Acute Effects of Caffeine Supplementation on Physical Performance, Physiological Responses, Perceived Exertion, and Technical-Tactical Skills in Combat Sports: A Systematic Review and Meta-Analysis

Inclusion criteria were applied following the PICOS (participants, intervention, comparators, outcomes, and study design) model. Articles were included if they were conducted on active combat sports athletes, using caffeine supplementation as an intervention in any form. The studies were retrieved if caffeine supplementation was controlled by a placebo intervention. Moreover, only the studies including performance measurement using combat simulation tasks and/or physical tests using a double-blind placebo-controlled design were included. These eligibility criteria resulted in the exclusion of studies that were conducted in non-combat sports athletes or did not include a caffeine supplementation protocol. In addition, articles were excluded if caffeine was administered in combination with other substances without a direct caffeine-placebo comparison. Trials that did not follow the double-blind placebo-controlled design were also excluded. Moreover, trials conducted during Ramadan were excluded as fasting can affect total sleep time [40], and induce dehydration [41], which could compromise performance. Furthermore, books, citations, conference proceedings, systematic reviews and narrative reviews were excluded.

The systematic literature search was performed up through 18 April 2022 following the 2020 Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [38] and the PRISMA guidelines adapted to sports science [42]. An a priori protocol was devised and can be requested from the corresponding author. PubMed, Scopus, Web of Science and Cochrane Library were searched without time limits or filters. The search syntax included the following keywords: Caffeine, coffee, combat sports, martial arts, judo, taekwondo, boxing, taekwondo, wrestling and jiu-jitsu. Appropriate Boolean connectors (AND, OR) were utilized to connect the various keywords and medical subject headings; (MeSH) terms were also used where appropriate. Combinations and search results on each database are presented in Table 1.

An additional search was performed by screening the reference lists of the included studies and related review papers using Google Scholar. Two authors performed the search independently. Any discrepancies between the authors in the study selection were resolved by consulting a third reviewer. No language restriction was applied in the search since this fact could introduce the risk of ignoring key data [42]. The search results were imported and treated using the software “Endnote 20” (Camelot UK Bidco Limited-Clarivate, London, UK).

The retrieved articles were firstly assessed for duplication using the software “Endnote 20” (Camelot UK Bidco Limited-Clarivate, London, UK) before being considered for inclusion. After duplicates were removed, a review of all relevant article titles was conducted before an examination of article abstracts and then full published articles. Two reviewers conducted the process independently. Consensus was used to resolve disagreements between the two reviewers. The reasons for the exclusion in each step were recorded.

For all studies meeting the inclusion criteria, data were summarized in a spreadsheet using Microsoft Excel software. A piloted data extraction form with the following items: Author(s), year of publication, study design, sample size and combat sport name, timing after caffeine supplementation, caffeine dose (≤3: low; 4–6: moderate; >6: high), the form of caffeine supplement (capsule; chewing gum), the exercises used to assess performance and the main results was used. The extracted data were presented in Table 2. Moreover, the characteristics of participants in each study were extracted. Data reporting sex (males; females), age (≤18 years: adolescent; >18 years; adult athlete), experience (<5 years: amateur; ≥5 years: elite) of athletes, as well as their habitual caffeine intake were presented in Table 3. The mean and standard deviation were retrieved using WebPlot Digitizer (Pacifica, CA, USA, Version: 4.5) (https://automeris.io/WebPlotDigitizer↗, accessed on 28 April 2022) when data were displayed inside a graphic and no further data was provided upon request. To avoid any selection bias and data extraction flaws, the data extraction process was double-checked by a second reviewer [43]. Otherwise, if extraction was impossible (e.g., the figure was not clear or no figure was presented), the corresponding authors were contacted. If we did not receive an answer, the trial was excluded from the analysis.

The 11-item PEDro scale was used to assess the methodological quality of the included studies [70,71]. The quality of studies was assessed based on the study conducted by Mielgo-Ayuso et al. [72] as follows: excellent quality (9–10 points), good quality (6–8 points), fair quality (4–5 points), and poor quality (<4 points). Two authors assessed the methodological quality independently and checked the studies that already have a score, with discussion and consensus over any observed differences.

