The ergogenic effects of caffeine are well documented for endurance events, with significant increases in the time to exhaustion or mean power output in time trials (9,11,15,24,30). There is also mounting evidence for enhanced performance during short-term intense exercise lasting 4–10 min (4,14,21). However, effects of caffeine on brief bouts of single or repeated sprints lasting less than 1 min are mostly unclear (3,7,17,27), and there has been no research on repeated bouts of high-intensity exercise over durations typical of team sports, which can last over an hour. In such sports, the outcome is often determined late in the game, when players are fatigued (16). Caffeine might enhance performance in these sports by attenuating fatigue. Therefore, the main purpose of the present study was to estimate the effects of caffeine on simulated team-sport performance, specifically the physical activity and skill demands of a rugby union game.
The mechanisms of the ergogenic effects of caffeine are unclear (15,22). In an early study Costill and coworkers (9) suggested that caffeine could improve endurance performance by elevating plasma free fatty acids, thereby sparing muscle glycogen. However, they also found that caffeine attenuated the rise in perceived exertion during exercise, which may have contributed to the ergogenic effect. Recent elegant experiments using animal models have provided evidence to support the notion that caffeine may act via the CNS (11). Furthermore, it has repeatedly been suggested that the effects of caffeine may be mediated indirectly via elevated plasma epinephrine (15,24), which in principle could act via the CNS or on peripheral tissues. Therefore, another purpose of the present study was to examine the relationship between changes in performance with caffeine and changes in plasma epinephrine concentration.
Eleven high-level amateur male rugby union players, of Polynesian or Polynesian-Caucasian ethnicity, were recruited from a team in the Auckland premier club competition. Owing to injuries independent of the present study two players did not complete the study. The subject characteristics were (mean ± SD, N = 9): age, 25 ± 4 yr; body mass, 98 ± 22 kg; and height, 181 ± 4 cm. The study was conducted 3 wk after the final championship game when the subjects reported a current training volume of 4.8 ± 0.8 h·wk−1. All subjects gave voluntary informed consent with protocols approved by the AUT ethics committee.
The study was a randomized crossover in which subjects performed a simulated rugby performance test. Sample size was determined by availability of subjects and resources and was similar to that of many studies of effects of caffeine. We could not determine a priori the minimum sample size for adequate precision of effects because the reliability of the performance measures was unknown. The subjects and research assistants who supervised the stations of the test were blind to the treatment. A familiarization and two treatment trials of either caffeine or placebo were each separated by a week. Owing to failure of the drive ergometer after the first half of the second trial, a fourth trial was added 1 wk after the third. Eight of the nine subjects returned for this fourth trial.
The subjects were weighed and a fluid sample was obtained transdermally 70 min before the start of the test. Subjects then ingested capsules containing either caffeine or placebo. A further fluid sample was taken 1 h later. Just before testing the subjects performed their normal warm-up (10 min), which included light jogging, stretching, and some touch rugby. The rugby test was then undertaken in two 40-min halves with a half-time 10-min rest period. Fluid samples were drawn at the end of each half and later assayed for caffeine and epinephrine concentrations.
A moderate dose of caffeine (6 mg·kg−1 body mass) was chosen, which has well-documented ergogenic effects for endurance events (15). Caffeine (Bronson and Jacobs, Sydney, Australia) or the same dose of placebo (dextrose) was weighed (589 ± 129 mg) and placed in gelatin capsules, then swallowed with water. This dose of dextrose is calculated to increase plasma glucose concentration by <0.3 mmol·L−1. All subjects stated that they were regular consumers of caffeine in their diet. They were provided with a list of foods and beverages that contained caffeine and asked to refrain from consuming these for 48 h before exercise testing. Subjects were also provided with a 48-h diet record questionnaire before their first testing session so that they could replicate their diet before each of the subsequent testing sessions.
This test was designed, based on time-motion analysis of first-class level rugby union games, to simulate the activities of a game. (For details, see Deutsch (12).) In brief, the test consisted of 14 circuits with each circuit made up of 11 stations for activities that included sprinting (straight-line and agility sprints), peak power generation in two consecutive drives, accuracy for passing balls, but also allowed for rest periods that included standing or walking (Table 1).
