Reliability of self-reported information by farmers on pesticide use

Objectives: Exposure assessment is a major challenge faced by studies that evaluate the association between pesticide exposure and adverse health outcomes. The objective of this study was to investigate the reliability of information that farmers self-report regarding their pesticide use. Methods: Twenty five items based upon existing questionnaires were designed to focus on pesticide exposure. In 2009, a self-administrated survey was conducted on two occasions four weeks apart among 205 farmers residing in Gyeonggi and Gangwon provinces. For a reliability measure, we calculated the percentage agreement, the kappa statistics and the intraclass correlation coefficient (ICC) between the two reports according to the characteristics of the subjects. Results: Agreement for ever-never use of any pesticide was 96.4% (kappa 0.61). For both 'years used' and 'age at the first use' of overall pesticides, high agreement was obtained (ICC: 0.88 and, 0.78, respectively), whereas those of 'days used' and 'hours used' were relatively low (ICC: 0.42 and, 0.66, respectively). The kappa value for the use of personal protective equipment ranged from 0.46 to 0.59, and hygiene activities came out at 0.19 to 0.37. The agreement for individual pesticide use ranged widely and there was relatively low agreement due to the low response rates. The reliability scores did not significantly vary according to gender, age, the education level, the types of crop or the years of farming. Conclusions: Our results support mat carefully designed, self-reported information on ever-never pesticide use among farmers is reliable. However, the reliability of data on individual pesticide exposure may be unstable due to low response rates and needs to be refined.