European Journal of Applied Physiology

, Volume 110, Issue 6, pp 1243–1250

Caffeinated chewing gum increases repeated sprint performance and augments increases in testosterone in competitive cyclists


    • Health and Sport ScienceEastern Institute of Technology
  • Timothy Lowe
    • School of Applied ScienceBay of Plenty Polytechnic
  • Athena Irvine
    • School of Sport and Exercise ScienceWaikato Institute of Technology
Original Article

DOI: 10.1007/s00421-010-1620-6

Cite this article as:
Paton, C.D., Lowe, T. & Irvine, A. Eur J Appl Physiol (2010) 110: 1243. doi:10.1007/s00421-010-1620-6


This investigation reports the effects of caffeinated chewing gum on fatigue and hormone response during repeated sprint performance with competitive cyclists. Nine male cyclists (mean ± SD, age 24 ± 7 years, VO2max 62.5 ± 5.4 mL kg−1 min−1) completed four high-intensity experimental sessions, consisting of four sets of 30 s sprints (5 sprints each set). Caffeine (240 mg) or placebo was administered via chewing gum following the second set of each experimental session. Testosterone and cortisol concentrations were assayed in saliva samples collected at rest and after each set of sprints. Mean power output in the first 10 sprints relative to the last 10 sprints declined by 5.8 ± 4.0% in the placebo and 0.4 ± 7.7% in the caffeine trials, respectively. The reduced fatigue in the caffeine trials equated to a 5.4% (90% confidence limit ±3.6%, effect size 0.25; ±0.16) performance enhancement in favour of caffeine. Salivary testosterone increased rapidly from rest (~53%) and prior to treatments in all trials. Following caffeine treatment, testosterone increased by a further 12 ± 14% (ES 0.50; ± 0.56) relative to the placebo condition. In contrast, cortisol concentrations were not elevated until after the third exercise set; following the caffeine treatment cortisol was reduced by 21 ± 31% (ES −0.30; ± 0.34) relative to placebo. The acute ingestion of caffeine via chewing gum attenuated fatigue during repeated, high-intensity sprint exercise in competitive cyclists. Furthermore, the delayed fatigue was associated with substantially elevated testosterone concentrations and decreased cortisol in the caffeine trials.




Caffeine is recognised as the World’s most commonly used drug and is frequently used by athletes as a nutritional ergogenic aid during training and competition. Indeed, in a recent scientific review of caffeine use in sports, Sokmen et al. (2008) provided athletes with a range of practical recommendations relating to the form, timing of ingestion and dosages required to enhance athletic performance.

A substantial (and growing) body of evidence supports caffeine’s ergogenic effects and its potential to enhance performance over a range of physical activities (see comprehensive reviews by Astorino and Roberson 2010 and Warren et al. 2010). While the improvements in exercise performance with caffeine have been predominantly associated with steady-state endurance exercise over a range of events including swimming, cycling, and running (Graham 2001), there is a growing interest in caffeine’s ability to enhance performance in shorter duration and repeated high intensity activities. However, the effects of caffeine on singular or repeated bouts of short-term high-intensity exercise and the mechanisms responsible for performance enhancement remain equivocal (Davis and Green 2009). Caffeine ingestion did not improve performance in subjects undertaking single (Collomp et al. 1991) or repeated Wingate tests (Greer et al. 1998), or in team sport athletes performing 10 repeated maximal sprints (Paton et al. 2001). In contrast, caffeine did enhance performance in a single 1 km time trial with trained cyclists (Wiles et al. 2006), in several tests of rugby performance (Stuart et al. 2005) and in repeated 12 × 30 m repeated running sprints (Glaister et al. 2008).

