Effects of Tea (Camellia sinensis) or its Bioactive Compounds l-Theanine or l-Theanine plus Caffeine on Cognition, Sleep, and Mood in Healthy Participants: A Systematic Review and Meta-Analysis of Randomized Controlled Trials

Abstract

Context

The bioactive compounds found in tea from Camellia sinensis, namely theanine, caffeine, and polyphenols, can potentially improve short-term and long-term health outcomes.

Objective

The aim of this study was to assess the effects of tea, theanine alone, or theanine plus caffeine on cognition, mood, and sleep outcomes, using data from randomized controlled trials (RCTs) in healthy participants.

Data sources

The Cochrane Database of Systematic Reviews, Cochrane Central Register of Controlled Trials, Embase, and Ovid Medline were searched up to and including August 2023.

Data extraction

Data relevant to the participants, interventions, comparisons, outcomes (ie, cognition, mood, and sleep), and study design were extracted.

Data analysis

Fifty RCTs were included in the review, of which 15 were eligible for at least 1 meta-analysis, most commonly performed by use of standardized mean differences (SMD), in random effects models. After intake of theanine plus caffeine, and of placebo, small-to-moderate differences were found between these interventions in the first hour (h1) and second hour (h2), that favoured theanine plus caffeine, with regard to cognition and mood outcomes such as choice reaction time (h1: SMD, −0.48; 95% CI, −1.01, 0.05), digit vigilance task accuracy (h2: SMD, 0.20; 95% CI, 0.02, 0.38), attention switching task accuracy (h2: SMD, 0.33; 95% CI, 0.13, 0.54), and overall mood (h2: SMD, 0.26; 95% CI,−0.10, 0.63). There was a small-to-moderate difference between the effects of theanine and placebo that favored theanine on choice reaction time (h1: SMD, −0.35; 95% CI,−0.61, −0.10). The CIs frequently highlighted the uncertainty surrounding the direction and magnitude of these differences.

Conclusions

This meta-analysis provides evidence that theanine plus caffeine, and theanine alone, could be beneficial for cognitive and mood outcomes. More research using tea beverages or tea-equivalent bioactive doses and research in free-living participants is needed.

PROSPERO registration no

CRD42022351601.

INTRODUCTION

Bioactive compounds in black tea and green tea (from Camellia sinensis), such as caffeine, l-theanine (theanine herein), and tea polyphenols,^1,2 have the potential to improve one’s health and well-being. Drinking black tea and supplementation with green tea extract (predominantly in capsule form) have been shown to improve levels of blood pressure and cholesterol.^3–6 Intake of black tea and green tea are associated with reduced risk of cardiovascular disease,^7,8 certain cancers,⁹ and cognitive diseases,¹⁰ and a dietary guideline has been proposed for flavan-3-ols—the broader classification for epigallocatechin (EGCG) and other catechins—recommending consumption of 400–600 mg/d, from foods and drinks rich in flavan-3-ols, such as green and black tea, to reduce risk of cardiovascular disease and diabetes.^11,12

Intake of tea, and its bioactive compounds studied in isolation, has also been causally linked with short-term improvements in cognitive function and certain mood outcomes. Intake of theanine plus caffeine improves accuracy in attentional tasks,¹³ and the European Food Safety Authority (EFSA) provided a positive evaluation of a health claim for the efficacy of black tea to improve attention.¹⁴ A cross-sectional study of university staff, studied in their natural work environment, found a positive association between tea intake and subjective ratings of tiredness and work performance.¹⁵ Caffeine intake has been shown to benefit drivers in certain situations,¹⁶ and reaction time and attention switching performance, as part of a broader test battery, has been independently shown to predict the likelihood of passing a driving assessment.¹⁷ While caffeine is a well-known stimulant, studies have found that theanine alone improves measures of stress and anxiety,¹⁸ objective measures of sleep quality,¹⁹ and neurophysiological outcomes associated with cognitive improvement and mood,^20–22 a finding that has also been observed after intake of EGCG.²³

Whilst current evidence points toward the beneficial effects of tea, or theanine plus caffeine, on cognition and mood, mostly being driven by caffeine, the presence and magnitude of positive effects observed appears to be inconsistent,^13,14,24,25 possibly a result of variations in assessment methods and doses of caffeine and theanine. Whether the combination and individual bioactive compounds offer distinguishable effects on cognition and mood outcomes is still to be fully understood. This review aims to quantify the effects of, and the differences between, tea, the combination of theanine and caffeine, or theanine alone, and placebo or each other, on short-term cognition and mood outcomes, and systematically assess their effects on sleep outcomes, in healthy participants.

