Acute caffeine supplementation and live match-play performance in team-sports: A systematic review (2000–2021)

ABSTRACT

Caffeine is a psycho-active stimulant that can improve physical and cognitive performance. We systematically reviewed the evidence on the effects of acute caffeine ingestion on physiological parameters, physical and technical-skill performance during high-performance team-sport match-play. Following PRISMA guidelines, studies were identified using scientific databases (PubMed, Web-of-Science, Scopus, and SPORTDiscus) in February 2021. Of 281 results, 13 studies met inclusion, totalling 213 participants. Included studies adopted the randomised double-blinded cross-over design, involving caffeine and control conditions. In studies reporting physiological variables, responses to caffeine included higher peak (n=6/ 8 [n/ total studies measuring the variable]) and mean (n=7/ 9) heart rates, increased blood glucose (n=2/ 2) and lactate (n=2/ 2) concentrations. Improvements in physical performance were widely documented with caffeine, including greater distance coverage (n=7/ 7), high-speed distance coverage (n=5/ 7) and impact frequencies (n=6/ 8). From three studies that assessed technical-skills, it appears caffeine may benefit gross-skill performance, but have no effect, or negatively confound finer technical-skill outcomes. There is compelling evidence that ingesting moderate caffeine doses (~3 to 6 mg·kg−1) ~60 minutes before exercise may improve physical performance in team-sports, whereas evidence is presently too scarce to draw confident conclusions regarding sport-specific skill performance.

               

KEYWORDS:

• Caffeine
• supplement
• stimulant
• team-sports
• ergogenic aid

Introduction

Caffeine (1,3,7-trimethylxanthine) is the most widely consumed central nervous system (CNS) stimulant. The pharmacological effects of caffeine include increased wakefulness, decreased reaction times and an increased ability to perform and sustain cognitive and physical activity (Keisler & Armsey, 2006). Mechanisms of action are believed to target a wide range of organ systems, highlighting the potential for caffeine to enhance various inter-related aspects of human performance, including cognition, strength, speed, endurance and skill. Cognitive effects of caffeine relate to central fatigue resistance and effort perception. The competitive binding of caffeine and paraxanthine to adenosine A₁ and A₂ receptors in the brain inhibits neuro-modulatory actions of adenosine, which upregulate CNS activity and sympathetic neurotransmitter release (e.g., dopamine; Graham & Spriet, 1995). Caffeine also stimulates the secretion of beta-endorphins which, due to their analgesic properties, may reduce the perception of pain (Laurent et al., 2000).

Peripheral effects of caffeine on skeletal muscle contraction and fatigue also involve an interference with calcium uptake and storage in the sarcoplasmic reticulum of striated muscle, an increased calcium ion (Ca++) translocation through the plasma membrane of muscle cells and an increased myofilament Ca++ sensitivity (Nehlig et al., 1992); thus optimising myofibrillar contractions. Further, clinical studies have proposed that ingesting caffeine before exercise can enhance lipolysis via inhibition of phosphodiesterase, along with direct effects on muscle glycogen sparing via inhibition of glycogen phosphorylase (Da Silva et al., 2018). Caffeine may also increase phosphorylation of AMP-activated protein kinase (AMPK) and enhance GLUT4 translocation to the plasma membrane of muscle cells and glycogen synthase (GS) activation, which can improve AMPK-dependent glucose uptake in skeletal muscle (Jensen et al., 2007). These factors may facilitate a decreased reliance on muscle glycogen as a metabolic fuel during exercise and simultaneously increase non-esterified fatty acid (NEFA) oxidation for energy provision (Arciero et al., 1995). Together, mechanisms at the central and peripheral level may therefore culminate in greater motor unit recruitment and power output, a delayed onset of fatigue and/or a decreased perception of effort with caffeine intake before endurance or high-intensity exercise. As such, caffeine is widely used as a supplement to improve sports performance.

Caffeine ingestion can benefit endurance and sustained high-intensity performance. Typical guidelines involve ingesting 3 to 6 mg · kg⁻¹ (i.e., of total body mass) one hour prior to exercise (Graham & Spriet, 1995). A meta-analysis on caffeine and endurance performance (measured with time-trial protocols) reported a mean ± SD improvement of 3.2 ± 4.3% with caffeine across 21 included studies with a total of 33 caffeine treatments (Ganio et al., 2009). However, the range of effects was heterogeneous (−0.3 to 17.3%) and one possibility is that habituation and training status may be important considerations when interpreting the ergogenic effects of caffeine intake. Caffeine intake can also benefit anaerobic exercise performance (Collomp et al., 1992), and this may be explained by the ability of caffeine to efficiently cross cellular membranes, including the blood-brain barrier (Keisler & Armsey, 2006). Indeed, caffeine appears to be an effective ergogenic aid across a range of physically taxing sporting activities.

Caffeine research in intermittent team-sports yields less consistent findings than that of endurance-type disciplines, and this is likely attributable to heterogeneity in study design and the complex nature of team-sports. Team-sports involve repeated high-intensity efforts and technical-skill performance over periods of ~1–2 hours (Baker et al., 2015). In competition, team-sport athletes must display high levels of strength, speed, agility and endurance. Team-sports are characterised by rapid transitions between high-intensity (anaerobic) and lower intensity (aerobic) efforts, both of which rely on an effective use and recovery of metabolic fuels (Narazaki et al., 2009). Further, efforts requiring high neuromuscular outputs, such as sprinting and jumping, are key predictors of success in team-sports (Gabbett, 2012). Indeed, acute caffeine intake has been observed to improve sprint performance in team-sport simulations and real match-play, both in terms of sprinting actions and distances (Lara et al., 2014; Stuart et al., 2005).

Moreover, a central aspect of team-sport performance is sport-specific skill execution. It is noteworthy that the evidence in relation to the effects of caffeine on technical-skill performance in team-sports are scarce and available studies have reported conflicting conclusions (Foskett et al., 2009; Portillo et al., 2017; Puente et al., 2017; Roberts et al., 2010). Perhaps, caffeine ingestion may be ergogenic to skill performance only in specific team-sport situations. Remarkably, no consensus currently exists on this topic. The dynamic physiological demands of team-sports present a rationale for investigating viable nutritional strategies to maximise performance.

Despite clear evidence for the ergogenic effects of caffeine on isolated aspects of team-sport performance – e.g., strength (Grgic et al., 2018) and endurance (Ganio et al., 2009) and skill (Foskett et al., 2009) – very few studies have examined the ecological validity of these effects during live competition. Thus, little is known on the effects of caffeine on performance under match-play specific fatigue (i.e., its ecological value). It is likely that current data on caffeine and team-sports performance are equivocal partly because the tests conducted in multiple studies are indirect/ proxy tests of match-play and so may be lacking internal validity. A major challenge for sports scientists is that the vast complexity of team-sports competition does not lend itself well to conduct physiological measurements during match-play. Further, game-to-game variability in physiological, physical, technical and tactical demands complicates standardisation of the environment chosen for tightly controlled cross-over trials which aim to generate precise comparisons.

Evidence on the effects of acute caffeine intake on physiological parameters, physical and technical-skill performance during live team-sport match-play has never been systematically reviewed. This is crucial given the prevalent use of caffeine and the fact that caffeine is a legal, un-regulated substance in most countries (Renda & De Caterina, 2020). Therefore, this study systematically reviewed the evidence addressing the effects of acute caffeine intake (alone, or as part of a product) in high-performance intermittent team-sport athletes on physiological parameters (e.g., blood glucose/lactate), physical (e.g., movement parameters) and technical-skill (e.g., skill execution) performance outcomes during live match-play.

