The Association between Green Tea Consumption and Cognitive Function: A Meta-Analysis of Current Evidence

Cognitive impairments (CoIs), including memory loss, learning difficulties, and diminished focus, are a significant public health concern, particularly as the global elderly population continues to rise [1]. Age-related cognitive disorders, such as dementia and Alzheimer’s disease (AD), are projected to become some of the greatest challenges to public health worldwide [2]. Currently, more than 55 million people worldwide are living with dementia, with nearly 10 million new cases reported each year. Over 60% of those affected reside in low- and middle-income countries [3]. Mortality rates from dementia, including AD, have risen by nearly 60% in the EU between 2011 and 2017 [4]. Dementia is now the seventh leading cause of death among older adults globally, contributing significantly to disability and dependency. While the search for effective treatment options continues, no randomized clinical trials have conclusively demonstrated that any intervention can prevent dementia. Therefore, the development of effective prevention and treatment strategies is of increasing importance [5].

The relationship between dietary factors and cognitive function has garnered substantial research interest, with growing evidence highlighting the role of diet in cognitive decline and dementia development. Notable dietary patterns studied include the Mediterranean diet, dietary approaches to stop hypertension, the Mediterranean diet -dietary approaches to stop hypertension neurodegenerative delayed intervention diet, and the ketogenic diet [6]. Additionally, evidence suggests that beverage consumption, including tea, may influence cognitive performance [7]. Tea, one of the most widely consumed beverages globally [8], originated in Asia and is an important part of many cultural diets. Recent systematic reviews [9, 10] have confirmed the cognitive benefits of tea, particularly for green, black, and oolong tea varieties, with green tea showing the most pronounced effects. Several studies support green tea’s neuroprotective effects, suggesting that its catechin polyphenols (a type of antioxidant found in tea) may help prevent age-related cognitive decline. These antioxidants are thought to work by enhancing the brain’s natural defenses, promoting the growth of new nerve cells, reducing brain inflammation, and regulating the processes that control cell survival and death [11, 12]. A comprehensive review [13] found that green tea consumption may help prevent dementia, AD, CoI, and MCI. A prospective study [14] further demonstrated a significant association between green tea intake and a reduced risk of cognitive decline, while no similar association was observed for black tea in relation to dementia or MCI. Despite the promising evidence, inconsistencies in findings across tea types and study designs highlight the need for a robust meta-analysis. For instance, a longitudinal study [15] found that black and oolong tea, but not green tea, were inversely associated with cognitive decline, while another cross-sectional study [16] showed that black, oolong, and green tea consumptions were associated with enhanced cognitive performance. Given the inconsistencies in existing findings and the lack of large-scale evidence, this meta-analysis aims to clarify the relationship between green tea consumption and the risk of CoI, including dementia, AD, and MCI. Subgroup analyses of study design, population characteristics, and gender differences were conducted to further elucidate potential associations.

This meta-analysis was registered in the PROSPERO platform (registration number: CRD42024595550). The study protocol was developed prior to study selection, providing a detailed outline for each step of the analysis. We adhered to the PRISMA guidelines for systematic reviews and meta-analyses, ensuring transparency and rigor throughout the process. Additionally, given the observational nature of the included studies, we also followed the MOOSE guidelines for epidemiological observational meta-analyses to ensure appropriate reporting of methods and results. The research questions were constructed using the Population, Intervention, Comparison, Outcome, Study Design (PICOS) framework, as detailed in Table 1. The planned statistical methods included the use of OR and 95% CI as the primary effect measures, with random-effects or fixed-effects models applied based on heterogeneity. Subgroup analyses were stratified by study design, type of CoI, population distribution, gender, and tea consumption frequency to control for potential confounders. All statistical analyses were performed using Stata 15.1 software.

A comprehensive literature search was conducted across four major databases: the Cochrane Library, PubMed, Embase, and Web of Science. The search period spanned from September 10, 2004, to September 10, 2024, using keywords such as “Green Tea,” “Green Tea Consumption,” “cognitive impairment,” “Dementia,” and “Alzheimer’s Disease,” among others. Green tea or tea terms were combined with each cognitive disorder term using the “AND” operator, with the search field set to “All Fields.” In addition to database searches, the reference lists of the retrieved studies and prior high-quality reviews were manually reviewed to ensure that no eligible studies were overlooked. The search was limited to studies published in English. Two researchers independently conducted the database searches to ensure accuracy and completeness.

