Recommendation of suitable sample size for exposure factors in deterministic exposure assessment of consumer products

Abstract

Introduction: The exposure of consumer products (CPs) can be estimated using deterministic and probabilistic approaches. Although deterministic approaches are simple to apply, there res no guidelines for selecting representative exposure factors. The purposes of this study were to determine the effect of sample size on estimates of exposure factors using a large database and recommend a suitable sample size for obtaining an acceptable margin of error. In addition, the exposure levels were calculated by the 75th percentile of the exposure factors and compared with exposure level of the parent population. Methods: This study used an existing database of 3,333 participants for the three exposure factors (frequency of use, amount of use, and duration of use) of 13 CPs, including cosmetics, cleaning products, and disinfectant. To examine how sample size affects the exposure factors by different sample sizes, the exposure factors were randomly resampled by ten different small samples. The three exposure factors of the parent population were compared with the corresponding values obtained from the smaller sample sizes. The inhalation and dermal exposure levels based on a small sample were calculated by the 75th and 95th percentiles of the exposure factors. The exposure levels by the 75th and 95th percentiles of the exposure factors were compared with the exposure distributions in the parent population. Results: As sample sizes increased, the exposure factors became more similar to those of the parent population for the most CPs. For 3 of the 13 CPs, the 75th percentiles of the exposure factors determined from sample sizes smaller than 50 were less than 10% different from those determined from the parent population. For 7 of the 13 CPs, the 75th percentiles of the exposure factors determined from sample sizes smaller than 100 were less than 10% different from those determined from the parent population. In addition, for the sample size of 100, the medians of inhalation and dermal exposure levels based on the 75th percentiles of the exposure factors were lower than the 95th percentiles of the exposure levels of the parent population. Conclusion: This study quantified the potential errors of exposure factors for 13 CPs if sample sizes for exposure factor was smaller. Using these results, the error of exposure factors derived from small sample sizes might be estimated quantitatively. . The inhalation and dermal exposure levels estimated by the 75th percentiles of the exposure factors in smaller samples were less than the 95th percentile exposure levels of the population for most CPs. Further studies are needed to determine the appropriate percentile for estimating exposure levels in high exposure groups.