METHODS

The protocol for this systematic review and meta-analysis was registered with PROSPERO (registration number CRD42022351601).²⁶ This study was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) checklist (Appendix S1),²⁷ and using methods stated in the Cochrane Handbook for Systematic Reviews of Interventions.²⁸

Search Strategy and Study Selection

The primary literature search was performed in May 2022 (Appendix S2) and updated in August 2023 (Appendix S3). Using the Ovid platform, four databases were searched: the Cochrane Database of Systematic Reviews, Cochrane Central Register of Controlled Trials, Embase, and Ovid MEDLINE (Epub Ahead of Print, In-Process, In-Data-Review & Other Non-Indexed Citations and Daily). No date or language restrictions were enforced. Bibliographies of included studies and other reviews were checked for additional relevant publications. The systematic search was formulated using the PICOS framework, as were the inclusion and exclusion criteria (Table 1). All titles and abstracts were screened by 1 author (E.P.), and 2 other authors (M.A.-M. and B.d.R.) each screened 10% to ensure consistency. After the initial filter, full texts were assessed for eligibility by 1 author (E.P.) before a final decision on the exclusion of publications was discussed between three authors (E.P., B.d.R. and A.G.).

Data Extraction and Synthesis

A data extraction sheet was designed for this review and 1 author (E.P.) performed data extraction. Relevant data linked to the PICOS framework such as study population, setting, intervention(s), control(s), outcome(s) assessed, and measurement tool(s) was extracted, as well as other characteristics such as the source of funding. The main characteristics of the studies were tabulated and summarized.

Data for outcomes pertaining to cognitive function at the behavioural level,²⁹ subjective mood, and subjective or objective sleep outcomes was sought. Outcomes reported by 2 or more studies were considered eligible for meta-analysis. Results data was extracted either as change-from-baseline, baseline and post-intervention, or post-intervention, means and standard deviations (SD), standard errors (SE), or 95% Cls, which were converted to SDs. Data presented in figures, with no data presented in the text or tables, was extracted using WebPlotDigitiser³⁰ and unreported data was sought by contacting the relevant authors.

All data from nine studies^31–40 was obtained through the respective publications. The authors of 2 included studies^41,42 provided the raw participant data for all outcomes measured. Raw participant data for the attention switching task, and summary data for alertness assessed with the Bond-Later (B-L), was available for 1 study.⁴³ Alertness data from 1 study⁴⁴ was provided by direct contact with the authors, and the authors of 1 study³⁶ were contacted to clarify an inconsistency in their publication.

Risk of Bias Assessment

Eligible studies were assessed for risk of bias using version 2 of the Cochrane risk-of-bias tool for randomized trials (RoB 2).⁴⁵ Data extraction and quality assessment were performed by 1 author (E.P.).

Statistical Analysis

Separate meta-analyses were performed using data from the first hour and second hour after intake of the interventions; only studies that provided short-term data were considered eligible for meta-analysis, ie, studies that delivered interventions on the test day(s), and not prior to the test days. Hours 1 and 2 were defined as any task or questionnaire administered between 1 and 59 minutes, and 60 and 119 minutes, after intake, respectively. Change-from-baseline values and post-intervention values adjusted for baseline values were used. Mean differences (MD) were calculated when the outcome measurement tool was the same across all studies, and the standardized mean difference (SMD) was calculated when the measurement tools differed, but the same behavioural outcome was being assessed. Calculations were obtained from two sources^28,46 and a detailed description of the calculations used is provided in an appendix (Appendix S4). Meta-analyses were conducted, and forest plots were produced, in R using the package “metafor”.⁴⁷ Tables were produced in Microsoft Word. Confidence intervals were calculated at the 95% level, and statistical significance was considered at P < .05. Random-effects models were fitted with the restricted maximum-likelihood estimation for τ²; the Knapp and Hartung adjustment was also applied.⁴⁸

