in future studies examining the effects of caffeine and pharmacologically stimulated ventilation on gas exchange limitations in EIH subjects may provide additional insight.
The global ergogenic effects of caffeine on exercise performance have been studied extensively; however, the concept of improving performance specifically via the ventilatory effects of the drug seems to be a novel approach. Caffeine-induced improvement in HbSaO2 during submaximal exercise may be beneficial in terms of increased oxygen transport to the periphery during prolonged exercise. However, a ventilatory-driven increase in HbSaO2 must be weighed against the potential metabolic cost of the increased ventilation. As changes in ventilatory work have been demonstrated to influence blood flow to the locomotor muscles during heavy exercise (17), it is unknown whether the improvement in HbSaO2 with increased V˙e holds an ergogenic advantage during prolonged submaximal exercise. During this type of exercise, caffeine has been theorized to be ergogenic primarily via an elevation of catecholamines, promoting greater fat oxidation and sparing glycogen stores (7). However, many studies demonstrating improved time to exhaustion at a constant, submaximal workload have failed to show a concomitant fall in RER with the observed rise in free fatty acid concentration or plasma catecholamine levels (6). Our data would suggest that the RER may not be sensitive enough to mark any changes in substrate use with caffeine, as potential reductions in the muscle respiratory quotient with caffeine may not be reflected in the pulmonary RER due to an increased ventilatory response to exercise and excess nonmetabolic CO2 production. Although the purpose of this study was not to investigate the ergogenic effects of caffeine per se, our results would suggest that future research on the effects of caffeine and exercise performance in endurance athletes should consider the changes in the ventilatory response to exercise.
Caffeine did not cause an increase in measures of resting ventilatory responsiveness in our group of EIH athletes (Fig. 2). In contrast to our data, D'Urzo et al. (15) found a significant 135% increase in the HVR with 650 mg of caffeine in a group of seven untrained, mostly caffeine-habituated males. These authors also found that caffeine caused a significant 28% increase in the HCVR (from 1.48 to 1.88 L·[min·mm Hg]−1) similar to our nonsignificant results. The lack of differences in ventilatory responsiveness we observed between treatments may be related to the unique characteristics of the population that our subjects represent. Several investigations have established that endurance-trained athletes demonstrate blunted ventilatory responses to hypoxic and hypercapnic stimuli at rest, compared to untrained individuals (2,29), and a previous investigation from our laboratory demonstrated that EIH athletes, in particular, have HVR and HCVR measures that are significantly lower than non-EIH athletes (16). The level of baseline ventilatory responsiveness may be an important characteristic in this case, as it has been shown that low-baseline ventilatory responsiveness may affect the ability of a pharmacological chemoreceptor stimulant to augment resting HVR and HCVR (32).
In a study by Stanley et al. (32), the authors reported a significant positive correlation between baseline HVR and the increase in HVR after the administration of the drug almitrine, which is a potent stimulator of the peripheral chemoreceptors. Thus, individuals with the lowest HVR values saw the smallest increase in HVR with pharmacological stimulation. However, it is important to note that of the 12 untrained subjects tested by Stanley et al., only 1 subject had a baseline HVR lower than the highest baseline HVR in our subject group. The mean HVR in our group of 0.30 L·[min·%HbSaO2]−1 is consistent with the blunted HVR values seen in other studies examining highly trained endurance athletes (2,16) and is substantially lower than the mean HVR values reported by investigations examining HVR in untrained subjects (29,32,35). Taken together, we believe it is reasonable to state that EIH athletes are generally insensitive to an increased gain in peripheral chemoresponsiveness with caffeine, perhaps due to the severely blunted baseline chemoresponsiveness typically seen in this population.
Several investigators have argued that the primary focus behind caffeine's ventilatory effect is central stimulation of the respiratory medullary complex (20,33). The finding of an increase in resting and exercise V˙E with CAF, without concomitant changes in resting ventilatory responsiveness, suggests that exercise ventilation was augmented in this subject group of EIH athletes through a central mechanism. Figure 4 displays the relationship between resting V˙E as a function of HbSaO2 during the HVR test, averaged for all eight subjects. If peripheral chemosensitivity were augmented with caffeine, we would expect to see a change in the slope of the V˙E-HbSaO2 relationship. However, the similar slopes and the higher V˙E in normoxia and at any HbSaO2 with CAF strongly suggest a central effect on ventilation within this population of athletes.
In summary, a moderate dose of caffeine was successful in augmenting exercise ventilation in athletes with EIH. Ventilatory stimulation was associated with an increase in end-tidal PO2 and HbSaO2 at submaximal workloads, however at V˙O2max, HbSaO2 was not significantly higher. The increase in exercise V˙E with caffeine in EIH subjects seems to result not from an increase in resting peripheral chemoresponsiveness but possibly from central or secondary effects of caffeine. A lack of peripheral chemoreceptor stimulation with caffeine may be unique to the portion of the highly trained population that demonstrates EIH and needs verification. At this time, it is unknown if caffeine-induced improvements in submaximal HbSaO2 enhance endurance exercise performance
This research study was supported by an Indiana University RUGS grant. The results of the present study do not constitute endorsement by ACSM.