Meta-analysis was conducted using the ComprehensiveMeta-Analysis software (version3; Biostat, Englewood, NJ, USA). To conduct meta-analysis, a minimum of three studies was requested. Standardized mean differences (Hedge’s g) and 95% confidence intervals (95% CI) were calculated between the placebo and caffeine trials using the variable mean and standard deviation values, the trial correlation, and the number of participants. Because no correlation values were published in any of the included investigations, a cautious 0.5 correlation was assumed across trials [73]. In the case of a multi-arms study (e.g., more than one caffeine dose), values were averaged across the different conditions [30]. The effects size (Hedge’s g) was interpreted as follows: Small (g < 0.50), medium (0.50 ≤ g < 0.80), and large (g ≥ 0.80) [74]. The inverse-variance random-effects model for meta-analysis was chosen since it assigns a proportionate weight to trials depending on the magnitude of their individual standard errors [75] and makes analysis easier while accounting for study heterogeneity [76]. This choice of weights minimizes the imprecision (uncertainty) of the pooled effect size [75]. To calculate heterogeneity among studies, the Q and I² statistics were used. Heterogeneity was indicated if the Q statistic reached a significance of p < 0.05 and I² > 50% [77]. The I² statistic was classified as low (<50%), moderate (50–75%) and high (>75%) [34]. Because there was an insufficient number of studies to conduct subgroup or meta-regression analysis, the different moderators that could explain the observed heterogeneity (i.e., dose, timing, form, sample age, sample sex, expertise level, tolerance to caffeine, genetic background, time of day) were discussed in a qualitative way to avoid false positive or negative effect due to a lack of information rather than a smaller (or absent) effect [75]. To identify publication bias, funnel plots’ potential asymmetries, Begg and Mazumdar’s rank correlation test [78], and Egger’s linear regression test [79] were used. Moreover, the Duval and Tweedie’s trim and fill method [80] was used to identify missing studies. The stability of the pooled ES of each study was assessed using sensitivity analyses and involved removing individual studies from the analysis and computing the impact of theexcluded study [81]. In addition, a cumulative meta-analysis, which aims to aggregate accumulating evidence with additional studies based on their chronological order [82], was conducted to further ensure the stability and reliability of the results. When data were included in qualitative synthesis (i.e., not in meta-analysis), the percentage of improvement was calculated using the equation proposed by Lopez-Gonzalez et al. [8] [% improvement = (value with supplementation—value with placebo)/value with placebo × 100)]. A significance of p < 0.05 was adopted for all analyses.

Figure 1 provides the flow diagram of the search process. A total of 92 records were found via the initial search. Among these, 48 were excluded as duplicates (40 using Endnote and eight manually), with 44 records remaining. Records’ titles were screened with 16 being removed. After screening the records’ abstracts for relevance, five wereremoved, and 23 reports remained and were assessed for eligibility. By applying the eligibility criteria, two full texts were excluded based on the study design (i.e., does not meet the double-blind placebo-controlled trial criterion) and population (see Table S1 for the list of excluded studies). After conducting an additional search, five articles were identified as potentially relevant studies, resulting in a total of 26 studies included in this systematic review.

The 26 studies included in this systematic review regrouped 17 trials [44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60] from grappling, eighttrials [61,62,63,64,65,66,67,68] from striking combat sports, and one trial [69] from mixed martial arts (MMA).

Within the included studies, caffeine was administered based on the combat sport athletes’ body mass (i.e., relative dose). Using a capsule form, the caffeine doses used in the trials were 2 mg·kg⁻¹ [65], 3 mg·kg⁻¹ [48,49,50,57,59,60,68], 4 mg·kg⁻¹ [45,54], 5 mg·kg⁻¹ [44,46,47,53,56,61,63,65,66,67,69], 6 mg·kg⁻¹ [48,50,53,56,57,60,62], 9 mg·kg⁻¹ [50], and 10 mg·kg⁻¹ [54]. In addition, both the repeated dose (5 × 2 mg·kg⁻¹) and selective dose (i.e., dose given based on performance decrement in the first trials) of caffeine were used [54]. From the 26 included studies, two trials [58,63] examined the effect of caffeine and sodium bicarbonate separately and in combination. Using a capsule form, the timing between caffeine supplementation and the subsequent performance testing was 30 min [44,62], 45 min [54], 50 min [63], 60 min [45,46,47,48,49,50,52,53,55,56,58,59,60,61,63,64,65,66,67,68,69], and 120 min (i.e., testing after a training session) [47]. However, when using a caffeinated chewing gums form, only one study [51] investigated the effect of this supplementation form using both 2.7 mg·kg⁻¹ and 5.4 mg·kg⁻¹ doses 15 min before testing.