Interpretation of the Results

The cognition and mood results presented in this meta-analysis mostly compare change-from-baseline results between 2 interventions. Effect sizes for comparisons including placebo were calculated as intervention value minus placebo value; effect sizes for comparisons written as A vs B were calculated as intervention A value minus intervention B value. Interventions were described as improving or not improving an outcome; however, in this study, improvements can be interpreted in 2 ways. For reaction time results, a difference could be interpreted as a greater positive change, ie, reaction times were reduced to a greater extent after one intervention than after another; or a smaller negative change, ie, reaction times increased to a lesser extent after one intervention than after another. When describing accuracy results, a difference could be interpreted as a greater positive change, ie, accuracy increased to a greater extent after one intervention than after another; or a smaller negative change, ie, accuracy decreased to a lesser extent after one intervention than after another. The magnitudes of effect sizes were described using Cohen’s guideline (1988) whereby 0.2 is a small effect, 0.5 is moderate effect, and 0.8 is large effect.⁴⁹ Effect sizes between these values were described as follows: 0.0 to 0.2, little-to-no effect; 0.2 to 0.5, small-to-moderate effect; 0.5 to 0.8, moderate-to-large effect.

RESULTS

The systematic searches performed in May 2022 and August 2023 identified 2535 and 249 unique articles, respectively; 64 were reviewed in full, and 15 were subsequently excluded (Figure 1). A total of 49 randomized trials presented in 48 references^{18,20,21,31–44,50–80} met the eligibility criteria for this review. A subset of 16 trials presented in 15 references^{21,31–36,38–44,51} were included in at least 1 meta-analysis (Table 2); 12 trials^{21,31–36,38,41–44} had a crossover design and 3 trials^39,40,51 had a parallel design. Across these trials, 484 participants were recruited (female participants, n = 229), 1 study³⁹ of 48 people did not report the gender split, and 1 study³⁶ recruited only male participants (n = 20). The median sample size was 27 participants (IQR, 23.75-35), with a weighted mean age of 32.00 years. Two studies used black tea, 2 used green tea, 1 delivered a “tea” intervention in capsule form, and 11 studied the compounds in isolation. In interventions comprising both theanine and caffeine, the median theanine dose was 82 mg (IQR, 48.5-175), and the median caffeine dose was 88.74 mg (IQR, 54.05-157.5). In theanine-only, low-caffeine, or caffeine-free interventions, the median theanine dose was 200 mg (IQR, 100.45-212.5). Industry was the most common funding source (n = 8), followed by academia (n = 4), and government (n = 2); however, 4 studies did not report their funding source(s).

A subset of 8 trials presented in 8 references^{56,57,65,74–77,80} reported sleep outcomes (Table 2). It was not possible to pool the data from these studies so instead their results were described in a narrative synthesis.

Cognition Outcomes

Simple Reaction Time

Six studies^{35–37,41,42,51} reported measuring simple reaction time (SRT) after intake of theanine plus caffeine, and placebo. One study³⁷ was excluded because the intervention was delivered prior to the test day. Of the remaining 5 studies, 3 studies^36,41,42 measured SRT in the first hour after intake, and 3 studies^35,41,51 in the second hour after intake. Three studies^35,41,42 used the SRT task, 1 study³⁶ used the simple visual reaction time task, and 1 study⁵¹ used a component of the Stroop test. There was little-to-no difference between the effects of theanine plus caffeine and placebo on SRT in the first hour after intake; the CI included a large effect favoring theanine plus caffeine and a moderate effect favoring placebo (SMD, −0.16; CI, −1.04, 0.72; P = .51; Figure 2A). The SMD was equivalent to approximately −4.79 ms (CI, −30.95, 21.38). However, there was a moderate difference between the effects of theanine plus caffeine, and placebo, on SRT in the second hour after intake and the CI was consistent with a moderate-to-large effect favoring theanine plus caffeine (SMD, −0.71; CI, −0.92, −0.50; P = .005; Figure 2B). The SMD was equivalent to approximately −26.54 ms (CI, −34.44, −18.65). Four studies^36,38,41,42 reported measuring SRT in the first hour after intake of theanine, and placebo, and 2^40,41 in the second hour. One study³⁸ used 3 different doses of theanine, and analyses were performed with the results following each dose. There was little-to-no difference between the effects of theanine, and placebo, on SRT in any of the analyses (Table 3).

