A large body of evidence has shown ingestion of caffeine in doses ranging from 3 to 13 mg·kg−1 body weight to be ergogenic for endurance exercise performance [for review, see Burke (13), Doherty and Smith (19), Ganio et al. (21), and Graham (25)] and, to a lesser degree, for short-term high-intensity anaerobic exercise performance [for review, see Astorino and Roberson (2) and Davis and Green (18)]. Despite these findings, the mechanism(s) underlying caffeine’s ergogenic effect(s) remain unclear. Costill et al. (16) initially proposed caffeine-enhanced endurance performance by promoting fat oxidation and sparing muscle glycogen. More recent evidence casts doubt on this hypothesis (27) because enhanced fat oxidation does not explain improvements in short-term high-intensity exercise where carbohydrate availability would not limit performance. Ingestion of lower doses of caffeine (2–3 mg·kg−1 of body weight) has also been shown to improve endurance performance (32) when no changes in fat oxidation would be expected. As such, focus has shifted to caffeine’s action within the central and peripheral nervous systems, which could alter sensations of effort (RPE) (19) and muscle pain (4,5,22–24,32,38,39,42) as well as the ability of the skeletal muscle to generate force [for review, see Warren et al. (51)].
Caffeine antagonizes both central and peripheral A1, A2a, and A2b adenosine receptors (1,46), which are known to be involved in nociceptive pathways (46). Ingestion before exercise has been shown to result in hypoalgesia of low- to moderate-intensity muscle pain during fixed-intensity exercise (i.e., exercise at a fixed percentage of V˙O2peak) ranging from ∼60% to 80% of V˙O2peak (22–24,38,39,42). Interestingly, hypoalgesia has not been observed during resistance exercise (3,28) or during fixed-time (32) and fixed-distance endurance time trials (5,6). Strikingly similar results have been shown for RPE. Caffeine ingestion attenuates RPE (8,16,17,19) during moderate, fixed-intensity exercise, but no effect has been observed during fixed-time (14,17,22,29,32) or fixed-distance cycling (5,6) and running (12,52) time trials. Despite a lack of effect on muscle pain and RPE during fixed-time and fixed-distance exercise, caffeine ingestion has consistently been ergogenic (5,6,12,14,17,22,29,32,52). On the basis of these findings, two plausible explanations present themselves: caffeine attenuates muscle pain and RPE during heavy and severe-intensity exercise common in endurance time trials, but this effect manifests as an increase in work performed at a given level of pain/effort because of participants pacing their work output based on sensations of pain and effort. Alternatively, the nociceptive inputs contributing to muscle pain and the host of factors contributing to RPE may be too great for antagonism of adenosine receptors alone to reduce pain and RPE. Thus, rather than reduced pain and/or RPE, some other mechanism (e.g., a direct effect on skeletal muscle or enhanced motor‐unit activation) or a combination of mechanisms (e.g., effects on pain/effort and effects on skeletal muscle and activation) may lead to improved endurance performance.
Caffeine has been shown to increase both cortical (35) and spinal (50) neuron excitability, which may plausibly explain the observed increases in motor-unit recruitment (34,35,44) and force generation during maximal voluntary contractions (MVC). In addition, millimolar concentrations of caffeine have been shown to enhance Ca2+ release from the sarcoplasmic reticulum via interaction with the ryanodine receptor (45) in vitro, suggesting that caffeine could enhance strength via a direct effect on skeletal muscle. Limited evidence supports a direct effect of micromolar concentrations in isolated muscle (30) (millimolar doses would be lethal) on skeletal muscle in vivo, but the effect seems to be most pronounced in fatigued muscle (17,37,49). Ascribing improved endurance performance to a specific mechanism(s) of action is difficult because of the limitations in the design of previous studies. Few studies have assessed muscle pain and RPE during endurance time trials (5,6,32), and to date, only one study has evaluated time-trial performance, muscular strength, and motor-unit recruitment after caffeine ingestion (17).
A recent meta-analysis (51) demonstrated that caffeine consistently increased motor-unit recruitment and MVC in the knee extensors but not in the smaller muscle groups of the arms. The lack of an effect in the smaller muscles of the arms was suggested (51) to be due to near-maximal (90%–99%) recruitment (48) before caffeine ingestion, whereas recruitment of the knee extensors is typically lower (85%–95%), providing a greater range for enhanced central excitability to manifest as an increase in motor-unit recruitment and thus improve strength. If improved endurance performance is due to increased strength, then it is plausible that caffeine would enhance performance during exercise primarily involving the knee extensors but not exercise primarily involving the arms. To our knowledge, no study has concomitantly assessed neuromuscular function (MVC, motor-unit recruitment, and muscle contractile properties), muscle pain, RPE, and endurance performance after ingestion of caffeine.
Two separate studies were performed. The purpose of study A was to examine the effects of a 5-mg·kg−1 dose of caffeine on leg and arm cycling time-trial performance, muscular strength, motor-unit recruitment, muscle contractile properties, and perceptions of muscle pain and effort. It was hypothesized that caffeine would improve leg but not arm cycling performance because of increased strength via enhanced motor-unit recruitment. The purpose of study B was to determine the effect of caffeine on muscle pain and RPE during heavy and severe fixed-intensity exercise, and it was hypothesized that caffeine would have no effect on the perceptions of pain and effort during exercise of this type.
For both studies, all participants were recreationally active, but not endurance trained, and all self reported a caffeine consumption of <40 mg·d−1. Participants were screened for medical or orthopedic conditions that would preclude performance of strenuous upper and lower body cycling. Potential participants who reported taking prescription pain and/or psychiatric medication as well as those who consumed >40 mg·d−1 of caffeine were excluded. Participants were asked to refrain from consuming any type of pain medication during the study and to refrain from consuming caffeine for at least 12 h before each testing session.
Sixteen healthy, college-age men and women initiated testing in the first study (study A). The experimental protocol for study A was approved by the Georgia College and State University’s Institutional Review Board, and participants gave written informed consent before study enrollment. Two female participants dropped out before completion of all testing sessions. One experienced severe cramping after completion of a testing session and voluntarily withdrew and the second was unable to complete a session because of dizziness and nausea. Twelve (six men and six women) college-age individuals participated in a follow-up study (study B)—none of which participated in study A. The protocol for study B was approved by the University of Mississippi and the University of Mississippi Medical Center Institutional Review Boards. For study A, data of 14 participants (9 women and 5 men) were used for analysis. A sample size of 14 was sufficient to detect a small (≥0.30 SD) difference in cycling performance using a two-tailed, dependent-measures t-test (43) at an α level of 0.05 and power of 0.80, assuming a correlation of ≥0.95 between repeated trials. Similarly, for study B, a sample size of 12 was sufficient to detect a small (≥0.25 SD) difference in quadriceps pain ratings using a repeated-measures ANOVA (43) at an α level of 0.05 and a power of 0.80, assuming a correlation of ≥0.90 between repeated trials.
A randomized, double-blind, placebo-controlled crossover design with each participant serving as his/her own control was used for both studies. Testing sessions for both studies occurred during the course of 2–3 wk with approximately 48-h intervals between sessions. All testing sessions were performed at approximately the same time of the day, with participants being instructed to eat a similar light meal (∼50% carbohydrates, 30% fat, and 20% protein) 2 h before each session. Participants were also encouraged to drink water liberally the day before testing and also consumed a 300-mL glass of water with caffeine/placebo capsule consumption. A schematic outline of the procedures for studies A and B can be seen in Figure 1.