Since the objective of each study was to enhance performance as much as possible, from the 26 included studies, six [45,48,49,52,57,63] were multi-arm studies, with variation in dose [48,49,52,57,63] or in timing [47].

To reveal any effect of caffeine supplementation on physical performance, several testing procedures have been proposed. More specifically, in both grappling and striking combat sports, a set of non-specific physical tests were performed. For the upper body, the arm ergometer test [44], bench-press [48,59], and handgrip strength [44,46,47,51,57,62] were the most used tests. For the lower body, the leg press [65], vertical jump tests (i.e., Sergeant and countermovement jump (CMJ) tests) [46,47,57,58,62,63], the repeated sprint anaerobic test [65], maximal static lift [48,49], and the Wingate test [64] were used. However, to examine the effect of caffeine supplementation on specific performance, specific physical fitness tests were used to mimic the technical and physical aspects of real competition. The special judo fitness test (SJFT) [45,47,50,51,52,55,58] and the judogi grip strength test [44,51,57] were used to evaluate grappling performance, while the dollyo-chagui kick test [61,67], the taekwondo specific agility test (TSAT), the 10s frequency speed of kick test (FSKT-10s) and its multiple version (FSKT-mult) in taekwondo [68], the karate specific aerobic test (KSAT) [63], the repeated punching test in mixed martial arts (MMA) [69], and the Pittsburgh Wrestling Performance Test (PWPT) in freestyle wrestling [54]. However, to identify the variation of technical-tactical aspects regarding caffeine supplementation, simulated combats were performed in taekwondo [66,67], boxing [62], judo [44,48,49,54,60], wrestling [54] and Brazilian jiu-jitsu (BJJ) [49].

To investigate the associated physiological and psychological responses to caffeine supplementation, heart rate (HR) [44,45,47,50,51,53,54,55,62,63,66,67], blood lactate concentration ([La]) [44,46,47,49,51,52,53,54,58,63,64,66,67], rating of perceived exertion (RPE) [44,45,47,49,50,51,52,53,56,58,62,63,65,66,67,68,69], rating of perceived fatigue (RPF) [44,54], perceived recovery status (PRS) [56], and pain perception [47,65] were measured.

In the 26 reviewed studies, there were a total of 323 combat sport athletes recruited. From the grappling modality, 223 athletes were recruited with 124 from judo [45,46,47,50,51,52,53,55,56,58,59], 88 from jiu-jitsu [46,47,51,55,58,81], and 11 from freestyle wrestling [54]. From striking sports, 89 athletes were recruited, with 18 from karate [63,65], 53 from taekwondo [61,66,67,68], and 18 from boxing [62,64]. From MMA, 11 athletes were recruited [69].

Regarding the sample sex distribution, a major participation of male athletes was noticed. Sixteen studies recruited male athletes (i.e., 187 athletes), two studies (i.e., 23 athletes) [55,65] recruited female athletes, four studies [55,57,58,66] combined both sexes (i.e., 35 males and 33 females, respectively), and four studies [51,59,61,81] did not identify their sample sex (i.e., 45 athletes).

After checking the score of 10 studies that have been already scored in previous studies [8,24] and assessing the score of the remaining ones, the 26 included studies showed high methodological quality, with one study [47] receiving a score equal to 9 points and all other received 10 points. The study conducted by Carmo et al. [47] failed to satisfy the “less than 15% dropout” criterion since the data were obtained from less than 85% of subjects who were initially recruited. The Pedro scores of the included studies are presented in Table 2.