Choice Reaction Time and Accuracy

Eight studies^{21,32,35,36,41,42,59,60} reported measuring choice reaction (CR) time after intake of theanine plus caffeine, and placebo. Two studies^32,59 were excluded because the results data necessary for meta-analysis were unreported and 1 study⁶⁰ was excluded because “total reaction time” results were unreported and therefore the results were considered aggregated or presented such that the data for distinct comparisons of individual interventions at each time point were unavailable. Three,^36,41,42 and three,^21,35,41 studies took measurements in the first, and second, hour after intake, respectively. Five studies^{32,35,36,41,42} reported measuring CR accuracy, one³² was excluded because the data necessary for meta-analysis was unreported. Three,^36,41,42 and 2,^35,41 studies took measurements in the first, and second, hour after intake, respectively. Three studies^35,41,42 used the CR task, and 2 studies^21,36 used the recognition visual reaction time task. There was a moderate difference between the effects of theanine plus caffeine and the effects of placebo on CR time in the first hour, that favored theanine plus caffeine, but the CI included no difference (SMD, −0.48; CI, −1.01, 0.05; P = .06; Figure 3A). The SMD was equivalent to approximately −24.71 ms (CI, −51.90, 2.48). The difference between the effects was equivalent but less precise in hour 2 (Figure 3B) and the SMD was equivalent to approximately −17.02 ms (CI, −46.42, 12.37). There was little-to-no difference between the effects of theanine plus caffeine and the effects of placebo on CR accuracy in the first hour (Figure 3C), while there was a small difference in the effects that favored theanine plus caffeine in the second hour, but the CI also included a moderate-to-large effect favoring placebo (Figure 3D). The SMDs were equivalent to approximately 0.20% (CI, −3.13, 3.54) and 0.92% (CI, −2.56, 4.39) in the first hour and second hour, respectively. Three^36,41,42 and 2^21,41 studies reported measuring CR time in the first and second hour, respectively, after intake of theanine plus caffeine, theanine, caffeine, and placebo; 3 studies also reported accuracy.^36,41,42 One study reported measuring CR time in the first hour after intake of 3 different theanine doses and intake of placebo doses.³⁸ Three studies used the CR task,^38,41,42, and 2 used the RVRT task.^21,36 In the first hour there were small-to-moderate differences favoring theanine between the effects of theanine and the effects of placebo on CR time, but, in each of the analyses, the CI also included little-to-no difference, or no difference (Table 3). There was little-to-no difference between the effects of theanine, and placebo, on CR accuracy (Table 3). In the second hour, there was a moderate-to-large difference that favored theanine between the effects of theanine and of placebo on CR time, but the CI also included a moderate-to-large effect in favor of placebo (Table 3). The other analyses of CR time in the second hour revealed CIs that included differences that favored both interventions being compared (Appendix S5).

Rapid Visual Information Processing Task

Two studies^41,42 reported using the rapid visual information processing (RVIP) task in the first hour after intake of theanine plus caffeine, theanine, caffeine, or placebo. One study⁴³ reported study participants using the RVIP task after intake of theanine plus caffeine or placebo, but that study was excluded from this study because the data necessary for the meta-analysis were unreported. Two studies^38,41 reported the same outcomes in the second hour after intake of theanine as after intake of placebo. In the first hour, the mean differences between the effects of an active intervention (theanine plus caffeine, theanine-only, or caffeine-only), and placebo for RVIP reaction time were between 25 and −25 ms. All of the CIs for these analyses included differences of at least 50 ms in favor of either of the included interventions (Tables 3 and 4, and Appendix S5). However, no minimally important difference (MID) appears to have been established for this task. In the second hour, there was either little-to-no difference or a small difference between the effects of theanine and effects of placebo, on RVIP reaction time, and little-to-no difference between the effects of theanine and the effects of placebo on RVIP accuracy. For both outcomes, the differences favored theanine (Table 3), and the CIs for the reaction time analysis, including the results from the 400-mg theanine dose from 1 study,³⁸ were consistent with a difference favoring theanine. However, the CIs for the other 5 analyses were always consistent with at least no difference between the effects of theanine and those of placebo.

Digit Vigilance Task

Four studies^35,37,41,51 reported measuring reaction time and accuracy during a sustained attention task in the second hour after intake of theanine plus caffeine. One study³⁷ was excluded because the intervention was delivered prior to the test day. Two studies^40,41 reported the same outcomes after intake of theanine and of placebo. Two studies^35,41 used the digit vigilance task (DVT), and 2 studies^40,51 used the continuous performance test. There was a small difference between the effects of theanine plus caffeine and the effects of placebo on DVT reaction time that favored theanine plus caffeine; however, the CI was consistent with a large difference in both directions (SMD, −0.23; CI, −1.70, 1.23; P = .56; Figure 4A). The SMD was equivalent to approximately −22.20 ms (CI, −161.97, 117.57). There was a small difference between the effects of theanine plus caffeine and those of placebo on DVT accuracy that favored theanine plus caffeine; however, the CI was consistent with little-to-no difference and a small-to-moderate difference (SMD, 0.20; CI, 0.02, 0.38; P = .04; Figure 4B). The SMD was equivalent to approximately 1.59% (CI, 0.17, 3.01). There was little-to-no difference between the effects of theanine and those of placebo on DVT reaction time or accuracy (Table 3).