In study A, participants completed eight (two familiarization and six experimental) testing sessions. During testing session 1, participants were screened, and consent documents were completed. They then completed a maximal exercise test on a cycle ergometer (Excalibur Sport; Lode BV, Groningen, The Netherlands), pedaling the cycle using their legs (V˙O2peak leg). During testing session 2, participants performed a maximal exercise test pedaling the ergometer using their arms (V˙O2peak arm). Arm crank exercise was performed on the same ergometer as leg cycling. Per the manufacturer’s instructions, the ergometer was rotated 90° such that its front (in leg cycling position) was resting on the ground. Participants were seated behind the ergometer, and seat height was adjusted so that the crank arms were at approximately shoulder height. The pedals were replaced by vertical mounted handgrips, and participants used a neutral grip position for all trials. During testing sessions 3 and 4, participants practiced the 10-min time-trial performance ride, pedaling the ergometer with the legs during testing session 3 and with the arms during testing session 4. Participants were also familiarized to the assessments of maximal voluntary isometric strength (MVC) and motor-unit recruitment in the knee extensors and elbow flexors across testing sessions 1–4. Testing sessions 5–8 were identical except for the muscle group used during exercise. Participants indicated their preference for arm or leg exercise for testing session 5, and then the muscle group was alternated for testing sessions 6–8. Each of these sessions lasted approximately 2.5 h. Participants reported to the laboratory and completed a 24-h history form to determine compliance with pretest diet, hydration, exercise, and pain medication instructions. MVC and muscle contractile properties were determined (PreCap) and then gelatin capsules containing either caffeine or placebo (all-purpose flour) were then consumed and participants rested quietly for 60 min. MVC and contractile properties were then reassessed (PostCap), and participants then cycled for 30 min at an intensity of ∼60% of previously determined V˙O2peak. Blood lactate was assessed immediately after exercise cessation. Next, participants performed a 10-min time-trial performance, and blood lactate levels were again assessed immediately after completion of the ride. Twenty minutes of quiet rest was provided, and then MVC and contractile properties were again assessed (PostEx).
During the initial testing session of study B, a leg cycling V˙O2peak test was performed. Testing sessions 2–5 were counterbalanced such that participants were randomly assigned to exercise twice at an intensity of 95% of gas exchange threshold (GET; heavy) or 70% of the difference between GET and of V˙O2peak (∼90% of V˙O2peak; severe); once after caffeine and once after placebo ingestion.
Assessment of V˙O2peak
For both studies A and B, graded cycling tests were performed on an electrically braked, computer-driven cycle ergometer to determine V˙O2peak. Participants were fitted to the ergometer and given instructions for providing RPE using the 6–20 Borg scale. Participants then performed a 5-min warm-up at a very light intensity, that is, 25 W (women) and 50 W (men) for leg cycling and 10 W (women) and 15 W (men) for arm crank cycling. After completion of the warm-up period, the initial work rate was set at 50 W (women) and 75 W (men) during leg cycling and 15 W (women) and 20 W (men) during arm crank cycling. Using a ramped protocol, work rate was progressively increased at 0.4 W·s−1 and 0.25 W·s−1 for leg and arm crank cycling, respectively. Participants cycled until volitional fatigue was reached. Strong verbal encouragement was provided throughout the exercise test. Oxygen uptake (V˙O2) and related gas exchange measures were assessed via open-circuit spirometry (TrueOne 2400 Metabolic Measurement System; Parvomedics, Sandy, UT) and averaged over 15-s intervals. The system was gas and flow calibrated according to manufacturer specifications. V˙O2 and V˙CO2 were standardized to STPD. HR was continuously measured using an HR monitor (Polar Electro Oy, Kempele, Finland). Work rate, HR, and RPE were recorded at the end of every minute during the test. Peak oxygen consumption was defined by the attainment of a plateau in oxygen uptake with an increase in work rate. In the absence of a plateau, peak oxygen consumption was defined as attainment of the following three criteria: 1) RER ≥ 1.1, 2) peak HR within 10 beats per minute of the age-predicted maximum for leg cycling and 20 beats per minute of the age-predicted maximum for arm crank cycling, and 3) peak RPE ≥ 18. All participants both studies A and B either demonstrated a plateau or met all three criteria during each test.
MVC and twitch interpolation
In study A, an interpolated-twitch electrical stimulation protocol was used to assess maximal isometric strength of the knee extensors (MVCKE) and the elbow flexors (MVCEF) and the percentage of motor-unit recruitment during MVC (%ACTKE and %ACTEF), essentially as described previously (17). All assessments were performed in the right knee extensor muscle group and the right elbow flexors. Knee extensor measurements were performed on a modified knee extension/leg curl machine (model GLCE-365; Body Solid, Forest Park, IL). Participants were seated with the hip at 90° of flexion, and the knee was fixed in a flexed position at an angle of 60° below horizontal. The lever arm of the machine was fixed to a force transducer (model SBO-750; Transducer Techniques, Temecula, CA) parallel to the line of pull and perpendicular to the lever arm, allowing for assessment of isometric torque. Inelastic straps were used to secure the participant’s right ankle to the lever arm. Stimulation electrodes (6.98 cm × 10.16 cm; Superior Silver; Unipatch, Wabasha, MN) were placed on the skin over the distal vastus medialis and the proximal vastus lateralis to enable electrical stimulation. Measurements made in the elbow flexors were performed as described previously (11). Force was assessed at 90° of elbow flexion using a modified preacher curl bench fixed to a force transducer (model SBO-750; Transducer Techniques) via a high-tension cable. Participants were seated on the bench with their feet flat on the floor and with their upper arm on the pad of the bench, such that the shoulder was fixed at approximately 45° of flexion. The arm was then secured at 90° of elbow flexion by the placement of a rigid, padded brace secured via inelastic straps. Participants grasped a wooden bar connected via cable to the force transducer. The force transducer was anchored into the floor approximately 3 ft in front of the preacher bench, such that the cable was parallel to the line of pull and perpendicular to the wooden bar. Stimulation electrodes (6.98 cm × 10.16 cm; Superior Silver; Unipatch, Wabasha, MN) were cut to a size of ∼3 cm × 4 cm and were placed over the proximal brachialis and distal brachialis for electrical stimulation. All electrode positions were marked with indelible ink to ensure similar placement for precycling and postcycling assessments, given that electrodes were removed during cycling, and to ensure similar placement during subsequent testing sessions.
On each test day, before assessment of MVC and %ACT, the stimulation current required to elicit a maximal torque value was determined by applying a series of brief electrical stimulations (paired pulses consisting of two 0.2-ms pulses with an interpulse interval of 10 ms). Stimulation was applied using a constant current stimulator (model DS7AH; Digitimer, Hertfordshire, England) controlled by a computer using an A/D interface board (model KPCI-3116; Keithley Instruments, Cleveland, OH) and a custom-written program using TestPoint software (version 6.0; Capital Equipment, Billerica, MA). Torque data were sampled at 5 kHz from the force transducer. The series of stimulations began with the current set at 40 mA, and the current was progressively increased by 20 mA until the measured torque plateaued; each contraction was separated by 20 s. The current eliciting the highest torque value was determined to represent supramaximal stimulation current and was used for all subsequent stimulations applied that day. Next, participants performed a 3-s MVC with the selected muscle group. Approximately 2.5 s into the contraction, paired-pulse stimulation was applied, and the increase in torque over MVC (interpolated-twitch torque; ITT) was determined. At 2 and 4 s after the completion of the MVC, additional paired-pulse stimulations were applied to the relaxed muscle. Peak electrically evoked twitch torque (EET) was determined as the average of the two post-MVC stimulations and used in subsequent analyses of muscle contractile properties. Percent motor–unit recruitment was calculated as 100% × (1 − ITT/EET). Participants were given strong verbal encouragement during each effort. Three trials were performed at the PreCap, PostCap, and PostEx time points with 2 min separating each trial. The two best trials were averaged and used as the criterion measure of MVC, EET, and %ACT.
