Comparison of questionnaire exposure data to land cover map from geographical information system to assess passive exposure to pesticides: a methodological study

Background: Exposure assessment based on questionnaires is frequently implemented in case-control studies, but possible information and recall bias could lead to misclassification of exposure. Methods: We evaluated passive exposure to pesticides as possible environmental risk factors for amyotrophic lateral scle-rosis (ALS) using a questionnaire mailed to participants in a case-control study in Emilia Romagna and Sicily. Results from questionnaire assessment were com-pared with a remote sensing methodology based on geographical information system, i.e. the land use within a circular 100-meter area around subjects' residence. Since land cover maps were made available only about once every ten years, we used the 2003 and 2009 maps for Emilia-Romagna and Sicily, respectively. Thus, we estimated the percent-age of 'recent' total crop density close to each participant's home, setting positive exposure above 10% of land use. Finally, we calculated the agreement between the two different methodologies using Cohen‟s kappa coefficients for all subjects, cases and controls. Results and Conclusions: Cohen's kappa was 0.364 (95% CI 0.158-0.569) in total population, 0.378 (0.056-0.700) in cases and 0.354 (0.090-0.618) in controls using the most recent land use map available close to year of case diagnosis. Although a moderate-to-low agreement could be seen between two exposure methods, similar results were found in both cases and controls, suggesting that no recall bias occurred in the most recent period. In the future, we plan to compare such agreement using historical residence over the 20-30 years prior to diagnosis, in order to validate the long-term exposure to pesticides in subjects.