Several potential mechanisms exist to explain caffeine’s performing enhancement effects during exercise. These mechanisms have been extensively reviewed for both aerobic (Graham 2001) and more recently anaerobic activities (Davis and Green 2009). The postulated mechanisms suggested for caffeine’s ergogenic effect in short-term high intensity activity includes increases in adenosine antagonism, hormonal stimulation and enhanced muscular excitation–contraction coupling. An area of particular interest is the effects of caffeine on the hormonal environment, and in particular its effect on the steroid hormones testosterone and cortisol. Pollard (1988) reported that caffeine ingestion acutely increases levels of testosterone in the rodent model. Similarly, caffeine has been shown to increase levels of cortisol in human subjects (Lin et al. 1997; Lovallo et al. 2005). In a recent study, using a circuit training programme of 1 h duration, caffeine was found to elevate salivary testosterone compared to a placebo treatment and at high doses also elevate cortisol (Beaven et al. 2008). The exact mechanism underlying the increase in testosterone in Beaven et al. (2008) study is unclear, however; the authors suggest that caffeine may produce an antagonistic effect on adenosine receptors in the central nervous system (CNS) and thus facilitate an increase in efferent motor command which enhances muscular work output. Since the release of testosterone in response to exercise is determined to some extent by the intensity of exercise (Kraemer et al. 1990, 2005) any increase in work output has the potential to further increase testosterone release. This response is of interest as testosterone is primarily an anabolic hormone, and chronic increases are linked with strength and muscle gains, moreover, there is also evidence that acute increases of testosterone in males have the potential to increase muscular force production through direct effects on the nervous system (Kraemer and Ratamess 2005).

Caffeine is typically administered in studies via the ingestion of tablets or capsules containing anhydrous caffeine. However, an alternative novel delivery method via chewing gum may provide some additional benefits to athletes. Drug delivery via chewing gum is widely accepted by the medical community and has been successfully used as a delivery medium for nicotine, aspirin, methadone, as well as caffeine. Kamimori et al. (2002) investigated the absorption rates of caffeine administered in capsules versus chewing-gum; the two delivery methods produced similar rates of relative bioavailability, although absorption rates were significantly higher with chewing gum. A faster absorption rate with chewing gum is one of the advantages of this drug delivery method and may prove beneficial in sporting situations. Chewing gum allows caffeine to be absorbed directly into the blood stream through the buccal mucosa (sublingual) thereby bypassing hepatic metabolism, and minimising the risk of gastrointestinal distress. It may also be possible to produce an equivalent effect with a lower dose as relative bioavailability is maintained (Rassing 1994).

Given the limited and inconsistent findings in the literature regarding the effects of caffeine on repeated high-intensity, intermittent exercise, the purpose of this study was to investigate the effects of caffeine, delivered via a novel method (chewing gum), on fatigue during repeated high-intensity sprints in well-trained cyclists. Furthermore, in an attempt to gain an understanding of the effects of caffeine on hormonal activity during intense exercise the acute salivary cortisol and testosterone responses were also investigated.



The study was a balanced, placebo controlled, double blind, cross-over trial. Ten subjects initially started the study but one subject failed to finish due to an illness unrelated to the study.


Nine well-trained male cyclists completed this study. The cyclists (mean ± SD) weight, height, age, maximum oxygen consumption, and peak aerobic power output were 79.3 ± 10.6 kg, 182 ± 6 cm, 24.1 ± 7.3 years, 62.5 ± 5.4 mL kg−1 min−1, and 351 ± 32 W, respectively. At the time of the study, the participants completed an average of 10.3 ± 4.0 h of cycling training per week. The investigation was conducted during the cyclists’ pre-competition phase of training. All cyclists were informed of the purpose and risks associated with participation before giving their written informed consent. The study was approved by the Institutes human ethics committee.

Dietary control

Prior to commencing the study subjects were provided with a comprehensive list of common dietary caffeine sources including food, beverages, medicines and supplements; subjects were asked to identify any caffeinated products that they regularly consumed as part of their diet. Cyclists were also required to complete a 3-day food diary in order to confirm that they were habitually low to moderate (<300 mg per day) caffeine users; regular high dose (>500 mg per day) caffeine users were excluded from the study. During the 24 h prior to the experimental trials, cyclists were instructed to prepare for the session as though it were a competition, and to abstain from caffeine consumption. Cyclists used a 24-h food diary to replicate their diet as closely as possible preceding each experimental session.