Inclusion criteria were as follows: (1) events – CoI, dementia, AD, and MCI; (2) exposure – limited to green tea consumption, with no restrictions on dosage; (3) comparison – comparison between green tea consumers and nonconsumers, or green tea compared to other types of tea (e.g., black tea, oolong tea); (4) outcomes – studies must report associations with the risk of CoI, dementia, AD, or MCI, providing OR, HR, or RR and their 95% CI, or sufficient data to calculate effect sizes; additionally, results must be adjusted for relevant covariates; (5) study design – cross-sectional and cohort studies that assess the association between green tea consumption and cognitive outcomes. Exclusion criteria include the following: (1) publication type – conference abstracts, editorials, commentaries, meta-analyses, and review articles were excluded; (2) insufficient data – excluded studies with incomplete data or those lacking sufficient information to extract or calculate effect sizes; (3) nonrelevant exposure – studies that did not specify green tea as the primary exposure, or only considered other types of tea (e.g., black tea, oolong tea) as the exposure, were excluded; (4) nonrelevant outcomes – studies that did not assess relevant outcomes were excluded; (5) unadjusted for covariates – studies that did not adjust results for potential confounding factors were excluded.

The initial records retrieved from the databases were imported into NoteExpress reference management software, where duplicates were removed using the software’s duplicate check function, followed by manual screening to retain the most comprehensive or up-to-date versions based on the completeness and relevance of the study information. In cases of uncertainty, the two independent reviewers (Shiyao Zhou and Yating Zhu) compared study details and made the final decision. The reviewers then screened the titles and abstracts to exclude irrelevant studies. The full-text articles of the remaining records were subsequently evaluated, and studies meeting the predefined inclusion and exclusion criteria were selected for meta-analysis. Any disagreements between the reviewers were resolved through discussion, and if consensus could not be reached, a third reviewer (Na Ren) provided an impartial judgment. However, we acknowledge that subjective bias may arise during the screening process, particularly in studies with smaller sample sizes or varying methodologies, which could affect the objectivity of the screening and the consistency of the data.

Two authors (Shiyao Zhou and Yating Zhu) independently extracted the following data from eligible studies: first author’s name, publication year, source, study design, country, age and gender distribution of participants, follow-up duration, sample size (overall and for each category), frequency of intervention, cognitive outcomes, exposure source, as well as OR, HR, and RR values for each study, and adjusted covariates. If data were published more than once, the first publication or the most comprehensive study was selected. For studies with multiple adjustment models, the risk estimate from the model with the greatest number of adjustments was used.

The quality of the included studies was assessed using the Newcastle-Ottawa Scale (NOS), with quality scores ranging from 0 to 9, where higher scores indicate better quality. Scores of ≥7, 4–6, and 0–3 were categorized as high, moderate, and low quality, respectively. The NOS evaluates cohort study quality across three domains: selection, comparability, and outcome. Any disagreements regarding data extraction and quality assessment were resolved through discussion until consensus was reached.

In this meta-analysis, all results were estimated as OR and 95% CI; RR or HR is regarded as mathematically approximate OR [17]. But when P₀ ≥ 0.1, the following formula $OR=RR1−P0+P0×RR$ was applied to convert RR to OR [18]. The pooled OR, derived from the 95% confidence intervals of multivariate models from original studies, was used to evaluate the relationship between green tea consumption and the risk of CoI. Subgroup analyses were stratified by the following factors to control for potential confounders: study design (cohort studies vs. cross-sectional studies), spectrum of CoI (CoI vs. dementia vs. AD vs. mild CoI), age groups (40–49 years, 50–59 years, 60–69 years, and ≥70 years), population distribution (Chinese vs. Japanese), gender (male vs. female), and tea consumption frequency (low vs. medium vs. high consumption groups). Heterogeneity among studies was assessed using the Q-test and I² statistic. An I2 value greater than 50% or a p value <0.05 was considered indicative of heterogeneity [19]. In such cases, a random-effects model was applied; otherwise, a fixed-effects model was used. Sensitivity analysis was performed by omitting one study at a time to verify the robustness of the results [20]. Publication bias was assessed using Begg’s test and Egger’s regression asymmetry test [21, 22], and a p value of less than 0.05 was considered indicative of publication bias. In cases of detected publication bias, the trim-and-fill method was applied [23] to reestimate the pooled risk estimates. This method corrects publication bias by identifying missing studies that cause an imbalance in the funnel plot. It removes studies that skew the results and imputes the missing studies to restore balance, thereby providing a more accurate and unbiased overall effect estimate. All statistical analyses were performed using Stata version 15.1 (StataCorp, College Station, TX). p values were two-sided, and values less than 0.05 were considered statistically significant.