Attention Switching Task

Six studies^{31–33,37,43,51} reported measuring accuracy during a selective attention task in the second hour after intake of theanine plus caffeine. One³⁷ was excluded because the intervention was delivered prior to the test day. Four studies^31–33,43 used the attention switching task (AST; also called the switch task), and 1 study⁵¹ used the shifting attention test. Three studies^31–33 also took measurements in the first hour. In the first hour there was a small-to-moderate difference in the effect of theanine plus caffeine, and placebo, on AST accuracy that favored theanine plus caffeine and the CI was consistent with a small, and moderate-to-large difference (SMD, 0.4; CI 0.24, 0.57; P = .004; Figure 4C). The SMD was equivalent to approximately 1.32% (CI 0.79, 1.86). The difference in the second hour was also small-to-moderate in favor of theanine plus caffeine, but the CI included little-to-no difference, and a small-to-moderate effect (SMD, 0.33; CI, 0.13, 0.54; P = .008; Figure 4D). The SMD was equivalent to approximately 1.10% (CI, 0.43, 1.77).

Serial 3 and 7 Subtractions

Two studies^41,42 reported measuring correct calculations for serial 3 and serial 7 subtractions in the first hour after intake of theanine plus caffeine, theanine, caffeine, or placebo. There was a mean difference in favor of theanine plus caffeine compared to theanine alone, equivalent to more than 4 correct subtractions during the serial 3 subtractions task; the CI was consistent with a difference of at least 1 correct subtraction in favor of theanine plus caffeine (MD, 4.13; CI, 1.74, 6.52; P = .029; Appendix S5). All other analyses of serial subtractions revealed CIs that included either no difference or differences that favored both interventions being compared (Tables 3 and 4, Appendix S5).

Mood Outcomes

Alertness

Eight studies^{31–33,39,41–44} reported measuring alertness. Alertness from the Bond-Lader visual analogue scales (B-L VAS) was measured 1 hour, and 2 hours, after intake in 5 studies^{31–33,39,41} and 6 studies,^{31–33,41–43} respectively. One study⁴² measured alertness 2 hours after intake using the Caffeine Research Visual Analogue Scales (CRVAS) and was included in the analysis. One study⁴¹ measured alertness using the CRVAS as well as the B-L VAS: The results below included the B-L data, and the results obtained with the CRVAS data are presented in Appendix S6. Two studies^31,43 measured alertness at 2 time points in the second hour after intake. The results below included the later hour-2 measurement, and the results with the earlier hour-2 measurement are presented in Appendix S6. In both the first hour and second hour after intake, there was a small-to-moderate difference in the effects of theanine and caffeine and the effects of placebo on alertness that favored theanine plus caffeine, but the CIs included a small difference and little-to-no difference in favor of placebo in the first and second hours, respectively (Figure 5A hour 1: SMD, 0.34; CI, −0.23, 0.91; P = .18; [B] hour 2: SMD, 0.28; CI, −0.08, 0.65; P = .11). The SMDs were equivalent to approximately 4.13 units (CI, −2.75, 11.00) and 2.94 units (CI, −0.87, 6.75), on the B-L VAS, in hour 1 and hour 2, respectively. One study⁴⁴ administered three different theanine doses and a placebo dose (on 4 separate test days) and measured alertness in the first hour after intake of each. Data from this study and 2 other studies^41,42 were included in analyses comparing the effects of theanine and of placebo on alertness. In each analysis there was little-to-no difference, or a small difference n favor of placebo between the effects of theanine and of placebo on alertness, but in each analysis the CI included at least little-to-no difference in favor of theanine (Appendix S7).

Overall Mood

Four studies^34,35,41,42 reported a subjective measurement of overall mood in the second hour after intake. Two studies^41,42 reported “overall mood” from the CRVAS, and 2 studies^34,35 reported total mood disturbance from the Profile of Mood States (POMS) questionnaire. There was a small-to-moderate difference between the effects of theanine plus caffeine and the effects of placebo on overall mood that favored theanine plus caffeine; however, the CI was also consistent with little-to-no difference, in favor of placebo (SMD, 0.26; CI, −0.10, 0.63; P = .11; Figure 5C). The SMD was equivalent to approximately 3.46 units (CI, −1.36, 8.28) on the CRVAS.