Exercise performance tests

All cyclists had previously participated in laboratory cycle-ergometer testing and were familiar with general exercise testing procedures. Cyclists reported to the laboratory on six separate occasions, over 3–4 weeks. During the initial visit to the laboratory, cyclists completed an incremental exercise test to exhaustion to determine their maximum aerobic capacity (VO2max), and aerobic peak power output (PPO). During the remaining laboratory sessions, athletes completed an experimental familiarisation trial followed by four experimental high-intensity interval sessions after receiving either placebo, or caffeine treatments. All testing was conducted in a laboratory under controlled environmental conditions. Air temperature and relative humidity were regulated to ~21°C and 50–60%, respectively. Testing was conducted using an adjustable friction-braked cycle ergometer (CycleOps Pro 300PT, Saris Cycling Group Inc., Madison, WI, USA). A Power tap hub integrated into the wheel of the ergometer measured and recorded power output at 1 s intervals throughout all experimental tests.

Maximal incremental exercise test

Initially, the cycle ergometer was adjusted to a position which was comfortable and resembled the set up of the subjects own bicycles. The selected dimensions were recorded, and replicated for subsequent tests. Cyclists performed a 10-min warm-up at a self-selected intensity followed by 5 min at a constant power of 100 W. Thereafter, power output was increased at a rate of 25 W min−1 until the cyclist reached volitional exhaustion. Oxygen uptake (VO2) was measured continuously with a calibrated metabolic cart (Vmax29, SensorMedics, Yorba, CA, USA). Maximum oxygen consumption (VO2max) was defined as the highest VO2 measured over a 60-s period during the test. PPO was defined as the highest continuous 60-s mean power output recorded on the ergometer during the test.

Experimental training sessions

All cyclists completed a familiarisation session which replicated the experimental procedure but without receiving any treatment. Following the familiarisation session, each cyclist completed four high-intensity experimental sessions. Each cyclist completed two trials under experimental (caffeine) and control (placebo) conditions. Half of the group commenced their experimental trials with the caffeine condition the remainder with the placebo condition; following each subjects first session their remaining trials were alternated between the two conditions. All sessions were conducted in the laboratory using the same friction-braked ergometer previously described. Experimental sessions were conducted at the same time of day for each individual in order to control for diurnal variation in hormone concentration. Experimental trials were separated by 3–5 days. Cyclists were requested to complete only light training of short duration (<90 min) in the 24 h preceding a session and were required to present in a euhydrated and non-carbohydrate depleted state. Throughout the experimental sessions, cyclists were cooled with standing floor fans and permitted to consume only water ad libitum.

Upon arrival to the laboratory, cyclists produced a baseline saliva sample, and completed a 10 min, self-paced, sub-maximal warm-up. To aid consistency all experimental sessions were controlled via pre-recorded audio cues indicating the beginning and end of each training set and recovery periods. The experimental procedure is outlined diagrammatically in Fig. 1. The high-intensity interval training session consisted of four exercise sets each of 5 min duration. During each set, cyclists performed five 30-s maximal effort sprints separated with 30-s active recovery periods. Following sets one and three, cyclists completed a 5-min active recovery period at ~100 W. Experimental treatments, of either caffeine or placebo, were provided to the cyclist during a 10-min recovery period after completion of the second exercise set. No performance information was made available to the cyclists during the experimental trials. Immediately following the interval training session cyclists were asked whether they thought they had received the caffeine or placebo dose to determine if they could accurately predict the provided treatment.
Fig. 1

Overview of the experimental trial design indicating exercise sets, with associated treatment and sampling times

Caffeine administration

Immediately following the second exercise set, after the collection of a second saliva sample, cyclists received either the placebo or caffeine treatment. Caffeine was administered as an absolute dose (240 mg), as six pieces of commercially available (spearmint flavoured) caffeinated chewing-gum (Jolt® caffeine-energy gum, Gum Runners, Hackensack, NJ, USA). The placebo was a similar looking and tasting, commercially available non-caffeinated chewing-gum (Spearmint Extra® professional, Wrigley’s, Chicago, IL, USA). Cyclists chewed the gum for 5 min and were then required to expectorate the chewed gum into a container. Verbal questioning during test trials indicated that athletes were unable to accurately distinguish between the caffeine and non-caffeine-containing chewing gums.