A total of 778 records were retrieved from four databases (304 articles from PubMed, 308 articles from Embase, 150 articles from Web of Knowledge, and 16 articles from Cochrane Library). After removing 253 duplicates, 64 studies were identified as potentially eligible based on the titles and abstracts. Following a full-text review, 46 studies were excluded based on inclusion and exclusion criteria. Ultimately, 18 studies [14, 24‒40] were included in this meta-analysis. Details of the literature screening process are presented in the PRISMA flow diagram (Fig. 1).

This study included 18 research articles published between 2006 and 2024, comprising seven cross-sectional studies and 11 cohort studies. A total of 58,929 participants from Asia and Europe were included. The median age of participants across the studies ranged from 50 to 85 years. In cohort studies, the follow-up period varied from 1 to 10.8 years, with an average duration of 4.6 years and a follow-up rate of over 75%. Each study considered various categories of tea intake, ranging from never or rarely consuming tea to as much as five cups per day. Most studies reported risk estimates adjusted for age, sex, and educational level. Additionally, some studies adjusted for other common risk factors, such as hypertension, diabetes, depression, APOE status, BMI, and cardiovascular disease. Outcome diagnoses were conducted according to established diagnostic criteria, including the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) for dementia; the National Institute of Neurological Disorders and Stroke and Alzheimer’s Disease and Related Disorders Association (NINCDS-ADRDA) criteria for AD; and the Mini-Mental State Examination (MMSE) for CoI. Other ancillary diagnostic criteria, such as the Montreal Cognitive Assessment (MoCA), Mini-Cog, and Petersen criteria, were also used to complement the primary diagnostic measures for CoI (detailed characteristics are presented in Table 2).

The NOS was employed to assess the quality of the studies. Of the 18 studies, one was considered to be of moderate quality, while the remaining 17 scored seven or above, indicating relatively high quality. Overall, the studies were conducted with robust methodologies and produced reliable results (see Table 3).

In the meta-analysis of 18 comparative studies, green tea consumption was found to be negatively associated with the occurrence of CoI OR 0.63 (95% CI: 0.54–0.73), with high heterogeneity (I² = 77.6%), as shown in Figure 2. Therefore, a random-effects model was employed in this meta-analysis. Sensitivity analysis revealed that the studies by Kitamura et al. [33] and Fischer et al. [35] contributed substantially to the observed heterogeneity (see online suppl. Fig. 1, 2; for all online suppl. material, see https://doi.org/10.1159/000543784). After excluding these two studies, the heterogeneity decreased to 60.5%, and the OR for the risk of CoI was 0.59 (95% CI: 0.51–0.69), as shown in Figure 3.

Egger’s test indicated no significant correlation between effect size and standard error (slope = 0.042, p = 0.469), suggesting the absence of small-study effects. However, the intercept (bias = −2.95, p = 0.000) was statistically significant, indicating the potential presence of publication bias. Asymmetry was observed in the funnel plot, as shown in Figure 4. The trim-and-fill method was applied to detect and adjust for publication bias. The results showed no significant asymmetry in the funnel plot, and no studies were trimmed or filled, suggesting that no significant publication bias was detected in the data. The adjusted pooled effect size remained consistent with the unadjusted effect size (statistically significant, p value = 0.000), indicating that the overall effect estimate is robust and the influence of publication bias on the results is minimal. This conclusion further strengthens the reliability and robustness of the study findings.

Based on study design, all studies were divided into two groups: cohort studies and cross-sectional studies (Fig. 5). Results from both groups showed a negative association between green tea consumption and CoI. In the population-based subgroup analyses (Fig. 6), green tea consumption was found to significantly reduce the risk of CoI in the Chinese OR 0.55 (95% CI = 0.44–0.68) and Japanese OR 0.74 (95% CI = 0.66–0.82) populations. The differing effect sizes in this subgroup may be influenced by cultural factors, such as traditional tea consumption habits [41]. In the subgroup analysis of CoI types (Fig. 7), green tea consumption was significantly associated with a reduced risk of CoI OR 0.59 (95% CI = 0.49–0.71), including a decreased risk of dementia OR 0.74 (95% CI = 0.56–0.99) and mild CoI OR 0.64 (95% CI = 0.43–0.96). However, no significant association was noted for AD OR 0.57 (95% CI = 0.21–1.59). Analysis by age group (Fig. 8) revealed the greatest benefits in individuals aged 50–59 years OR 0.52 (95% CI: 0.38–0.72) and 60–69 years OR 0.65 (95% CI: 0.54–0.77). In the gender-specific analysis (Fig. 9), a notable reduction in risk was detected among women OR 0.51 (95% CI = 0.28–0.95) and men OR 0.47 (95% CI = 0.28–0.80). These results suggest that both genders may benefit from green tea consumption, with differences in effect sizes potentially influenced by genetic, sociopsychological, and cultural factors, such as age-related hormonal changes and variations in education and career opportunities [42, 43]. With regard to different consumption levels (Fig. 10), no significant effect was observed in the low consumption group OR 0.70 (95% CI = 0.47–1.04). Conversely, there was an apparent trend toward reduced CoI risk in both moderate OR 0.73 (95% CI = 0.60–0.89) and high green tea consumption groups OR 0.64 (95% CI = 0.50–0.82).