For sprint tasks, the subjects began approximately 20 cm behind the start line in a static position and sprint time (to the nearest 0.001 s) was measured using electronic timing lights (Speed-Light, Swift Performance Equipment, Goonellabah, Australia) set at knee height at the start and finish gates. The offensive sprint involved a forward run with swerving, while carrying a ball, then making a tackle on a tackle bag after the final timing gate. The defensive sprint involved running three arcs, first forward then backward. The tackle sprint involved making a tackle on a tackle bag, picking up a ball and running backward, placing the ball, making another tackle, and then running forward. Total distance covered with sprinting over the test was 1904 m. For the dynamic drive task, a dynamometer cart (GRUNT 3000, School of Physical Education, University of Otago, Dunedin, New Zealand) measured peak power via transducers for speed of the cart and for force (load cell) attached to a bungee tether. Subjects began 1 m behind the cart then drove into the cart as quickly as possible for 5 s using shoulders, arms and legs (Drive 1), with the drive repeated from the starting position approximately 5 s later (Drive 2). For the passing accuracy task, the subjects were instructed to pass a ball as rapidly as possible at a target (dimensions of 1 × 1 m), placed 4 m from the player with its center being 2 m above the ground. After each pass the subject picked up another ball from the ground and repeated the pass. The time for this activity was approximately 10–15 s. The number of successful target hits out of five balls passed was counted.
The 14 circuits were split into two 40-min halves with a 10-min half-time rest to simulate a game. There were seven circuits in each half, with a 2-min rest after circuit 4 (first half) and circuit 11 (second half). Subjects began at Station 1 and proceeded through the 11 stations at 30-s intervals. Once the task at each station was complete, the subject had the remainder of the 30 s to rest and move to the next station. Research assistants were present at each station to verbally encourage the subjects and record performance measures. Water was available for consumption at Station 11. The tests were performed in the early evening in an indoor sports stadium on a wooden floor at an ambient temperature of 21–23°C.
Plasma caffeine and epinephrine concentrations.
Fluid samples were collected transdermally from the forearm using the noninvasive technique of electrosonophoresis (8), then analyzed for caffeine and epinephrine concentrations by high-performance liquid chromatography (HPLC). Plasma concentrations were predicted from the concentrations in the samples using equations derived in separate calibration studies. For the caffeine calibration, venous blood and transdermal fluid samples were drawn concurrently from 12 men on up to 8 d sampled twice daily (129 samples in total). Some subjects varied their caffeine intake. Validity analysis of the paired samples (using a spreadsheet available at http://newstats.org) revealed that log transformation produced satisfactory uniformity of percent error over the range of plasma concentrations (1.9 to 8.2 μg·mL−1). The calibration equation was P = aTb, where P is the plasma concentration, T is the transdermal concentration, a = 9.93, b = 0.996, with an SE of the estimate of 6.8%. There was insufficient caffeine in the transdermal sample for accurate measurement by HPLC, when the corresponding plasma concentrations were less than 2.0 μg·mL−1. For the epinephrine calibration, eight men undertook five cycle ergometry sessions of varying duration (30, 45, and 60 min) with each session 2–3 d apart. Venous blood and transdermal samples were collected concurrently immediately before and after the exercise. A validity analysis similar to that for caffeine showed uniformity of the percent error over the range of observed plasma epinephrine concentrations (0.79–4.98 nmol·L−1). The calibration equation was P = aTb, where a = 7.89, b = 0.988, with an SE of the estimate of 6.0%.
Data from all trials were included in the analyses. All variables representing performance were log transformed before analysis to reduce nonuniformity of error and to express effects as percent changes (20). Repeated-measures analyses were performed with a mixed-modeling procedure (Proc Mixed) in the Statistical Analysis System (Version 8.2 SAS Institute, Cary, NC). Fixed effects in the mixed model were trial (first to fourth), treatment (familiarization, caffeine and placebo), and circuit (first to fourteenth). Trial was included only as a main effect to account for familiarization; the other effects were included with their interactions. Random effects were the identity of the athletes and terms representing within-athlete variation in performance between trials and between circuits within trials. Analyses for epinephrine and caffeine concentrations were similar to those for measures of performance, except that the fixed and random effects for circuit were replaced with time (with levels pre-, mid-, and postexercise). The possibility that individual responses in epinephrine and caffeine concentrations accounted for individual responses in performance was investigated by deriving correlations of changes in performance between caffeine and placebo conditions with changes in plasma caffeine and changes in plasma epinephrine. For plasma caffeine, the correlations were derived for the concentration averaged for pre-, mid-, and postexercise in the caffeine condition, the concentrations in the placebo condition being zero. For plasma epinephrine, separate correlations were derived for the preexercise caffeine-placebo change and for the average of the mid- and postexercise changes. Magnitudes of correlations were interpreted using Cohen’s thresholds (<0.1, trivial; 0.1–0.3, small; 0.3–0.5, moderate; >0.5, large) (6).
