Hernán · Annals of neurology 2002 · systematic review and meta-analysis · n=48 studies for smoking, 13 studies for coffee

A meta-analysis of coffee drinking, cigarette smoking, and the risk of Parkinson's disease.

Cited 817 times in the scientific literature.

Level 3 - non-randomized controlled study

Systematic review of non-randomized observational (cohort and case-control) studies

PubMed 12205639 · doi:10.1002/ana.10277 · record verified 2026-08-29

What was done

A systematic review and meta-analysis assessing the association between cigarette smoking, coffee consumption, and physician-confirmed Parkinson's disease. Case-control and cohort studies reporting relative risk were pooled by inverse-variance weighting of study-specific log relative risks. The smoking analysis included 44 case-control and 4 cohort studies; the coffee analysis included 8 case-control and 5 cohort studies.

What was found

Compared with never smokers, the pooled relative risk (RR) of Parkinson's disease was 0.59 (95% CI, 0.54-0.63) for ever smokers, 0.80 (95% CI, 0.69-0.93) for past smokers, and 0.39 (95% CI, 0.32-0.47) for current smokers. Each additional 10 pack-years conferred an RR of 0.84 (95% CI, 0.81-0.88) in case-control studies and 0.78 (95% CI, 0.73-0.84) in cohort studies. For coffee, drinkers had an RR of 0.69 (95% CI, 0.59-0.80) compared with non-drinkers. Each additional three cups per day yielded an RR of 0.75 (95% CI, 0.64-0.86) in case-control studies and 0.68 (95% CI, 0.46-1.00) in cohort studies.

Why it matters

This meta-analysis quantitatively demonstrates a robust, dose-dependent inverse association between both smoking and coffee consumption and the risk of Parkinson's disease across observational studies.

Limits

The evidence relies entirely on observational studies and is heavily weighted toward retrospective case-control designs (44 of 48 smoking studies, 8 of 13 coffee studies), which are prone to recall and selection bias. The analysis cannot establish causality, exclude reverse causation, or rule out residual confounding, and total participant sample size was not reported in the abstract.

Cited by