Estimating hazardous concentrations by an informative Bayesian approach.

Area of Science:

• Environmental Toxicology
• Ecotoxicology
• Risk Assessment

Background:

• The species sensitivity distribution (SSD) approach is standard for chemical risk assessment but requires extensive ecotoxicological data.
• Frequentist methods for SSD and hazardous concentration 5th percentile (HC5) derivation are data-intensive, limiting their practical application.
• Existing Bayesian approaches for HC5 estimation often use non-informative priors, which may not be optimal for small datasets.

Purpose of the Study:

• To develop and evaluate an alternative Bayesian approach for estimating HC5 from small ecotoxicological datasets.
• To compare the performance of an informative Bayesian method against frequentist approaches and non-informative Bayesian methods.
• To assess the robustness and reliability of HC5 estimates derived from limited data.

Main Methods:

• Developed a Bayesian approach for HC5 estimation incorporating informative priors on species sensitivity variance.
• Tested the method using actual ecotoxicological data for 21 substances.
• Performed cross-validation comparing Bayesian HC5 values with frequentist results using complete and truncated datasets.

Main Results:

• The informative Bayesian approach yielded HC5 values comparable to frequentist methods, even with truncated data.
• The probability of overestimating HC5 using the Bayesian method was found to be limited.
• The approach demonstrated robustness, providing reliable HC5 estimates from smaller datasets.

Conclusions:

• The informative Bayesian approach is a robust and practical surrogate for deriving HC5 values when extensive ecotoxicological data is unavailable.
• This method facilitates obtaining more robust risk estimates from limited data without compromising regulatory protection.
• The findings encourage the collection of new ecotoxicological data by demonstrating the utility of smaller datasets with advanced statistical methods.