Rise time and half-relaxation time (HRT), as well as the rate of torque development (+dT/dt; N·m·s−1) and rate of torque relaxation (−dT/dt; N·m·s−1), were calculated from the evoked muscle twitches as indices of muscle contractile properties and fatigue. Rise time was calculated as the time required for torque to move from 20% to 80% of its peak value after stimulation. HRT was calculated as the time from peak torque to a value of 50% of peak. Rate of torque development was calculated as the slope of the torque–time curve corresponding to the rise in torque from 20% to 80% of peak torque. The rate of torque relaxation was calculated as the slope of the torque–time curve between peak torque values and when the value fell to 50% of peak. Calculations were performed on both evoked twitches after performance of MVC, and values from each twitch were averaged and used for analysis.
Participants consumed gelatin capsules containing a 5-mg·kg−1 body weight dose of caffeine (Caffeine USP Anhydrous, PCCA, Houston, TX) or placebo (general purpose flour). On the basis of previous findings (26), this dose likely results in plasma caffeine concentrations ranging from ∼30 to 40 μM·L−1. The placebo dose was an equal number of capsules. Caffeine was administered in a double-blind manner to minimize potential participant and experimenter expectancy effects. In study A, participants were randomly assigned, with separate randomization for the leg and arm trials, to receive caffeine or placebo during each of the initial trials, and capsule content was reversed for the second leg and arm trial. In study B, participants were randomly assigned to receive caffeine or placebo for the initial trial of moderate- to heavy-intensity and severe exercise, and the content was then reversed for the second trial. The caffeine dose was chosen because it has been shown to be ergogenic and to reduce muscle pain during fixed-intensity exercise. Sixty minutes was allowed after ingestion before exercise to allow for peak blood concentrations to be reached (26).
Leg muscle pain ratings
In studies A and B, quadriceps muscle pain intensity was measured using a previously validated 0 to 10 category scale (15). The instructions for the pain scale as used in the current study have been previously described (15).
In study A, participants cycled for 30 min at a power output estimated to elicit 60% of their previously determined V˙O2peak for leg and arm crank cycling. During this phase, the ergometer was in the hyperbolic mode so that power output was independent of pedal rate. Oxygen consumption and related gas exchange measures as well as HR, muscle pain intensity, and RPE were collected every 5 min. During the initial leg and arm crank cycling sessions, work rate was initially set to 45% of maximal power output during the corresponding V˙O2peak test. Work rate was adjusted every fifth minute, as needed, to maintain exercise intensity at ∼60% of V˙O2peak. During the second leg and arm crank cycling sessions, work rate was set and adjusted to match that used during the initial sessions.
After completion of the 30-min fixed-intensity submaximal bout, participants rested for 3–4 min, during which the ergometer was reprogrammed to linear mode. A blood sample was collected via finger stick in a 25-μL capillary tube, 3 min after cessation of exercise, and blood lactate concentration was determined (YSI Sport 2600; Yellowsprings Instuments, Yellowsprings, OH) immediately. On the basis of pilot testing, the Lode ergometer linear value was set to 0.042 and 0.035 for the leg cycling time trial for men and women, respectively, and to 0.030 and 0.025 for the arm crank cycling time trial for men and women, respectively. Participants were instructed to ride as hard as possible for 10 min, in a manner to simulate an all-out effort at the end of a cycling race. Work performed (kJ) during the time trial was used as a measure of performance. This type of test was used as it has been shown to yield reliable (coefficient of variation of ∼ 3.5%) results in recreationally active individuals after a single familiarization trial (47). During the 10-min ride, HR, RPE, muscle pain intensity, pedal rate (RPM), and work output were recorded during the last 5 s of each minute, and gas exchange measures were measured continuously throughout the 10-min ride. After completion of the time trial, a second blood sample was obtained, and blood lactate was again assessed.
In study B, participants cycled for 15 min at a power output that corresponded to 95% of the power output at GET as determined by the during their V˙O2peak test (heavy exercise). GET was determined independently by two investigators by identifying the V˙CO2 and power output values at which V˙CO2 increased disproportionately to V˙O2 and V˙E/V˙O2 increased without a concomitant increase in V˙E/V˙CO2 (7). Exercise at 70% of the difference between GET and V˙O2peak was prescribed to elicit severe-intensity exercise, and participants cycled until volitional exhaustion at this intensity. Prescribing exercise in this manner allowed for a similar relative intensities of “heavy” and “severe” exercise to be prescribed to each participant regardless of their training status.
For Study A, mean values of V˙O2peak during leg and arm crank cycling were compared using a dependent-measures t-test. The mean individual percent changes from PreCap to PostCap and from PostCap to PostEx between caffeine and placebo conditions for MVC, EET, and %ACT were compared using a dependent-measures t-test. A 2 (condition) × 3 (time) within-participant repeated-measures ANOVA was used to investigate differences in MVC, EET, rise time, HRT, +dT/dt, and −dT/dt. Data from the 30-min cycling trial were analyzed using a 2 (condition) × 6 (time), and data from the 10-min time-trial performance were analyzed using a 2 (condition) × 10 (time) within-participants repeated-measures ANOVA. For study B, data from the 15-min trial at 95% GET and at 70%Δ were analyzed using a 2 (condition) × 5 (time) within-participant repeated-measures ANOVA. For all ANOVAs, if a significant interaction was found, it was followed up by a one-way ANOVA to examine differences over time and between conditions. Dependent measures t-tests with a Bonferroni α correction were used for planned comparisons of means. If main effects were found in the absence of significant interaction, main comparisons were performed either between conditions (caffeine vs placebo) or by comparing values to the initial time point during each trial. For both studies A and B, mean values for all measures across time in each exercise bout (Bout Mean) were compared using a dependent-measures t-test. Statistical significance for studies A and B was set a priori at an α level of ≤0.05 for a two-tailed test. Effect sizes were calculated as a Cohen d statistic, that is, the difference in means divided by the pooled SD of the means. All data are presented as mean ± SD, and all testing was performed using SPSS (version 19).
Participants in study A were 21.5 ± 1.9 yr old with a height of 173.9 ± 13.8 cm and weight of 74.6 ± 17.9 kg. V˙O2peak was significantly higher during leg cycling compared to arm crank cycling (40.9 ± 6.7 vs 30.6 ± 6.9 mL·kg−1·min−1, P < 0.001). In study B, participants were 21.7 ± 2.8 yr old, 175.7 ± 8.6 cm tall, and weighed 78.8 ± 18.3 kg with a cycling V˙O2peak of 36.6 ± 7.3 mL·kg−1·min−1. GET occurred at 22.7 ± 3.1 mL·kg−1·min−1, which corresponded to 65.1% ± 9.4% of V˙O2peak.
Muscle contractile properties (study A)
The mean individual percent change in MVC, EET, and %ACT of the knee extensors and elbow flexors from the PreCap to PostCap time points are shown in Figures 2A and B, respectively. Percent change in MVC, EET, and %ACT from the PostCap to PostEx are shown in Figure 2C for the knee extensors and in Figure 2D for the elbow flexors, respectively. Caffeine significantly increased MVCKE (P = 0.014, d = 1.18 SD) and %ACTKE (P = 0.013, d = 0.71) but not EETKE (P = 0.54, d = 0.18) 60 min after ingestion compared to placebo in the knee extensors (Fig. 2A). Caffeine ingestion did not alter MVCEF (P = 0.24, d = 0.45), EETEF (P = 0.44, d = 0.31), or %ACTEF (P = 0.22, d = 0.55) compared to placebo in the elbow flexors (Fig. 2B). After leg cycling, MVCKE was reduced to a similar extent in the caffeine and placebo conditions (P = 0.22, d = 0.30; Fig. 2C). EETKE (P = 0.25, d = −0.46) and %ACTKE (P = 0.32, d = 0.27) were also reduced to a similar extent in the caffeine and placebo conditions (Fig. 2C). Similarly, MVCEF (P = 0.95, d = −0.02), EETEF (P = 0.97, d = 0.02), and %ACTEF (P = 0.67, d = −0.14) were reduced to a similar extent in the caffeine and placebo conditions after the arm crank cycling time trial (Fig. 3D).