Means for measures of performance shown in tables and figures are least-squares means (i.e., means adjusted for any missing values using the fixed-effects model). Between-subject variations for measures of performance in the tables are coefficients of variation derived from the statistical model; these represent typical variation between subjects in either group for the measure in any one circuit. They were converted to SD for display in the figures. The between-subject variations for caffeine and epinephrine concentrations in the figures were derived directly from raw data.
To make inferences about true (population) values of the effect of caffeine on performance, the uncertainty in the effect was expressed as 90% confidence limits and as likelihoods that the true value of the effect represents substantial change (harm or benefit) (20). An effect was deemed unclear if its confidence interval overlapped the thresholds for substantiveness, that is, if the effect could be substantially positive and negative or beneficial and detrimental. The smallest substantial change in sprint performance was assumed to be a reduction or increase in sprint time of more than 0.8% (27). The between-subject SD for drive power and accuracy of passing were used to convert the log-transformed changes in performance into standardized (Cohen) changes in the mean. The smallest standardized change was assumed to be 0.20 (6). Inferences about the correlations between plasma caffeine, plasma epinephrine, and performance were made with respect to a smallest worthwhile correlation of 0.10 (6). Based on eight observations, only correlations >0.56 were conclusive (true correlation likely to be substantial of the same sign and very unlikely to be substantial of the opposite sign).
Before each test, all subjects expressed uncertainty about the identity of the treatment they had received. Table 2 shows the mean performance values for the various tasks obtained by averaging all circuits of the test in the placebo condition. The rugby test involved brief (3–14 s) all-out exercise bouts. The total exercise time for each circuit amounted to approximately 60 s; the remainder of each 5.5-min circuit consisted of periods of rest or walking (20–55 s).
In the caffeine condition there were enhancements of mean performance for all measures except Drive 2 power, although the effects were unclear for this measure and for the 20-m and offensive sprints (Table 3). The effects on performance in each circuit are shown in Figure 1 for tackle sprints (which had the greatest mean enhancement and greatest chance of benefit with caffeine), the two drives, and passing accuracy. Most measures showed an enhancement of performance with caffeine in the first circuit, as can be seen for the measures in Figure 1. Analysis of performance for the first circuit revealed a possible beneficial effect for Drive 1 power (5.5%; 90% confidence limits, ±5.7%) and a likely beneficial effect for passing accuracy (12%; ±13%). Effects for the other measures in the first circuit were unclear; the observed values were beneficial for offensive sprint speed (1.6%; ±4.7%), Drive 2 power (1.9%; ±9.6%), defensive sprint speed (2.7%; ±3.6%), tackle sprint speed (2.1%; ±3.0%), and 30-m sprint speed (2.0%; ±4.1%), but detrimental only for 20-m sprint speed (−1.6%; ±3.3%).
It is also apparent in Figure 1 and for the other sprints (data not shown) that the mean enhancements with caffeine arose partly through a reduction in fatigue as the test progressed. We quantified fatigue in the test by calculating the decline in the performance in the second half relative to the first. In the placebo condition, the declines were: 20-m sprint speed, 2.1% (90% confidence limits, ±1.4%); offensive sprint speed, 2.8% (±1.4%); Drive 1 power, 3.7% (±2.6%); Drive 2 power, 5.0% (±3.7%); defensive sprint speed, 0.9% (±1.2%); tackle sprint speed, 2.0% (±1.4%); passing accuracy, 9.0% (±4.7%); and 30-m sprint speed, 5.4% (±1.8%). The effects of caffeine on fatigue were decisive for all measures: possible through very likely reductions in fatigue for all but power in Drive 2, where a detrimental effect of caffeine was possible (Table 3).
