Lifestyle, Family History, and Risk of Parkinson Disease
Lifestyle, Family History, and Risk of Parkinson Disease
In the present study of 1,808 patients with verified IDP, we found a strong inverse association between cigarette smoking and risk of IPD, even when smoking habits before the first cardinal symptom were lagged 10 years. Although we saw a downward trend with increasing number of years of smoking, there appeared to be no additional contribution from tobacco consumption of more than 10 cigarettes per day. In a large case-control study of the associations of both the duration and intensity of smoking with the risk of PD, Chen et al. also found that duration was more important than intensity in modulating the smoking-PD relationship.
Although inconsistent results regarding the association between caffeine intake and risk of PD among women have been reported in previous studies, we found that caffeine intake was associated with a lower risk of IPD in both men and women. The previous studies might have been limited by their lower average coffee consumption; intakes are comparatively high in Denmark. Inconsistent results were also obtained in previous studies of the risk of PD in relation to alcohol consumption, but in a meta-analysis of 8 prospective studies and 24 case-control studies, Zhang et al. reported an overall pooled relative risk for PD of 0.75 (95% confidence interval: 0.66, 0.85) among those with the highest alcohol consumption when compared with those with the lowest level. Further analyses showed that the inverse association was mostly driven by beer rather than by wine or liquor. In a prospective cohort study, Hernán et al. also observed an inverse association between beer consumption and risk of PD, but only in men. In the present study, we found inverse associations with IPD for all types of alcohol in men. As in the case of smoking, we did not observe any strengthening of the inverse association with a consumption greater than a moderate intake, which again suggests that intensity is less important in mediating these lifestyle associations with IPD.
In our analysis of smoking in combination with caffeine consumption or alcohol consumption, we observed a stronger negative odds ratio for IPD in smokers with medium or high intakes of coffee or alcohol than for the individual lifestyle habits. To the best of our knowledge, the combined association of cigarette smoking and alcohol consumption with PD has not been reported before, although the joint associations of cigarette smoking and caffeine intake have been examined in a few studies. Whereas no clear interaction was observed in the study by Liu et al., Powers et al. reported findings similar to ours, that is, a lower risk for PD in subjects who used tobacco and had high intakes of caffeine than in subjects with only 1 of these habits.
Despite the abundance of literature inversely linking in particular tobacco smoking with the risk of PD, discussion continues about whether the decreased risk reflects a true association or a premorbid change in personality, with avoidance of addictive behavior, by people who later develop PD. Animal models of parkinsonism have demonstrated that both caffeine and nicotine can prevent neurotoxin-induced loss of neurons in the substantia nigra. Furthermore, because polyphenols in wine and ethanol have potent antioxidant effects and nicotine and alcohol share nicotinic acetylcholine receptors as their biological target, it is conceivable that the negative associations observed in the present and other studies represent causal inverse links between these lifestyle habits and IPD. Although we also observed inverse associations when lagging exposure 10 years from first cardinal symptom, reverse causality cannot be excluded if persons who later develop PD are less likely to engage in addictive behavior in the first place or if they quit much earlier.
Because there was a large number of IPD patients included in the present study, we were able to examine the joint associations of cigarette smoking and a family history of PD. The higher odds ratio associated with a family history of PD and the inverse association with tobacco smoking were essentially the same in analyses stratified by smoking habits and family history. Thus, although each single exposure was associated with a marked deviation in risk, when examining the variables in combination, we observed a more modest increase in the odds ratio of 1.6 comparing smokers with a family history of PD to never smokers without a family history. Studies of gene-environment interactions have shown that the inverse association between smoking and PD diminishes or is eliminated in people who carry risk alleles, indicating that the smoking-IPD relationship is less strong in those at high risk because of a genetic predisposition. Joint exposure to tobacco smoking and a family history of PD was previously examined in only 1 case-control study of limited size. Somewhat surprisingly, Elbaz et al. did not find an overall reduction in the risk of PD in smokers; however, they did find that people with a family history who had ever smoked had the highest risk of PD. After stratification by age, the finding was restricted to subjects older than 75 years of age. The latter finding was, however, based on only 6 PD patients and 1 control subject.