Materials and methods

Search strategy

This review followed the Preferred Reporting Items for Systematic-Reviews and Meta-Analysis (PRISMA) guidelines (Moher et al., 2009). The literature search was conducted using PubMed, Web-of-Science (all databases), Scopus and SPORTDiscus. The search was restricted to publications between the 1^st of January 2000 and the 22^nd of February 2021. This timeframe was appropriate as the majority of investigations examining caffeine’s effect on intermittent sport performance have been conducted onwards from 2005 (Burke, 2008).

The search strategy used was: (concept 1) (caffeine OR coffee) AND (concept 2) (supplement OR supplementation OR “ergogenic aid”) AND (concept 3) (basketball OR football OR rugby OR soccer OR hockey OR handball OR netball). Search terms included Medical Subject Headings (MeSH) and Text Words. The search criteria for PubMed is displayed in Table 1. The search conditions were identical for Web-of-Science, Scopus and SPORTDiscus, but terms were not included as MeSH. The search results (Titles & abstracts) were downloaded to Endnote (X8; Clarivate Analytics, UK) and duplicates were eliminated manually. Titles and abstracts were screened for full-text review by two investigators (AAS & IT), independently. Disagreements were resolved through discussion. Where access to a manuscript was not possible – e.g., due to recent journal acceptance (Raya-González et al., 2021) – authors were contacted directly, and the manuscript and data were obtained.

Table 1. Search criteria as conducted in the PubMed database

(“Caffeine“[MeSH Terms] OR “Coffee“[MeSH Terms] OR “Energy Drinks“[MeSH Terms] OR “Caffeine“[Text Word] OR “caffeinated“[Text Word] OR “coffee*“[Text Word] OR “energy drink*“[Text Word]) AND (“Dietary Supplements“[MeSH Terms] OR “Performance-Enhancing Substances“[MeSH Terms] OR “supplement*“[Text Word] OR “supplementation“[Text Word] OR “ergogenic aid*”[Text Word]) AND (”Basketball”[MeSH Terms] OR ”Football”[MeSH Terms] OR ”Soccer”[MeSH Terms] OR ”Hockey”[MeSH Terms] OR ”basketball*”[Text Word] OR ”football*”[Text Word] OR ”rugby”[Text Word] OR ”Soccer”[Text Word] OR ”Hockey”[Text Word] OR ”handball*”[Text Word] OR ”netball*”[Text Word])

Study selection: inclusion and exclusion criteria

Figure 1 displays the search and selection process. To warrant inclusion, studies had to meet the following criteria: a) an experimental trial published in a peer-reviewed journal, or a doctoral or master’s thesis; b) assessed the effects of caffeine ingestion (in capsule, liquid, gum or gel format) on performance during match-play in the physical setting that corresponds to the sport (e.g., basketball court, rugby pitch, etc.). This included: live structured match-play (i.e., real matches), semi-structured match-play (e.g., training scrimmages) or modified match-play (e.g., small-sided games); c) included measurement of performance variables relevant to the sport (e.g., running distance, body impacts, points scored, etc.); d) used a repeated measures design where an experimental condition involving caffeine ingestion was compared to an identical condition without caffeine. This was necessary as the control condition serves as a reference value to assess the effect of the caffeine intervention whilst controlling for inter-individual heterogeneity in responses to caffeine as much as possible; e) reported information on caffeine administration (i.e., dose per kilogram of body mass [mg · kg⁻¹], timing of intake, format, etc.); f) investigated high-performance team-sport athletes, defined as: Athletes competing at academy, collegiate (highest level only; e.g., NCAA Division-1, BUCS First Teams), semi-professional or professional levels in one of the following disciplines: basketball, soccer, handball, rugby, ice or field hockey, American football, Futsal or netball; g) investigated humans without disease or injury. Exclusion criteria applied to reject potential studies, included: a) violating any of the inclusion criteria; b) studies not related to sports nutrition; c) not including a caffeine supplementation protocol or using caffeine doses <2 mg · kg⁻¹ – based on past research reporting no effect with <2 mg · kg⁻¹ of caffeine on exercise performance (Astorino et al., 2010); d) studies involving chronic supplementation (i.e., multiple days).

Figure 1. Flow diagram of the search and study selection process following the Preferred.

Data extraction and synthesis

The following information was tabulated on a pre-defined spreadsheet using Microsoft Excel (Version 16.41, WA, USA) for all studies: a) reference: author, publication year, and study design; b) participant profile: sample, sex, age, sport and level; c) supplementation strategy: form, dose, and time of ingestion; d) control: constituents of product ingested for control conditions; e) assessment: match-play protocol used with detail concerning duration, environment, rules and regulations; f) outcome measure(s): physiological, physical and technical-skill measures, and units of measurement. When values were not available, authors were contacted directly to request these (e.g., Madden et al., 2019); g) results: means and standard deviation (SD) for control and caffeine conditions, and mean differences with pooled SD; h) Cohen’s d was calculated using the equation: = ((mean value with caffeine – mean value with control)/(average (caffeine SD, control SD))). Effect sizes as operated in (Hopkins et al., 2009): ≤0.20 (trivial), 0.20 to 0.59 (small), 0.60 to 1.19 (moderate), 1.20 to 1.99 (large), 2.00 to 3.99 (very large), ≥4.00 (extremely large), and p-values; i) methodological quality score.

Methodological quality

Two researchers (AAS & AH) independently appraised methodological quality of each study using the Physiotherapy Evidence Database (PEDro) Scale (Appendix). Studies were rated according to whether they met each criterion of the PEDro scale (e.g., yes, no, unsure), and the overall score for each study (i.e., the sum of “yes” answers) was allocated. Disagreements were resolved through discussion. The scale encompasses information about randomisation, blinding, statistical analysis and presentation of results. The PEDro scale has been reported as valid and reliable in assessing the internal validity of randomised-controlled trials (Maher et al., 2003). Included studies were classified by strength of evidence as follows: 9–10 (excellent), 6–8 (good), 4–5 (fair) or < 4 (poor).

Results

Study selection

The database search yielded 281 studies, of which 273 were published after 2000. As seen in Figures 1, 70 full-text articles were identified for this review. Of these, only 13 (soccer [n = 4]; basketball [n = 3]; handball [n = 1]; ice hockey [n = 1]; field hockey [n = 1]; Rugby Sevens [n = 2]; Rugby Union [n = 1]) remained after applying exclusion criteria.

Study characteristics

Table 2 contains a summary of the 13 studies. The studies were published between 2012 and 2021. All 13 investigations adopted the randomised double-blind cross-over design. The total number of participants was 213 (79 soccer, 34 basketball, 31 handball, 14 ice hockey, 13 field hockey, 16 rugby sevens, and 26 rugby union athletes). To note, the study by Puente et al. (2018) is a sub-analysis of Puente et al. (2017) investigating the influence of the CYP1A2-163 C > A polymorphism on caffeine’s ergogenicity in basketball. Similarly, Portillo et al. (2017) conducted a secondary analysis of accelerometer data from Del Coso et al. (2013a). No outcome measures were duplicated in the final analyses.

