The effects of tea on psychophysiological stress responsivity and post-stress recovery: a randomised double-blind trial
- First Online:
- Cite this article as:
- Steptoe, A., Gibson, E.L., Vounonvirta, R. et al. Psychopharmacology (2007) 190: 81. doi:10.1007/s00213-006-0573-2
Tea has anecdotally been associated with stress relief, but this has seldom been tested scientifically.
To investigate the effects of 6 weeks of black tea consumption, compared with matched placebo, on subjective, cardiovascular, cortisol and platelet responses to acute stress, in a parallel group double-blind randomised design.
Materials and methods
Seventy-five healthy nonsmoking men were withdrawn from tea, coffee and caffeinated beverages for a 4-week wash-out phase during which they drank four cups per day of a caffeinated placebo. A pretreatment laboratory test session was carried out, followed by either placebo (n = 38) or active tea treatment (n = 37) for 6 weeks, then, a final test session. Cardiovascular measures were obtained before, during and after two challenging behavioural tasks, while cortisol, platelet and subjective measures were assessed before and after tasks.
The tasks induced substantial increases in blood pressure, heart rate and subjective stress ratings, but responses did not differ between tea and placebo treatments. Platelet activation (assessed using flow cytometry) was lower following tea than placebo treatment in both baseline and post-stress samples (P < 0.005). The active tea group also showed lower post-task cortisol levels compared with placebo (P = 0.032), and a relative increase in subjective relaxation during the post-task recovery period (P = 0.036).
Compared with placebo, 6 weeks of tea consumption leads to lower post-stress cortisol and greater subjective relaxation, together with reduced platelet activation. Black tea may have health benefits in part by aiding stress recovery.
KeywordsTeaStressCortisolHeart rateBlood pressurePlatelet activationCaffeineMood
Drinking tea has traditionally been associated with stress relief, and many people say that drinking tea helps them relax after facing the vicissitudes of everyday life. However, scientific evidence for the relaxing properties of tea is very limited. Black tea drinking has been found to enhance positive moods acutely (Quinlan et al. 2000), and to maintain alertness over the day (Hindmarch et al. 2000). Steptoe and Wardle (1999) showed significant associations between tea drinking and positive mood in a diary study, particularly among respondents who reported high social support.
If tea does have relaxing properties, these might derive either from direct biological effects or from the social context in which it is consumed. Animal studies indicate that tea flavonoids have antagonistic effects on sympathetic nervous system activity (Perez-Vizcaino et al. 2002) and reduce blood pressure in spontaneously hypertensive rats (Duarte et al. 2001). Henry and Stephens-Larson (1984) demonstrated in a mouse psychosocial stress model that 3 to 5 months of decaffeinated tea consumption reduced blood pressure and a number of psychobiological markers of stress compared with drinking water. Components of tea also have central nervous system effects. For example, oral administration of theanine (an amino acid found in tea) stimulates increased α-wave activity in the occipital and parietal regions in human volunteers within 40 min (Juneja et al. 1999). The principal catechin in tea, epigallocatechin gallate (EGCG), has sedative effects, and reduces corticosterone responses to separation stress (Adachi et al. 2006). These responses were attenuated by the γ-aminobutyric acid (GABA)A receptor antagonist, picrotoxin, suggesting partial mediation through GABAnergic pathways. At the same time, tea is frequently consumed under conditions that are conducive to relaxation that may, themselves, be responsible for the apparent benefits. The most rigorous way to rule out this possibility is to carry out a placebo-controlled double blind trial, so that the direct impact of tea can be evaluated.
A widely accepted method of assessing stress responsivity and post-stress recovery in humans is psychophysiological stress testing, in which subjective and physiological variables are recorded while people are administered standardised behavioural challenges (Schneiderman and McCabe 1989). Psychophysiological stress testing has been used to evaluate cardiovascular disease risk (Steptoe 1997), psychological characteristics such as depression (Kibler and Ma 2004), habitual physical activity levels (Hamer et al. 2006a), and other factors, including the influence of functional foods on biological responsivity (Hamer et al. 2005). The method has largely been used to analyse cardiovascular and neuroendocrine responses. However, it has been extended to evaluate immune and inflammatory processes and biological responses relevant to cardiovascular disease such as platelet activation (Brydon et al. 2006).