Caffeine ingestion has been reported to modify perceived exertion (RPE), which can result in altered performance [19]. In the present meta-analysis, there were no significant impacts of caffeine supplementation on RPE post-anaerobic exercise and simulated combats. These findings supported those reported in previous meta-analyses [27,29,32], showing that caffeine did not affect RPE. However, these results contradict those reported by Glaister and Gissane [28] and Grgic and Mikulic [102], who revealed caffeine supplementation has a suppressive effect on RPE following submaximal exercise and a 1RM barbell back squat performance. Moreover, it was reported that the RPE decrease explains up to 29% of the ergogenic effect of caffeine on sub-maximal aerobic exercise performance [19]. Since the reduction in blood flow to the brain during exercise has been identified as a factor promoting fatigue [103], the previously observed effects of caffeine supplementation on RPE are mediated by caffeine’s impact on cerebral oxygenation [89] via crossing the blood-brain barrier [104] and inhibiting the A₁ and A_2a adenosine receptors [16,18,19]. Nevertheless, the discrepancy across studies could be related to several factors including the variation in the testing procedure and the participant profiles. It has been theorized that exercise complexity may affect the amount of fatigue, given that complex, multi-joint exercises stimulate more muscle groups and, hence, demand higher exertion [105]. Under simulated combat conditions, caffeine served to increase combat load, while fatigue perception was unaffected. Thus, it seems possible to confirm caffeine may have the capacity to enhance the physical performance of combat sports athletes without producing higher values of fatigue. The increased time to fatigue may be attributed to the reduction of the central fatigue rate [106]. Moreover, because the specific testing procedures are most similar to competition and mimic the demands of training and in-competition sessions, athletes were familiarized and there was no effect of learning that could increase fatigue levels. Additionally, given that the RPE scale was created to reflect both the heart rate response to exercise and the level of exertion [107], it is perhaps not surprising to find any difference in RPE, as HR post-combats did not appear to be altered by caffeine. Furthermore, the lack of RPE variation in response to high-intensity exercise could be related to other extrinsic factors (i.e., timing of evaluation, athlete profile, time of day, familiarity with the testing procedure).

It should be noted that the use of scales to quantify RPE has been criticized in numerous ways [27]. For instance, Davis and Green [17] demonstrated RPE measures can be coarse in detecting perceptual responses during high-intensity exercise. In addition, Gergic et al. [105] attributed the lack of difference in RPE following caffeine ingestion to the timing of evaluation as, in high-intensity exercise, RPE is evaluated almost exclusively at exercise termination. In combat sports, the high-intensity efforts are intercepted by low-intensity periods and do not follow a constant load pattern, which could explain the lack of caffeine effect on RPE found in the present meta-analysis. This may be evident, as caffeine ingestion reduced RPE recorded during exercise, but does not affect RPE obtained at the end of exhausting exercise [19]. Caffeine appears to delay autonomic recovery following exercise [108], as throughout judo matches, RPE increased and RPR decreased independent of performance enhancement [56]. All studies included in our meta-analysis only included an RPE assessment at the end of the test or combat, but it remains interesting for future studies to consider RPE throughout exercise.