Two studies^41,42 reported feelings of relaxation, jitteriness, tiredness, tenseness, and “headache”, measured using the CRVAS, 2 hours after intake of theanine plus caffeine or theanine, caffeine, or placebo alone (Appendix S8). For feelings of relaxation, there was a mean difference of at least 3 units on a VAS between the effects of placebo, and theanine, caffeine, or both; however, the CI in each analysis included at least no difference between the effects of each treatment and placebo. For feelings of tiredness, the mean difference for caffeine vs placebo was −12.34 (CI, −15.11, −9.58; P = .011) and for theanine vs caffeine was 13.66 (CI, 3.02, 24.3; P = .039). For feelings of jitteriness and “headache”, the mean differences between theanine plus caffeine and placebo alone were −3.25 (CI, −3.81, −2.68; P = .009) and −6.83 (CI, −9.51, −4.15; P = .02), respectively. No MID appears to have been established for any of these outcomes.

Sleep Outcomes

Five studies reported sleep outcomes in free-living adult participants.^57,74–77 One study⁵⁷ reported that 200 mg of theanine per day for 4 weeks, compared with a placebo, was associated with significant improvements in sleep latency, sleep disturbances, and the use of sleep medication, compared to placebo, when measured using the Pittsburgh Sleep Quality Index. Another study⁷⁶ reported that intake of low-caffeine green tea, but not standard green tea, showed a significant positive correlation with total combined time spent in sleep stages N2 and N3 (the latter 2 stages of non–rapid eye movement sleep), measured over 3 nights. However, the total combined time was not significantly different between the interventions. Three studies of students undertaking pharmacy practice reported no significant differences in subjective sleep time between groups consuming green tea, theanine, or placebo.^74,75,77 One polysomnography study found no difference in sleep parameters after intake of oolong tea, caffeine, or placebo.⁸⁰ In children with ADHD (attention deficit hyperactivity disorder), 400 mg theanine per day for 6 weeks, compared with a placebo, significantly improved objective sleep parameters (the percentage of time spent in restful sleep and occasions of night-time activity).⁶⁵ The number of minutes spent awake after the initial onset of sleep (wakefulness after sleep onset) tended to be improved (reduced) by the theanine treatment, but this effect did not reach statistical significance (P = .058). No improvements in subjective sleep parameters were reported. In children with Autism Spectrum Disorder, 1 study⁵⁶ reported that objective sleep parameters were not significantly different between the oolong tea and placebo conditions, or the green tea and placebo conditions.

Risk of Bias

Overall, zero studies were judged to be at low risk of bias, 12 were deemed to be of “some concerns,”, and 25 were judged to be at high risk of bias (Figure 6).⁸¹ Eighteen studies failed to report their methods for participant randomization or intervention sequence/allocation concealment and were judged to be of some concern in domain one.

DISCUSSION

The main findings of this systematic review and meta-analysis are that theanine plus caffeine likely confers small-to-moderate improvements, compared to placebo, in attentional task performance and may benefit certain mood outcomes, primarily in the second hour after intake, and that theanine alone may confer small-to-moderate improvements, compared to placebo, on cognitive performance measures during the first and second hours after intake, in healthy adults. However, the CIs were frequently consistent with little-to-no difference in favor of the intervention, or placebo, indicative of uncertainty surrounding the direction and magnitude of these differences. The evidence regarding the effects of tea, theanine plus caffeine, or theanine alone on sleep outcomes was inconclusive.

