# Risk Assessment of Dibutyl Phthalate (DBP) and Bis(2-Ethylhexyl) Phthalate (DEHP) in Hot Pot Bases with a Hybrid Modeling Approach

## TL;DR

This study assesses the risk of phthalate contamination in hot pot bases using advanced statistical models, finding that children aged 7–13 face the highest risk despite overall acceptable exposure levels.

## Contribution

The study introduces a hybrid modeling approach to better capture extreme phthalate contamination events, improving risk assessment accuracy.

## Key findings

- DEHP and DBP contamination levels were best modeled using advanced statistical distributions that capture extreme values.
- Children aged 7–13 had the highest calculated risk, with a maximum DEHP hazard quotient of 0.68.
- Hybrid models outperformed traditional methods in identifying sporadic high-exposure events.

## Abstract

(1) Background: Hot pot bases are susceptible to phthalate (PAE) contamination due to their high lipid content. Standard risk models often fail to capture extreme values, leading to biased exposure estimates. This study characterized dibutyl phthalate (DBP) and bis(2-ethylhexyl) phthalate (DEHP) contamination using a hybrid modeling framework to ensure precise risk profiling. (2) Methods: A total of 91 samples were analyzed via GC-MS. Concentration data were fitted using traditional parametric, extreme value mixture (EVMM), and finite mixture models. Probabilistic dietary risks were assessed for Chinese demographic groups using 10,000-iteration Monte Carlo simulations. (3) Results: DEHP (detection rate: 55%) and DBP (32%) were best modeled by a two-component Gamma mixture and a Lognormal–Generalized Pareto distribution, respectively. These advanced models significantly outperformed conventional distributions in capturing upper-tail extremes. Crucially, all hazard quotients (HQs) remained below the safety threshold of 1, indicating acceptable risk, although children aged 7–13 exhibited the highest calculated risk (Max DEHP HQ = 0.68). (4) Conclusions: Although current exposure levels are within safe limits, the heavy-tailed distributions identify potential sporadic high-exposure events that traditional models overlook, specifically highlighting the relative vulnerability of children aged 7–13. This study validates that hybrid statistical approaches offer superior precision for analyzing skewed contamination data. Consequently, these findings provide a critical scientific basis for refining regulatory monitoring and implementing targeted source-tracking measures to mitigate long-tail food safety risks.