Although our finding of decreased odds ratios among the most frequent welders might be a chance finding related to the low number of exposed which resulted in wide confidence intervals, the finding is consistent with a recent meta-analysis showing a summary risk ratio of 0.86 (95% CI 0.80e0.92) for PD in welders. Explanations brought forward for the risk estimates below unity include subtle, sub-clinical effects that appear long before diagnosis and that might lead to a self-selection of affected persons into specific jobs [16]. The time interval between onset of motor symptoms and diagnosis is around one to two years [19,20]. Nonmotor preclinical manifestations (e.g. constipation) of PD may occur much earlier, where time frames between a few years and more than two decades prior to diagnosis have been described [21].
However, because onset of PD is late in life, this means that in order to fully explain the reduced risk estimates, such non-motor manifestations would have to appear much earlier than previously reported and to have a relatively strong impact on career-choices, which appears unlikely. Also, it has been hypothesized that a premorbid parkinsonian personality predispose cases to select for white collar jobs associated with low exposures, such as teaching or