This meta-analysis showed that the combination of theanine and caffeine likely improves cognitive performance to a greater extent than placebo. The differences found in favor of theanine plus caffeine for simple reaction time and accuracy in the choice reaction, digit vigilance, and attention switching tasks, adds to the evidence from previously published studies^13,14 that theanine plus caffeine likely improves performance in tasks requiring attention. Based on a previous meta-analysis,¹³ the authors concluded that any beneficial effects of the combination of theanine and caffeine on attention are likely attributable to caffeine, and EFSA suggested that a health claim for black tea would be based on its caffeine content. In the present study, the results of the analyses of theanine alone compared to caffeine alone on cognitive outcomes suggested that caffeine improved SRT, CR time and accuracy, RVIP reaction time and accuracy, and serial 3 and 7 subtractions to a greater extent than theanine because the overall effect sizes consistently favored caffeine (ie, positive for reaction time results and negative for accuracy results; Appendix S5). However, the wide CIs and lack of statistical significance indicated that the precise differences between the effects of theanine and caffeine on cognitive outcomes are not fully understood. Interestingly, the analyses of theanine alone compared to placebo (Table 3) revealed that theanine may in fact be beneficial for cognitive outcomes at a behavioral level. However, the lowest dose of theanine used in any study included in these analyses was 50 mg,⁴² which is equivalent to approximately 2 cups of black tea,⁸² and all other studies used doses >100 mg,^36,38,40,41 and no study delivered their interventions as tea or black or green tea–flavored drinks. One study³⁶ did use a tea intervention (loose-leaf tea) but did not report the type of tea, nor the quantities of theanine and caffeine in the delivered interventions; hence, the data from this intervention were not included in the analyses. A recent functional magnetic resonance imaging study²¹ has also provided evidence showing that theanine and caffeine could improve attention by reducing activity in brain regions associated with mind wandering—also termed spontaneous thought⁸³—and 1 study found positive effects of 100 mg theanine on behavioral cognition outcomes,⁵² although 2 other studies found no effect,^55,62 and 1 study showed an effect in favor of placebo compared to theanine,⁵⁴ using doses ≥100 mg.

A high degree of variability in the magnitude of effect sizes and breadth of CIs was a defining feature of the findings from this meta-analysis. These results could be, in part, a consequence of the small number of studies included in each analysis, which may be overcome if future investigators opt to use outcome measures that have been used previously and report a complete summary of their results. The risk of bias analysis also showed that future studies should aim to provide greater detail in the Methods and Results sections to reduce the risk of bias associated with the results. There was also variation in the reported effects of theanine and caffeine between studies using similar doses or outcome measures. Two studies^36,41 that used similar doses of theanine and caffeine delivered in combination demonstrated contradictory effects on SRT (Figure 3A) and CR accuracy (Figure 4C) in the first hour after intake—but the tasks used to assess these outcomes were not identical. A similar observation can be seen in the alertness analysis (Figure 5A and B), although in this case, the 2 studies used the same mood questionnaire.^32,33 Variation in observed effects between studies may also be explained by biological and biochemical factors. While caffeine-containing interventions might be expected to exert effects in the first and second hours after intake given caffeine’s absorption and retention properties^84,85—although peak plasma concentration does not necessarily correlate with behavioural improvements⁸⁶—variation in observed effects between studies of caffeine-containing interventions could be influenced by interindividual genetic variation linked to caffeine metabolism. A recent meta-analysis revealed that hetero- and homozygosity for a single-nucleotide polymorphism (C → A) within the gene encoding the caffeine-metabolizing protein CYP1A2 resulted in a greater ratio of paraxanthine to caffeine, an effect accentuated by tobacco smoking.⁸⁷ Subsequently, 1 study showed that caffeine improved sustained attention performance in A-allele homozygotes compared to C-allele carriers, but no effect of genotype on the paraxanthine-to-caffeine ratio was observed.⁸⁸

The current review revealed that the difference between the effects of theanine plus caffeine, and placebo, on alertness was not as conclusive as has been suggested before. A previous meta-analysis¹³ revealed that theanine plus caffeine significantly improved alertness in the first and second hours after intake, but only after excluding the data from a study deemed to be a statistical outlier. In the present analysis, no adverse clinical or methodological diversity was found to justify this exclusion. While a small-to-moderate difference between the effects on alertness of theanine plus caffeine or placebo that favored the former was found in both the first and second hours after intake, both CIs included little-to-no difference in favor of placebo, and the prediction intervals showed that future studies may find no difference between theanine plus caffeine or placebo, or a difference favoring either.

It has long been hypothesized that tea induces feelings of relaxation,²² and research from the last 20 years has sought to elucidate these effects from behavioral and neurophysiological perspectives. While the results of the present meta-analysis suggested that theanine (≥50 mg), caffeine (≥75 mg), and the combination may improve feelings of relaxation compared to placebo, data from only 2 studies^41,42 were available, and the CIs were consistent with both no difference and a difference favoring placebo, evidence of a high degree of uncertainty in the MDs. In studies not eligible for meta-analysis, 1 study showed that tea intake improved subjective relaxation following a psychophysiological stress test,⁷³ but studies that have assessed calmness via the B-L VAS following a cognitive test battery have not found statistically significant effects of theanine-caffeine combinations,^31–33,43 and 1 study found no differences in mood between participants given a single cup of tea, placebo, or control.⁸⁹ In studies that have used EEG (electroencephalography) to measure alpha power, a brain wave associated with relaxation, there appeared to be a division between recordings taken while participants were engaged in a cognitive task, and while participants were resting. In the resting state, studies have shown that theanine increased alpha power at 50 mg⁶⁶ and 200 mg,^18,22 whereas in active and anticipatory conditions, studies have shown that theanine reduced alpha power at 100 mg,⁶² and 1 study⁵² found no effect of theanine at 100 mg—and 250 mg,^54,55 which is thought to correspond to cognitive improvements.²¹ Further research exploring this relationship, particularly from a behavioral perspective and using tea beverage interventions or doses of theanine that reflect the levels found naturally in tea, would be informative.