Table 2. Summary of studies included in the systematic review on acute caffeine supplementation and match-play performance in high-performance team-sports

Reference (Study design)	Participant profile	Supplementation strategy	Caffeine format	Control	Performance assessment
Muniz Guttierres et al. (2013) (RDB)	20 elite soccer players (age: 16.1 ± 0.7 years) (sex: N/A)	250 mg/L^−1 20 min before match. 5 mL·kg^−1 every 15 min of match. Average intake: 7.2 mg·kg^−1	Caffeinated sports drink	Carbohydrate drink (7% carbohydrate)	2 x full (90 min) soccer games with 48-hour interval in between
Pettersen et al. (2014) (RDB)	22 male regional or national-level Soccer athletes (age: 17.6 ± 1.1 years)	6 mg·kg^−1 50 min before warm-ups	Gelatin capsule with caffeine	Identical capsule with dextrose	2 x in-season matches separated by 7 days
Del Coso et al. (2012) (RDB)	19 semi-professional Soccer athletes (age: of 21 ± 2 years) (sex: N/A)	3 mg·kg^−1 60 min before testing	Caffeine drink (sugar-free Red Bull®)	Control drink (sugar-free decaffeinated Pepsi®)	2 x 40 min games
Del Coso et al. (2013a) (RDB)	16 female Rugby athletes from Spain’s National Rugby Sevens team (age: 23 ± 2 years)	3 mg·kg^−1 60 min before competition	Caffeine energy drink (Fure, ProEnergetics,)	Same drink with no caffeine (placebo)	2 x competition days (Mon. & Thurs.): 3 matches per day
Del Coso et al. (2013b) (RDB)	26 male professional Rugby Union athletes (age: 25 ± 2 years)	3 mg·kg^−1 60 min before testing	Caffeine energy drink (Fure, ProEnergetics)	Same drink with no caffeine (placebo)	2 x 30 min games
Del Coso et al. (2016) (RDB)	13 male professional Field Hockey players (age: 23.2 ± 3.9 years)	3 mg·kg^−1 60 min before testing	Caffeine energy drink (Fure, ProEnergetics)	Same drink with no caffeine (placebo)	2 x 50 min games
Lara et al. (2014) (RDB)	18 female elite Soccer players (age: of 21 ± 2 years)	3 mg·kg^−1 60 min before testing	Caffeine energy drink (Fure, ProEnergetics)	Same drink with no caffeine (placebo)	2 x 40 min games
Madden et al. (2019) (RDB)	14 male collegiate Ice Hockey players (age 22.6 ± 0.4 years)	3 mg·kg^−1 60 min before testing	Gelatin capsule with caffeine	Identical placebo capsules (maltodextrin NF)	2 x scrimmages (3 x 8 min periods)
Muñoz et al. (2020) (RDB)	31 professional Handball players (age: 23.7 ± 2.8 years) (Females: 15)	3 mg·kg^−1 60 min before testing	Gelatin capsule with caffeine	Identical capsule with cellulose	2 x 40 min games
Portillo et al. (2017) (RDB) ‡	16 female athletes, Spain’s Rugby Sevens team (age: 23 ± 2 years)	3 mg·kg^−1 60 min before competition	Caffeine energy drink (Fure, ProEnergetics)	Same drink with no caffeine (placebo)	2 x competition days (Mon. & Thurs.): 3 matches per day
Puente et al. (2017) (RDB)	10 professional female (age: 27.9 ± 6.1 years) and 10 semi-professional male (27.1 ± 4.0 years) basketballers	3 mg·kg^−1 60 min before testing	Unidentifiable caffeine capsule	Identical capsule with cellulose	2 x 20 min games
Puente et al. (2018) (RDB) †	19 elite basketballers (age: 28.1 ± 4.75 years) (Females: 9)	3 mg·kg^−1 60 min before testing	Unidentifiable caffeine capsule	Identical capsule with cellulose	2 x 20 min games
Raya-González et al. (2021) (RDB)#	14 professional male basketballers (age: 21 ± 2 years)	6 mg·kg^−1 60 min before testing	Caffeine in 250 mL of water	Placebo (sucrose) in 250 mL of water	2 x games (3 x 7 min periods)

Key – N/A: Not Available; RDB: randomised double-blinded study; ‡: secondary analysis of (Del Coso et al., 2013); †: sub-analysis of (Puente et al., 2017); #: not presently available to public [data/ manuscript retrieved from authors directly].

Competitive levels ranged from academy to professional. The caffeine doses consumed were: 3 mg · kg⁻¹ (n = 10), 6 mg · kg⁻¹ (n = 2) and 7.2 mg · kg⁻¹ (n = 1). Caffeine was consumed at different times: 60 minutes prior to testing (n = 11), 50 minutes prior to testing (n = 1), and 20 minutes before a match followed by every 15 minutes during the match (n = 1). In 6 studies caffeine trials involved ingesting an energy drink containing additional ingredients, such as maltodextrin, taurine, _L-carnitine, B-group vitamins and sodium bicarbonate (Table 2). To note, some of these ingredients may have ergogenic properties and affect physiological and physical outcomes. However, in all said studies the placebo drink was identically matched for all ingredients except caffeine, therefore isolating (as feasible) caffeine’s effect on variables of interest (Table 2).

Several performance assessments were employed. Pettersen et al. (2014), Portillo et al. (2017), and Del Coso et al. (2013a) analysed the effects of caffeine during real match-play competitions. Muniz Guttierres et al. (2013) conducted two full soccer matches with a 48-hour interval in between. The remaining nine studies assessed performance during modified match-play where the playing duration was reduced compared to an official match in the sport (Table 2). In all trials, participants were allocated into their habitual playing positions and match-play was officiated by a coach or referee.

Study results

Tables 3–5 detail the results of the 13 studies reviewed. Two studies reported no effects on any outcome measure with acute caffeine ingestion (Muñoz et al., 2020; Raya-González et al., 2021). Eleven studies reported differences in some, but not all measures, and no study reported differences in all measures. The range of effects of caffeine on outcome measures across studies was heterogeneous (ES range: −1.18 [moderate, favouring placebo] to 1.27 [large, favouring caffeine]). Only one large effect size was observed for one variable (blood glucose post-second half [mmol · L⁻¹]; ES = 1.27) in one study (Pettersen et al., 2014).

Table 3. Effects of acute caffeine supplementation on physiological measures during match-play in high-performance team-sports

