Aggregate exposure assessment using cosmetic co-use scenarios: I. Establishment of aggregate exposure scenarios

The "co-use" patterns of cosmetics would be critical for accurate aggregate exposure assessment. The Korean national representative exposure factor database, which includes simultaneous usage patterns of 31 cosmetics by 1001 subjects, was used to analyze the co-use patterns by Koreans. Three analytical methods were applied to determine the co-use patterns. Cohen's kappa coefficient was used to analyze the correlation between pairs. This method revealed an effect of gender on the cosmetics co-use pattern. Hierarchical clustering analysis was performed using the binary linkage distance method. The clusters were divided into one large cluster and small clusters of one or two cosmetics. Frequent pattern mining was performed using the eclat algorithm. The number of cosmetics used and co-use pattern were influenced by gender and age of the population. The co-use patterns exhibited an additive property in that new cosmetics were added to previous cosmetic combinations. A co-use scenario was proposed using the rank of occurrence frequency in co-use patterns and percentile values of the number of cosmetics used. The 16 co-use scenarios represented to the 25th, 50th, 75th, and 95th percentiles of the co-use patterns for four gender-age groups. These could be applied to aggregate exposure assessment as exposure scenarios.