ABSTRACT
Objective: To investigate the risk of Parkinson disease (PD) associated with exposure to pesticides and solvents using meta-analyses of data from cohort and case-control studies.
Methods: Prospective cohort and case-control studies providing risk and precision estimates relating PD to exposure to pesticides or solvents or to proxies of exposure were considered eligible. The heterogeneity in risk estimates associated with objective study quality was also investigated.
Results: A total of 104 studies/3,087 citations fulfilled inclusion criteria for meta-analysis. In prospective studies, study quality was not a source of heterogeneity. PD was associated with farming and the association with pesticides was highly significant in the studies in which PD diagnosis was self-reported. In case-control studies, study quality appeared to be a source of heterogeneity in risk estimates for some exposures. Higher study quality was frequently associated with a reduction in heterogeneity. In high-quality case-control studies, PD risk was increased by exposure to any-type pesticides, herbicides, and solvents. Exposure to paraquat or maneb/mancozeb was associated with about a 2-fold increase in risk. In high-quality case-control studies including an appreciable number of cases (>200), heterogeneity remained significantly high (>40%) only for insecticides, organochlorines, organophosphates, and farming; also, the risk associated with rural living was found to be significant.
Conclusions: The literature supports the hypothesis that exposure to pesticides or solvents is a risk factor for PD. Further prospective and high-quality case-control studies are required to substantiate a cause-effect relationship. The studies should also focus on specific chemical agents.
DISCUSSION
Exposure to pesticides and solvents appears to be a risk factor for PD. Our evidence also supports the involvement of specific compounds, such as paraquat, maneb/mancozeb family, as well as proxies of exposure. However, it could be argued that the evidence collected is still limited, or at least inconclusive, because there was no definitive agreement between cohort and case-control studies.
Indeed, most of the evidence found relied on data from case-control studies. To investigate an etiologic relationship, the use of cohort studies is preferable. However, the incidence of PD is low and usually occurs in the elderly; large populations, a large number of cases, and a long follow-up are required to achieve adequate statistical power. Accordingly, most neuroepidemiologists resort to case-control studies, which are practical and, despite their retrospective nature, have the advantage of more detailed exposure assessment.
We have also partly explained the sources of heterogeneity in individual study results. In prospective studies, differences in estimates of exposure to pesticides appeared to depend on the method of ascertainment of PD. This factor is less likely to have been a source of bias in case-control studies because, although different sets of well-accepted diagnostic criteria were used, in most cases secondary causes of PD were excluded in patients recruited at movement disorders clinics. However, we did not assess the effect of this feature and we recognize that this is a possible limitation of our study. Heterogeneity in case-control studies appeared to be due mainly to study quality and size. With respect to study quality, our results are consistent with previous suggestions.e86 However, the issue of sample size analysis was addressed by only a few authors.e25,e60,e66,e80 To detect an OR of 2 and an exposure frequency of 20% we calculated that at least 200 case-control pairs would be needed.
A meta-analysis investigating several sources of heterogeneity in risk estimates has recently shown that study design (case-control vs prospective), source of controls (community vs non-community controls), type of exposure (occupational vs non-occupational/others type), adjustment for potential confounders, or geographical area do not appear to be important determinants. Accordingly, no consistent explanation of heterogeneity has been provided. The only factor that appeared to contribute was the method used for exposure assessment, as the use of job title–based exposure matrix resulted in a higher risk than assignment based on self-reported exposure.17 Unfortunately, this method could not be applied with sufficient accuracy to specific working occupations.
Despite our methodologic approach to quantitative synthesis, notable heterogeneity, probably affecting the evidence of an increased risk associated with exposure, was still present for insecticides, organochlorines, organophosphates, and farming. There are some possible explanations for this observation. With respect to farming, we observed that exposure was assessed either by open questions or by specific industry coding systems. Moreover, the choice of controls may introduce bias. Few studies have considered the effect of geography (area/region of residence; the additional criteria for “comparability” in our quality assessment process) in study design or adjustment of analyses. Regional controls may be preferable for the evaluation of direct exposure but both these and neighboring areas may affect the assessment of risk associated with this proxy measure of exposure. Insecticides are a heterogeneous class of compounds to which most organochlorines and organophosphates belong. Indeed, among organochlorines the most frequently used insecticide is DDT and the risk associated with this compound was found to be nonsignificant. In some cases, frequency of exposure is low and there may be difficulties in recalling specific product names. Given the advanced age of patients with PD, impairment of cognitive function is possible. Although poor cognition has been considered as an exclusion criterion during recruitment in some studies, only one research group adjusted for this covariate.e63,e68 Exposure to insecticides also appears to be closely correlated with exposure to herbicides.16
Finally, there may be residual confounders that we were not able to address. In some cases, data derived from proxy respondents were pooled with those reported by cases,29,e10,e44,e51,e79,e84 probably introducing misclassification bias.e6,e8
Confounding also could be secondary to the use of protective equipment and compliance with suggested, or even recommended, preventive practices. Only one study addressed this issue.26 Prevalent exposure was heterogeneous among the populations investigated and was likely to be higher in certain working categories. Inclusion bias should be taken into consideration, because in some studies selection of cases or controls was performed by linking to professional and insurance databases26,e7,e15,e38,e39,e45,e63,e68,e77,e79,e84 or in geographical areas characterized by extensive use of pesticides. A few studies investigated the role of genetic susceptibility. Although mechanisms of action at the molecular level are largely unknown, there is a growing body of evidence progressively substantiating the hypothesis of a gene–environment interaction.14 Finally, positive exposure was defined according to different levels (e.g., number of chemicals or times of usage over a period), types (particularly for toxin application), or durations.
The present study highlights unresolved issues with implications for health policies. From a preventive perspective, we observed that the route of exposure (e.g., inhaled or transcutaneous) and the method of toxin application (e.g., spraying or mixing) has never been investigated. Risk appears to increase as the duration of exposure increases. Since several compounds are likely to be used by the same people, different routes of exposure may act synergistically in increasing the risk. Unfortunately, it was not possible to investigate the issue of a dose–response relationship and to provide a cutoff for exposure.
The literature supports the hypothesis that exposure to pesticides or solvents is a risk factor for PD. However, further prospective and high-quality case-control studies are required to substantiate a cause-effect relationship. Although some compounds have been withdrawn from the market in industrialized countries, they are still in use in developing parts of the world. According to our review of the sources of funding, interest in this issue should come also from chemicals manufacturers. This should be emphasized because an interest in the adverse effects of specific compounds appears justified.
Source: http://www.neurology.org
44.276672
-85.874794