Reference	Outcome measure (units)	Control (mean ± SD)	Caffeine (mean ± SD)	Mean difference (∆ ± pooled SD)	Cohen’s d	Effect size	Direction of effect	p-value
Muniz Guttierres et al., 2013	Mean heart rate (b·min^−1)	154.00 ± 8.90	158.00 ± 8.20	4.00 ± 8.55	0.47	small	favours caffeine	N/A
		Blood glucose (mg·dL^−1) [post-first half]	143.00 ± N/A	163.00 ± N/A	20.00 ± N/A	N/A	N/A	favours caffeine	0.01*
		Blood glucose (mg·dL^−1) [post-second half]	124.00 ± N/A	131.00 ± N/A	7.00 ± N/A	N/A	N/A	favours caffeine	0.87
		Blood lactate (mmol·L^−1) [post-first half]	3.90 ± N/A	5.10 ± N/A	1.20 ± N/A	N/A	N/A	favours caffeine	< 0.05*
		Blood lactate (mmol·L^−1) [post-second half]	3.20 ± N/A	4.90 ± N/A	1.70 ± N/A	N/A	N/A	favours caffeine	< 0.05*
Pettersen et al., 2014	Mean Heart Rate (b·min^−1)	168.00 ± 8.00	166.00 ± 11.00	−2.00 ± 9.50	−0.21	small	favours placebo	0.67
		Peak Heart Rate (b·min^−1)	192.00 ± 5.00	192.00 ± 8.00	0.00 ± 6.50	0.00	no effect	no effect	> 0.05
		Weight loss (kg^−1) §	1.20 ± 0.60	1.20 ± 0.60	0.00 ± 0.60	0.00	no effect	no effect	> 0.05
		Blood glucose (mmol·L^−1) [post-first half]	6.30 ± 1.60	7.50 ± 2.10	1.20 ± 1.85	0.65	moderate	favours caffeine	< 0.05*
		Blood glucose (mmol·L^−1) [post-second half]	5.40 ± 0.40	6.10 ± 0.70	0.70 ± 0.55	1.27	large	favours caffeine	0.07
		Blood lactate (mmol·L^−1) [post-first half]	6.10 ± 3.60	7.90 ± 4.80	1.80 ± 4.20	0.43	small	favours caffeine	< 0.05*
		Blood lactate (mmol·L^−1) [post-second half]	5.70 ± 2.90	5.50 ± 2.60	−0.20 ± 2.75	−0.07	trivial	favours placebo	> 0.05
Del Coso et al., 2012	Mean heart rate (b·min^−1)	161.00 ± 12.00	160.00 ± 10.00	−1.00 ± 11.00	−0.09	trivial	favours placebo	> 0.05
		Peak heart rate (b·min^−1)	197.00 ± 12.00	196.00 ± 7.00	−1.00 ± 9.50	−0.11	trivial	favours placebo	> 0.05
		Sweat rate (L·h^−1) [during match]	1.00 ± 0.30	1.10 ± 0.30	0.10 ± 0.30	0.33	small	favours caffeine	> 0.05
Del Coso et al., 2013	Mean heart rate (b·min^−1)	164.00 ± 4.00	168.00 ± 7.00	4.00 ± 5.50	0.73	moderate	favours caffeine	< 0.05*
		Peak heart rate (b·min^−1)	188.00 ± 9.00	189.00 ± 10.00	1.00 ± 9.50	0.11	trivial	favours caffeine	0.62
		Sweat rate (L·h^−1)	1.40 ± 0.50	1.40 ± 0.30	0.00 ± 0.40	0.00	no effect	no effect	0.99
		Dehydration level (%) [post-match]	0.60 ± 0.60	0.50 ± 0.50	−0.10 ± 0.55	−0.18	trivial	favours placebo	0.67
Del Coso et al., 2013	Peak heart rate (b·min^−1)	185.00 ± 12.00	189.00 ± 12.00	4.00 ± 12.00	0.33	small	favours caffeine	0.92
		Mean heart rate (b·min^−1)	145.00 ± 8.00	151.00 ± 11.00	6.00 ± 9.50	0.63	moderate	favours caffeine	0.19
		Sweat rate (L·h^−1) [during match]	1.80 ± 0.60	1.80 ± 0.80	0.00 ± 0.70	0.00	no effect	no effect	> 0.05
		Dehydration level (%) [post-match]	1.30 ± 0.60	1.30 ± 0.90	0.00 ± 0.75	0.00	no effect	no effect	> 0.05
Del Coso et al., 2016	Peak heart rate (b·min^−1) [first half]	181.00 ± 9.00	185.00 ± 11.00	4.00 ± 10.00	0.40	small	favours caffeine	0.03*
		Peak heart rate (b·min^−1) [second half]	190.00 ± 9.00	187.00 ± 11.00	−3.00 ± 10.00	−0.30	small	favours placebo	0.44
		Mean heart rate (b·min^−1) [first half]	154.00 ± 14.00	158.00 ± 16.00	4.00 ± 15.00	0.27	small	favours caffeine	0.34
		Mean heart rate (b·min^−1) [second half]	156.00 ± 13.00	158.00 ± 20.00	2.00 ± 16.50	0.12	trivial	favours caffeine	0.60
		Sweat rate (L·h^−1) [during match]	1.00 ± 0.30	0.90 ± 0.40	−0.10 ± 0.35	−0.29	small	favours placebo	0.30
		Dehydration level (%) [post-match]	−0.10 ± 0.70	−0.20 ± 0.80	0.10 ± 0.75	0.13	trivial	favours caffeine	0.50
Lara et al., 2014	Mean heart rate (b·min^−1)	152.00 ± 13.00	158.00 ± 12.00	6.00 ± 12.50	0.48	small	favours caffeine	0.10
		Peak heart rate (b·min^−1)	182.00 ± 6.00	186.00 ± 5.00	4.00 ± 5.50	0.73	moderate	favours caffeine	0.16
		Dehydration level (%) [post-match]	0.20 ± 0.80	0.20 ± 0.80	0.00 ± 0.80	0.00	no effect	no effect	0.86
		Sweat rate (L·h^−1) [during match]	0.60 ± 0.10	0.50 ± 0.10	−0.10 ± 0.10	−1.00	moderate	favours placebo	0.47
Puente et al., 2017	Mean heart rate (b·min^−1)	157.00 ± 13.00	161.00 ± 10.00	4.00 ± 11.50	0.35	small	favours caffeine	0.30
		Peak heart rate (b·min^−1)	185.00 ± 12.00	188.00 ± 10.00	3.00 ± 11.00	0.27	small	favours caffeine	0.50
Puente et al., 2018 †	Mean heart rate (b·min^−1) [AA homozygotes]	158.00 ± 9.00	160.00 ± 10.00	2.00 ± 9.50	0.21	small	favours caffeine	0.72
		Mean heart rate (b·min^−1) [C-allele carriers]	161.00 ± 13.00	163.00 ± 9.00	2.00 ± 11.00	0.18	trivial	favours caffeine	0.82
		Peak heart rate (b·min^−1) [AA homozygotes]	187.00 ± 12.00	188.00 ± 13.00	1.00 ± 12.50	0.08	trivial	favours caffeine	0.22
		Peak heart rate (b·min^−1) [C-allele carriers]	182.00 ± 7.00	185.00 ± 6.00	3.00 ± 6.50	0.46	small	favours caffeine	0.46

Key – N/A: Not Available; * = significant difference (p < 0.05); ∆: mean difference = CAFFEINE – CONTROL; † = sub-analysis of (Puente et al., 2017); § weight loss was not different, but caffeine group consumed 0.5 ± 0.4 L more fluid compared to placebo group (p < 0.05) (Pettersen et al., 2014).

Table 4. Effects of acute caffeine supplementation on technical-skill performance during match-play in high-performance team-sports

