Response accuracy of hazard identification in migrant farm workers

This study assessed potential biases introduced by using maternal proxies for reporting work hazards, illness symptoms, and acute agricultural injuries in studies of migrant farmworker families. A convenience sample of 79 mother/spouse and mother/oldest child pairs was obtained from a two-year cohort study of migrant farmworker families from Starr County, Texas. Pairs completed an interviewer-administered survey including 27 close-ended items describing work history, illness symptoms, and acute injuries during the 2001 migration season. Data analysis included prevalence, sensitivity, and specificity measures. Maternal proxy reports of subjective symptoms of a spouse or child were approximately 50% lower than spouse or child self-reports. Sensitivity and specificity of mothers' responses regarding hazard items were higher for spouses than for children. For items measuring illness symptoms, sensitivity of mothers' responses for spouses and oldest child pairs was generally less than 30%; however, specificity was generally above 90%. For acute agricultural injury, the mother/spouse sensitivity was 75%, the mother/child 40%, and specificities were 100% and 97%, respectively. Sensitivity and specificity for mother/ spouse pairs was generally higher than mother/child pairs for work hazards, illness symptoms and acute injuries. While mothers may be acceptable proxies for spouses in this population, efforts should be made to collect data directly from adolescent children in studies of agricultural injuries.