65-year follow-up of British Doctors Study shows association, but findings still questioned

The risk of death from Parkinson’s disease (PD) was inversely associated with the amount of tobacco smoked, a study of British male physicians between 1951 and 2016 suggested.

In an analysis limited to baseline smoking habits — cigarette, pipe, or cigar — men who were current smokers in 1951 had a 30% lower risk of PD mortality (RR 0.71, 95% CI 0.60-0.84) compared with never-smokers, reported Robert Clarke, MD, of the University of Oxford, in England, and coauthors in Neurology.

For men who quit smoking, the protective effect was attenuated with increasing time since quitting. A model updated with changed smoking habits over time showed a 40% lower risk of PD mortality compared with never smokers (RR 0.60; 95% CI 0.46-0.77).

An inverse dose-response relationship also was seen between smoking and PD mortality risk that was significant in both baseline (P=0.0006) and updated (P=0.002) models, they added.

“This study of 30,000 male British doctors, which minimized the risk of reverse causality by having very prolonged followup, demonstrated an inverse association of current tobacco smoking with the risk of PD,” wrote Clarke and coauthors.

“In contrast to previous suggestions, the present report demonstrates a causally protective effect of current smoking on the risk of PD, which may provide insights into the etiology of PD,” they added.

While the study demonstrated that tobacco smoking, smoking amount, and time since cessation all were associated with PD risk, the results “do not provide a mechanism by which this effect occurs, so it is premature to assume that there is a direct causative relationship between tobacco use and a reduced risk of PD,” wrote Yuan Cheng, MD, and Yan-Jiang Wang, MD, PhD, both of the Third Military Medical University, in Chongqing, China, in an accompanying editorial.

“For example, there could be a common genetic predisposition to smoking and to PD,” Cheng and Wang noted.

Risks are higher in smokers compared to nonsmokers for most noncommunicable diseases, including ischemic heart and brain disease, diabetes, chronic lung disease, cirrhosis, and multiple malignancies, but Parkinson’s disease has been reported to be inversely associated with smoking.

Other reported inverse associations with PD include caffeine, serum urate, physical activity, and ibuprofen, while positive associations include head injury, pesticide exposure, and consumption of dairy products.

Much research about smoking and PD risk has been case-control studies with smoking data collected after PD onset, making reverse causality a concern. One large, recent study suggested that the inverse risk seen in studies was a reverse causality artifact due to decreased prodromal responsiveness to nicotine that prompted people to stop smoking before PD was diagnosed.

In their study, Clarke and colleagues analyzed data of 29,737 male physicians who participated in the British Doctors Study. The study started in 1951; of those contacted initially, about 59% provided full information about their smoking habits. Men with a pre-existing diagnosis of PD were excluded. Data were updated six times by survey between 1958 and 1998 with response rates of 94% to 98%, and cause-specific mortality was followed until December 2016. Doctors who withdrew before the end of the study or were lost to follow-up were included until absent, then censored. The first 10 years of data were excluded.

“To minimize reverse causality bias, the authors excluded participants with prior disease at enrollment and ruled out subjects with preclinical PD by excluding the first 10 years of follow-up data for all analyses, based on the idea that disease pathogenesis starts 2 decades before the first motor manifestation of PD,” Cheng and Wang observed.

Average followup was 35 years, and cause-specific mortality included 283 deaths (1.1%) recorded as due to PD. Those 283 men died after average followup of 42 years at a mean age of 82. Men whose deaths were ascribed to other causes had average followup of 35 years and died at a mean age of 77.

The relative risks of PD (and 95% confidence intervals) were estimated using Cox models for smoking habits (smoking status, amount smoked, and years since quitting) at baseline or updated habits at resurvey.

The prevalence of current smoking declined progressively for all ages and with each survey between 1951 and 1998; for example, in those 65 to 69 years old, current smoking declined from 67% to 8% over that interval.

Time since quitting, dichotomized as 10 or more years versus 0 to 9 years, showed those quitting 10 or more years previously had a 14% lower PD risk, compared with a 29% lower risk in those who quite more recently.

Limitations include analyses based on a relatively small number of PD cases, all of whom were male British doctors, which limits generalizability. Data about family history and other risk factors for PD were not included in the study. Concomitant smoking-related disease and unrecognized cases of secondary parkinsonism may have affected the results, including the cause of death data.

The PD incidence also was likely underestimated in this study, the editorialists noted. “Strictly speaking, the present study demonstrates an association of tobacco’s protective effect on the risk of death from PD instead of the risk of PD incidence,” they wrote.

  1. In an analysis limited to baseline smoking habits — cigarette, pipe, or cigar — men who were current smokers in 1951 had a 30% lower risk of Parkinson’s disease (PD) mortality compared with never-smokers; for those who quit smoking, the effect was attenuated with increasing time since quitting.

  2. Be aware that the mechanism of the proposed effect remains unclear, and “it is premature to assume that there is a direct causative relationship between tobacco use and a reduced risk of PD,” the editorialists noted.

Paul Smyth, MD, Contributing Writer, BreakingMED™

The Medical Research Council has supported the study since 1951 and continues to do so by supporting the Clinical Trial Service Unit and Epidemiological Studies Unit, University of Oxford, with the British Heart Foundation and Cancer Research UK.

The researchers reported no disclosures.

The editorialists reported no disclosures.

Cat ID: 37

Topic ID: 82,37,730,37,143,489,925