Mean absolute values for MVC, EET, rise time, HRT, +dT/dt, and −dT/dt are shown in Table 1. A significant condition–time interaction (P = 0.014) was found for MVCKE. Results from one-way ANOVA for time were significant (P ≤ 0.001) for both the caffeine and placebo conditions. In the caffeine condition, MVCKE increased from PreCap to PostCap (P = 0.049), and PostEx values were reduced compared to PostCap (P = 0.003). No differences were found between PreCap and PostCap (P = 0.542) in the placebo condition, but values from PostEx were reduced compared to PostCap time points (P ≤ 0.001). Interestingly, values from PreCap in the caffeine condition were reduced from PreCap values in the placebo condition (P = 0.022). A main effect for time (P ≤ 0.001) was found for MVCEF, with values from PostEx being reduced from PreCap and PostCap (P ≤ 0.001). EETKE was reduced PostEx (main effect for time, P ≤ 0.001) compared to PreCap and PostCap (P ≤ 0.001 for each). No changes were observed for EETEF over time (P = 0.47 for interaction and P = 0.09 for main effect) or between conditions (P = 0.30 for main effect). Mean values for rise time were unchanged between conditions and over time in both the knee extensors and elbow flexors (P ≤ 0.09 for interactions and main effects). No differences were found between caffeine and placebo conditions for HRT (P = 0.25 and P = 0.13 for interactions for the knee extensors and elbow flexors, respectively; P = 0.29 and P = 48 for main effect for conditions for the knee extensors and elbow flexors, respectively). A significant main effect for time was found (P ≤ 0.002 for each), with HRT being reduced at PostEx compared to PreCap and PostCap in the knee extensors (P ≤ 0.002) and elbow flexors (P ≤ 0.008). The rate of force development was unaffected by caffeine ingestion in both the knee extensors (P = 0.32 and P = 0.54 for interaction and main effect for condition, respectively) and the elbow flexors (P = 0.34 and P = 0.52 for interaction and main effects for condition, respectively). +dT/dt decreased (slowed) after exercise in both the knee extensors (−28% vs PreCap and PostCap; P ≤ 0.004) and the elbow flexors (−27% vs PreCap and −18% vs PostCap; P ≤ 0.02). Despite the 17% (compared to PreCap) and 21% (compared to PostCap) increases (slowing) in –dT/dt observed after exercise in the knee extensors, the changes were not significant (P = 0.54 for interaction and P = 0.34 for main effect for time). Caffeine ingestion also had no effect on –dT/dt (P = 0.29 for main effect for condition). Similar findings were observed in the elbow flexors where neither caffeine (P = 0.23 for main effect for condition) nor exercise (P = 0.73 for interaction and P = 0.59 for main effect for time) had an effect on –dT/dt.
30-min submaximal cycling bout (study A)
Power output was matched during the caffeine and placebo conditions for both leg cycling and arm crank cycling (Table 2). Oxygen consumption did not differ between conditions for leg cycling (P = 0.47 for interaction and P = 0.61 for main effect), with the mean response being 62.1% ± 4.1% and 62.4% ± 3.9% of V˙O2peak leg during leg cycling for the caffeine and placebo conditions, respectively. Responses during arm crank were 60.5% ± 3.2% and 59.6% ± 3.8% of V˙O2peak arm for the caffeine and placebo conditions, respectively (P = 0.42 for interaction and P = 0.15 for main effect; Table 2). Significant main effects for time (P ≤ 0.004) were found for both cycling modalities, with V˙O2 values from the 10th and 15th minutes being greater than values from the 5th minute for both modalities (P ≤ 0.022 and P ≤ 0.039 for leg and arm crank, respectively; Table 2). HR did not differ between conditions for leg cycling (P = 0.33 for interaction and P = 0.60 for main effect) or arm crank cycling (P = 0.23 for interaction and P = 0.15 for main effect; Table 2). Significant main effects for time (P ≤ 0.002) were observed for both cycling modalities. During leg cycling, HR values from the 15th, 20th, 25th, and 30th minutes were greater than values from the 5th minute (P < 0.001); whereas during arm crank cycling, values from the 10th to the 30th minutes were elevated compared to values from the 5th minute (P < 0.001). RER values did not differ between conditions for leg cycling (P = 0.23 for interaction and P = 0.21 for main effect; Table 2) and arm crank (P = 0.45 for interaction and P = 0.32 for main effect; Table 2). Significant main effects for time were seen during leg and arm crank cycling (P = 0.001 and P ≤ 0.001 for leg and arm, respectively), with RER values decreasing over time. For leg cycling, values from the 25th and 30th minutes of cycling differed from values from the 5th minute (P ≤ 0.05). During arm crank cycling, values from the 15th to the 30th minute were reduced compared to values from the 5th minute (P ≤ 0.05). A significant main effect for condition (P = 0.002 and P = 0.039 for leg and arm crank, respectively) was found for RPE (Table 2), with a lower perception of effort after caffeine ingestion (d = −0.90 and −0.67 SD for leg and arm crank, respectively). In addition, significant main effects for time (P ≤ 0.002) were found. During leg cycling, RPE values from the 10th to the 30th minute were significantly higher (P ≤ 0.005) than values from the 5th minute. For arm crank cycling, values from the 10th to the 30th minute were also elevated compared to the 5th minute (P ≤ 0.02). Caffeine ingestion reduced (d = −0.78 SD for the mean across the 30-min trial) muscle pain in the quadriceps during leg cycling (P = 0.004 for interaction effect) compared to placebo at the 15th, 20th, 25th, and 30th minutes (P ≤ 0.005 for each; Table 2). A significant main effect for condition was found during arm crank cycling (P = 0.007, d = −0.77), with caffeine resulting in reduced pain ratings compared to placebo (Table 2). Blood lactate concentration was elevated immediately after the 30-min exercise bout in the caffeine condition for both leg (P = 0.019) and arm crank cycling (P = 0.003; Table 2).