The present meta-analysis revealed caffeine supplementation did not improve the number or duration of attacks. This result was surprising as caffeine ingestion has been associated with anincrease in anxiety arousal, energy, impulsiveness, and risk-taking [65,107,108], which could affect high-intensity offensive activities. Our findings were partially (i.e., only for the number of offensives actions) in line with those reported in a previous meta-analysis [37], showing caffeine supplementation did not affect the number of offensive actions during the first combat. However, the previous meta-analysis reported caffeine improved the number of offensive actions in the second combat (SMD = 0.40) and the duration of offensive actions (SMD = 0.58). As we selected data only from the first combat, the discrepancy between our results and those of Diaz-Lara et al. [37] concerning the effect of caffeine supplementation on the duration of offensive actions may be related to the inclusion of data from the first and the second combat in their meta-analysis. The lack of performance enhancement after caffeine supplementation and the high heterogeneity across studies maybe related to the divergence in combat duration and modality (i.e., striking vs. grappling). In fact, it was suggested that match-derived judo performance variables were not sensitive enough to distinguish between small or medium effects of caffeine [46,56]. This has been explained by the fact that judo combat is composed by acyclic actions and is dependent on the opponent’s level and experience and his/her technical-tactical strategies used during the fight [109]. Moreover, it seems that combat duration and level play a crucial role. In fact, a low dose (i.e., 3 mg·kg⁻¹) of caffeine was sufficient to increase the number of offensive actions by 28% in 8 min of Brazilian jiu-jitsu simulated combat [49]. This increase in high-intensity actions was linked to a trend toward greater advantages, scoring actions, and submissions [49]. Furthermore, in male taekwondo athletes, Santos et al. [67] reported that 5 mg·kg⁻¹ of caffeine 1 h before two simulated taekwondo combats did not increase the number of attacks in the first combat; however, a significant improvement of 37% was recorded in the second combat. Contrarily, using the same dose and respecting the same timing after supplementation, Lopes-Silva et al. [66] did not report time-motion variables’ improvement during simulated taekwondo combat. These contradictory results were explained by the fact that combat in the study conducted by Santos et al. [67] was performed between supplemented athletes (i.e., caffeine versus caffeine), while Lopes-Silva et al. [66] used supplemented athletes confronting non-supplemented ones. Through this, it was suggested that during simulated taekwondo combat, it is difficult to observe any performance variables’ improvement when caffeine is supplemented from one side (i.e., only one opponent receives caffeine) [66]. This result may be explained by the fact that in a 1 vs. 1 competitive situation, combat athletes’ behavior is directly influenced and constrained by that of the opponent [5]. Moreover, despite similarities between combat sports (i.e., intermittence), there are considerable differences between grappling and striking in terms of the technical-tactical characteristics and the effort-to-pause ratio [1,63]. For example, grappling combat sports require strength-endurance and power to perform grappling and throwing techniques [4]. However, striking combat sports are characterized by high-speed attack and defense movements to perform kicking and punching techniques [3,110]. These differences could explain, in part, the contradictory results between the two modalities, even when the same dose and timing after supplementation were provided (i.e., [67] vs. [46,56]). Therefore, and as the effort to pause ratio is modality-dependent [1], results from striking cannot be transferred to grappling contests, particularly, for technical-tactical skills.

When conducting caffeine supplementation studies, inter-individual variability (i.e., sex, age, habitual caffeine consumption and training status) and supplementation protocols (i.e., dose, timing, form and time of day of measurement) should be taken into consideration.

Excessive doses of caffeine consumption are known to induce a high incidence of side effects [11,124], such as tachycardia, anxiety, gastrointestinal discomfort, tremors, hypertension, and insomnia [119]. Among the included studies, Lopes-Silva et al. [66] reported caffeine produced an increase in the tachycardia sensation in 20% of athletes and anxiety/nervousness in 10% of athletes. In addition, Felippe et al. [58] investigated the possible gastrointestinal discomfort caused by a combined supplementation of sodium bicarbonate (NaHCO₃) and caffeine. In this study, the authors showed that only one athlete reported flatulence following supplementation [58]. However, this gastrointestinal discomfort was attributed to the ingestion of NaHCO₃, but not to caffeine. To avoid the negative effects of NaHCO₃ when it was administrated alone or with caffeine, it was suggested that ingesting divided NaHCO₃ doses is the best strategy [58,63]. However, among freestyle wrestlers, the supplementation with high doses of caffeine alone increased urine volume and the dehydration index when compared to all other conditions [54]. The same study showed the scores of gastrointestinal complaints and gastrointestinal discomfort were higher with the high dose of caffeine and the repeated doses of caffeine, in comparison with the other conditions [54]. Moreover, in BJJ athletes, it was reported that caffeine administration increased the rate of fatigue sensation even when low doses (i.e., 3 mg·kg⁻¹) were administered [48].

As assessed using the Pedro scale, the included studies are classified as being of good or excellent methodological quality. However, in these studies, there was minor participation of females. Possibly, the assessment of the effect of caffeine on female athletes is considered complex as both oral contraceptives and the different phases of the menstrual cycle are designated to be factors that may affect caffeine metabolism [140]. Therefore, the menstrual cycle phases are one methodological element to keep in mind when conducting research on female athletes [140]. However, among the six studies including female athletes [53,55,57,58,63,66], only one [68] has indicated that menstrual cycle phases were considered as an exclusion/inclusion criterion for female athletes. Moreover, despite sex beingan effective moderator that can influence the outcomes, four studies [51,59,61,81] did not indicate the sex of their athletes. Thus, future studies should identify the sex of participants explicitly.

