In this relatively large, population-based, case–control study of incident PD from western Washington State, our findings do not provide strong support for the hypothesis that workplace factors affect the risk of PD. The fact that our results replicate the commonly observed inverse association between smoking and PD provides some reassurance of internal validity [Checkoway et al., 2002
; Ritz et al., 2007
]. Although some of our findings are consistent with previous reports, we found no statistically significant associations, so risk estimates must be interpreted cautiously.
Our results () are consistent with a previous report which examined major divisions of the DOT and identified decreased risk from “service” work [Kirkey et al., 2001
]. The explanation for this observation remains uncertain, as classifications based on occupational titles capture a wide range of work activities and potential exposures and there is no apparent biological mechanism to link service work and PD.
Agricultural exposures are among the workplace factors most consistently reported to be associated with PD [Hertzman et al., 1990
; Seidler et al., 1996
; Gorell et al., 1999
; Tuchsen and Jensen, 2000
; Baldi et al., 2003
; Park et al., 2005
; Ascherio et al., 2006
; Kamel et al., 2007
]. Although several studies have indicated pesticides may be the causal factor, specific agents have not been consistently identified [Priyadarshi et al., 2000
]. Specific pesticides of a priori interest, based on previously reported associations, include organophosphate insecticides such as parathion, malathion, and diazinon [Bhatt et al., 1999
; Arima et al., 2003
; Firestone et al., 2005
], whose mechanism of action specifically targets the nervous system; organochlorines such as DDT [Seidler et al., 1996
], whose lipid solubility and biopersistence intensify and extend exposures to the nervous system; herbicides such as paraquat [Hertzman et al., 1990
], whose chemical structure mimics that of the parkinsonian neurotoxin MPTP; and 2,4-D [Kamel et al., 2007
], whose reputation is notorious as one of the primary constituents in Agent Orange. Our results () do not provide strong support for the hypothesis that exposure to any of these pesticides affect the risk of PD. Although the data do suggest increased risk for parathion, a potent organophosphate that is no longer used in United States, that risk estimate is not significant.
There has been considerable interest in the possible association between PD and occupational exposures to neurotoxic metals in various metal working activities. In particular, exposure to manganese-oxide dust in mining and ore processing has been linked with parkinsonism [McMillan, 1999
]. Although it is likely that manganese-related parkinsonism is clinically distinct from PD [Olanow, 2004
; Jankovic, 2005
; Perl and Olanow, 2007
], it has been suggested that PD may be associated with exposure to manganese in welding fumes [Racette et al., 2005
]. Our results () showing reduced risk in metal workers are consistent with several other studies showing no significant risk of PD from metalworking occupations including welding [Frigerio et al., 2005
; Fryzek et al., 2005
; Goldman et al., 2005
; Park et al., 2005
; Fored et al., 2006
Some metal exposures have been previously associated with PD [Gorell et al., 1999
; Coon et al., 2006
]. However, our results () did not corroborate these findings, as we found no increased risk for lead or manganese exposure or for combined lead and copper exposures, even after 20 years. Though these differences may relate to different study designs, it is likely that the small number of exposed subjects in each of these studies compromises the stability of statistical results.
Solvent exposures have also been associated with PD [Pezzoli et al., 2000
; McDonnell et al., 2003
]. Although our results () suggest increased risk for solvent exposures in women, this was not statistically significant. The small number of subjects in these exposure categories and lack of statistical significance prevents firm conclusions.
Our results () suggest increased risk from work in teaching and health care, as previously reported from studies in Korea [Park et al., 2005
] and Canada [Tsui et al., 1999
]. Though these observations may reflect ascertainment bias resulting from better access to care, they would also be consistent with the neuroinflammatory hypothesis of PD pathogenesis [Barcia et al., 2003
; McGeer and McGeer, 2004
; Whitton, 2007
], as workers in these occupations have relatively frequent exposures to infectious agents. However, the lack of association with work as a nurse argues against this hypothesis.
The difference in observed risk between some occupational titles fits with the idea of career self-selection based on early determinants of susceptibility with deterioration of dopaminergic pathways producing subtle, sub-clinical effects long before diagnosis. Thus, for example, we identified decreased risk of PD from jobs with high physical demands, such as “construction and electrical work,” and increased risk from jobs with low physical demands, such as “administration, computing, and sales” ().
That the risk estimates for men and women were different for several workplace factors fits with the idea that patterns of work exposures differ between men and women. Thus, for example, risk estimates for men tended to be lower than for women working in farming and related occupations ().
The case–control study design is efficient for studying relatively uncommon diseases such as PD. However, when exposure strata contain few subjects even slight misclassification of diagnosis or exposure may substantially alter observed associations. Misclassification of diagnosis may occur since PD diagnosis relies on clinical criteria whose specificity only approaches 80% [Hughes et al., 1992
; Tolosa et al., 2006
]. Misclassification of exposure may occur since retrospective exposure assessment does not allow confirmatory environmental monitoring, even though recall bias was limited by blinding the subjects to specific hypotheses. Because diagnostic uncertainty and inaccurate recall are likely to similarly affect cases and controls, the resulting non-differential misclassification may mask true associations. This study’s power to detect significant differences is further limited by the small number of subjects in some exposure groups, as this was not an industry-based but rather a population-based study. Fortunately this means that the findings can be reasonably generalized, as the demographic characteristics of our study population are representative of GHC enrollees, who in turn reflect the demographics of western Washington State ().
The growing scientific consensus is that PD is not a single disorder, but instead reflects a common pathological endpoint resulting from the interaction of various environmental and genetic risk factors. Thus, the risk attributable to any single factor is likely to be small and to differ depending on specific population characteristics. An epidemiological approach to identify etiologic factors with small attributable risk for PD, such as from workplace factors, will require very large study populations. Conclusive studies will likely require inter-disciplinary, multi-center collaborations.