For some measures, fatigue was not evident in the last circuit of the placebo condition relative to the first circuit. Indeed, for the 20-m sprint, there appeared to be an increase in performance of 5.5% (±3.2%) relative to the first circuit, and there were also substantial although less clear improvements for defensive, tackle and 30-m sprints. The other measures showed declines in performance, which were unclear for the offensive sprint and Drive 2, but possible for Drive 1 (6.5%; ±7.8%), and very likely for passing accuracy (18%; ±11%). It was only for these latter two measures that the effect of caffeine was clear in the last circuit: a likely benefit for the first drive (12%; ±11%) and an almost certain benefit for passing accuracy (28%; ±19%).
Plasma caffeine and epinephrine.
Figure 2 shows that 1 h after ingestion of caffeine, the plasma caffeine concentration increased to 8.2 μg·mL−1 (equivalent to 42 μmol·L−1) and then declined somewhat during the rugby test. The caffeine concentration before ingestion was below the level of detection in the assay for all subjects; the value is shown in the figure as 1 μg·mL−1. Caffeine concentration was also below the level of detection for all subjects at all time points in the placebo condition. Figure 2 also shows the plasma epinephrine concentration before, during, and after the rugby test for the caffeine and placebo conditions. Averaged over these three time points, the epinephrine concentration was 51% higher (90% confidence limits ±11%) in the caffeine condition than in the placebo condition.
There was a strong positive correlation (r = 0.59; 90% confidence limits −0.06 to 0.89) between plasma caffeine concentration during the caffeine condition and the change in epinephrine concentration between placebo and caffeine conditions; both concentrations were averaged in each subject over the pre-, mid-, and postexercise assays. Correlations between plasma caffeine and changes in performance were inconclusive, with the exception of a strong positive correlation with tackle sprint speed (r = 0.63) and a strong negative correlation with Drive 1 power (r = −0.80). All but one of the correlations of changes in epinephrine concentration before and during the test with changes in sprint speed were substantial and positive, and two were conclusive (range −0.03 through 0.64). All the correlations of changes in epinephrine with changes in drive power were substantial and negative, but none was conclusive (range −0.27 through −0.49). There was also a strong negative correlation between passing accuracy and epinephrine concentration before the performance test (r = −0.63), and the correlation during the test was negative but unclear (r = −0.27).
Effects of caffeine on exercise performance.
The present study is the first to report that caffeine has multiple beneficial effects on the physical and skill activities required in an intermittent high-intensity team sport. Such effects were demonstrated for five sprint tasks (high-speed requirement), a power task (high-force requirement), and an accuracy task performed rapidly (high motor skill requirement). The enhancements of performance were apparently not due to a placebo effect at the start of the rugby test because no subjects stated that they were confident about what treatment they had received.
Our conclusions are based on the approach to inferential statistics that emphasizes precision of estimation rather than null-hypothesis testing. To that end, we have followed recommendations to show and interpret the practical importance of confidence limits (e.g., Altman et al. (2) and Sterne and Smith (29)), which represent the uncertainty in the true value of each effect. We have built on these recommendations by enunciating a rule for deciding when an effect is clear or unclear and by making quantitative assertions about likelihood that the effect is beneficial or detrimental (20).
For the sprint tasks, most of the effects of caffeine averaged over the entire test were clear and all were positive (Table 3). The improvements of 0.5–2.9% were of similar magnitude to those effects reported for endurance exercise (9,15,24) or high-intensity exercise of 4- to 10-min duration (4,14,21). The positive effects in the present study cannot be compared with those of other studies on repeated sprints (17,27) because the duration of the test was considerably longer. This point is important because the influence of caffeine on the sprints was more apparent in the second half of the test (Fig. 1). We demonstrated that caffeine protects against fatigue during sprints, first from our caffeine versus placebo analysis of the first versus second half sprint performances (Table 3), and second because caffeine was without substantial effect when there was no apparent fatigue, as seen during the efforts for the sprints of the last circuit (Fig. 1).
Caffeine increased the peak power of Drive 1 over the entire test by 5% (Table 3, Fig. 1). Specifically, this involved enhancement in the first circuit and diminished fatigue as shown over the two halves and in the last circuit. Caffeine has also been shown to improve other activities requiring high force production such as competitive rowing (4) and maximum isometric voluntary contractions (MVC) (22,23). The peak MVC force for the quadriceps can increase by 5% (23), but such effects are only detected in situations where there is submaximal voluntary activation (23,25,31). The only clear detrimental effect with caffeine in the present study was for Drive 2, which showed increased fatigue in the second half compared with the first half. Similar detrimental effects have been shown for the latter sprints during repeated Wingate testing (17).
