Bioassay predictive values for chemical health risks in drinking water.

Area of Science:

• Environmental Chemistry
• Toxicology
• Risk Assessment

Background:

• Bioanalytical tools are crucial for assessing drinking water chemical quality.
• Water managers need to understand the reliability of bioassay responses exceeding health-based trigger values.

Purpose of the Study:

• To present and apply a Bayesian framework for calculating risk probabilities of bioassay responses.
• To evaluate the predictive values of bioassays under various chemical mixture scenarios relevant to drinking water.

Main Methods:

• Utilized Bayes' theorem to derive risk probabilities for bioassay responses.
• Evaluated six in vitro assays from the ToxCast database using in silico chemical mixtures.
• Assessed performance based on negative predictive value (NPV) and positive predictive value (PPV).

Main Results:

• Achieved 100% NPV for all assays, confirming absence of risk when below the effect-based trigger value (EBT).
• Positive predictive value (PPV) ranged from 40-80% for concentrations above 2 µg/L, varying by assay.
• PPV remained low for complex, low-concentration mixtures, indicating potential for false positives.

Conclusions:

• Bioassays reliably rule out chemical risk below the EBT.
• Exceeding the EBT has limitations in confirming harmful concentrations, particularly for diverse low-concentration mixtures.
• Adjusting trigger values can optimize predictive values for water quality management.