Evaluation of individual-based and group-based exposure estimation of microbial agents in health effects associated...

We evaluated attenuation in linear associations between microbial exposure and respiratory symptoms occurring when individual measurements of microbial agents were used for estimating employees' exposure compared with group means. Symptoms, which improved when away from the building (building-related, BR), and measurements of culturable fungi, ergosterol, and endotoxin in floor dust were obtained between 2001 and 2007 from four cross-sectional studies on occupants of a water-damaged building. We compared odds ratios from longitudinal health effect models using individual measurements at employees' workstations with those using floor (group) means. Estimated odds for BR respiratory symptoms in group-based analyses increased by 2 to 5 times compared with those from individual-based analyses for culturable fungi and ergosterol, although they were less precise. For endotoxin, we found substantially increased and significant odds in group-based analyses, while we found no associations in individual-based analyses for various symptoms. Our study suggested that the building floor was useful in constructing exposure groups for microbial agents in this water-damaged building for epidemiologic analysis. Our study showed that group-average exposure estimation provides less attenuated associations between exposures to microbial agents and health in damp indoor environments where measurement error and intrinsic temporal variability are often large.