The largest and possibly most exciting new finding with caffeine was a 10% improvement in the ability to pass balls accurately while pressured to pass rapidly (simulating game conditions). This was observed both early in the test and later when the subjects were fatigued (Table 3, Fig. 1). The importance of this observation is that over the course of a game, the subjects would deliver a pass successfully 90% of the time with caffeine compared with 83% in the placebo condition or a difference of about five passes out of 70. This supplement-induced improvement in a skill task requiring coordinated movements of several muscle groups is not found in current literature.
Ingestion of 6 mg·kg−1 caffeine produced a peak plasma concentration of approximately 8 μg·mL−1, which is similar to that previously reported for the same dose of caffeine (15,17). Before administration of treatments, the subjects showed levels of caffeine below the 2 μg·mL−1 threshold of detection for the assay. For the average subject, who weighed 98 kg, the dose was 600 mg of caffeine, which is equivalent to approximately six cups of coffee (30). The detection threshold is therefore equivalent to approximately 1.5 cups of coffee consumed 1 h before the assay. The assay was therefore not sensitive enough to confirm rigorous compliance with the instruction to abstain from caffeine-containing products. In a meta-analysis of the effects of caffeine on performance (G. R. Stuart and W. G. Hopkins, unpublished observations), abstention from caffeine-containing food and drink for at least several days produced an additional performance enhancement of approximately 1.5%. The effects we have observed on performance could therefore be underestimates of the full effect expected with abstention.
Mechanisms for the effects of caffeine on performance.
Caffeine exerted its beneficial effects by causing obvious reductions in fatigue and in some cases by enhancement in the first circuit. However, apart from the first station in the first circuit (the 20-m sprint), all bouts were performed after prior exercise and therefore some fatigue could be involved. We discuss first the extent and nature of fatigue for the different activities and then how caffeine may exert its performance enhancing effects by reducing fatigue or enhancement when the subjects were fresh.
The nature of the rugby test was one of repeated high-intensity exercise, with fatigue being apparent in every type of activity in the placebo condition (Fig. 1). The rugby test involved a much greater volume of supramaximal exercise (~14 min) compared with other studies on intermittent exercise, yet compared with typical endurance studies the volume was much less. The etiology of fatigue in repeated intermittent exercise could in principle involve several factors (16). It should also be noted that fatigue during any one activity in the rugby test is likely to depend on the cumulative effects imposed by all other previous physical activities. Fatigue mechanisms that are plausible for rugby include changes of muscle high-energy phosphates (16), acidosis (13,16), lowered muscle glycogen levels (16,17), rundown of electrolyte gradients such as for potassium (K+) (17,24), and/or a reduced motor drive from the CNS, i.e., central fatigue (11,22).
With the sprint tasks, an astounding finding was that in the very last circuit the subjects were able to restore sprint times relative to that of the first circuit (or even improve the time). This observation (Fig. 1) strongly suggests that the working muscles could still function maximally and therefore the slowing of sprints over the course of the test resulted from a diminished motor drive from the CNS. The slowing may be related to an inability to sustain exercise at the high levels of perceived exertion that occur in the latter stages of intermittent high-intensity shuttle running (10,14,26) or with endurance exercise (9).
The magnitude of fatigue in the drives (Fig. 1) is comparable to the reduction in peak MVC force (~10%) seen for leg muscles after a soccer test (28). Indeed, the drives are likely to require near maximal recruitment of motor units in the working muscles as is necessary for MVC. Several neural or muscular mechanisms already alluded to could be involved in fatigue of the drives but the present study cannot discriminate between the possibilities. The fact that peak power for Drive 2 was always less than for Drive 1 (by 10–15%) throughout the test and even in the last circuit (Table 1, Fig. 1), could be explained by a substantial contribution from processes in muscle and may involve high-energy phosphates that had not recovered in the 5-s rest period between drives (16).
Impairment of the ball-passing accuracy task (Fig. 1) could involve either a loss of concentration or another effect via the CNS that diminishes coordinated motor drive to the muscle groups involved in passing the ball. Other studies involving intermittent high-intensity exercise tests have revealed a loss of ground-stroke hitting accuracy in a tennis performance test (10) or diminished ability to dribble a soccer ball quickly (26). However, the impaired performance in both of these tests may have been due to players moving more slowly, which did not contribute to the skill impairment in the present study.