The results of the present review showed no consistent effects of either theanine, caffeine, or the combination, on sleep. Another review¹⁹ reported results from 2 original studies that were published in Japanese. The review that included these studies, 1 of men (n = 10) and 1 of postmenopausal women (n = 20), showed that 200 mg of theanine, but not placebo, 1 hour before going to bed, was significantly beneficial for certain objective, but not subjective, sleep (in men), and sleep-associated (in women), outcomes. It is worth mentioning that the 3 studies that reported positive effects on sleep used high doses of theanine and/or objective or validated measurement techniques. One of these studies used both a high dose and objective measures,⁶⁵ 1 study used a high dose and validated questionnaire,⁵⁷ and 1 study used objective measures.⁷⁶

The key strength of this meta-analysis was the inclusion of analyses comparing theanine with placebo, caffeine, and the combination of theanine and caffeine, something that to our knowledge had not been done at the meta-analytical level before. The overarching limitation of this meta-analysis was the inability to draw definitive conclusions about the true effects of the theanine-caffeine combination, and theanine-only, interventions, a point highlighted by the large prediction intervals. This limitation stemmed from the extent of clinical and methodological diversity between the included studies, particularly the variety of outcomes assessed, and the measurement tools used, which restricted the number of studies that could be included in any meta-analysis; only 3 analyses included more than 4 studies. This was a problem accentuated by unreported results. The second key limitation was the uncertainty in the source of variation in the observed effects reported in the studies in each analysis. The I² statistic in analyses of a few studies is considered to be uninformative^90,91; therefore, the I² CIs were factored in. Frequently encompassing a lower bound of zero and an upper bound greater than 75%, the proportion of variation that is attributable to variation in the true effects or to sampling error was often unclear.

CONCLUSION

In conclusion, the combination of theanine and caffeine, compared to placebo, likely improves performance in attentional tasks such as the AST and DVT, and may benefit mood outcomes such as alertness and overall mood; theanine alone may improve performance in the CR and RVIP tasks. However, it was not always possible to rule out little-to-no difference between treatments, or differences favoring placebo. Most studies included in this review were conducted in the laboratory, and few studies used tea interventions or tea-equivalent doses/ratios of bioactive compounds, the investigation of which would produce results of greater relevance to tea drinkers and would better complement the guideline intake recommendation for flavan-3-ols, as this was a food-based proposition.¹¹ Further research exploring the effects of commercially available tea, or tea-equivalent bioactive doses, at an individual level and in free-living contexts, would provide greater insight into the potential benefits of tea consumption for cognitive and mood outcomes.

Acknowledgments

We acknowledge David Cooper, of the University of Aberdeen, for valuable insights into conducting a meta-analysis and interpreting the results.

Author Contributions

B.d.R. and A.G. conceptualized the study. All authors were involved in designing the study. E.P., B.d.R., M.A.-M., and A.G. were involved in the systematic search process. E.P. collected the data, performed the meta-analyses, and wrote the manuscript. All authors were involved in critically reviewing the manuscript, and all authors approved the final manuscript.

Supplementary Material

Supplementary Material is available at Nutrition Reviews online.

Funding

This research forms part of a PhD studentship to E.P., funded by Lipton Teas and Infusions. The research of B.d.R. and of M.A-M. is funded by the Scottish Government Rural and Environment Science and Analytical Services Division (RESAS), grant RI-B5-06.

Conflicts of Interest

The PhD studentship of E.P. is funded by Lipton Teas and Infusions. B.d.R supervises the PhD studentship of E.P. but is herself not directly funded by Lipton Teas and Infusions. J.D. is an employee of Lipton Teas and Infusions. A.G. is an employee of Unilever, a company that produces and markets tea and tea-based products. M.A.-M. has no interests to declare.

References