To pool or not to pool: That is the question in microbial kinetics

Area of Science:

• Microbiology
• Food Science
• Statistical Modeling

Background:

• Heat inactivation is crucial for Salmonella control in food.
• Existing models often oversimplify variability in inactivation data.
• Understanding parameter variation is key for accurate microbial inactivation modeling.

Purpose of the Study:

• To characterize variation in Salmonella heat inactivation using multilevel modeling.
• To compare different data pooling strategies (no pooling, complete pooling, partial pooling).
• To assess the impact of experimental conditions on inactivation parameters.

Main Methods:

• Applied multilevel modeling and Bayesian regression with the Weibull model.
• Analyzed two case studies: single-temperature repetitions and multi-temperature experiments.
• Investigated parameter variation at different data hierarchy levels.

Main Results:

• Partial pooling significantly reduced parameter variation compared to complete pooling.
• Temperature strongly affected the rate parameter, while shape parameter estimates varied with analysis method.
• Multilevel global modeling yielded narrower distributions for decimal reduction times.

Conclusions:

• Multilevel modeling is a powerful tool for quantifying variation in microbial heat inactivation.
• Partial pooling is recommended for prediction-focused modeling of inactivation data.
• This approach provides more accurate and unbiased parameter estimates.