1. Brown DD, Knowlton R, Sullivan J, Sanjabi P. Effect of caffeine ingestion on alveolar ventilation during moderate exercise. Aviat Space Environ Med
2. Byrne-Quinn E, Weil JV, Sodal I, Filley G, Grover R. Ventilatory control in the athlete. J Appl Physiol
3. Chapman RF, Emery M, Stager JM. Extent of expiratory flow limitation in normoxia influences the increase in exercise ventilation in mild hypoxia. Respir Physiol
4. Chapman RF, Emery M, Stager JM. Degree of arterial saturation in normoxia influences the decline in V˙O2max
in mild hypoxia. Med Sci Sports Exerc
5. Chardon K, Bach V, Telliez F, et al. Effect of caffeine on peripheral chemoreceptor activity in premature neonates: interaction with sleep stages. J Appl Physiol
6. Conlee RK. Amphetamine, caffeine, and cocaine. In: Williams DR, Williams MH, editors. Ergogenics: Enhancement of Sports Performance in Exercise and Sport
. Indianapolis, IN: Brown & Benchmark; 1991. p. 285-330.
7. Costill DL, Dalasky G, Fink W. Effects of caffeine ingestion on metabolism and exercise performance. Med Sci Sports Exerc
8. Dejours P. Principles of Comparative Respiratory Physiology
. New York, NY: Elsevier; 1981. p. 105-8.
9. Dempsey JA. Exercise-induced imperfections in pulmonary gas exchange. Can J Appl Sport Sci
10. Dempsey JA, Hanson PE, Henderson KS. Exercise induced arterial hypoxemia in healthy persons at sea level. J Physiol
11. Dempsey JA, Johnson BD, Saupe KW. Adaptations and limitations in the pulmonary system during exercise. Chest
12. Derchak PA, Stager JM, Tanner DA, Chapman RF. Expiratory flow limitation confounds ventilatory response during exercise in athletes. Med Sci Sports Exerc
13. Dodd SL, Herb RA, Powers SK. Caffeine and exercise performance: an update. Sports Med
14. Dodd SL, Brooks E, Powers SK, Tulley R. The effects of caffeine on graded exercise performance in caffeine naive versus caffeine habituated subjects. Eur J Appl Physiol
15. D'Urzo AD, Jhirad R, Jenne H, et al. Effect of caffeine on ventilatory responses to hypercapnia, hypoxia, and exercise in humans. J Appl Physiol
16. Harms CA, Stager JM. Low chemoresponsiveness and inadequate hyperventilation contribute to exercise-induced hypoxemia. J Appl Physiol
17. Harms CA, Wetter TJ, McClaran SR, et al. Respiratory muscle work compromises leg blood flow during maximal exercise. J Appl Physiol
18. Johnson BD, Saupe K, Dempsey JA. Mechanical constraints on exercise hyperpnea in endurance athletes. J Appl Physiol
19. Jones JE. Caffeine and Health
. San Diego, CA: Academic Press; 1991. p. 24-69.
20. Le Messurier DH. The site of action of caffeine as a respiratory stimulant. J Pharmacol Exp Ther
21. McParland C, Mink J, Gallagher CG. Respiratory adaptations to dead space loading during maximal incremental exercise. J Appl Physiol
22. Naeije R, Melot C, Niset G, Delcroix M, Wagner PD. Mechanisms of improved arterial oxygenation after peripheral chemoreceptor stimulation during hypoxic exercise. J Appl Physiol
23. Persson C. Overview of the effects of theophylline. J Allergy Clin Immunol
24. Pianosi P, Grondin D, Desmond K, Coates A, Aranda J. Effect of caffeine on the ventilatory response to inhaled carbon dioxide. Respir Physiol
25. Powers SK, Lawler J, Dempsey JA, Dodd S, Landry G. Effects of incomplete pulmonary gas exchange on V˙O2max
. J Appl Physiol
26. Rall TW. Central nervous system stimulants. In: Pharmacological Basis of Therapeutics
. New York: MacMillian; 1985. p. 589-603.
27. Read DJC. A clinical method for assessing the ventilatory response to carbon dioxide. Australas Annals Med
28. Richmond FH. Action of caffeine and aminophylline as respiratory stimulants in man. J Appl Physiol
29. Scoggin CH, Doekel RD, Krueger MH, Zwillich CW, Weil JV. Familial aspects of decreased hypoxic drive in endurance athletes. J Appl Physiol
30. Sharp JT. Theophylline in chronic obstructive pulmonary disease. J Allergy Clin Immunol
31. Smyth RJ, D'Urzo AD, Slutsky AS, Galko BM, Rebuck AS. Ear oximetry during combined hypoxia and exercise. J Appl Physiol
32. Stanley NN, Galloway JM, Flint KC, Campbell DB. Increased respiratory chemosensitivity induced by oral almitrine in healthy men. Br J Dis Chest
33. Trippenbach T, Zinman R, Milic-Emili J. Caffeine effect on breathing pattern and vagal reflexes in newborn rabbits. Respir Physiol
34. Weil JV, Byrne-Quinn E, Sodal IE, et al. Hypoxic ventilatory drive in normal man. J Clin Invest
35. Weil JV, Byrne-Quinn E, Sodal IE, Kline JS, McCullough RE, Filley GF. Augmentation of chemosensitivity during mild exercise in normal man. J Appl Physiol
36. West JB. Respiratory Physiology-the Essentials
. Baltimore, MD: Lippincott Williams & Wilkins; 1985. p. 11-85.