Saliva collection and analysis

Saliva samples were collected upon arrival to the laboratory (baseline), and immediately following each set of five sprints. Prior to giving the first sample, cyclists were instructed to rinse their mouths with water. Cyclists chewed one piece of sugar-free chewing-gum (Extra® Original, Wrigley’s, Chicago, IL, USA) for 30-s to aid salivation. Cyclists swallowed the initial saliva cud, and passively drooled into a 10 mL collection tube until they had produced approximately 2 mL of sample. A further four samples were collected immediately following sets one to four. Saliva samples were immediately frozen at −20°C until radioimmunoassay (RIA). Saliva samples were analysed in triplicate for testosterone and cortisol using radioimmunoassay. The methods were modified from those described by Granger et al. (1999) and Morelius et al. (2004). Briefly, standards from serum diagnostic kits (Diagnostic Systems Laboratories, USA) were diluted in phosphate buffer saline (Sigma P4417) to cover the range 0–500 and 0–51.2 nmol L−1, for cortisol and testosterone, respectively. Saliva sample sizes of 50 and 100 μl were used for cortisol and testosterone, respectively. The antibodies were diluted in a phosphate-buffered saline solution containing 0.05% bovine serum albumin. Kit standards were diluted so that approximately 50% binding was achieved compared with the total counts. These were 10,000 and 4,500 CPM for cortisol and testosterone, respectively. Detection limits for the assays were 0.55 and 0.035 nmol L−1, for cortisol and testosterone, respectively. The intra-assay coefficients of variation were 5.6% for cortisol and 6.4% for testosterone.

Statistical analysis

Simple descriptive statistics are shown as mean ± between-subject standard deviations. To reduce random error, the means of the two repeat trials for each condition were used for subsequent data analysis. Mean effects of caffeine, on performance and hormones, and the 90% confidence limits (CL) were estimated with a spreadsheet (Hopkins 2003) via the unequal-variances t statistic computed for change scores between caffeine and placebo conditions. All measures were log transformed to reduce bias arising from non-uniformity of error and back transformed to obtain changes in means as percents. Performance effects were analysed by comparing changes in mean power between the first (combined set 1 and 2) and second half (combined set 3 and 4) of the experimental session (pre-treatment vs. post-treatment).

Effects of caffeine on performance and salivary hormone concentrations were expressed in raw units, as percentage changes, and as standardised effects. Magnitudes of the standardised effects were interpreted and reported using the modified Cohen effect thresholds of: 0.2, 0.5, and 0.8 for small, moderate, and large, respectively, in accordance with the recommendations and guidelines of Hopkins (2002). The smallest substantial change in cycling sprint performance was assumed to be a reduction or increase in sprint time of >1% in accordance with the recommendations of Paton and Hopkins (2006).


Power output

Mean power output for each exercise set is shown in Fig. 2a, and the relative difference in power output between placebo and caffeine treatments is shown in Fig. 2b. Mean power output decreased from the first to second exercise sets in both the placebo and caffeine treatment groups with only trivial differences between the treatments. The main finding for power output was that following the chewing of caffeinated gum, subjects showed a mean percentage decrease (±90% CL) of 0.4%; ±7.7% over the third and fourth exercise sets, whereas in the placebo trial the respective decrease was 5.8%; ±4.0%. The difference between the observed reduction in power output between the placebo and caffeine trials was 5.4%; ±3.6% in favour of caffeine. The observed effect of caffeine on power output was equivalent to a small effect size of 0.25; ±0.16.
Fig. 2

a The mean ± SD power output for the two pre- and two post-treatment exercise sets. b The relative percent change in power output between placebo and caffeine treatments. Data are mean and 90% CL. The grey-shaded area represents the threshold for the smallest estimated worthwhile effect

Salivary testosterone

Mean testosterone concentration following each exercise set is shown in Fig. 3a, and the relative difference in testosterone between treatments between sets is shown in Fig. 3b. There was an initially rapid increase in testosterone of 57 ± 27% for the placebo, and 48 ± 26% for the caffeine treatments above resting values following the first two exercise sets (Fig. 3a). Subsequently, there were further increases during the third and fourth exercise sets with values above resting of 89 ± 28% for the placebo, and 100 ± 23% in the caffeine treatment. The main finding for testosterone was that following the chewing of caffeinated gum, subjects experienced a mean increase of 12%; ±14% (90% CL) relative to the placebo condition with a moderate effect size of 0.50; ± 0.56 (90% CL).
Fig. 3

a The between subject testosterone concentrations for resting levels and the two pre- and two post-treatment exercise sets. b The percent change of testosterone concentration between placebo and caffeine treatments. The grey-shaded area represents the threshold for the smallest estimated worthwhile effect