10-min time-trial performance (study A)
Caffeine ingestion resulted in a small (d = 0.13 SD) but consistent (13 of 14 participants showed an improvement) increase (4.9%; P = 0.034) in total work during the leg cycling time trial compared to placebo (Table 3). However, the effect was not found during arm crank cycling (2.1%; P = 0.28, d = 0.07 SD). Mean work performed during each minute of the time trial was greater after caffeine ingestion during leg cycling (P = 0.035 for main effect) but not arm crank cycling (P = 0.19 for interaction and P = 0.28 for condition main effect; Table 3). Oxygen consumption did not differ between the caffeine and placebo conditions during leg cycling (P = 0.76 for interaction and P = 0.75 for main effect; Table 3). A main effect for time was found (P < 0.001), with V˙O2 for all time points differing from values from the first minute. For arm crank cycling, V˙O2 values were greater in the caffeine condition (P = 0.036 for main effect; Table 3) compared to placebo, and a significant main effect for time was also found with all values differing from the first minute (P < 0.001). HR did not differ between the caffeine and placebo conditions for either leg cycling (P = 0.83 and P = 0.32 for interaction and main effect, respectively; Table 3) or arm crank cycling (P = 0.34 and P = 0.10 for interaction and main effect, respectively; Table 3). A significant main effect for time was found for HR during both leg cycling (P < 0.001) and arm crank cycling (P < 0.001), with HR increasing (P < 0.001 for all time points in both leg cycling and arm crank cycling) progressively over time compared to the first minute of the time trial. Caffeine ingestion had no effect on RER during leg cycling (P = 0.14 for interaction and P = 0.09 for main effect for condition) and arm crank cycling (P = 0.40 for interaction and P = 0.83 for main effect for condition). A main effect for time (P ≤ 0.001 for both leg cycling and arm crank cycling) was found, with RER values progressively increasing over time. During leg cycling, values from the 2nd and 4th–10th minutes were elevated from values from the 1st minute (P ≤ 0.05). Similarly, during arm crank cycling, values from the 2nd and 4th–10th minutes were elevated from values from the 1st minute (P ≤ 0.05). A significant condition–time interaction (P = 0.014) was found for RPE during leg cycling. However, caffeine ingestion only reduced RPE during the 2nd minute of the time trial (P = 0.02; Table 3). RPE did not differ between conditions during arm crank cycling (P = 0.74 and P = 0.44 for the interaction and condition main effect, respectively; Table 3). RPE increased progressively over time during leg cycling in both the caffeine and placebo conditions, with all values from the 2nd to the 10th minute differing from the 1st minute (P < 0.001 for both caffeine and placebo conditions). For arm crank cycling, a significant main effect for time was found (P < 0.001; Table 3), with each time point differing from the 1st minute (P < 0.001). Muscle pain ratings did not differ between the caffeine and placebo conditions during either leg cycling (P = 0.63 for interaction and P = 0.58 for condition main effect; Table 3) or arm crank cycling (P = 0.60 for interaction and P = 0.59 for condition main effect; Table 3). Significant main effects for time were found for both leg cycling (P < 0.001) and arm crank cycling (P < 0.001), with pain ratings increasing progressively over time during the time trial and with values from the 2nd to the 10th minute differing (P < 0.001) from values from the 1st minute. Blood lactate levels after completion of the time trial was higher in the caffeine condition (P = 0.003) after leg cycling (Table 3), whereas no differences were found between conditions after arm crank cycling (P = 0.11; Table 3).
Exercise at 95% of GET (study B)
Power output was matched between the caffeine and placebo conditions (Table 4). Oxygen consumption was higher (main effect for condition, P = 0.02) in the caffeine condition and a significant main effect for time (P ≤ 0.003) was found, with values from the 6th, 9th, 12th, and 15th minutes being greater than values from the 3rd minute of exercise (Table 4). HR neither differed between conditions (P = 0.31 for the interaction and P = 0.12 for the main effect) nor differed across time (P = 0.10; Table 4). RPE did not differ between conditions (P = 0.71 for the interaction and P = 0.78 for the main effect; Table 4) but did increase with exercise time, with rating during the 6th–15th minutes being greater than the rating from the 3rd minute (P ≤ 0.001 for each). Similar results were observed for ratings of muscle pain, with no differences found between the caffeine and placebo conditions (P = 0.68 for the interaction and P = 0.90 for the main effect; Table 4). Pain did increase over time (P ≤ 0.001 for main effect), with the 6th–15th minutes being elevated (P ≤ 0.001) compared to the 3rd minute of exercise.
Exercise at 70%Δ (study B)
Power output was matched between the caffeine and placebo conditions (Table 4). Oxygen consumption was similar (P = 0.65 for the interaction and P = 0.62 for the main effect) between the caffeine and placebo conditions, and a significant main effect for time (P ≤ 0.001) was found, with values from the 40%, 60%, 80%, and 100% of exercise time being greater than 20% of exercise time (Table 4). HR did not differ between conditions (P = 0.55 for the interaction and P = 0.30 for the main effect) but was found to increase over time (P ≤ 0.001 for main effect), with all values differing from each other (P ≤ 0.05; Table 4). RPE did not differ between conditions (P = 0.15 for the interaction and P = 0.79 for the main effect; Table 4) but did increase with exercise time, with rating from 40% to 100% of exercise time being greater than rating from 20% of time (P ≤ 0.001 for each). Again, similar results were observed for ratings of muscle pain, with no differences found between the caffeine and placebo conditions (P = 0.22 for the interaction and P = 0.77 for the main effect; Table 4). Pain did increase over time (P ≤ 0.001 for main effect), with the 40%–100% of exercise time being elevated (P ≤ 0.001) compared to the 20% time point.
Caffeine ingestion has repeatedly been shown to be ergogenic for endurance exercise performance (13,19,21,25). Despite the wealth of evidence supporting caffeine’s ergogenic effects on endurance exercise, data linking improved performance to a mechanism(s) are limited. As the idea of glycogen sparing consequent to enhanced fat oxidation during exercise has fallen out of favor (27), an alternative mechanism(s) has been sought. Actions in the central nervous system resulting in reductions in muscle pain, sense of effort, and enhanced muscular force production, as well as enhanced force from a direct effect on skeletal muscle, have been postulated. To date, no study has concomitantly assessed muscle pain and perceived effort, neuromuscular function, and endurance performance after caffeine ingestion. A novel study design using both leg cycling and arm crank cycling was used in an effort to manipulate caffeine’s effects on force production and motor-unit recruitment. This was done because both have been shown to be augmented in the knee extensors but not in the smaller muscles of the arms (51). In addition, a follow-up study was performed to further examine caffeine’s hypoalgesic effects during heavy- to severe-intensity exercise. The primary findings were as follows: 1) caffeine improved MVC and motor–unit recruitment in the knee extensors but not the elbow flexors; 2) caffeine reduced muscle pain and RPE during fixed-intensity submaximal cycling in both the legs and arms, but this effect was eliminated during the self-paced performance trial (study A) and heavy to severe fixed-intensity cycling (study B); and 3) caffeine yielded a small but consistent ergogenic effect on endurance time-trial performance during leg cycling but not arm crank cycling.
Caffeine ingestion resulted in consistent improvement in total work performed during leg cycling. The magnitude of improvement was similar to the mean improvement (i.e., 4.9% vs 4.7%) reported in a review (21) of 16 trials examining caffeine’s effects on endurance performance when a maximal-effort time trial was preceded by a period of fixed-intensity submaximal exercise. It has been hypothesized that caffeine could improve endurance performance by enhancing fat oxidation leading to a sparing of muscle glycogen (16). Recent work has refuted this idea (27) and is supported by our finding that caffeine had no effect on RER during exercise. In addition, the relatively short exercise time (40 min) likely limited fatigue occurring because of glycogen depletion. Consistent with other studies (8,17,52), the greater work rate after caffeine ingestion was not accompanied by changes in oxygen consumption, suggesting that the ergogenic effect was not due to an alteration in aerobic metabolism. Thus, the effect is likely the result of enhancements of muscular strength, reductions in muscle pain and RPE, or some combination.
