9 Risk of amyotrophic lateral sclerosis and passive residential exposure to pesticides: comparison of questionnaire-based with gis-based exposure assessment methods

Abstract

Background/aim Amyotrophic lateral sclerosis (ALS) is a progressive neurodegenerative disease with still unknown aetiology. Among environmental factors, pesticides have been investigated due to their potential neurotoxic effects. Within a population-based case-control study conducted in two Italian regions, we aimed to investigate ALS risk due to passive residential exposure to pesticides using two methodologies. Methods The exposure assessment was carried with an individual questionnaire, which collected information of the entire residential history of subjects, focusing on rural residence or in the vicinity of agricultural areas. It was compared with assessment based on geographical information system (GIS), avoiding direct contact with study subjects. To do that, we computed the percentage (≥50%) of rural land use within the 100 m round buffer around each subjects’ residence, according to cover maps of two periods available from the Department of Agriculture, recent (2003–2009) and historical (1978–1989) ones. Risk for passive residential exposure to pesticides was computed using a sex and age adjusted logistic regression model for both methods, and their agreement was assessed using Cohen’s kappa (k). Results The odds ratio (OR) with their 95% confidence intervals (CI) for passive residential exposure to pesticides was 1.67 (95% CI 0.87 to 3.20) from the questionnaire-based assessment, while ORs from the GIS-based assessment were 1.05 (0.40 to 2.73) and 1.13 (0.49 to 2.63) for the recent and historical period, respectively. The agreement between two methods considering all participants was generally moderate to high, with k of 0.564 (95% CI: 0.361 to 0.767) and 0.648 (0.494–0.802) for recent and historical periods, respectively. Analyses divided between cases and controls yielded similar results, with k of 0.468 (0.133–0.803) in cases and 0.630 (0.382–0.879) in controls for recent period, and 0.642 (0.380–0.904) in cases and 0.652 (0.464–0.840) in controls for historical one. Conclusion Our results showed a slight increased risk of passive exposure to pesticides using the questionnaire-based assessment, with less conclusive results from the GIS-based one. The similar agreement either between periods and case/control status, suggested also that no substantial information bias and differential exposure misclassification occurred when assessing pesticides exposure in our population.