Caffeine intake and anxiety: a meta-analysis

Coffee is widely consumed globally and contains multiple bioactive compounds, notably caffeine, a central nervous system stimulant that antagonizes adenosine A1 and A2A receptors implicated in anxiety modulation. Blockade of these receptors can alter neurotransmission (glutamate, GABA, acetylcholine, serotonin, dopamine) and physiological functions (e.g., vasoconstriction, microglial activity), potentially provoking anxiety. Anxiety disorders are highly prevalent worldwide and increased further during the COVID-19 pandemic, with lifestyle changes including potentially increased caffeine intake. Prior epidemiologic studies on caffeine/coffee and anxiety report inconsistent findings, ranging from protective associations to increased anxiety risk. Given the public health relevance and the lack of a meta-analysis focused on healthy populations, the authors aimed to clarify the association between caffeine intake and anxiety risk in individuals without psychiatric disorders through a systematic review and meta-analysis.

The paper highlights mixed evidence from prior studies: an Iranian cross-sectional analysis reported lower odds of anxiety symptoms with weekly or greater coffee consumption, and a Singapore cohort linked tea consumption to lower anxiety symptoms. Conversely, a cross-sectional study among Florida State University students and randomized controlled trials examining caffeine’s effects on mood reported increased anxiety. Previous meta-analyses have mainly addressed caffeine’s effects in patients with panic disorder or its relationship with depressive symptoms, leaving a gap concerning anxiety in healthy populations. These inconsistencies and gaps motivated the current meta-analysis.

The meta-analysis followed PRISMA guidelines. Databases searched included PubMed, Web of Science, Cochrane Library, Embase, CNKI, WANFANG DATA, SinoMed, and VIP from inception to December 1, 2022. Search terms combined controlled vocabulary and free-text terms for coffee/caffeine and anxiety; a detailed PubMed query was provided. Inclusion criteria: randomized clinical trials, prospective cohort, case-control, and cross-sectional studies; healthy populations without psychiatric disorders; caffeine consumption as the exposure/intervention (trial groups received coffee/caffeinated beverages; controls received decaffeinated coffee or placebo); and availability of mean and standard deviation of validated anxiety scale scores. Exclusion criteria: participants with pre-existing anxiety or in specific acute stress situations; concurrent psychotropic drug interventions; duplicate publications (most detailed retained); ambiguous/incomplete or non-transformable data; lack of original data. Two reviewers independently screened titles/abstracts and full texts, resolving disagreements by discussion or third reviewer. Data extracted included authors, year, region, population characteristics, sample size, study design, caffeine consumption details, anxiety assessment methods, outcomes (means and SDs), and adjustment factors. Methodological quality was assessed using the Cochrane risk-of-bias tool for RCTs (selection, performance, detection, attrition, reporting, other biases) and JBI PACES for cross-sectional studies (10-item scale; 0–2 per item). Statistical analysis used RevMan 5.4 and Stata 12.0. Because different anxiety scales were used, standardized mean differences (SMD) with 95% confidence intervals were computed. Heterogeneity was assessed using Q-test and I-squared. Fixed-effects models were used when p>0.1 and I2<50%; otherwise, random-effects models with subgroup and sensitivity analyses were applied. Statistical significance was set at p<0.05. Cohen’s thresholds categorized effect sizes: small (0.2–0.5), medium (0.5–0.8), large (>0.8). Publication bias was assessed by funnel plot symmetry.

- Study selection: 5,365 records identified; 3,895 after removing duplicates; 34 full texts assessed; 8 studies included in the meta-analysis. Designs: seven RCTs and one cross-sectional study; total participants ≈546; regions included Europe, Tunisia, the United States, and Korea. - Anxiety assessment tools included BAI, POMS, STAI, and SAS. Risk-of-bias for RCTs was generally moderate; the cross-sectional study scored high quality on JBI. - Overall effect: Caffeine consumption significantly increased anxiety compared with controls with a large effect size (SMD = 0.94; 95% CI: 0.28 to 1.60; p<0.05), with substantial heterogeneity (I2 = 94.7%; p<0.001). - Sensitivity and subgroup analyses suggested dose as a key source of heterogeneity. Subgroups by dose: • Low-dose caffeine intake (<400 mg): SMD = 0.61; 95% CI: 0.42 to 0.79; p<0.05; low heterogeneity (I2 ≈ 0–10%). • High-dose caffeine intake (≥400 mg): SMD = 2.86; 95% CI: 2.50 to 3.22; p<0.05; moderate heterogeneity (I2 ≈ 45.9%). - Publication bias: Funnel plot appeared symmetrical, suggesting no major publication bias.

This meta-analysis indicates that caffeine intake is associated with increased anxiety in healthy individuals, with stronger effects at higher doses (≥400 mg). These findings align with prior experimental and pharmacokinetic literature suggesting an anxiogenic effect of caffeine. Sensitivity analyses implicated dose as a major contributor to between-study heterogeneity. The results have clinical implications: caffeine’s antagonism of adenosine A1/A2A receptors and interaction with dopaminergic systems may provoke or exacerbate anxiety symptoms. Individual susceptibility may vary due to genetic polymorphisms (e.g., ADORA2A), although none of the included studies assessed these variants, which could introduce residual heterogeneity. Considering the potential for caffeine to interact with psychiatric treatments and the inclusion of caffeine withdrawal in DSM-5, clinicians should assess and counsel on caffeine intake when addressing anxiety. While self-limiting behavior may reduce intake among sensitive individuals, results support caution with higher daily doses. Further mechanistic work, including neurobiological and genetic studies, and examination of brain activity/connectivity, is warranted.

Caffeine consumption is associated with an increased risk of anxiety in healthy populations, with more pronounced effects at intakes ≥400 mg. These results suggest limiting caffeine intake in healthy individuals and highlight the need for future research on mechanisms spanning genetic polymorphisms to phenotypic risk, as well as studies employing a wider dose range to clarify dose–response relationships.

- High between-study heterogeneity in the primary analysis, addressed partially via sensitivity and subgroup analyses. - Potential confounding not fully controlled (e.g., sex differences); some evidence suggests caffeine may increase anxiety in males but not females. - Limited caffeine dose range across included studies (0–460 mg) precluded formal dose–response analysis. - Possible confusion by side effects and unmeasured factors that may bias results.