Skip navigation
Link to Back issues listing | Back Issue Listing with content Index | Subject Index

Smoking, coffee, and Parkinson's Disease

Systematic review
Cigarette smoking
Coffee drinking

Healthy living messages are usually relatively simple, and involve not smoking, eating fruit and vegetables, taking exercise and perhaps the odd glass of wine. That message does for heart disease, and cancer, and bone density and a raft of other things. It all gets more difficult when some of those lifestyle elements thought not to be good for us are actually shown to have some benefits. An example is the association between smoking and coffee drinking and the risk of Parkinson's disease [1].

Systematic review

An extensive search of several databases sought to identify studies associating smoking, coffee and Parkinson's disease. For inclusion studies had to have a case-control or cohort design, present original data, have Parkinson's disease diagnosed by a physician as the outcome, and attempt to ascertain exposure before the diagnosis.

Cigarette smoking

The 44 case-control studies involved 6,814 cases and 11,791 controls. Controls were often friends or relatives, patients with other diseases, or community controls, or a combination of these. The four cohort studies involved 409 cases in a total cohort size of just under 190,000.

Compared with never having smoked, cigarette smoking reduced the risk of developing Parkinson's disease (Table 1). There was a greater reduction for current smokers than for all smokers or past smokers. The magnitude of the risk reduction was similar for case-control and cohort studies. Each additional 10 pack years smoked was associated with a risk reduction of about 15%.

Table 1: Summary of results on smoking and coffee drinking and risk of Parkinson's disease


Type of study

Number of studies

Relative risk
(95% CI)

Ever smoked All studies 45 0.58 (0.54 to 0.63)
Case-control 41 0.59 (0.55 to 0.65)
Cohort 4 0.52 (0.42 to 0.64)
Past smokers All studies 16 0.80 (0.69 to 0.93)
Current smokers All studies 18 0.39 (0.32 to 0.47)
Coffee drinker All studies 12 0.69 (0.59 to 0.80)
Case-control 8 0.66 (0.52 to 0.83)
  Cohort 4 0.70 (0.56 to 0.88)
Relative risk is by random effects model

Coffee drinking

The eight case-control studies involved 1,440 cases and 4,016 controls. The four cohort studies involved 321 cases in a total cohort size of just under 190,000.

Compared with people who did not drink coffee, drinking coffee reduced the risk of developing Parkinson's disease (Table 1). The magnitude of the reduction was the same in case-control and cohort studies, and studies that adjusted for smoking. The evidence concerning the effect of the amount of coffee was mixed, though the authors estimated a risk reduction of 10% for each additional cup of coffee per day.


The results of this meta-analysis suggest that current smokers have a 60% lower risk of Parkinson's disease and coffee drinkers a 30% lower risk. The results were consistent across study design and geographical setting. Smokers who drink a lot of coffee will be given succour by these findings, but not for long.

First, Parkinson's disease occurs in only 1 in 200 of the elderly population, and 1 in 1,000 of the adult population. With cigarette smoking, the balance of risk is still negative, taking increased risks of cancer, and heart and respiratory disease into account.

Second, the association found in this meta-analysis does not prove that there is a cause and effect. There may be, but other possibilities exist and there is an interesting discussion about them in this paper. For instance, the diagnosis of Parkinson's disease could be more frequently omitted from death certificates and medical records of smokers (information bias). Another explanation may be that there is an increased mortality of younger smokers from causes other than Parkinson's disease (selection bias). Or again, smokers and sufferers of Parkinson's disease may share common genetic or environmental causes of which we are presently unaware (confounding). Perhaps patients with subclinical Parkinson's disease are less likely to start smoking or more likely to stop.

The authors of the paper argue persuasively against these possibilities, and suggest that smoking might protect against Parkinson's disease. We need to watch this space.


  1. MA Hernán et al. A meta-analysis of coffee drinking, cigarette smoking, and the risk of Parkinson's disease. Annals of Neurology 2002 52: 276-284.
previous or next story in this issue