Occupational exposures and risk of aplastic anemia in Thailand

In developing countries exposure concentration data are seldom available. Self-reported information on occupational exposure obtained from case-control studies are prone to differential misclassification. The current study was designed to explore a semi-quantitative exposure assessment strategy to estimate the risk of aplastic anemia (AA) associated with occupational exposures. We evaluated the reliability of experts' assessment on benzene, 6 different pesticides including organophosphate, carbamate, organochlorine, paraquat, pyrethrins, and fungicide, glycol ether, cadmium and mercury. We also compared pesticide exposure obtained by self-report and experts' assessment, and evaluated patterns of discrepancy between the two methods. We also developed a standardized instrument based upon exposure probability, frequency, intensity, and confidence rating. For each of 476 job titles, we used expert judgments to assign levels of exposure. Individual information for subjects was based on a population-based case-control study of AA in Thailand, which was comprised of 541 cases of AA and 2261 controls.; The experts' assessment showed fair to good inter-rater reliability for glycol ether, carbamate, fungicide, paraquat, and organophosphate, but poor for organochlorine, pyrethrins, benzene, cadmium and mercury. The experts' agreement on pesticide exposures was high in agriculture as well as benzene in art industry. There was little or no improvement on reliability of expert when the analysis was restricted to high confident rating. Low agreement between self-reported and experts' assessment was observed primarily due to large different prevalence of exposure obtained by the two methods of data collection. There was tendency that cases, men, younger, and high income workers were more likely to report the exposures.; There was an increased risk of AA associated with major pesticides but not with glycol ether. Pesticide exposures were not associated with risk of AA in non-agriculture industry. Adding self-reported exposure in the analysis elevated the risks of AA: OR for high/very high exposure to organophosphate, carbamate, organochlorine and paraquat were 3.2 (95% CI:1.87-5.46), 4.75 (95% CI:1.92-11.75), 6.04 (95% CI: 1.31-27.84), and 2.17 (95% CI: 1.11-4.25), respectively. However, the risks of AA diminished in multi pesticides model indicate the uncertainty of which pesticide play a role. In summary, the present study supports the evidence of risk of AA associated with pesticides exposure. Improvement of the experts approach may increase the reliability of the method.