Caffeine is known to influence many processes that could explain performance enhancement. These processes include several aspects of excitation–contraction coupling in skeletal muscles (1,5), attenuation of the rundown of the K+ gradient during exercise (18,24), mobilization of free fatty acids from adipose tissue (4,9), slowing of muscle glycogen depletion (15), and modifications to CNS activity (11,22). Moreover, caffeine may act indirectly via inhibition of adenosine receptors (11,15,22), inhibition of cyclic AMP phosphodiesterase activity (15), elevation of the plasma epinephrine concentration (9,24), or via theophylline or paraxanthine the breakdown products of caffeine (15,18). The only unlikely mechanisms include some of the excitation contraction coupling processes (1,5) and phosphodiesterase inhibition (15), which require much higher concentrations of caffeine than in the present study.
We looked for a possible role of epinephrine but did not systematically investigate other mechanisms. In the placebo condition, the plasma epinephrine concentration exceeded 2 nmol·L−1 throughout the rugby test (Fig. 2), which is comparable to that seen with intense exercise (17,21). Also, in the caffeine condition, plasma epinephrine was elevated both at rest and throughout the test (Fig. 2), making it a potential mediator for the enhancements. However, the correlation analysis revealed that epinephrine could be a mediator of performance improvement but only for the sprint activities. On the contrary, the higher epinephrine levels tended to dampen the beneficial effects of caffeine with the drives and the passing accuracy task.
We attributed the fatigue during repeated sprints to impaired motor drive. Caffeine could act either by attenuating this central fatigue or by acting directly on muscle. However, because the sprint time in the last circuit showed no fatigue in the placebo condition and caffeine did not improve this performance measure (Fig. 1), we propose that the entire effect of caffeine on the earlier sprints is mediated via the CNS. If this is the case, then one explanation could be that caffeine inhibits the binding of adenosine to adenosine receptors in the brain (11,15) and thereby reduces perception of exertion (9,14). Another possibility, based on the positive correlation between changes in plasma epinephrine and sprints speed, is that a restoration of motor drive to the exercising leg muscles with caffeine may actually be mediated by epinephrine acting on the CNS.
Caffeine may have improved Drive 1 performance through increased motor drive (23), changes in metabolism (9,15), or direct effects on muscle (24,25,31). This effect is unlikely to be mediated via epinephrine acting on muscle to potentiate excitation-contraction coupling (5) because the correlation analysis suggested that elevated epinephrine is, if anything, detrimental for this response. The mechanism for the decline in performance of Drive 2 with caffeine is uncertain. Nevertheless, this negative effect is of minor concern given the notable gain in power output with caffeine during Drive 1.
It is difficult to imagine that the improved passing accuracy with caffeine would involve anything other than an effect somewhere in the motor control areas in the CNS. This effect could involve an increased level of attention (19) or arousal (15). The negative correlation between change in epinephrine concentration and change in performance suggests that epinephrine attenuates this effect. This effect of caffeine is likely to be important in competitive sports where the ability to perform skills like passing balls, hitting balls, or shooting goals with accuracy late in the event is a key to successful performance.
Caffeine provides several valuable performance-enhancing effects on simulated intermittent high-intensity team-sport performance, making it a potentially useful supplement for games such as rugby, football, soccer, hockey, basketball, and tennis. Although the mechanisms for the effects of caffeine are not fully understood, we speculate that caffeine influences several processes in the CNS to reduce fatigue with repeated sprint and permit a higher level of motor drive and motor skills throughout games.
1.Allen, D. G., and H. Westerblad. The effects of caffeine on intracellular calcium, force and the rate of relaxation of mouse skeletal muscle. J. Physiol.
2.Altman, D. G., K. F. Schultz, D. Moher, et al. The revised CONSORT statement for reporting randomized trials: explanation and elaboration. Ann. Intern. Med.
3.Bell, D. G., I. Jacobs, and K. Ellerington. Effect of caffeine and ephedrine ingestion on anaerobic exercise performance. Med. Sci. Sports Exerc.
4.Bruce, C. R., M. E. Anderson, S. F. Fraser, et al. Enhancement of 2000-m rowing performance after caffeine ingestion. Med. Sci. Sports Exerc.
5.Cairns, S. P., and A. F. Dulhunty. β-adrenoceptor activation slows high-frequency fatigue in skeletal muscle fibers of the rat. Am. J. Physiol.