Reference	Outcome measure (units)	Control mean ± SD	Caffeine mean ± SD	Mean difference ∆ ± pooled SD	Cohen’s d	Effect size	Direction of effect	p-value
Portillo et al., 2017 ‡	Tackle frequency (count·min^−1)	0.25 ± 0.12	0.17 ± 0.09	−0.08 ± 0.11	−0.76	moderate	favours placebo	0.14
		Ruck frequency (count·min^−1)	0.27 ± 0.13	0.24 ± 0.10	−0.03 ± 0.12	−0.26	small	favours placebo	0.62
		Pass frequency (count·min^−1)	0.34 ± 0.25	0.34 ± 0.11	0.00 ± 0.18	0.00	no effect	no effect	0.84
		Pass receive frequency (count·min^−1)	0.32 ± 0.18	0.29 ± 0.12	−0.03 ± 0.15	−0.20	trivial	favours placebo	0.56
		Ball carry frequency (count·min^−1)	0.17 ± 0.13	0.16 ± 0.12	−0.01 ± 0.13	−0.08	trivial	favours placebo	0.67
		Tackle quality (Skill-Index rating)	2.51 ± 0.49	2.57 ± 0.46	0.06 ± 0.48	0.13	trivial	favours caffeine	0.72
		Ruck quality (Skill-Index rating)	3.16 ± 0.39	3.40 ± 0.34	0.24 ± 0.37	0.66	moderate	favours caffeine	0.07
		Pass quality (Skill-Index rating)	3.14 ± 0.49	3.09 ± 0.32	−0.05 ± 0.41	−0.12	trivial	favours placebo	0.78
		Pass receive quality (Skill-Index rating)	2.99 ± 0.26	3.08 ± 0.21	0.09 ± 0.24	0.38	small	favours caffeine	0.31
		Ball carry quality (Skill-Index rating)	3.34 ± 0.64	3.28 ± 0.62	−0.06 ± 0.63	−0.10	trivial	favours placebo	0.74
Madden et al., 2019	Conquered puck (unadjusted IR) §	57.50 ± 20.10	59.50 ± 20.50	2.00 ± 20.30	0.10	trivial	favours caffeine	0.79
		Received pass (unadjusted IR) §	47.00 ± 18.20	28.00 ± 14.10	−19.00 ± 16.15	−1.18	moderate	favours placebo	< 0.01*
		Lost puck (unadjusted IR) §†	31.50 ± 15.00	33.50 ± 15.50	2.00 ± 15.25	0.13	trivial	favours caffeine	0.73
		Offensive pass (unadjusted IR) §	42.50 ± 17.40	32.00 ± 15.10	−10.50 ± 16.25	−0.65	moderate	favours placebo	0.55
		Successful shot (unadjusted IR) §	23.50 ± 13.00	16.50 ± 11.00	−7.00 ± 12.00	−0.58	small	favours placebo	0.09
		Neutral pass (unadjusted IR) §	7.00 ± 7.30	5.50 ± 6.50	−1.50 ± 6.90	−0.22	small	favours placebo	0.11
Puente et al., 2017	Total points made (count)	8.20 ± 6.90	8.80 ± 6.10	0.60 ± 6.50	0.09	trivial	favours caffeine	0.35
		Total rebounds (count)	2.50 ± 2.00	3.70 ± 2.60	1.20 ± 2.30	0.52	small	favours caffeine	0.03*
		Total assists (count)	1.10 ± 0.90	2.10 ± 1.60	1.00 ± 1.25	0.80	moderate	favours caffeine	0.02*
		PIR	8.40 ± 8.30	11.60 ± 7.30	3.20 ± 7.80	0.41	small	favours caffeine	0.04*
		Accuracy in 2-point field goals (%)	54.70 ± 30.50	52.90 ± 37.20	−1.80 ± 33.85	−0.05	trivial	favours placebo	0.45
		Accuracy in 3-point field goals (%)	27.40 ± 31.50	23.70 ± 27.50	−3.70 ± 29.50	−0.13	trivial	favours placebo	0.33
		Accuracy in free throws (%)	71.40 ± 40.50	73.80 ± 20.70	2.40 ± 30.60	0.08	trivial	favours caffeine	0.44

Key – N/A: Not Available; * = significant difference (p < 0.05); ∆ = mean difference = CAFFEINE – CONTROL; ‡: secondary analysis of (Del Coso et al., 2013); IR: Incidence Rate; §: SD calculated from 95% confidence intervals using the equation: SD = √N × (UCI – LCI) / 3.92 [Higgins et al., 2019; Cochrane Handbook for Systematic Reviews]; †: outcome negatively related to technical-skill performance; PIR: Performance-Index Rating = ((points + total rebounds + assists + steals + blocks committed + fouls received) − (missed shots + turnovers + fouls committed)).

Table 5. Effects of acute caffeine supplementation on physical performance during match-play in high-performance team-sports