The current meta-analysis included 7 cross-sectional studies and 11 cohort studies, involving a total of 58,929 participants. These studies provide strong evidence that green tea intake or consumption is associated with a reduced risk of CoI. Our findings are similar to those of a recent systematic review [13], which included three cohort studies and five cross-sectional studies, which suggested that green tea consumption may reduce the risk of dementia, AD, MCI, or cognitive decline. Green tea is rich in polyphenols [44], and the ability of polyphenols to cross the blood-brain barrier makes them an important class of compounds for the treatment of neurodegenerative diseases. The neuroprotective effects of green tea are mainly observed in AD, Parkinson’s disease, cerebral ischemia, and brain injury [45]. In recent years, an increasing number of studies have also demonstrated the potential of green tea and its constituents in improving CoI, largely due to their antioxidant and anti-inflammatory properties, protein kinase C activation, and acetylcholinesterase inhibition. Several potential mechanisms have been identified through both in vitro and in vivo studies [46, 47]. For instance, green tea can prevent hippocampal neuronal apoptosis by inhibiting the JNK/MLCK pathway, ultimately improving cognitive function in diabetic rats [48]. For rats with sleep deprivation-induced CoI, green tea and its primary catechin, EGCG, have been shown to prevent memory deficits during TSD 6 h [49]. In age-related cognitive decline, green tea catechins act by increasing the expression of immediate early genes involved in synaptic and neuronal circuit plasticity, thereby inhibiting cognitive decline [50]. However, it is important to note that much of the evidence supporting these mechanisms comes from animal models, and further research in human populations is needed to confirm the applicability of these findings to human health.

In addition, green tea consumption was associated with women in subgroup analyses. This finding is consistent with a recent study by [51] which showed that acute supplementation with decaffeinated green tea extract improved working memory in women aged 50–63 years. And in contrast to the results of a previous study, which found that the benefits of tea consumption were more pronounced in men than in women [52]. We also found that higher frequencies of green tea consumption were associated with stronger protective effects against cognitive decline. In previous studies, Ran et al. [53] found that drinking 1 cup/day of tea reduced the incidence of CoI by 6%, while drinking 2 cups/day reduced it by 11%. Similarly, Zhu et al. [54] reported that for each additional cup/day of tea consumed, the risk of cognitive decline decreased by 11%. When considering age stratification, we observed that individuals aged 50–69 years appeared to benefit more from green tea consumption compared to those aged 40–49 years and those aged 70 and older. This difference may be attributed to age-related metabolic and physiological changes, lifestyle factors, and other relevant variables [55, 56]. A significant reduction in the risk of CoI with green tea consumption was observed in the Chinese and Japanese populations, but not in the European population. This may be due to cultural and dietary differences, as green tea is more commonly consumed in Asian populations. In contrast, its consumption is less frequent in Europe, where diets tend to be higher in processed foods and lower in antioxidants, which may limit its benefits [57]. Furthermore, due to the inclusion of only 1 study with German patients [35], they were not included in the subgroup analyses. It is worth noting that although no significant results were seen in patients with AD, green tea consumption was found to have the potential to reduce the prevalence to some extent in patients with dementia, MCI, and CoI. The overall trend continues to point to the potential cognitive health benefits of green tea.

The strengths of our study include adherence to both the MOOSE checklist and the PRISMA guidelines, ensuring transparency and scientific rigor. The study was also registered with PROSPERO, enhancing methodological transparency. Notably, we are the first to conduct a meta-analysis on the impact of green tea consumption on CoI risk. Additionally, subgroup analyses based on factors such as study design, outcome measures, participant sex, ethnicity, and green tea intake dosage further strengthened our findings.