Salivary cortisol

Mean cortisol concentration following each exercise set is shown in Fig. 4a, and the relative difference in cortisol between treatments is shown in Fig. 4b. There were only minimal increases in cortisol of 5 ± 34% for the placebo, and 4% ± 33% for the caffeine treatments above resting values following the first two exercise sets. Subsequently, there were increases during the third and fourth exercise sets with values above resting of 83 ± 93% for the placebo, and 42 ± 93% for caffeine conditions. The main finding for cortisol was that following the chewing of caffeinated gum, subjects showed a relative decrease of 21%; ±31% compared to placebo, equivalent to a moderate effect size of −0.30; ±0.34.
Fig. 4

a The between subject cortisol concentrations for resting levels and the two pre- and two post-treatment exercise sets. b The percent change of cortisol concentration between placebo and caffeine treatments. Data are mean and 90% CL. The grey-shaded area represents the threshold for the smallest estimated worthwhile effect


The key findings in this study were that caffeine, administered via a novel method, led to substantial improvements in performance during repeated sprint cycling by reducing the rate of accumulated fatigue over four exercise sets. The decrease in fatigue following caffeine ingestion was associated with moderate increases in testosterone of ~12% whereas cortisol showed a moderate decrease of ~21% relative to the placebo condition.

The gains in performance following caffeine ingestion were relatively small (equivalent to an enhancement in power output of ~5.5%) but based on previous research (Paton and Hopkins 2006) of a magnitude likely to provide a worthwhile enhancement in real competition, especially in events requiring repeated high-intensity efforts such as road race, criterium and mass start track cycling events. The magnitude of improvement in performance following caffeine ingestion in our study is consistent with the findings of two other recent caffeine studies. Schneiker et al. (2006) reported an increase in mean and peak power of ~7% with team sport athletes completing sets of exercise, consisting of eighteen, 4-s sprints on a cycle ergometer. Similarly, Stuart et al. (2005), reported performance improvements of up to 5% in a variety of rugby-specific drills, including sprint, force, power, and motor skill-oriented tasks following caffeine ingestion. Stuart et al. (2005) reported that caffeine appeared protective against fatigue, as the ergogenic effect became more apparent in the latter half of repeated tests. Therefore, the results of our current study lend further support to the role of caffeine as a performance enhancing substance which reduces the rate at which fatigue accumulates.

During the placebo trials in our study, there was a clear and gradual reduction in sustainable power of ~6% over the first three exercise sets, though interestingly power output tended to recover slightly during the last exercise set. The recovery in power output during this final set suggests that the predominant fatigue mechanism was not related to muscle acidosis or damage, but most likely due to a reduction in efferent motor command (Enoka and Stuart 1992) and central drive (Stuart et al. 2005). Power output may have been regulated in response to efferent command signals or ‘teloanticipation’ due to the ‘closed loop’ nature of the training session (Billaut et al. 2005). Therefore, as the fatigue observed during the current investigation was probably caused by reduced central drive, it is likely that the performance enhancing effect of caffeine involved mechanisms within the CNS. Potential cellular mechanisms responsible for CNS stimulation include the inhibition of phosphodiesterase (Graham 2001), mobilisation of intracellular calcium, and the inhibition of adenosine (Nehlig and Debry 1994). It is uncertain which if any of these mechanisms led to the reduction in fatigue in our study; however, previous authors’ (Doherty et al. 2004) have proposed that the inhibition of adenosine is responsible for the alterations in the rating of perceived exertion and subsequent performance enhancement during repeated cycling sprint activity. Adenosine is regulated by ATP metabolism and inhibits the release of excitatory neurotransmitters such as dopamine (Fredholm et al. 1999). Caffeine is similar in structure to adenosine; as it is lipophilic, it easily crosses the blood brain barrier and binds adenosine receptors. A reduction in adenosine activity could result in greater motor unit recruitment thereby enhancing power output (Williams 1991), and reducing the perception of effort for a given workload (Doherty et al. 2004).