Reference	Outcome measure (units)	Control mean ± SD	Caffeine mean	Mean difference ∆ ± pooled SD	Cohen’s d	Effect size	Direction of effect	p-value
Pettersen et al., 2014	Total distance (m)	9854.00 ± 901.00	10,062.00 ± 916.00	208.00 ± 908.50	0.23	small	favours caffeine	0.13
		High-speed distance (m) [>19.8 km·h^−1]	642.00 ± 240.00	557.00 ± 178.00	−85.00 ± 209.00	−0.41	small	favours placebo	0.14
		Distance sprinting (m) [>25.2 km·h^−1]	112.00 ± 69.00	109.00 ± 58.00	−3.00 ± 63.50	−0.05	trivial	favours placebo	0.84
		Accelerations (count) [> 2 m·s^−2]	126.00 ± 24.00	123.00 ± 31.00	−3.00 ± 27.50	−0.11	trivial	favours placebo	0.67
		Peak speed (km·h^−1) [first half]	28.40 ± 1.90	28.40 ± 1.70	0.00 ± 1.80	0.00	no effect	no effect	> 0.05
		Peak speed (km·h^−1) [second half]	28.30 ± 1.50	27.90 ± 1.50	−0.40 ± 1.50	−0.27	small	favours placebo	> 0.05
Del Coso et al., 2012	Total distance (m)	7352.00 ± 881.00	7782.00 ± 878.00	430.00 ± 879.50	0.49	small	favours caffeine	< 0.05*
		Mean distance (m·min^−1)	92.00 ± 11.00	98.00 ± 11.00	6.00 ± 11.00	0.55	small	favours caffeine	N/A
		Distance at high-speed (m) [> 13.1 km·h^−1]	1205.00 ± 289.00	1436.00 ± 326.00	231.00 ± 307.50	0.75	moderate	favours caffeine	< 0.05*
		Distance sprinting (m) [> 18.0 km·h^−1]	376.00 ± 122.00	489.00 ± 172.00	113.00 ± 147.00	0.77	moderate	favours caffeine	< 0.05*
		Sprint bouts (count) [> 18 km·h^−1, > 1 s^−1]	24.00 ± 8.00	30.00 ± 10.00	6.00 ± 9.00	0.67	moderate	favours caffeine	< 0.05*
Del Coso et al., 2013	Mean running pace (m·min^−1)	87.50 ± 8.30	95.40 ± 12.70	7.90 ± 10.50	0.75	moderate	favours caffeine	< 0.05*
		High-speed pace [18–20 km·h] (m·min^−1)	3.70 ± 1.40	4.40 ± 1.40	0.70 ± 1.40	0.50	small	favours caffeine	< 0.05*
		Sprinting pace [> 20 km·h] (m·min^−1)	4.60 ± 3.30	6.10 ± 3.40	1.50 ± 3.35	0.45	small	favours caffeine	< 0.05*
Del Coso et al., 2013	Total running distance (m) [match]	4749.00 ± 589.00	5139.00 ± 475.00	390.00 ± 532.00	0.73	moderate	favours caffeine	0.01*
		Running distance (m) [first half]	2327.00 ± 232.00	2366.00 ± 263.00	39.00 ± 247.50	0.16	trivial	favours caffeine	0.51
		Running distance (m) [second half]	2416.00 ± 443.00	2772.00 ± 308.00	356.00 ± 375.50	0.95	moderate	favours caffeine	0.01*
		High-speed distance (m) [18–20 km·h^−1]	107.00 ± 70.00	143.00 ± 86.00	36.00 ± 78.00	0.46	small	favours caffeine	0.01*
		Distance sprinting (m) [> 20 km·h^−1]	184.00 ± 38.00	208.00 ± 38.00	24.00 ± 38.00	0.63	moderate	favours caffeine	0.01*
		Sprint actions (count) [> 20 km·h^−1, > 5 m)	10.00 ± 7.00	12.00 ± 7.00	2.00 ± 7.00	0.29	small	favours caffeine	< 0.05*
		Total impacts (count) [> 5 g]	481.00 ± 352.00	641.00 ± 366.00	160.00 ± 359.00	0.45	small	favours caffeine	0.01*
		Total impacts (count) [> 5 g] [first half]	227.00 ± 181.00	296.00 ± 175.00	69.00 ± 178.00	0.39	small	favours caffeine	0.23
		Total impacts (count) [> 5 g] [second half]	254.00 ± 179.00	346.00 ± 198.00	92.00 ± 188.50	0.49	small	favours caffeine	0.01*
Del Coso et al., 2016	Total distance (m)	6035.00 ± 451.00	6055.00 ± 499.00	20.00 ± 475.00	0.04	small	favours caffeine	0.87
		High-intensity actions [> 19 km·h^−1, > 1 s^−1]	23.00 ± 5.60	27.40 ± 7.00	4.40 ± 6.30	0.70	moderate	favours caffeine	0.02*
		Sprint actions [> 23 km·h^−1, > 1 s^−1]	4.50 ± 1.90	6.30 ± 2.90	1.80 ± 2.40	0.75	moderate	favours caffeine	0.05
		High-speed distance (m) [19–22.9 km·h^−1]	303.00 ± 67.00	358.00 ± 117.00	55.00 ± 92.00	0.60	moderate	favours caffeine	0.05
		Distance sprinting (m) [> 23 km·h^−1]	85.00 ± 41.00	117.00 ± 55.00	32.00 ± 48.00	0.67	moderate	favours caffeine	0.02*
		Total body impacts (count)	697.00 ± 285.00	618.00 ± 221.00	−79.00 ± 253.00	−0.31	small	favours placebo	0.15
Lara et al., 2014	Total distance (m)	6631.00 ± 1618.00	7087.00 ± 1501.00	456.00 ± 1559.50	0.29	small	favours caffeine	< 0.05*
		High-speed distance (m) [13.1–18.0 km·h^−1]	680.00 ± 334.00	773.00 ± 281.00	93.00 ± 307.50	0.30	small	favours caffeine	> 0.05
		Distance sprinting (m) [> 18.0 km·h^−1]	161.00 ± 99.00	216.00 ± 103.00	55.00 ± 101.00	0.54	small	favours caffeine	< 0.05*
		Sprint bouts (count) [> 18 km·h^−1, > 1 s^−1]	16.00 ± 9.00	21.00 ± 13.00	5.00 ± 11.00	0.45	small	favours caffeine	< 0.05*
		Maximum speed (km·h^−1)	23.80 ± 2.50	24.20 ± 2.40	0.40 ± 2.45	0.16	trivial	favours caffeine	0.25
Madden et al., 2019	Skate time (s^−1) #	361.36 ± 34.72	324.38 ± 47.93	−36.99 ± 41.32	−0.90	moderate	favours placebo	0.04*
		Shifts (count) #	9.57 ± 1.16	8.71 ± 1.64	−0.86 ± 1.40	−0.61	moderate	favours placebo	0.11
		Time (s^−1) per shift #	37.91 ± 2.00	37.52 ± 2.91	−0.39 ± 2.46	−0.16	trivial	favours placebo	1.00
		Mean skate acceleration (g)#	1.28 ± 0.05	1.29 ± 0.05	0.01 ± 0.05	0.20	trivial	favours caffeine	0.18
		Peak head accelerations [> 1.85 g, GMSA] (count·s^−1)	4.10 ± 0.90	4.40 ± 0.90	0.30 ± 0.90	0.33	small	favours caffeine	0.02*
		Maximum peak head accelerations (g)	10.00 ± 5.10	12.30 ± 4.80	2.30 ± 4.95	0.46	small	favours caffeine	0.14
		Trunk contact [level 1] (unadjusted IR)	4.50 ± 6.10	0.50 ± 3.40	−4.00 ± 4.75	−0.84	moderate	favours placebo	0.04*
		Trunk contact [level 2] (unadjusted IR)	24.50 ± 13.30	19.00 ± 11.70	−5.50 ± 12.50	−0.44	small	favours placebo	0.24
		Trunk contact [level 3] (unadjusted IR)	5.00 ± 6.30	5.00 ± 6.30	0.00 ± 6.30	0.00	no effect	no effect	1.00
		Trunk contact [level 4] (unadjusted IR)	3.00 ± 5.10	3.00 ± 5.10	0.00 ± 5.10	0.00	no effect	no effect	1.00
		Trunk contact [level 5] (unadjusted IR)	0.00 ± 0.00	0.00 ± 0.00	0.00 ± 0.00	N/A	no effect	no effect	N/A
Muñoz et al., 2020	Accelerations (count·min^−1)	18.75 ± 1.57	19.13 ± 1.16	0.38 ± 1.37	0.28	small	favours caffeine	0.18
		Decelerations (count·min^−1)	19.00 ± 1.37	18.72 ± 1.42	−0.28 ± 1.40	−0.20	trivial	favours placebo	0.05
		Impacts (count·min^−1)	23.87 ± 11.46	26.25 ± 13.57	2.38 ± 12.52	0.19	trivial	favours caffeine	0.56
Portillo et al., 2017 ‡	Impacts [< 5.00 to 6 g] (count·min^−1)	16.10 ± 4.90	20.80 ± 9.90	4.70 ± 7.40	0.64	moderate	favours caffeine	< 0.05*
		Impacts [6.01 < 6.50 g] (count·min^−1)	12.20 ± 10.60	16.20 ± 15.20	4.00 ± 12.90	0.31	small	favours caffeine	< 0.05*
		Impacts [6.51 < 7 g] (count·min^−1)	3.80 ± 1.50	4.70 ± 2.60	0.90 ± 2.05	0.44	small	favours caffeine	< 0.05*
		Impacts [7.01 < 8 g] (count·min^−1)	N/A ± N/A	N/A ± N/A	N/A ± N/A	N/A	N/A	favours placebo	0.28
		Impacts [8.01 < 10 g] (count·min^−1)	0.80 ± 0.40	1.10 ± 0.60	0.30 ± 0.50	0.60	moderate	favours caffeine	< 0.05*
		Impacts [> 10 g] (count·min^−1)	N/A ± N/A	N/A ± N/A	N/A ± N/A	N/A	N/A	favours placebo	0.23
Puente et al., 2017	Impacts (count·min^−1)	396.00 ± 43.00	410.00 ± 41.00	14.00 ± 42.00	0.33	small	favours caffeine	< 0.01*
Puente et al., 2018 †	Impacts (count·min^−1) [AA homozygotes]	385.00 ± 48.00	401.00 ± 36.00	16.00 ± 42.00	0.38	small	favours caffeine	0.02*
		Impacts (count·min^−1) [C-allele carriers]	401.00 ± 36.00	415.00 ± 35.00	14.00 ± 35.50	0.39	small	favours caffeine	0.01*
Raya-Gonzalez et al., 2021#	Total distance (m·min^−1)	79.30 ± 21.50	80.00 ± 6.40	0.70 ± 13.95	0.05	trivial	favours caffeine	0.07
		High-speed distance [18.01–24.0 km·h^−1] (m·min^−1)	2.60 ± 1.10	2.60 ± 1.90	0.00 ± 1.50	0.00	no effect	no effect	0.66
		Distance sprinting (m·min^−1) [>24 km·h^−1]	1.10 ± 2.00	1.10 ± 0.20	0.00 ± 1.10	0.00	no effect	no effect	0.19
		Maximum speed (km·h^−1)	21.60 ± 3.90	19.40 ± 1.20	−2.20 ± 2.55	−0.86	moderate	favours placebo	0.13
		Accelerations (m·min^−1)	6.60 ± 1.90	6.40 ± 0.90	−0.20 ± 1.40	−0.14	trivial	favours placebo	0.38
		Decelerations (m·min^−1)	6.20 ± 1.70	6.22 ± 0.70	0.02 ± 1.20	0.02	trivial	favours caffeine	0.12
		Impacts (count·min^−1)	8.20 ± 6.30	7.50 ± 2.90	−0.70 ± 4.60	−0.15	trivial	favours placebo	0.29
		Jumps (count·min^−1)	4.00 ± 1.70	3.60 ± 1.50	−0.40 ± 1.60	−0.25	small	favours placebo	0.18
		Player load (AU·min^−1)	1.40 ± 0.30	1.20 ± 0.20	−0.20 ± 0.25	−0.80	moderate	favours placebo	0.07