On the basis of the findings from the meta-analysis of Warren et al. (51), it was hypothesized that caffeine ingestion would improve strength and motor-unit recruitment in the knee extensors but not the elbow flexors. A large (d = 1.18 SD) increase in MVCKE and a moderate-to-large (d = 0.71 SD) increase in %ACTKE were observed 60 min after caffeine ingestion, whereas no effect was found for MVCEF and %ACTEF. It has been hypothesized that caffeine may enhance force production via actions in the central nervous system and/or direct actions on skeletal muscle. Caffeine has been shown to antagonize both central and peripheral adenosine A1, A2A, and A2B receptors (46), increase neurotransmitter and dopaminergic transmission in the CNS (20), and enhance motor neuron excitability as measured by the H-reflex (50). In addition, micromolar doses of caffeine have been shown to enhance force production in isolated muscles (30), and several studies (17,37,49) have demonstrated attenuated fatigue during electrically stimulated exercise, suggesting that doses in this range may exert a direct effect(s) on skeletal muscle in vivo. In the present study, the lack of an effect of caffeine on EET, rise time, HRT, and rate of torque development and relaxation in both muscle groups suggests that a direct action on skeletal muscle did not occur. The observed increase in %ACTKE suggests that the increase in MVCKE was likely the result of an action in the CNS, although it was beyond the scope of this study to establish whether the effect occurred at a spinal and/or supraspinal level. Because the knee extensor data support a central mechanism, how then do we explain the lack of improved strength and motor-unit recruitment in the elbow flexors? Warren et al. (51) suggest that a ceiling effect may limit the ability of enhanced central excitability to manifest as improved motor-unit recruitment and thus improve strength in the smaller muscles of the arms. Consistent with this hypothesis, mean %ACTEF was near maximal (∼97%) before caffeine ingestion, whereas %ACTKE was lower (∼83%) in the present study. It should also be noted that, in the knee extensors, PreCap values for MVC were reduced during the caffeine trial compared to the placebo trial. After caffeine ingestion, MVC increased such that it approximated PreCap values in the placebo condition. This finding is very similar to that of Plaskett and Cafarelli (44), but it is unclear as to whether this was due to a larger-than-expected day-to-day variation in MVC or fatigue from a previous testing day. It is important to note that our participants were not resistance trained, and this also may have influenced the MVC and %ACT results.
Interestingly, caffeine exhibited no effects on muscle contractile properties after exercise. Cureton et al. (17) found that a caffeinated sports drink (total caffeine dose across the exercise protocol ∼5.3 mg·kg−1 body weight) attenuated the decline in MVC and electrically evoked torque after 135 min of cycling exercise. The authors found no difference in motor–unit recruitment between the caffeinated and placebo conditions and suggest that reduced intrinsic muscular fatigue may explain their findings. In the present study, MVCKE and MVCEF declined after cycling, clearly indicating fatigue. Consistent with the findings of Cureton et al. (17), PostEx %ACT was unaffected by caffeine ingestion. EETKE declined ∼20% in both the caffeine and placebo conditions, indicating that the decline in MVCKE was likely due to fatigue intrinsic to the muscle rather than a failure of recruitment. Significant declines in EETEF were not observed, although torque was reduced 5% and 10% in the placebo and caffeine conditions, respectively. This highlights a potential limitation of the study—the elbow flexors, while active during arm crank, may contribute to the total work performed during arm crank cycling to a smaller relative degree compared to the contribution of the knee extensors to leg cycling. When examined over a 360° range of motion, EMG analysis has shown the biceps to be the most active muscle during arm crank cycling (53). It has also been shown to experience the largest change in tissue deoxygenation assessed via near-infrared spectroscopy (36). However, it is clear that muscles such as the triceps brachii and anterior deltoids are also active during arm crank cycling (36,53). This could have contributed to our finding of a lack of decline in EETEF. If the force required during arm crank exercise was spread over a host of muscles (in comparison to leg cycling where the knee extensor predominate), it could have limited fatigue in the elbow flexors. It is interesting to note that lactate concentrations were elevated in the caffeine condition during submaximal cycling with both the arms and legs although power output was matched. Although caution should be taken in interpreting a small (∼0.7–1.0 mM) difference in lactate concentration, this finding could suggest a slight shift in recruitment patterns during cycling toward the less metabolically efficient Type II fibers.
Previous studies have found that caffeine reduces both muscle pain (22–24,38,39,42) and RPE (8,16,17,19,22) during submaximal, fixed-intensity exercise between 60% and 80% of V˙O2peak. Based in large part on these observations, reductions in pain and effort have been suggested (18,19,38,51) as potential mechanisms of caffeine’s ergogenic effect on endurance exercise performance—regardless of exercise intensity. However, this hypothesis is not supported by evidence from studies (5,6,8,12,14,17,22,29,32,52), examining performance during high-intensity, fixed-time or fixed-distance performance. Indeed, no study has demonstrated an improvement in performance concomitant to reduced pain or effort. The unique design of the present study, cycling for 30 min at 60% of V˙O2peak followed by a time-trial, allowed for a clear demonstration of reduced pain and RPE after caffeine ingestion that was eliminated as exercise transitioned to the maximal-effort performance trial. To our knowledge, the results of study A are the first to demonstrate the elimination of hypoalgesia as exercise intensity transitions from moderate to heavy/severe. When coupled with the improved work performed during leg cycling, but not arm crank cycling, it supports the notion that reductions in pain and effort per se may not be the primary mechanism by which caffeine improves endurance performance.
Two possible explanations could account for the disparate results observed between submaximal fixed-intensity exercise and fixed-time or fixed-distance performance trials: 1) a threshold intensity of pain and/or perceived exertion exists, above which caffeine exerts no effect, or 2) participants pace their work output during fixed-time or fixed-distance trials based on sensations of muscle pain and RPE, masking any effect of caffeine, but allowing for greater work to be performed at a given intensity of pain and/or effort. Some evidence exists in support of caffeine not being hypoalgesic during exercise evoking higher-intensity muscle pain. Studies finding no hypoalgesic effect during submaximal fixed-intensity cycling (32), fixed-distance performance trials (5,6), resistance exercise (3,28), and fixed-time performance trials (32) have found pain ratings of 5 or 6 units (“strong” and “very strong,” respectively) or higher on a 0 to 10 rating scale (15). In contrast, the majority of studies demonstrating a hypoalgesic effect of caffeine (23,24,38,39,42) during submaximal cycling tended to report pain ranging from 0 to 3 (“mild” to “moderate”) on a 0 to 10 rating scale (15). In the present study, caffeine was hypoalgesic during submaximal leg and arm cycling when pain was reported to be “mild” to “moderate,” but this effect disappeared during the performance trial when pain was much higher. A host of biochemical (hydrogen ions, bradykinin, substance P, PGE2) and mechanical signals (pressure from muscle contractions) are thought to contribute to sensations of muscle pain during exercise (9,15,40). It is plausible that, during high-intensity exercise, the collective nociceptive inputs are too great for caffeine’s antagonism of adenosine receptors alone to result in measureable hypoalgesia.
We performed a second experiment to explore this idea further (study B). Fixed-intensity cycling was performed at 95% of GET (to approximate moderate- to heavy-intensity exercise) and at 70% of the difference between GET and V˙O2peak (severe-intensity exercise) to evoke higher levels of muscle pain. Percentage of GET was used to ensure that similar relative exercise intensities were prescribed to each participant, regardless of fitness level. Caffeine had no effect on muscle pain or RPE compared to placebo at either exercise intensity. The bout at 95% of GET produced V˙O2 values ranging from ∼65% to 75% of V˙O2peak over the 15-min bout, with pain ratings reaching values of 4 units, “somewhat strong” to 5 units, and “strong” in most participants. This finding is roughly similar to that of Jenkins et al. (32) who found no effect of caffeine on pain or RPE in highly trained cyclists during exercise at 80% of V˙O2peak. Values for oxygen consumption were consistently at or above 90% of V˙O2peak during the severe-intensity bout, with pain ratings reaching “very strong” (6 units) or above. Thus, the results of study B lend additional support to the idea high-intensity exercise may result in muscle pain that is too great for caffeine to be hypoalgesic. Several studies have linked ratings of muscle pain to RPE during aerobic exercise (10,15,31). Our findings of caffeine-induced reductions in both muscle pain and RPE in the absence of changes in V˙O2 and HR during submaximal exercise further support a role of muscle pain in informing perception of effort. Given this apparent link, it is not surprising that minimal effects of caffeine were observed on RPE during the performance trials, that is, exercise at 95% GET and 70%Δ where similar levels of muscle pain are found between caffeine and placebo conditions.
