Conclusion

As this case study shows, aggregate exposure assessments can be quite complex, particularly if the material of interest is ubiquitous, being present for example in several consumer products, foods and the environment.

Such assessments are, by necessity, an iterative process, which should be conducted using a tiered strategy, where the lowest tier (0) consists of a rough sum of exposure to each product, the mid-tier (1) tends to be a semi quantitative estimate, such as a deterministic estimate with conservative assumptions, and the higher tier (2) is a more realistic estimation of population exposure that is modelled using probabilistic methods and a person-orientated approach. The rough or low tier estimates can be calculated quickly yielding conservative exposure values, and if this approach is lower than the “safe” toxicological exposure dose, then it may not be necessary to move to a higher tier. While exposure assessments at tier 2, using person orientated probabilistic approaches to estimate exposure in populations, can be data-intensive and time consuming, they produce more refined and accurate estimates of population exposure, enabling the risk assessor to feel confident that the risk assessment is applicable to the population of interest.

The estimated refined internal exposures can be compared with the experimental biomonitoring data estimated from urine excretion. As discussed in Troutman et al, 2015, three human biomonitoring studies (Fromme et al, 2013; Goeen et al, 2001; Garlantézec et al, 2012) that include a total of 637 subjects report levels of the major metabolite of phenoxyethanol, PhAA, in urine ranging from 0.12 to 47.4 mg/L, with the highest values reported by Fromme et al (2013). In the Fromme study, the measured concentration of PhAA in urine collected from 44 subjects was 0.80 mg/L (median), 23.6 mg/L (95th percentile) and 47.4 mg/L (maximum). From this the corresponding external dose level of phenoxyethanol was estimated at 0.015 (median), 0.43 (P95) and 0.86 (maximum) mg /kg/day by assuming that the urine samples were collected under steady-state conditions from the general population with corresponding body weight and urine output values of 70 kg and 1.4 L urine/day, respectively (Davies and Morris, 1993). This estimated P95 internal exposure value of 0.43 mg/kg/day from biomonitoring data, can be compared to the tier 2 P95 internal exposure estimates from the PACEM and Creme Care and Cosmetics assessment tools (1.3 mg/kg/day and 0.38 mg/kg/day) and are similar. However, aggregation of the exposure from household products yields an internal exposure value of 9.67 mg/kg/day. Comparison of this value with the internal exposure value from biomonitoring data demonstrates that the lower tier tools do not provide a realistic estimate of general population exposure that can be aggregated with other exposures. Investment is needed in the collation/generation of data (e.g. co-use and chemical occurrence data) for household products to facilitate the development of higher tier more realistic aggregate exposure estimates across consumer product categories.