Caffeine habituation is difficult to be quantified accurately [141]. In the 26 included studies, it seems that the dose of caffeine consumed per day was given as an approximation. However, measuring objective markers, such as urine caffeine production, plasma caffeine, or caffeine metabolite levels, could be a solution [142]. Otherwise, future studies should ensure that a validated questionnaire for assessing habitual caffeine intake is used (e.g., the questionnaire proposed by Bühler et al. [143]). Additionally, there is no clear classification concerning habitual caffeine intake that can help to categorize athletes into light, moderate and high consumers. Only in two studies [50,68], there was a clear classification. In this context, Durkalec-Michalski et al. [50] considered athletes who consumed less than 160 mg·kg⁻¹ as light consumers, whereas, those who consumed more were classified as high consumers. However, Filip et al. [144] proposed a recent classification according to individual’s habitual caffeine consumption with six categories [i.e., Naïve consumer (<25 mg/day); Low consumer (25 mg/day to 0.99 mg/kg/day); Mild consumer (1.00–2.99 mg/kg/day); Moderate consumer (3.00–5.99 mg/kg/day); High consumer (6.00–8.99 mg/kg/day); Very high consumer (>9.00 mg/kg/day)]. Thus, future researchers should consider this categorization to minimize possible methodological bias.

Regarding power analysis, it is interesting to highlight the fact that sample sizes recruited in the majority of the included studies were less than 20 participants, which often limited the statistical power to detect significant effects [145]. Therefore, it is recommended to recruit larger samples in future investigations.

Finally, in some circumstances, giving a placebo with the implication that it is caffeine has been demonstrated to improve performance and reduce RPE [105]. In this regard, Grgic [30] attributed the discrepancy between studies to the effect of an error in the measurement rather than the effects of the condition; the success of blinding could determine the result. Among the 26 included studies, only three have tested the success of blinding. Therefore, future research should consider the effectiveness of blinding.

This is the first systematic review with meta-analysis that addressed the effects of caffeine supplementation on combat sports physical performance, technical-tactical skills, perceived exertion and physiological response. The strengths of the present investigation include a comprehensive exposure of the available literature and a careful appraisal of its value. Moreover, the search was conducted without any language or year restriction. In addition, the study was conducted according to the updated PRISMA guidelines. Furthermore, the recommendations issued from this study can help to draw strong investigations into the topic of caffeine supplementation and athletic performance, particularly, in the context of combat sports. However, the present systematic review presents some limitations related to the inclusion criteria. Indeed, investigations in which caffeine was co-ingested with other substances (i.e., sodium bicarbonate) were selected, which can lead to other effects. Moreover, meta-analytical calculations were not possible for the different outcomes, due to the limited number of studies, and diverse methodology. Furthermore, meta-regression and subgroup analysis were not performed, respecting the Cochrane handbook recommendations [39]. Further studies, such as those using a mixed-sex sample and undertaking between-group comparisons, are warranted. In addition, recruiting samples with different competitive levels and ages could help to generalize the effect of caffeine administration on combat sports’ performance atdifferent expertise levels or ages. Moreover, in the included studies, only three compared the effects of low and moderate doses of caffeine. Thus, future protocols should be designed to compare the effects of caffeine depending on the dose and form, particularly in striking combat sports. Furthermore, because combat sports are contact sports where tactical behaviors are needed, further investigations are necessary to determine the effects of caffeine supplementation in real competition outcomes.

The improvement in performance following caffeine supplementation, without affecting the metabolism process in some studies [53,63], could be informative of other physiological responses than glycogen use. For example, the improvement in physical performance was associated with a 12% increase in testosterone concentration [146]. This hormone is responsible for the increase in risk-taking behaviors and aggressiveness [146], which are considered necessary responses to improve performances in both striking [67] and grappling [49] contests. However, information about testosterone variation in response to caffeine supplementation remains speculative, and future works are warranted.

Data that support the findings of this study are available from the corresponding author upon reasonable request.