A major strength of our study is the systematic review of medical records undertaken to identify the subgroup of patients with IPD. Because an independent re-evaluation of a random sample of 50 medical records of patients initially considered to have IPD confirmed the diagnosis for 48 patients (96%), diagnostic misclassification was minimal. The present study was, however, limited by the fact that patients with less severe IPD were potentially excluded from our patient group because they were not registered with a diagnosis in the hospital registry.
Case-control studies are prone to recall bias because patients often search for explanations for their disease and are more likely to report exposures. In general, however, recall bias leads to overestimation of risks, because patients focus on potential risk factors for disease rather than protective aspects of lifestyle, which we examined in this study. It is therefore unlikely that the lower odds ratios associated with smoking, caffeine intake, and alcohol consumption observed in this study are a result of biased recall of exposures. Because our study population was elderly and we asked about habits over a lifetime, the reporting of exposures might have some inaccuracy. Nondifferential misclassification of exposure information might move odds ratio estimates toward the neutral value of 1, typically leading to underestimation of an increase or decrease. The information on PD among family members was also self-reported, and patients with PD have been found to report the occurrence of PD in first-degree relatives more often than control subjects. Nevertheless, the odds ratio for PD for subjects with a family history of PD was 2.8 in our study, which is compatible with the risk estimate of 2.9 reported in a recent meta-analysis.
The participation rate was high among the patients (81%) but moderate among controls (53%). Further examination of all eligible control subjects revealed that more nonparticipating controls died 1 year or less after contact for interview than those who participated (13.5% versus 5.9%). In addition, the nonparticipating controls were more often hospitalized for diseases related to tobacco smoking, such as ischemic heart disease or chronic obstructive pulmonary disease (19.3% versus 16.5%), indicating a slight underrepresentation of smokers among control subjects. Thus, although we cannot rule out the possibility of biased selection of study controls, the finding of a lower frequency of smoking-related disease among participating controls indicates that the observed decrease in risk for IPD associated with smoking in our study is slightly underestimated rather than exaggerated.
In conclusion, we found that duration of smoking, caffeine intake, and alcohol consumption seem to be more important than intensity of use as determinants of associations with IPD. When we combined information on smoking with that on caffeine and alcohol, the associations were stronger than among participants who reported just 1 of these exposures. Finally, compared with never smokers without a family history of PD, never smokers with a positive family history had the highest odds ratio for PD and smokers with a family history had a weaker association.
Discussion
In the present study of 1,808 patients with verified IDP, we found a strong inverse association between cigarette smoking and risk of IPD, even when smoking habits before the first cardinal symptom were lagged 10 years. Although we saw a downward trend with increasing number of years of smoking, there appeared to be no additional contribution from tobacco consumption of more than 10 cigarettes per day. In a large case-control study of the associations of both the duration and intensity of smoking with the risk of PD, Chen et al. also found that duration was more important than intensity in modulating the smoking-PD relationship.
Although inconsistent results regarding the association between caffeine intake and risk of PD among women have been reported in previous studies, we found that caffeine intake was associated with a lower risk of IPD in both men and women. The previous studies might have been limited by their lower average coffee consumption; intakes are comparatively high in Denmark. Inconsistent results were also obtained in previous studies of the risk of PD in relation to alcohol consumption, but in a meta-analysis of 8 prospective studies and 24 case-control studies, Zhang et al. reported an overall pooled relative risk for PD of 0.75 (95% confidence interval: 0.66, 0.85) among those with the highest alcohol consumption when compared with those with the lowest level. Further analyses showed that the inverse association was mostly driven by beer rather than by wine or liquor. In a prospective cohort study, Hernán et al. also observed an inverse association between beer consumption and risk of PD, but only in men. In the present study, we found inverse associations with IPD for all types of alcohol in men. As in the case of smoking, we did not observe any strengthening of the inverse association with a consumption greater than a moderate intake, which again suggests that intensity is less important in mediating these lifestyle associations with IPD.
In our analysis of smoking in combination with caffeine consumption or alcohol consumption, we observed a stronger negative odds ratio for IPD in smokers with medium or high intakes of coffee or alcohol than for the individual lifestyle habits. To the best of our knowledge, the combined association of cigarette smoking and alcohol consumption with PD has not been reported before, although the joint associations of cigarette smoking and caffeine intake have been examined in a few studies. Whereas no clear interaction was observed in the study by Liu et al., Powers et al. reported findings similar to ours, that is, a lower risk for PD in subjects who used tobacco and had high intakes of caffeine than in subjects with only 1 of these habits.