The present study investigated the influence of 6 weeks of black tea compared with matched placebo on subjective, cardiovascular, cortisol and platelet responses. Caffeine intake is positively associated with cortisol and cardiovascular stress responses (James 2004; Lovallo et al. 2005), so the tea and placebo conditions were matched for caffeine levels. In view of the notion that tea has relaxing properties, we assessed both acute biological responses and rate of post-stress recovery. A double-blind methodology was adopted, so that the influence of tea could be evaluated independently of the normal cues and circumstances surrounding tea drinking.
Materials and methods
The study involved an initial laboratory assessment session that is not described here, but has been analysed in relation to coffee consumption (Hamer et al. 2006b). It was followed by a 4-week washout phase, during which participants were withdrawn from ordinary tea, coffee and other caffeinated beverages, and drank fruit-flavoured caffeinated placebo tea. Use of aspirin, ibuprofen, caffeine and dietary supplements was prohibited. Various fruits and vegetables rich in flavonoids were either excluded from participants’ diets altogether, or restricted to no more than one portion per day (citrus fruits and berries, apples and chocolate products). Otherwise, participants continued with their usual diets. The washout phase was designed both to standardise tea and caffeine conditions before randomisation and to screen out volunteers who did not comply with instructions. At the end of the washout phase, the baseline psychophysiological assessment was carried out, and participants were then randomised to 6 weeks tea or placebo in a double-blind parallel group design.
Composition of the treatments in the study
Caffeine (naturally occurring)
Black tea formulation (Galenic form)
Community volunteers were recruited through printed and email invitations to local employers and commercial outlets. We recruited apparently healthy, male, nonsmoking tea drinkers aged 18–55 years. People who were diabetic or who reported significant medical or psychiatric histories were excluded, as were those who were currently prescribed medication or who followed vegetarian or restricted diets. Volunteers were told that the aim of the study was to understand how tea influences mental and physical function, particularly the blood vessels and substances circulating in the blood. One hundred five individuals were recruited, of whom 13 were discharged during the washout phase for lack of compliance. Seventeen other participants for whom there were insufficient data were excluded from all analyses. The remaining 75 were randomized to tea (37) or placebo (38). Participants who completed the study were given an honorarium of £250.
Psychophysiological stress testing
Testing was carried out in the morning, in a quiet, air-conditioned room. Before arrival in the laboratory, participants consumed one serving of their allocated beverage so as to avoid any confound due to caffeine withdrawal during testing (James and Rogers 2005). Weight and height were measured for the calculation of body mass index (BMI), and the first blood sample was drawn from the antecubital fossa. Participants then rested for 10 min while blood pressure and heart rate were measured continuously using a Finapres (TNO Biomedical Instrumentation, Amsterdam, The Netherlands) which derives measures from the finger using the vascular unloading technique. The Finapres provides beat by beat data, so it detects the full profile of cardiovascular responses that cannot be provided by intermittent measures using sphygmomanometry. The last 5 min of data were averaged to constitute the baseline. Participants then completed ratings of stress and relaxation using 7-point scales where 1 = low to 7 = high. The baseline was followed by two standardised behavioural challenges that have been used extensively in previous research (Steptoe et al. 2003; Strike et al. 2006). The first was a socially evaluating speech task in which participants were given one of three stressful situations assigned at random (threat of unemployment, a shop lifting accusation and an incident in a nursing home involving a close relative). They were instructed to prepare a verbal response for 2 min and then to speak for 3 min, and performance was recorded on a video camera. The second task was mirror tracing, involving the tracing of a star with a metal stylus, which could only be seen in a mirror image (Lafayette Instruments, Lafayette, IN, USA), and this was also carried out for 5 min. Cardiovascular monitoring continued throughout each task, and ratings of perceived stress were obtained immediately after each task. Participants also rated the difficulty and controllability of tasks and task involvement on 7-point scales. A second blood sample was drawn 10 min post-task, and recovery heart rate, blood pressure were then monitored for a further 5 min, and recovery ratings of stress and relaxation were obtained. Saliva samples were obtained at baseline, immediately after tasks, and then at 15, 30 and 50 min post-stress for the assessment of salivary cortisol.