The second possible explanation for our findings was that any effect of caffeine on muscle pain and/or RPE was masked because of participants pacing their work output based on sensations of muscle pain and RPE. Research indicates that individuals are capable of manipulating exercise intensity to produce a prescribed level of muscle pain (41). Thus, it is plausible that the participants manipulated their work rate to produce a level of pain and effort they felt would allow them to not only complete the performance ride but also maximize their total work output during the trial. Caffeine has consistently been shown to increase work output at a given level of muscle pain (5,6,32) and RPE (5,6,14,17,29,32) as occurred during the leg cycling performance trial in study A. However, this relationship did not hold during the arm crank cycling trial. Because caffeine resulted in a similar reduction in muscle pain and RPE during submaximal leg and arm cycling, it would be plausible to expect that caffeine would exert a similar effect during both performance rides. Taken together with the results of study B, these findings cast doubt on the idea that caffeine was indeed hypoalgesic during the performance ride and thus allowed for a greater work rate to be performed at a given level of pain and exertion. Given that caffeine enhanced strength and motor-unit recruitment in the knee extensors, but not the elbow flexors of our participants, then it is plausible the enhanced strength, rather than a hypoalgesic effect per se, could be responsible for the greater work performed at a similar rating of muscle pain and RPE.
Several limitations of these studies are worth noting. Although our participants were recreationally active, they had no recent experience/training with resistance training and cycling. Their general lack of knowledge and experience with pacing their performance during high-intensity cycling could have limited the reliability of their performance during the time trials. Prior familiarization to an endurance time trial has been shown to improve day-to-day reliability in untrained cyclists (47), yielding a relatively similar coefficient of variation (∼3.5%) to that observed in trained cyclists (33). In addition, arm crank cycling is not a commonly used exercise modality, especially when compared to traditional leg cycling and this could also have affected our ability to detect small ergogenic effects of caffeine, if they existed. Part of our rationale for using non–cycling-trained participants was to minimize any potential differences in familiarity between leg cycling and arm crank cycling. In addition, familiarization was provided to improve reliability. All data were checked for an order effect, which might be expected if a significant “learning” effect was present, and no effect was found. Plasma caffeine concentrations were not measured, and therefore, we cannot be certain that differences in plasma level of caffeine did not play a role in the lack of effect on strength and motor-unit recruitment in the elbow flexors and during arm crank cycling. Finally, it should be pointed out that caffeine acts at multiple sites in the central and peripheral nervous system. As such, it is impossible to rule out several mechanisms acting synergistically to produce enhanced endurance performance. Small changes in muscle pain and RPE, which may not be detectible given the subjective nature of the measurements, could act in concert with changes in strength and motor-unit recruitment to be ergogenic. Future research attempting to block caffeine’s effects at specific sites of action and/or at specific adenosine receptors using targeted antagonists could provide further insight into this possibility.
The goal of this study was to gain a better understanding of the mechanism(s) by which caffeine enhances endurance exercise performance. To our knowledge, this was the first study to combine measurements of muscle strength, motor-unit recruitment, and muscle pain and endurance time-trial performance. Our primary novel findings were as follows: 1) caffeine ingestion enhanced strength and motor-unit recruitment in the knee extensors, but not the elbow flexors; 2) caffeine ingestion reduced muscle pain and RPE during moderate-intensity submaximal exercise that was subsequently eliminated when exercise intensity was increased during the time-trial performance; and 3) caffeine ingestion improved endurance performance during leg cycling but not arm crank cycling. Hypoalgesia has been suggested as a mechanism by which caffeine is ergogenic. The disappearance of the effect as exercise transitioned from submaximal to very high-intensity combined with our findings that caffeine did not reduce muscle pain during fixed-intensity exercise at moderate-to-heavy and heavy-to-severe metabolic intensities cast doubt on the ability of caffeine to be hypoalgesic during high-intensity exercise and thus for hypoalgesia per se to underlie caffeine’s ergogenic effects. Although greater strength and motor-unit recruitment may not be the sole mechanism of action for caffeine’s ergogenic effect on endurance exercise, when taken with the lack of effect on pain and effort, enhanced strength could represent a plausible mechanism by which caffeine is ergogenic.
Portions of this work were completed while one of the authors (C.B.) was an assistant professor at Georgia College and State University and the University of Mississippi.
The authors gratefully acknowledge Julia Borland, Dallas Dixon, Justin Hill, Matt Nelson, Katie Rich, and Jessica Sethman for assistance with data collection and preparation of caffeine treatments. The authors also thank Gordon Warren for providing assistance and sharing his technical expertise regarding the interpolated-twitch technique and Manning Sabatier for helpful discussion of the data. No funding was received for this study and the authors declare no conflicts of interest. The results of the present study do not constitute endorsement by the American College of Sports Medicine.
1. Abo-Salem OM, Hayallah AM, Bilkei-Gorzo A, Filipek B, Zimmer A, Muller CE. Antinociceptive effects of novel A2B
adenosine receptor antagonists. J Pharmacol Exp Ther
. 2004; 308 (1): 358–66.
2. Astorino TA, Roberson DW. Efficacy of acute caffeine ingestion for short-term high-intensity exercise performance: a systematic review. J Strength Cond Res
. 2010; 24 (1): 257–65.
3. Astorino TA, Terzi MN, Roberson DW, Burnett TR. Effect of caffeine intake on pain perception during high-intensity exercise. Int J Sport Nutr Exerc Metab
. 2011; 21 (1): 27–32.
4. Astorino TA, Cottrell T, Lozano AT, Aburto-Pratt K, Duhon J. Increases in cycling performance in response to caffeine ingestion are repeatable. Nutr Res
. 2012; 32 (2): 78–84.
5. Astorino TA, Cottrell T, Talhami Lozano A, Aburto-Pratt K, Duhon J. Effect of caffeine on RPE
and perceptions of pain, arousal, and pleasure/displeasure during a cycling time trial
in endurance trained and active men. Physiol Behav
. 2012; 106 (2): 211–7.
6. Astorino TA, Roupoli LR, Valdivieso BR. Caffeine does not alter RPE
or pain perception during intense exercise in active women. Appetite
. 2012; 59 (2): 585–90.
7. Beaver WL, Wasserman K, Whipp BJ. A new method for detecting anaerobic threshold by gas exchange. J Appl Physiol (1985)
. 1986; 60 (6): 2020–7.
8. Bell DG, Jacobs I, Ellerington K. Effect of caffeine and ephedrine ingestion on anaerobic exercise performance. Med Sci Sports Exerc
. 2001; 33 (8): 1399–403.
9. Black CD. Muscle pain
during and following exercise. In: The Oxford Handbook of Exercise Psychology
. New York, NY: Oxford; 2012. p. 144.
10. Black CD, Dobson RM. Prior eccentric exercise augments muscle pain
and perception of effort during cycling exercise. Clin J Pain
. 2013; 29 (5): 443–9.
11. Black CD, Herring MP, Hurley DJ, O’Connor PJ. Ginger (Zingiber officinale
) reduces muscle pain
caused by eccentric exercise. J Pain
. 2010; 11 (9): 894–903.
12. Bridge CA, Jones MA. The effect of caffeine ingestion on 8 km run performance in a field setting. J Sports Sci
. 2006; 24 (4): 433–9.
13. Burke LM. Caffeine and sports performance. Appl Physiol Nutr Metab
. 2008; 33 (6): 1319–34.
14. Cole KJ, Costill DL, Starling RD, Goodpaster BH, Trappe SW, Fink WJ. Effect of caffeine ingestion on perception of effort and subsequent work production. Int J Sport Nutr
. 1996; 6 (1): 14–23.
15. Cook DB, O’Connor PJ, Eubanks SA, Smith JC, Lee M. Naturally occurring muscle pain
during exercise: assessment and experimental evidence. Med Sci Sports Exerc
. 1997; 29 (8): 999–1012.