Despite the abundance of literature inversely linking in particular tobacco smoking with the risk of PD, discussion continues about whether the decreased risk reflects a true association or a premorbid change in personality, with avoidance of addictive behavior, by people who later develop PD. Animal models of parkinsonism have demonstrated that both caffeine and nicotine can prevent neurotoxin-induced loss of neurons in the substantia nigra. Furthermore, because polyphenols in wine and ethanol have potent antioxidant effects and nicotine and alcohol share nicotinic acetylcholine receptors as their biological target, it is conceivable that the negative associations observed in the present and other studies represent causal inverse links between these lifestyle habits and IPD. Although we also observed inverse associations when lagging exposure 10 years from first cardinal symptom, reverse causality cannot be excluded if persons who later develop PD are less likely to engage in addictive behavior in the first place or if they quit much earlier.
Because there was a large number of IPD patients included in the present study, we were able to examine the joint associations of cigarette smoking and a family history of PD. The higher odds ratio associated with a family history of PD and the inverse association with tobacco smoking were essentially the same in analyses stratified by smoking habits and family history. Thus, although each single exposure was associated with a marked deviation in risk, when examining the variables in combination, we observed a more modest increase in the odds ratio of 1.6 comparing smokers with a family history of PD to never smokers without a family history. Studies of gene-environment interactions have shown that the inverse association between smoking and PD diminishes or is eliminated in people who carry risk alleles, indicating that the smoking-IPD relationship is less strong in those at high risk because of a genetic predisposition. Joint exposure to tobacco smoking and a family history of PD was previously examined in only 1 case-control study of limited size. Somewhat surprisingly, Elbaz et al. did not find an overall reduction in the risk of PD in smokers; however, they did find that people with a family history who had ever smoked had the highest risk of PD. After stratification by age, the finding was restricted to subjects older than 75 years of age. The latter finding was, however, based on only 6 PD patients and 1 control subject.
A major strength of our study is the systematic review of medical records undertaken to identify the subgroup of patients with IPD. Because an independent re-evaluation of a random sample of 50 medical records of patients initially considered to have IPD confirmed the diagnosis for 48 patients (96%), diagnostic misclassification was minimal. The present study was, however, limited by the fact that patients with less severe IPD were potentially excluded from our patient group because they were not registered with a diagnosis in the hospital registry.
Case-control studies are prone to recall bias because patients often search for explanations for their disease and are more likely to report exposures. In general, however, recall bias leads to overestimation of risks, because patients focus on potential risk factors for disease rather than protective aspects of lifestyle, which we examined in this study. It is therefore unlikely that the lower odds ratios associated with smoking, caffeine intake, and alcohol consumption observed in this study are a result of biased recall of exposures. Because our study population was elderly and we asked about habits over a lifetime, the reporting of exposures might have some inaccuracy. Nondifferential misclassification of exposure information might move odds ratio estimates toward the neutral value of 1, typically leading to underestimation of an increase or decrease. The information on PD among family members was also self-reported, and patients with PD have been found to report the occurrence of PD in first-degree relatives more often than control subjects. Nevertheless, the odds ratio for PD for subjects with a family history of PD was 2.8 in our study, which is compatible with the risk estimate of 2.9 reported in a recent meta-analysis.
The participation rate was high among the patients (81%) but moderate among controls (53%). Further examination of all eligible control subjects revealed that more nonparticipating controls died 1 year or less after contact for interview than those who participated (13.5% versus 5.9%). In addition, the nonparticipating controls were more often hospitalized for diseases related to tobacco smoking, such as ischemic heart disease or chronic obstructive pulmonary disease (19.3% versus 16.5%), indicating a slight underrepresentation of smokers among control subjects. Thus, although we cannot rule out the possibility of biased selection of study controls, the finding of a lower frequency of smoking-related disease among participating controls indicates that the observed decrease in risk for IPD associated with smoking in our study is slightly underestimated rather than exaggerated.
In conclusion, we found that duration of smoking, caffeine intake, and alcohol consumption seem to be more important than intensity of use as determinants of associations with IPD. When we combined information on smoking with that on caffeine and alcohol, the associations were stronger than among participants who reported just 1 of these exposures. Finally, compared with never smokers without a family history of PD, never smokers with a positive family history had the highest odds ratio for PD and smokers with a family history had a weaker association.