Platelet activation was measured with flow cytometric assessment of circulating leukocyte–platelet aggregates as described previously (Steptoe et al. 2003). Blood was drawn using a butterfly needle with Luer adapter connected to a sodium citrate Vacutainer, and the first 2 ml of each sample were discarded. Whole blood samples (10 μl) were diluted with 90 μl Hepes buffered saline and then incubated at room temperature for 20 min with 10 μl each of fluorescein isothiocyanate-conjugated mouse antihuman CD45 monoclonal antibody (clone H130, BD PharMingen, Oxford) and R-Phycoerythrin (PE)-conjugated mouse antihuman CD42a monoclonal antibody (clone ALMA.16, BD PharMingen) which recognize leukocytes and platelets, respectively. An isotype matched PE-conjugated antibody was used as a negative control. After fixation, total platelet–leukocyte aggregates and their subsets were quantified using a Becton Dickinson FACScan Flow Cytometer and Cellquest software. Results are presented as percentages of monocytes, neutrophils and total leukocytes bound to platelets. Saliva free cortisol was assessed at the University of Dresden using a time-resolved immunoassay with fluorescence detection.
Blood pressure and heart rate data were averaged into four 5-min trials (baseline, speech, mirror tracing and recovery). Pretreatment psychophysiological responses were analysed using repeated measures analysis of variance with group (tea, placebo) as the between-subject factor and trial as the within-subject factor. The analysis of cortisol involved five trials within sessions (baseline, immediately post-task, and 15, 30 and 50 min post-task). The Greenhouse–Geisser correction for degrees of freedom was applied when appropriate, and post hoc tests were made using Tukey’s least significant difference (LSD) test. Additionally, total cortisol output over each session was computed using area under the curve methods (Pruessner et al. 2003). The effect of tea was assessed with repeated measures analysis of covariance of posttreatment data, using corresponding pretreatment values for each trial as covariates. Complete data from all 75 participants were available for BP and platelet activation, but missing values resulted in one person missing from heart rate analyses, two from subjective rating analyses and four from cortisol analyses. Results are presented as means±standard deviation.
Characteristics of participants in the two experimental groups
Tea group (n = 37)
Placebo group (n = 38)
33.2 ± 8.6
33.1 ± 8.1
Ethnicity: White (%)
Marital status: married (%)
Education: GCSE/A-level (%)
Degree and above
Body mass index (kg m2)
25.6 ± 3.2
25.4 ± 3.1
Former smokers (%)
Vigorous physical activity (2 weeks)
4 times or more
Baseline tea consumption (cups/day)
2.19 ± 1.8
2.20 ± 1.9
Pretreatment stress responsivity
Biological and subjective responses pretreatment
Post-task 1 min
Post-task 10 min
Post-task 15 min
Post-task 20–25 min
Post-task 30 min
Post-task 50 min
Systolic BP (mmHg)
121.4 ± 11.5a
160.9 ± 22.2b
156.0 ± 21.2c
130.2 ± 14.9d
Diastolic BP (mmHg)
72.6 ± 8.9a
94.8 ± 12.2b
93.7 ± 13.9b
78.0 ± 10.8c
Heart rate (bpm)
68.5 ± 12.1a
83.4 ± 16.8b
75.6 ± 14.6c
65.8 ± 11.0d
Stress rating (1–7)
2.36 ± 1.1a
4.23 ± 1.3b
4.51 ± 1.2b
2.05 ± 1.0c
Relaxation rating (1–7)
5.36 ± 1.1a
5.72 ± 0.89b
Platelet–monocyte aggregates (%)
5.91 ± 1.4a
6.38 ± 1.5b
Platelet–neutrophil aggregates (%)
4.60 ± 0.95a
4.88 ± 1.0c
Platelet–leukocyte aggregates (%)
5.02 ± 0.94a
5.23 ± 1.0b
11.7 ± 6.3a
11.0 ± 5.8a
10.9 ± 6.3a
9.63 ± 5.7b
7.21 ± 4.1c
Effects of tea and placebo treatments
Treatment effects were also evident in subjective relaxation ratings. Repeated measures analyses of covariance of baseline and recovery ratings revealed a significant group by trial interaction (F(1,70) = 4.54, p = 0.037). The two groups did not differ in baseline relaxation ratings, but the change in relaxation between baseline and recovery was positive in the tea and negative in the placebo condition (p = 0.036). The rating of relaxation increased by 6.26% in the tea condition but fell by 3.19% in the placebo group. This corresponds to just under a 10% difference in change in relaxation rating. There were no significant changes over sessions or differences between groups in ratings of task difficulty, controllability or involvement.
