A comparison of microbial dose-response models fitted to human data

Area of Science:

• Microbiology
• Toxicology
• Risk Assessment

Background:

• Mathematical models are crucial for microbial risk assessment.
• Human volunteer studies provide essential infectivity and illness data for pathogens.
• Evaluating model variability is key to determining appropriate complexity for dose-response data.

Purpose of the Study:

• To assess variability among eight mathematical dose-response models for microbial risk assessment.
• To determine if two-parameter models are sufficient or if three-parameter models are generally required for human microbial dose-response data.
• To evaluate model performance using estimated effective dose levels (ED01 and ED10).

Main Methods:

• Utilized infectivity and illness data from published human volunteer studies.
• Compared eight mathematical dose-response models.
• Measured model variability using estimated ED01s and ED10s, representing 1-10% risk levels.
• Investigated the ranking of ED01 and ED10 values across different models.

Main Results:

• Two-parameter models captured comparable uncertainty to three-parameter models for the examined microbial dose-response data.
• No single "best" two-parameter model was identified, but relative behaviors were elucidated.
• The findings reinforced model uncertainty analysis using multiple two-parameter models.

Conclusions:

• Two-parameter models appear adequate for representing uncertainty in human microbial dose-response data.
• Simpler models may be suitable for microbial risk assessment, reducing complexity without sacrificing significant uncertainty estimation.
• Further insights into the comparative performance of two-parameter models were gained.