Key – N/A: Not Available OR Not Applicable; * = significant difference (p < 0.05); ∆ = mean difference = CAFFEINE – CONTROL; IR: Incidence Rate; GMSA: global mean skating acceleration; #: not presently available to the public [data/ manuscript was retrieved from the authors]; AU: Arbitrary Units; g: gravitational force equivalent; ‡: secondary analysis of (Del Coso et al., 2013); †: sub-analysis of (Puente et al., 2017).

Nine studies reported on physiological measures. Commonly observed effects with caffeine included: higher peak and mean heart rates and increased blood glucose and lactate concentrations (Table 3). However, variability existed with studies observing no changes in outcomes (e.g., mean heart rates and sweat rates) where other studies did observe changes (e.g., (Del Coso et al., 2013; Lara et al., 2014; Pettersen et al., 2014). In 3 studies reporting technical-skill outcomes, 13 measures indicated favourable performances with placebo and 9 favoured caffeine (ES range: −1.18 to 0.80; Table 4, Figure 2); only one measure (pass frequency) in one study (Portillo et al., 2017) indicated no effect. In 12 studies reporting on physical performance, data from 18 measures indicated favourable performances with placebo, 42 favoured caffeine and 6 showed no effect (ES range: −0.90 to 0.95, Table 5).

Figure 2. Effects of acute caffeine supplementation on technical-skill performance outcomes during live match-play in high-performance team-sports. Key: † = outcome measure negatively related to technical-skill performance.

Cohen’s d was calculated using the equation: d = ((mean value with caffeine – mean value with control) / (average (caffeine SD, control SD))); Effect sizes are operationalised as per Hopkins et al. (2009), pp. – i.e., trivial, ≤ 0.20; small, 0.20 to 0.59; moderate, 0.60 to 1.19; large, 1.20 to 1.99; very large, 2.00 to 3.99; extremely large, ≥ 4.00.

Study methodological quality

The mean PEDro methodological quality score was 9.0 (range: 7 to 10). One study (Muniz Guttierres et al., 2013) was categorised as being of “good” methodological quality (PEDro score = 7), while all other studies were classified as being of “excellent” quality (i.e., 9 or 10). Further, none of the studies declared any conflicts of interest.

Discussion

This is the first systematic review of how acute caffeine ingestion affects match-play performance in high-performance team-sports. Overall, there is convincing evidence that ingesting a moderate dose of caffeine (3 to 6 mg · kg⁻¹) 50 to 60 minutes before performing improves physical performance in team-sports. However, there are a limited number of studies involving team-sport athletes that have assessed technical-skill outcomes during match-play (Figure 2), thus it is still not well-understood whether the benefits to physical performance come at the expense of, or help in addition to, technical-skill performance.

The widely reported increases in mean and peak heart rates in the present review suggest an augmented sympathetic response with caffeine ingestion and/or may be a reflection of greater exercise intensities achieved with caffeine (Table 3). Blood lactate and glucose concentrations were generally higher during match-play with caffeine than placebo, which is consistent with prior research (Mahdavi et al., 2015). The higher heart rates and blood lactate concentrations may reflect a larger contribution from anaerobic glycolysis to energy turnover, as a product of improved fatigue resistance and higher work intensities. However, some studies reported no change in high-intensity performance variables, despite elevated blood lactate and/or heart rates; suggesting that this biological mechanism may not entirely explain caffeine’s ergogenicity in team-sports (Pettersen et al., 2014). The higher blood glucose concentrations observed with caffeine intake may indicate a larger catecholamine response, which may have multiple performance-enhancing effects (Cooper et al., 2008), or short-term impairment of insulin-mediated glucose disposal (Burke, 2008). Specifically, Mohr et al. (2011) reported a correlation between higher blood glucose levels after caffeine intake with improvements in muscle interstitial K+ accumulation, an acknowledged fatigue mechanism in high-intensity exercise. This poses the question whether caffeine may facilitate a more efficient uptake of exogenous fuels (e.g., creatine, carbohydrate) by muscle tissue, which is an important mechanism to study in future, as athletes often ingest caffeine alongside other nutrients in team-sports settings. Alternatively, there is some evidence of an impaired insulin-mediated glucose disposal in response to caffeine (Van Dam & Hu, 2005) and it is plausible that, if caffeine acutely impaired glucose disposal, this could impair re-synthesis of muscle glycogen during recovery from exercise (Burke, 2008). However, Battram et al. (2004) reported that ingesting 6 mg · kg⁻¹ of caffeine before and during glycogen-depleting exercise did not affect glycogen re-synthesis rates during a 5-hour recovery when adequate carbohydrate was consumed. Therefore, if carbohydrates are consumed with caffeine during recovery from exercise, it is unlikely that muscle glycogen re-synthesis would be impaired. Caffeine induces physiological responses in team sports consistent with the effects observed in other sports and modes of exercise.

Caffeine ingestion has been proposed to affect hydration status. For instance, Gonzalez-Alonso et al. (1992) observed increased urine production and reduced rehydration after exercise with caffeine compared with a carbohydrate-electrolyte drink. A potential concern is that the diuretic effects of caffeine may harm performance and induce greater physiological strain to the athletes; in the past this has led to recommendations on limiting caffeine intake prior to performance or travel, or to compensate with larger fluid intakes (Burke, 2008). However, more recent evidence has contended this position. In a dose–response experimental study involving a chronic caffeine dose of 0 (placebo), 3 or 6 mg · kg⁻¹·day⁻¹, Dias et al. (2005) observed no between-group differences immediately and 16 hours after an exercise-heat tolerance test (37.7°C) in total plasma protein, haematocrit, body mass, or urine osmolality, specific gravity, colour, and volume. Generally, studies in this review that assessed hydration markers during and post-match-play, reported no or trivial effects of caffeine on hydration status (Table 3). This concurs with a review by Armstrong (2002) reporting that caffeine doses within the range said to be ergogenic (i.e., 3 to 6 mg·kg⁻¹) do not alter sweat rates, urine loss or hydration indices during exercise. Thus, it is probable that team-sport athletes do not need to alter their habitual fluid intake with caffeine ingestion.

The results of this review suggest that acute ingestion of moderate caffeine doses may lead to trivial-to-small, in some cases moderate, improvements in physical performance during team-sport match-play. Commonly observed outcomes with caffeine included: increased total distances, increased high-speed running distances, higher peak speeds, more frequent body impacts and accelerations (Table 5). This may be explained by caffeine’s action as a sympathomimetic agent in elevating circulating adrenaline levels, which can result in muscle and central adaptations, such as increasing muscle contractile force and velocity (Giráldez-Costas et al., 2020), inducing analgesia (Ward et al., 1991), delaying fatigue and reducing effort perception (Bangsbo et al., 1992; Figure 3). Interestingly, some studies reported trends favouring placebo for outcomes including impacts (Del Coso et al., 2016), accelerations (Pettersen et al., 2014), decelerations (Muñoz et al., 2020) and jumps (Raya-González et al., 2021). These discrepancies highlight the difficulty of obtaining data relevant to match-play and reflect a need for more ecological repeated-measures studies so that sample power may overcome measurement noise, allowing for more robust conclusions. In the present review, when placebo was favoured over caffeine ingestion, effects were mostly trivial and studies included only two trials which is not necessarily enough to account for variability in the match-play environment (Hopkins, 2000). Overall, caffeine ingestion seems to improve physical performance outcomes related to endurance, strength, power and sustained high-intensity performance in team sports, but any improvements are small to moderate.

