The negative association between smoking and the risk of Parkinson's disease (PD) has been a controversial issue since it was first reported by Dorn in 1959. 1 Recent meta-analysis of prospective studies showed that smoking protects against PD with pooled relative risk of 0.51 (95% confidence interval 0.43-0.61). Pooling prospective studies that investigated that association was not conclusive in regard to women because localized cohort studies, except the Framingham and Rotterdam studies, included or reported relative risks only for men. The results of the Framingham study reported significant inverse association between smoking and PD risk in men but not in women. Meanwhile, the Rotterdam study showed unclear protective effect of smoking against PD with relative risk of 0.39% and 95% confidence interval overlapping unity (0.14 -1.08). 2 Several investigators have tried to find explanations for the unclear association between PD and smoking in women, suggesting that estrogens could be an independent confounder between PD and smoking. 2,3 Gonadal hormones, estrogens, and testosterone, are correlated with PD. It has been proven that estrogen has an essential role in maintaining the integrity of the nigral dopamine system involved in muscle control and higher brain functions. 3 However, the associations between novelty-seeking behavior, smoking, and estrogens in PD are mere speculations.
Because the meta-analysis of prospective studies did not clarify the controversial negative association between smoking and the risk of PD, we have reviewed all retrospective studies that evaluated that association in women. Fifteen retrospective studies were identified: 1 cross-sectional and 14 case-control studies. These studies were carried out between 1971 and 2000 (Table 1).
There was an obvious protective effect of smoking in the pooled odds ratio [odds ratio 0.60 (95% confidence interval 0.50 -0.72)]. 4 -18 Finally, The validity of pooling the odds ratios was tested using the test of homogeneity ( 2 test). A violation of this test implies that the studies being pooled differ from one another. 19 The homogeneity test of our meta-analysis was not significant (P ϭ 0.09), denoting homogeneity across the pooled studies.
This quantitative review demonstrates a negative relationship between premorbid smoking habits and PD in women, as in men.