Poole · BMJ (Clinical research ed.) 2017 · umbrella review of meta-analyses · n=218 meta-analyses (201 observational, 17 interventional)

Coffee consumption and health: umbrella review of meta-analyses of multiple health outcomes.

Cited 822 times in the scientific literature.

Level 1 - systematic review of randomized trials

Umbrella review synthesizing systematic reviews and meta-analyses of observational and interventional studies

PubMed 29167102 · doi:10.1136/bmj.j5024 · record verified 2026-08-30

What was done

This umbrella review searched PubMed, Embase, CINAHL, and the Cochrane Database of Systematic Reviews to synthesize evidence across meta-analyses of observational and interventional studies examining associations between coffee consumption and any health outcome in adult populations. Studies of genetic polymorphisms in coffee metabolism were excluded.

What was found

The review included 201 meta-analyses of observational studies covering 67 unique outcomes and 17 meta-analyses of interventional studies covering nine unique outcomes. The largest relative risk reductions occurred at 3–4 cups per day versus none for all-cause mortality (RR 0.83, 95% CI 0.79 to 0.88), cardiovascular mortality (RR 0.81, 95% CI 0.72 to 0.90), and incident cardiovascular disease (RR 0.85, 95% CI 0.80 to 0.90). High versus low intake was associated with an 18% lower risk of incident cancer (RR 0.82, 95% CI 0.74 to 0.89) alongside lower risks for several specific cancers, neurological, metabolic, and liver disorders. In pregnancy, high consumption was associated with low birth weight (OR 1.31, 95% CI 1.03 to 1.67), pregnancy loss (OR 1.46, 95% CI 1.06 to 1.99), and preterm birth in the first (OR 1.22, 95% CI 1.00 to 1.49) and second (OR 1.12, 95% CI 1.02 to 1.22) trimesters. Coffee intake was also associated with increased fracture risk in women but not men.

Why it matters

This comprehensive synthesis demonstrates that moderate habitual coffee intake (3–4 cups per day) is generally safe and consistently associated with reduced risk of major chronic diseases and mortality, while establishing clear contraindications in pregnancy and potential risk for fracture in women.

Limits

The evidence base is overwhelmingly observational (201 of 218 meta-analyses), which limits causal inference due to potential residual confounding from lifestyle factors like diet and smoking. Interventional data were sparse (only 17 meta-analyses covering nine intermediate outcomes), and heterogeneity across included meta-analyses in exposure assessment and adjustment variables was not fully eliminated.

Cited by