Figure 3. Schematic to summarise the evidence included in this systematic literature review on the effects of acute caffeine ingestion on physiological, physical and technical-skill performance outcomes during live intermittent team-sport match-play.

Caffeine is a potent stimulant that may decrease precision, increase nervousness/anxiety, gastrointestinal distress and insomnia (Pallarés et al., 2013); which could have ergolytic effects in sports requiring high-accuracy skills. In contrast, it has been postulated that caffeine intake could benefit technical-skill performance by attenuating the effects of fatigue on precision, decision-making, reaction times and cognition (Russell et al., 2011). Encouragingly, researchers have observed improved rates of positive technical actions with caffeine during match-play in elite volleyball (Del Coso et al., 2014). A challenge in investigating technical-skill outcomes in team-sports, is that certain sports require a greater number of fine (e.g., free-throws in basketball) or gross (e.g., tackle in rugby) skills compared to others, and these can be differentially affected by caffeine (Burke, 2008). One of three studies in this review that assessed technical-skill performance observed increases in total rebounds (ES = 0.52) and assists (ES = 0.80) with caffeine; both of which represent relatively gross skills in basketball (Puente et al., 2017). However, two-point (ES = −0.05) and three-point (ES = −0.13) shot accuracy showed trivial effects favouring placebo; both representing finer skills. Although effects were trivial, there are presently insufficient studies in basketball match-play to ascertain that caffeine is not harmful to technical-skill performance, and certainly not to suggest that it is favourable. Adding to the complexity, most team-sports require a combination of fine and gross skills, and the competitive value that each represents may be influenced by a team’s tactical style of play (e.g., high vs. low physicality; Mitrotasios et al., 2019). This opens scope for highly innovative research on the effects of caffeine ingestion in relation to tactical strategies in team-sports.

Similarly confounding results can be observed for technical-skill performance outcomes from match-play in rugby sevens (Portillo et al., 2017) and ice hockey (Madden et al., 2019). In their study in rugby sevens Portillo et al. (2017) reported that frequency of tackles, rucks, passes received and ball carries all decreased in caffeine conditions (Table 4, Figure 2). Negative changes on quality of passes (as per a 5-point Likert scale) (ES = −0.12) were also observed with caffeine, but not for tackles, rucks or passes received. This data may suggest that some of caffeine’s ergogenic effects on technical-skill performance weaken at, or do not translate to, the ecological level. However, this cannot be ascertained until more studies quantify technical-skill outcomes during rugby match-play. In ice hockey, Madden et al. (2019) concluded that caffeine had limited impact on skill performance. However, the outcomes during live scrimmages displayed a trend towards better technical-skill performance in placebo, despite greater physicality in caffeine trials (Table 5). With placebo, there were greater incidence rates of received passes (ES = −1.18), offensive passes (ES = −0.65), successful shots (ES = −0.58), neutral passes (ES = −0.22) and reduced lost puck incidence rates (ES = 0.13). Greater rates of conquered pucks were reported with caffeine, possibly resulting from increased physicality, however effects were trivial (ES = 0.10). It appears that although caffeine may benefit gross efforts, it may have no effect, or negatively confound finer technical-skill performance in ice hockey. This is crucial because although more impacts may indicate greater player involvement, if technical-skill performance is impaired the net effects may not be positive.

Whilst there is some evidence of improved technical-skill performance in sport simulations with caffeine (e.g., Foskett et al., 2009; Roberts et al., 2010) findings are less clear at the ecological level (Figure 2). To date, few studies have investigated elite athletes in match-play, thus it is not well understood how much of the variability with caffeine across technical-skill variables is explained by the performer’s level or the nature of the assessment (Burke, 2008). Also, it is not well understood if the benefits to physical performance come at the expense of a deteriorated technical-skill performance, or whether ergogenic responses may vary across disciplines, between the sexes or according to genotype in team-sports (Muñoz et al., 2020; Puente et al., 2018). Future prospective longitudinal investigations assessing technical-skill outcomes in match-play are necessary to advance consensus on the potential of caffeine in team-sports.

Limitations

There are potential limitations worth noting. Firstly, it was not possible to draw consensus regarding the effects that habitual caffeine intake may have on the ergogenic properties of caffeine in team-sport match-play. Prior evidence is in support that caffeine habituation status may have confounding effects on the ergogenicity of caffeine on exercise performance (Collomp et al., 1992). However, we could not confidently present on habituation status in this review as most included studies did not collect data for this variable. Prior dietary habits, aside from ingestion of caffeine in the hours preceding the assessments, was not controlled in any of the 13 studies. This may have confounded the effects of caffeine on match-play performance, particularly if other possibly ergogenic or ergolytic substances were ingested (e.g., creatine, carbohydrate, bicarbonate, alcohol, etc.; Burke, 2008; Roberts et al., 2010). Moreover, although in studies where energy drinks included additional ingredients an identically matched drink with no caffeine was provided for placebo conditions, it is impossible to ascertain that the other ingredients played no additional effect on performance (i.e., that only caffeine’s isolated effect was measured). Potent interactive effects of caffeine with other ingredients, such as creatine, have been reported to further enhance physical performance outcomes including intermittent high-intensity sprint performance (Lee et al., 2011). At present, it is not possible to ensure that similar effects did not occur with other ingredients in the 6 studies that used commercial energy drinks instead of caffeine alone (Table 2). Finally, although the review was restricted to studies involving high-performance team-sport athletes (defined in our inclusion criteria), training status was not analysed as a variable in this review and could have affected performance-related effects. In future studies researchers may wish to assess the effect of training status within high-performance team-sport populations (i.e., years in high-performance) on caffeine’s ergogenicity in team-sports. Particularly given that outside of high-performance team-sports, level of proficiency is thought confound the effects of caffeine (Burke, 2008).

Summary and conclusions

This is the first systematic review of evidence regarding the effects of acute caffeine ingestion on match-play performance in intermittent team-sports. Overall, there is good ecological evidence in support that ingesting a moderate dose of caffeine (3 to 6 mg · kg⁻¹) 50 to 60 minutes before match-play may yield a moderate benefit to physical performance in team-sports. Team-sport athletes may be confident that supplementing with caffeine prior to performing will likely yield superior performances in competition for parameters related to endurance, strength, power and sustained high-intensity performance. Trials involving high-performance team-sport athletes that have assessed the effects of acute caffeine intake on technical-skill outcomes during match-play are scarce, and thus it is difficult to conclude whether or not caffeine provides a performance benefit. It is also not well understood whether these effects may vary according to the athletic discipline or to the regime of muscle involvement of individual skills. Future studies are necessary to quantify technical-skill outcomes using video analysis of match-play, categorising these into fine and gross skills, to better understand what team-sport situations caffeine supplementation may be best suitable for.

Disclaimer

The views expressed in this article are of the authors and not an official position of the institution. No sources of funding or external support were granted for this work.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The author(s) reported there is no funding associated with the work featured in this article.

Appendix.

PEDro Scale