However, there are several limitations to our study. First, the inclusion of only observational studies limits our ability to establish a causal relationship between green tea consumption and reduced CoI risk. Although an association was observed, it remains uncertain whether green tea directly prevents cognitive decline. Observational studies are inherently limited by their inability to control for all potential confounders, leaving room for unmeasured confounding factors to influence the results. For example, green tea intake is associated with various lifestyle factors and genetic differences, such as physical activity, mental health, sleep quality, and the APOE genotype, all of which may affect cognitive function [58, 59]. Moreover, reliance on self-reported data introduces recall and reporting biases, which could lead to either over- or underestimation of actual green tea consumption, thereby affecting the accuracy of the findings and obscuring true associations. Second, the limited number of included studies, particularly the lack of studies involving non-Asian populations, might have constrained the statistical power of our analysis. Additionally, variations in study origin, geographic location of the study populations, and sample sizes for different disease types could have influenced the pooled results. In our study, publication bias was detected. The Egger test indicated the presence of publication bias, whereas the trim-and-fill method suggested its absence. These discrepancies likely arise from the inherent characteristics of the two methods. The Egger test is highly sensitive to outliers, detecting not only publication bias but also asymmetries caused by factors such as heterogeneity or true small-study effects. In contrast, the trim-and-fill method assumes a specific pattern for missing studies and may underestimate bias if the sample size is small or if the missing studies deviate from this pattern. Publication bias also implies that smaller studies or those with insignificant or negative results may not be published or included. To enhance the interpretability and generalizability of future findings, it is crucial to incorporate more gray literature and unpublished studies to minimize the impact of bias [60, 61]. Third, the self-reported tea consumption by participants was not standardized, and the frequency of green tea intake may not accurately reflect the actual intake of its components. Additionally, not all studies conducted follow-up assessments on tea consumption after the baseline evaluation. This limitation overlooks potential confounding factors such as tea-drinking frequency and duration. Finally, we observed significant heterogeneity in our analyses. The main factors contributing to this heterogeneity, as shown by the subgroup analyses, were study design and subgroups by type of CoI. In the study design subgroups, the heterogeneity in the prospective studies can be attributed to differences in follow-up duration, loss-to-follow-up bias, and differences in sample size. In cross-sectional studies, the observed heterogeneity may stem from unclear causality, inadequate control for confounders, and differences in measurement instruments. Within the CoI-type subgroups, differences in measurement tools, including the MMSE scoring range (0–30 points), assessment dimensions, and the degree of standardization, were identified as the main contributors to the observed heterogeneity. Furthermore, participants in the low tea consumption group were found to have a higher risk of CoI. Potential confounding factors, such as tea-drinking frequency and lifestyle factors, may partially account for this heterogeneity.

In summary, this meta-analysis suggests that green tea consumption may be associated with a lower risk of CoI. Regular green tea drinkers, both men and women, appear to have a reduced risk of cognitive decline. These findings imply that simple lifestyle changes, such as drinking green tea, could help mitigate the risk of CoI. While these results highlight green tea’s potential benefits, the limitations of observational studies and the heterogeneity in the data underscore the need for large-scale, well-designed randomized controlled trials to confirm causality and further investigate the dose-response relationship and long-term effects of green tea on cognitive health. In future research, the role of green tea in personalized nutrition, particularly in interventions tailored to genetic predispositions, should be explored to optimize its cognitive health benefits.

The authors would like to thank the Faculty of Chinese Medicine, State Key Laboratory of Quality Research in Chinese Medicine, and University Hospital, Macau University of Science and Technology, Macao, China; Zhuhai People’s Hospital (The Affiliated Hospital of Beijing Institute of Technology, Zhuhai Clinical Medical College of Jinan University); Faculty of Medicine, Macau University of Science and Technology, Macao, China; Institute of Chinese Medical Sciences, University of Macau, Taipa, Macau, China; and State Key Laboratory of Quality Research in Chinese Medicine, University of Macau, Taipa, Macau, China, for their support in this study. This study was supported by the Science and Technology Program of Guangzhou.

Ethical approval was not required for this study, as it is a meta-analysis of previously published studies and did not involve any new experiments on human or animal subjects.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

This work was supported by the Science and Technology Program of Guangzhou (No. 2024A03J0450). The funder had no role in the study design, data collection, data analysis, data interpretation, or writing of the manuscript.

Shiyao Zhou and Yating Zhu contributed equally to the study design and the development of the literature search strategy; designed data extraction forms; performed statistical analysis using Stata software; and jointly drafted the manuscript, who also carried out data extraction and quality assessment, utilizing the Newcastle-Ottawa Scale for quality assessment. Shiyao Zhou, Yating Zhu, and Na Ren conducted the literature search and performed study selection based on predefined inclusion and exclusion criteria. Yu Liu and Mishan Wu provided critical revisions to the manuscript and served as corresponding authors. All authors have read and approved the final version of the manuscript and agreed to be accountable for all aspects of the work to ensure its accuracy and integrity.

The datasets used and analyzed in the current study are available from the corresponding author upon reasonable request.