147 Association between mixed exposures to cleaning chemicals and asthma outcomes

Introduction Cleaning products are complex mixtures of chemicals and their use in healthcare is associated with elevated prevalence/odds of asthma outcomes in models with single exposure variables. To better represent workplace mixed exposures, we explored the use of Bayesian Kernel Machine Regression (BKMR) combined with prior toxicological knowledge on asthmagenic potential to understand the combined effect of correlated mixtures. Methods Healthcare workers from nine occupations completed a questionnaire with modules on cleaning and disinfecting tasks and product use. Quantitative exposure to total and eight specific volatile organic compounds were assessed at five different hospitals and were assigned to participants based on predictive statistical models and responses on questionnaires. Exposures were classified as asthmagens or non-classified based on toxicological evidence. A five-level asthma outcome variable was created by hierarchical clustering of 27 respiratory symptom/care variables. Exposure-response relationships for asthma clusters were explored using BKMR for mixed exposures. All models were adjusted for age, sex, education, smoking and allergic status. Results Group probability of inclusion was always higher for asthmagens compared to non-classified for all asthma clusters. Within asthmagens, acetone and chloroform had the higher probabilities of inclusion for several asthma clusters. Higher levels of acetone had the highest risk for all health clusters when evaluating the effect of mixed asthmagens. Among non-classified, BTEX chemicals and methyl-methacrylate had the higher probabilities of inclusion for several asthma clusters. Conclusions Multipollutant models combined with toxicological knowledge is a more realistic approach to model workplace mixed exposures although the effect of correlated variables cannot be entirely eliminated.