Presence Probabilities
For this study, presence probabilities were derived from an online available database called Codecheck.info (Table 6), which lists the labelling information for a large number of products in a number of cosmetic categories, gathered from crowd-sourced data. While the representativeness and uncertainty associated with using such a database is open to debate, the goal in this assessment is to understand the impact of using chemical occurrence data in a probabilistic exposure assessment to account for instances when a substance is not present in a given product category a certain portion of the time.
To derive the likely presence of triclosan in the products used in this assessment, the total number of products in each category was counted, and the proportion of these that contain triclosan based on whether it was listed on the label or not was used to estimate the chemical occurrence. Note that this approach therefore assumes equal market shares; if a product that does or does not contain triclosan has a large market share then the chemical occurrence in reality will be reduced or increased accordingly. However, for simplicity an equal market share approach is used initially as market shares or sales volumes are not readily available. To reduce uncertainty and err on the side of conservatism, presence probabilities were rounded to the nearest upper 10%.
Table 6: Chemical occurrence values for triclosan derived from Codecheck.info database
| Product Type | Total Number of Products | Number of Products with Triclosan | Original Chemical Occurrence (between 0 and 1) | Chemical Occurrence (between 0 and 1) |
| Toothpaste | 552 | 35 | 0.063405797 | 0.1 |
| Mouthwash | 328 | 3 | 0.009146341
|
0.1 |
| Deodorant Stick | 284 | 19
|
0.066901408
|
0.1 |
| Body Lotion | 1710 | 8 | 0.004678363
|
0.1 |
| Face Powder | 835 | 2 | 0.00239521
|
0.1 |
| Blemish Concealer | 435 | 3 | 0.006896552
|
0.1 |
| Hand Soap | 171 | 5 | 0.029239766
|
0.1 |
| Shower Gel/Body Soap | 4000 | 7 | 0.00175
|
0.1 |
Finally, only European consumers, both male and female, were selected for the assessments. This gave rise to a population sample size of n = 26,209.
As a distribution of exposure is the resulting output of the assessment, characterising this distribution can be done in a number of ways. In the following, two exposure statistics are presented. The first is the arithmetic mean of exposure to represent the average exposure, and the other is the 95th percentile of exposure is used to represent the upper exposure in the population. Additionally, when using real habits and practices data, exposure statistics can be calculated over two populations. The first is the Total Population, i.e. all subjects in the survey, and the other is the Exposed Population, i.e. all those consumers who are exposed to the substance.
Given that there are two potential populations that can be used to characterise exposure, an immediate question posed is to determine which one is most appropriate. One guiding principle is that for aggregate exposure resulting from multiple sources that are used by the majority of the population (e.g. all categories of cosmetics and personal care products or most foods in the diet), then exposure is well represented by the Total Population. However, if exposure is due to a small number of products or is due to an infrequently occurring substance, then the Exposed Population is likely the more appropriate set to use. This is so that there is not inappropriate dilution of exposure statistics by the inclusion of large number of zeroes in their calculation. The initial problem formulation step of the assessment should consider this aspect by defining if a general population exposure or the exposures to the population of product users is the assessment goal.