Hertzman and colleagues (1994) studied 127 Canadian PD patients identified by their physicians and compared them with cardiac-disease controls and with controls selected from electoral lists. The study made the most aggressive attempt to assess exposure to specific agents. It was conducted in a region with a high prevalence of orchard chemicals and an agricultural office, which provided records of specific pesticides and dates of marketing. From that information, the investigators created cue cards with full descriptions of 79 agricultural chemicals to aid subjects’ memories. Analysis was performed for pesticides in general, insecticides in general, and specific insecticide classes. The study found a positive association between pesticide exposure and PD. But the opposite—a protective effect—was found specifically for insecticide exposure: the confidence intervals for PD were entirely below 1 for both men and women with past insecticide exposure compared with electoral-list controls. No positive associations were found for exposures to OPs, organochlorines, or carbamates. The study was limited by differential ascertainment between cases and controls. One control group was restricted to voters, but cases were not restricted in this manner. That might be important if cases included migrant workers or other noncitizens working in orchards; such inclusions would tend to increase the likelihood of exposure in the cases. The finding of a protective effect of insecticides is inconsistent with findings from other studies discussed in this part of the chapter.
Seidler and colleagues (1996) studied 380 PD patients from nine neurology clinics in Germany. The authors did not find positive associations between insecticide exposure and PD when they used neighborhood controls, but they did when they used regional controls (although not at the highest dose). They also found a positive association with exposure to organochlorine and alkylated phosphates or carbamates. The study’s advantages were large size, good diagnostic criteria, blinding of interviewers to the study hypothesis, and categorization of pesticide exposures by a toxicologist. Its limitations were recruitment of cases from clinics with a “special interest in PD,” which might result in inclusion of unusual cases. Regional (but not neighborhood) controls were recruited by random selection from electoral rolls. If there were demographic differences between cases and controls, they may be related to exposure, thereby increasing the likelihood of observing a positive association. Another problem was that cases were asked about exposure that occurred at any time before diagnosis, but controls were asked about exposure only one year prior to interview. Thus, cases were asked to remember exposure over a longer period, which might have attenuated effects. Another limitation was that the results from analyzing exposure with a job-exposure matrix were not reported.
Semchuk and colleagues (1992) studied Canadian patients from a population-based registry and community controls. They found a positive association with insecticides, but the association disappeared after adjustment for occupational herbicides. The study had the advantages of being population-based, of excluding people with dementia, and of restricting analysis to exposures with long latency. One limitation was having the primary author review occupational history, which introduced the possibility of bias. Another was the long disease duration in cases (mean duration from onset to evaluation, 7.8 years), which introduced the problems associated with patients’ recall of predisease risk factors and the potential for survival bias.
Stern and colleagues (1991) studied patients with early-onset and late-onset PD. They found no positive association with insecticide use in the home, yard, or garden, but