16. Costill DL, Dalsky GP, Fink WJ. Effects of caffeine ingestion on metabolism and exercise performance. Med Sci Sports
. 1978; 10 (3): 155–8.
17. Cureton KJ, Warren GL, Millard-Stafford ML, et al. Caffeinated sports drink: ergogenic effects and possible mechanisms. Int J Sport Nutr Exerc Metab
. 2007; 17 (1): 35–55.
18. Davis JK, Green JM. Caffeine and anaerobic performance: ergogenic value and mechanisms of action. Sports Med
. 2009; 39 (10): 813–32.
19. Doherty M, Smith PM. Effects of caffeine ingestion on rating of perceived exertion during and after exercise: a meta-analysis. Scand J Med Sci Sports
. 2005; 15 (2): 69–78.
20. Fredholm BB, Dunwiddie TV. How does adenosine inhibit transmitter release? Trends Pharmacol Sci
. 1988; 9 (4): 130–4.
21. Ganio MS, Klau JF, Casa DJ, Armstrong LE, Maresh CM. Effect of caffeine on sport-specific endurance performance: a systematic review. J Strength Cond Res
. 2009; 23 (1): 315–24.
22. Ganio MS, Johnson EC, Lopez RM, et al. Caffeine lowers muscle pain
during exercise in hot but not cool environments. Physiol Behav
. 2011; 102 (3–4): 429–35.
23. Gliottoni RC, Motl RW. Effect of caffeine on leg-muscle pain
during intense cycling exercise: possible role of anxiety sensitivity. Int J Sport Nutr Exerc Metab
. 2008; 18 (2): 103–15.
24. Gliottoni RC, Meyers JR, Arngrimsson SA, Broglio SP, Motl RW. Effect of caffeine on quadriceps muscle pain
during acute cycling exercise in low versus high caffeine consumers. Int J Sport Nutr Exerc Metab
. 2009; 19 (2): 150–61.
25. Graham TE. Caffeine and exercise: metabolism, endurance and performance. Sports Med
. 2001; 31 (11): 785–807.
26. Graham TE, Spriet LL. Metabolic, catecholamine, and exercise performance responses to various doses of caffeine. J Appl Physiol
. 1995; 78 (3): 867–74.
27. Graham TE, Helge JW, MacLean DA, Kiens B, Richter EA. Caffeine ingestion does not alter carbohydrate or fat metabolism in human skeletal muscle during exercise. J Physiol
. 2000; 529 (Pt 3): 837–47.
28. Hudson GM, Green JM, Bishop PA, Richardson MT. Effects of caffeine and aspirin on light resistance training performance, perceived exertion, and pain perception. J Strength Cond Res
. 2008; 22 (6): 1950–7.
29. Ivy JL, Costill DL, Fink WJ, Lower RW. Influence of caffeine and carbohydrate feedings on endurance performance. Med Sci Sports
. 1979; 11 (1): 6–11.
30. James RS, Kohlsdorf T, Cox VM, Navas CA. 70 microM caffeine treatment enhances in vitro force and power output during cyclic activities in mouse extensor digitorum longus muscle. Eur J Appl Physiol
. 2005; 95 (1): 74–82.
31. Jameson C, Ring C. Contributions of local and central sensations to the perception of exertion during cycling: effects of work rate and cadence. J Sports Sci
. 2000; 18 (4): 291–8.
32. Jenkins NT, Trilk JL, Singhal A, O’Connor PJ, Cureton KJ. Ergogenic effects of low doses of caffeine on cycling performance. Int J Sport Nutr Exerc Metab
. 2008; 18 (3): 328–42.
33. Jeukendrup A, Saris WHM, Brouns F, Kester ADM. A new validated endurance performance test. Med Sci Sports Exerc
. 1996; 28 (2): 266–70.
34. Kalmar JM, Cafarelli E. Effects of caffeine on neuromuscular function. J Appl Physiol
. 1999; 87 (2): 801–8.
35. Kalmar JM, Cafarelli E. Central excitability does not limit postfatigue voluntary activation of quadriceps femoris. J Appl Physiol
. 2006; 100 (6): 1757–64.
36. Lusina SJ, Warburton DE, Hatfield NG, Sheel AW. Muscle deoxygenation of upper-limb muscles during progressive arm-cranking exercise. Appl Physiol Nutr Metab
. 2008; 33 (2): 231–8.
37. Mohr T, Van Soeren M, Graham TE, Kjaer M. Caffeine ingestion and metabolic responses of tetraplegic humans during electrical cycling. J Appl Physiol (1985)
. 1998; 85 (3): 979–85.
38. Motl RW, O’Connor PJ, Dishman RK. Effect of caffeine on perceptions of leg muscle pain
during moderate intensity cycling exercise. J Pain
. 2003; 4 (6): 316–21.
39. Motl RW, O’Connor PJ, Tubandt L, Puetz T, Ely MR. Effect of caffeine on leg muscle pain
during cycling exercise among females. Med Sci Sports Exerc
. 2006; 38 (3): 598–604.
40. O’Connor PJ, Cook DB. Exercise and pain: the neurobiology, measurement, and laboratory study of pain in relation to exercise in humans. Exerc Sport Sci Rev
. 1999; 27: 119–66.
41. O’Connor PJ, Cook DB. Moderate-intensity muscle pain
can be produced and sustained during cycle ergometry. Med Sci Sports Exerc
. 2001; 33 (6): 1046–51.
42. O’Connor PJ, Motl RW, Broglio SP, Ely MR. Dose-dependent effect of caffeine on reducing leg muscle pain
during cycling exercise is unrelated to systolic blood pressure. Pain
. 2004; 109 (3): 291–8.
43. Park I, Schutz RW. “Quick and easy” forumulae for approximating statistical power in repeated measures ANOVA. Meas Phys Educ Exerc Sci
. 1999; 3: 249–70.
44. Plaskett CJ, Cafarelli E. Caffeine increases endurance and attenuates force sensation during submaximal isometric contractions. J Appl Physiol
. 2001; 91 (4): 1535–44.
45. Rousseau E, Ladine J, Liu QY, Meissner G. Activation of the Ca2+
release channel of skeletal muscle sarcoplasmic reticulum by caffeine and related compounds. Arch Biochem Biophys
. 1988; 267 (1): 75–86.
46. Sawynok J. Adenosine receptor activation and nociception. Eur J Pharmacol
. 1998; 347 (1): 1–11.
47. Sewell DA, McGregor RA. Evaluation of a cycling pre-load time trial
protocol in recreationally active humans. Eur J Appl Physiol
. 2008; 102 (5): 615–21.
48. Shield A, Zhou S. Assessing voluntary muscle activation with the twitch interpolation
technique. Sports Med
. 2004; 34 (4): 253–67.
49. Tarnopolsky M, Cupido C. Caffeine potentiates low frequency skeletal muscle force in habitual and nonhabitual caffeine consumers. J Appl Physiol
. 2000; 89 (5): 1719–24.
50. Walton C, Kalmar J, Cafarelli E. Caffeine increases spinal excitability in humans. Muscle Nerve
. 2003; 28 (3): 359–64.
51. Warren GL, Park ND, Maresca RD, McKibans KI, Millard-Stafford ML. Effect of caffeine ingestion on muscular strength
and endurance: a meta-analysis. Med Sci Sports Exerc
. 2010; 42 (7): 1375–87.
52. Wiles JD, Bird SR, Hopkins J, Riley M. Effect of caffeinated coffee on running speed, respiratory factors, blood lactate and perceived exertion during 1500-m treadmill running. Br J Sports Med
. 1992; 26 (2): 116–20.
53. Zehr EP, Chua R. Modulation of human cutaneous reflexes during rhythmic cyclical arm movement. Exp Brain Res
. 2000; 135 (2): 241–50.