QMRA for Drinking Water: 2. The Effect of Pathogen Clustering in Single-Hit Dose-Response Models

Area of Science:

• Microbial risk assessment
• Quantitative microbiology
• Statistical modeling

Background:

• Standard quantitative microbial risk assessment (QMRA) assumes pathogen counts follow a Poisson distribution.
• Pathogen spatial or temporal clustering can violate this assumption, potentially invalidating risk estimates.
• Single-hit dose-response models are commonly used in QMRA but may be sensitive to clustering.

Purpose of the Study:

• To investigate the theoretical impact of pathogen clustering on single-hit dose-response models in QMRA.
• To evaluate the suitability of the stuttering Poisson distribution for modeling clustered pathogen doses.
• To develop and apply a risk bound applicable to overdispersed dose distributions.

Main Methods:

• Employed the stuttering Poisson distribution, a flexible count distribution modeling pathogen clustering.
• Utilized probability generating functions to formulate dose-response models.
• Derived an upper bound for risk using Jensen's inequality for overdispersed distributions.

Main Results:

• Theoretical single-hit risk is lower with stuttering Poisson than with Poisson distribution for identical mean doses.
• Risk insensitivity to moderate clustering observed, with greater effect at lower mean doses.
• Derived risk bound approximates exact theoretical risk for highly overdispersed doses.

Conclusions:

• Pathogen clustering can reduce theoretical risk estimates in QMRA compared to standard Poisson assumptions.
• The stuttering Poisson distribution is a suitable model for pathogen clustering in QMRA.
• The derived risk bound offers a practical tool for risk characterization with overdispersed pathogen doses.