This study used laboratory psychophysiological testing to evaluate the influence of chronic tea intake on cardiovascular, neuroendocrine, platelet and subjective stress responses. It was carried out using placebo-controlled, double-blind methodology. The main findings are that tea had no effect on blood pressure or heart rate stress reactivity or recovery. Participants in the tea condition had lower platelet activation, an effect that was present both in baseline and stress samples. Tea also led to lower cortisol towards the end of the post-task recovery period and to greater subjective relaxation post-tasks. Tea, therefore, appears to influence the effectiveness of post-stress recovery, rather than the magnitude of stress responses themselves. These effects cannot be attributed to bias on the part of participants or investigators.
Studying the psychophysiological effects of tea presents particular difficulties. Unlike substances such as Ginkgo biloba that can readily be presented in the form of placebo capsules (Jezova et al. 2002), a placebo drink that tastes like black tea cannot be manufactured. Many studies of psychological and biological responses have, therefore, compared tea with water (e.g., Duffy et al. 2001; Hindmarch et al. 1998; Quinlan et al. 1997). Under these conditions, both the participants and investigators are aware of treatment condition, so, bias cannot be ruled out. In the present study, we overcame this problem by administering tea and placebo in fruit flavoured forms. These not only masked any sensory changes when switching from placebo to active tea but also removed the beverage from the normal set of sensory cues associated with tea drinking, so that the effects of tea could be studied independently of potential confounders.
The pattern of subjective and physiological responses to behavioural tasks was typical of results obtained with these stimuli (Steptoe et al. 2002, 2003). Blood pressure, heart rate and subjective ratings of stress all increased acutely, returning toward baseline during the post-task recovery period. Recovery of systolic and diastolic BP was incomplete 20 min post-stress, as noted in previous investigations. The increase in cardiovascular activity was substantial, with BP rises of more than 30% (Table 2). By contrast, although platelet activation increased reliably, the magnitude of responses was relatively small (<10%). We were only able to sample blood twice during experimental sessions, and the timing of the post-stress blood draw (10 min after tasks) may not have been optimal for all individuals. The small magnitude of responses limited the scope for observing reduced stress responsivity following tea treatment. The absence of a stress-free condition prevented assessment of any increase in cortisol in response to the tasks, though it could be argued that the expected diurnal decline in cortisol was arrested by the stress tasks, as evidenced by the lack of cortisol decline between immediate and 15 min post-stress samples (Fig. 3). Cortisol responses to acute challenges vary greatly in relation to situational factors and the type of demand imposed (Dickerson and Kemeny 2004), and we have observed small responses in previous studies using these stimuli (Kunz-Ebrecht et al. 2003). Furthermore, the placebo and active tea drinks contained equal amounts of caffeine, so that any differential effects were independent of caffeine, unlike many previous studies of effects of tea.
The first hypothesis tested in this study was that chronic tea administration would reduce stress-induced cardiovascular and platelet activation. This hypothesis was not confirmed, as we found no evidence that the magnitude of stress responses was systematically reduced by tea administration. There was a tendency for heart rate to be lower following tea than placebo administration, but set against this is the slightly higher level of diastolic BP in the tea condition (Fig. 1). One explanation of the lack of differences may be that caffeine intake was equated across experimental groups. Caffeine is a central nervous system stimulant and promotes modest increases in blood pressure and cortisol (James 2004; Lovallo et al. 2005). This effect may have outweighed any influence of the more specific constituents of tea.
There was a significant group difference posttreatment in platelet activation, as shown in Fig. 2. This was evident both in baseline and post-task samples, so was not an effect on stress reactivity but on tonic levels. We have discussed the baseline difference elsewhere (Steptoe et al. 2006). However, the fact that platelet–leukocyte aggregate levels were lower during both baseline and stress samples following tea treatment may be significant for cardiovascular health. Platelets play an important role in the development of coronary atherosclerosis and in the acute processes underlying acute coronary syndrome (Brydon et al. 2006; Monaco et al. 2005). Earlier studies of stress and platelet activation have used indirect measures such as aggregation in response to physiological agonists (collagen, adenosine diphosphate (ADP)) and measurement of a concentration of platelet products in plasma (von Kanel et al. 2001). An important limitation to these techniques is the sample manipulation required before analysis, and plasma separation can lead to artefactual platelet activation. Whole blood flow cytometry involves little sample manipulation as cells are maintained in their native milieu.