6.Cohen, J. Statistical Power Analysis for the Behavioral Sciences
, 2nd ed. Hillsdale, NJ: Lawrence Erlbaum, 1988, p. 25.
7.Collomp, K., S. Ahmaidi, M. Audran, J.-L. Chanal, and C. Prfaut. Effects of caffeine ingestion on performance and anaerobic metabolism during the Wingate Test. Int. J. Sports Med.
8.Cook, C. Rapid non invasive measurement of hormones in transdermal exudates and saliva. Physiol. Behav.
9.Costill, D. L., G. P. Dalsky, and W. J. Fink. Effects of caffeine ingestion on metabolism and exercise performance. Med. Sci. Sports Exerc.
10.Davey, P. R., R. D. Thorpe, and C. Williams. Fatigue decreases skilled tennis performance. J. Sports Sci.
11.Davis, J. M., Z. Zhao, H. S. Stock, K. A. Mehl, J. Buggy, and G. A. Hand. Central nervous system effects of caffeine and adenosine on fatigue. Am. J. Physiol.
12.Deutsch, M. U. The Physiology of Rugby: Effects of Playing Position, Playing Level and Creatine Supplementation
. Dunedin: University of Otago, School of Physical Education, 2001.
13.Deutsch, M. U., G. J. Maw, D. Jenkins, and P. Reaburn. Heart rate, blood lactate and kinematic data of elite colts (under-19) rugby union players during competition. J. Sports Sci.
14.Doherty, M., P. M. Smith, M. G. Hughes, and R. C. Davison. Caffeine lowers perceptual response and increases power output during high-intensity cycling. J. Sports Sci.
15.Graham, T. E. Caffeine and exercise: metabolism, endurance and performance. Sports Med.
16.Green, H. J. Mechanisms of muscle fatigue in intense exercise. J. Sports Sci.
17.Greer, F., C. McLean, and T. E. Graham. Caffeine, performance, and metabolism during repeated Wingate exercise tests. J. Appl. Physiol.
18.Hawke, T. J., R. G. Willmets, and M. I. Lindinger. K+ transport in resting rat hind-limb skeletal muscle in response to paraxanthine, a caffeine metabolite. Can. J. Physiol. Pharmacol.
19.Hogervorst, E., W. J. Riedel, E. Kovacs, F. Brouns, and J. Jolles. Caffeine improves cognitive performance after strenuous physical exercise. Int. J. Sports Med.
20.Hopkins, W. G. Probabilities of clinical or practical significance. Sportscience 6:sportsci.org/jour/0201/wghprob.htm
21.Jackman, M., P. Wendling, D. Friars, and T. E. Graham. Metabolic, catecholamine, and endurance responses to caffeine during intense exercise. J. Appl. Physiol.
22.Kalmar, J. M., and E. Caffarelli. Caffeine: a valuable tool to study central fatigue in humans? Exerc. Sport Sci. Rev.
23.Kalmar, J. M., and E. Caffarelli. Effects of caffeine on neuromuscular function. J. Appl. Physiol.
24.Lindinger, M. I., T. E. Graham, and L. L. Spriet. Caffeine attenuates the exercise-induced increase in plasma [K+] in humans. J. Appl. Physiol.
25.Lopes, J. M., M. Aubier, J. Jardim, J. V. Aranda, and P. T. Macklem. Effect of caffeine on skeletal muscle function before and after fatigue. J. Appl. Physiol.
26.McGregor, S. J., C. W. Nicholas, H. K. A. Lakomy, and C. Williams. The influence of intermittent high-intensity shuttle running and fluid ingestion on the performance of a soccer skill. J. Sports Sci.
27.Paton, C. D., W. G. Hopkins, and L. Vollebregt. Little effect of caffeine ingestion on repeated sprints in team-sport athletes. Med. Sci. Sports Exerc.
28.Rahnama, N., T. Reilly, A. Lees, and P. Graham-Smith. Muscle fatigue induced by exercise simulating the work rate of competitive soccer. J. Sports Sci.
29.Sterne, J. A. C., and G. D. Smith. Sifting the evidence–what’s wrong with significance tests. BMJ
30.Tarnopolsky, M. A. Caffeine and endurance performance. Sports Med.
31.Tarnopolsky, M. A., and C. Cupido. Caffeine potentiates low frequency skeletal muscle force in habitual and nonhabitual caffeine users. J. Appl. Physiol.