Further evidence for the role of adenosine as the mediator of performance enhancement in our study is suggested by the greater testosterone response observed following the administration of caffeine. Acute increases in testosterone are reportedly related to exercise selection and intensity, nutritional status and experience of the athlete (Linnamo et al. 2005). Testosterone responses are also related to the mechanical stimulus of the muscle; that is, exercise protocols which involve a greater amount of active muscle mass, with higher loading (greater time under tension or force) elicit a greater testosterone response (Kraemer and Ratamess 2005). It is possible that caffeine acted on the CNS to stimulate a greater recruitment of motor neurons, which resulted in a greater degree of active muscle mass. Subsequently, a greater force was produced and testosterone concentration was elevated. Additionally, an acute increase in testosterone can directly influence force production by facilitating neurotransmitter release (Nagaya and Herrera 1995). Thus, the ingestion of caffeine, together with acute elevations in plasma testosterone may act synergistically on the CNS to directly increase force production and improve performance.

Interestingly, we found a large and rapid initial increase in testosterone following just the first exercise bout; this is a novel finding. The release of testosterone in response to exercise is now routinely reported and exercise protocols which involve a greater amount of active muscle mass, with higher loading (greater time under tension or force) generally elicit a greater testosterone response (Kraemer and Ratamess 2005), but these studies report testosterone changes typically at the conclusion of the exercise session. In the current study, testosterone was elevated by ~50% within 5 min of initialising the first intense exercise bout. This result is of interest as this timing indicates that testosterone is being elevated by mechanisms other that the classical hypothalamus–pituitary–gonadal axis where the endocrine response has been reported to have lag phase of ~40 min from stimulus to increased concentration of testosterone (Spratt et al. 1988). Recent research in vivo (Selvage et al. 2004) has identified direct neural links between the para-ventricular nucleus of the hypothalamus and the testes, and this adds further supports to the idea that the rapid increase in testosterone seen here could have been elicited via a direct neural pathway.

In contrast to the rapid rise in testosterone, cortisol did not increase substantially in either the placebo or caffeine condition until after the third exercise set. The timing of this increase is in context with the dynamics of cortisol release via the hypothalamus–pituitary–adrenal axis. Interestingly, caffeine ingestion led to a blunted cortisol increase relative to the placebo treatment; this may benefit the athlete as previous research has reported that suppressing cortisol during exercise is associated with greater gains in muscle mass (Bird et al. 2006). A decrease in cortisol has also been reported when athletes perform 60 min of resistance training exercise, following supplementation with low doses (200 mg) of caffeine; however, at higher caffeine doses (400–800 mg) cortisol showed large increases (Beaven et al. 2008). The physiological reason for this relative reduction in cortisol following caffeine ingestion is unclear as we are unaware of any studies that indicate a direct interaction between caffeine and cortisol.

The unique aspect of this study was the administration of the caffeine dose via a commercially available chewing gum. The caffeine chewing gum utilised in this study was indistinguishable from most regular non-caffeinated chewing gums and had the additional advantage of causing no symptoms of gastrointestinal distress in our subject sample. The use of chewing gum appears an effective and convenient method of caffeine ingestion for athletes. Administration of caffeine by this method may be particularly advantageous for team sport athletes during game period breaks or for endurance athletes requiring a rapid enhancement in performance during the final phases of competition.

In conclusion, results of the study show that a moderate dose (~3 mg kg−1) of caffeine delivered by chewing gum is ergogenic and can delay fatigue during repeated high-intensity, intermittent sprint exercise. The mechanisms for the actions of caffeine are unclear, although the results suggest that caffeine may act on the CNS, possibly by increasing central drive. Further, caffeine ingestion has a positive effect on the testosterone to cortisol ratio and may provide additional benefits by enhancing muscles anabolic processes. The administration of caffeine during the later stages of exercise may help to attenuate fatigue associated with decreased neural drive.


The authors gratefully acknowledge the funding provided by the Waikato Institute of Technology to enable this study, and the assistance of Dr. David Rowlands on manuscript preparation.

Conflict of interest

The authors received no assistance from any commercial company whose products were used in the study, and report no conflict of interest with this study and its results.

Copyright information

© Springer-Verlag 2010