Michelson et al. (2001) have argued that the measurement of platelet–leukocyte aggregates by flow cytometry provides a more accurate assessment of platelet activation than other measures. The polyphenols in black tea have previously been shown to inhibit platelet activation in vitro (Formica and Regelson 1995; Neiva et al. 1999), but two studies comparing tea and water administration over 4 weeks did not show any effect on platelet aggregation stimulated by collagen, ADP or thrombin receptor activating peptide (Duffy et al. 2001; Hodgson et al. 2001). Our results may have been due to the longer period of administration (6 weeks) or the method of assessing platelet activation. The fact that platelet–leukocyte aggregate levels during stress were diminished after tea treatment suggests that this might be a mechanism through which tea drinking contributes to reduced cardiovascular disease risk (Peters et al. 2001).
The second hypothesis was that tea consumption would enhance psychophysiological recovery following stress. The recovery arm of physiological responses has attracted increasing attention over recent years, as prolonged activation may be an important disease-related process (Brosschot et al. 2005). In the allostatic model developed by McEwen (McEwen 1998), impaired post-stress recovery is an indicator of chronic allostatic load. Slow recovery following acute stress has been associated with increased cardiovascular disease risk and heightened mortality in prospective investigations (Cole et al. 1999; Steptoe and Marmot 2005).
Evidence that tea enhances post-stress recovery was provided by two measures. First, ratings of relaxation during the recovery period were more positive in the tea than in the placebo conditions. As there were no differences in subjective stress responses to the tasks or in cognitive appraisal of tasks, this finding suggests that tea drinking had a specific effect on subjective recovery from stress. Second, cortisol showed a greater decline following stress in the tea condition (Fig. 3). By the final sample 50 min post-stress, cortisol had fallen to 53% of baseline in the tea compared with 73% of baseline in the placebo condition. It is notable that this response took an extended period to evolve. Post-stress recovery of BP and heart rate were measured 20 min after tasks, and this may have been too early to detect beneficial effects of tea consumption.
Both natural and synthetic flavonoids have affinities for the benzodiazepine binding sites of GABAA receptors, and this may be an important mechanism in the effects of tea flavonoids on the central nervous system (Dekermendjian et al. 1999; Paladini et al. 1999). As noted in the Introduction section, another component of tea that could contribute to enhanced recovery from stress is theanine, which may inhibit action of excitatory amino acid neurotransmitters and has been found to reduce anxiety under resting conditions but not in response to experimental stress (Lu et al. 2004). Preliminary evidence suggests that theanine may potentiate some actions of caffeine on cognitive function and arousal, and this might explain the dissociation between the effect of tea on post-task cortisol recovery and relaxation and the lack of effect on cardiovascular reactivity (Haskell et al. 2005).
The strengths of the study were the randomised double-blind placebo-controlled design, so that neither the participants nor the investigators were aware of group assignment, the use of standardised psychophysiological stress testing, assessment of platelet activation of whole blood flow cytometry, continuous beat to beat measures of blood pressure and heart rate and treatment conditions that lasted 6 weeks. Most studies of the biological effects of chronic tea consumption have involved a 4-week intervention phase, and this may not be long enough for differential response patterns to develop. The limitations of the study include the fact that participants were young, apparently healthy men, so we do not know whether other groups would respond similarly. Platelet stress responses were small, and there was no absolute increase in cortisol level. Future studies would benefit from a non-stress control condition. Additionally, we relied on participants to report deviations from compliance with the dietary restrictions imposed, and adherence to these restrictions was not assessed objectively.
In summary, 6 weeks of drinking the equivalent of four cups of black tea per day lead to lower post-stress cortisol and greater subjective relaxation under laboratory conditions, together with reduced platelet activation, compared with placebo. These results suggest that black tea consumption may have benefits to health, in part, by aiding recovery from stress, mediated through psychoneuroendocrine and inflammatory pathways.
We are grateful to Peirluigi Giacobazzi and Kesson Magid for their assistance in data collection and biological assays. Leigh Gibson is now at Roehampton University, London, Raisa Vounonvirta at the Institute of Cancer Research, Sutton, UK, and Jorge Erusalimsky is at the University of Wales Institute, Cardiff.