Related Experiment Videos
Uncertainty in most probable number calculations for microbiological assays
Graham B McBride1, Judith L McWhirter, Matthew H Dalgety
1National Institute of Water and Atmospheric Research, PO Box 11-115, Hamilton, New Zealand. g.mcbride@niwa.co.nz
Journal of AOAC International
|November 25, 2003
Summary
Accurate microorganism concentration estimation in microbiological assays is improved using exact Most Probable Number (MPN) calculation methods. Bayesian statistics offer superior credible intervals for uncertainty quantification in MPN estimates.
Area of Science:
- Microbiology
- Statistics
- Bioinformatics
Background:
- Microbiological assays frequently employ dilution series and Most Probable Number (MPN) estimations.
- Existing MPN tables exhibit variability due to approximate calculations, rounding, and differing interval methods.
Purpose of the Study:
- To address discrepancies in published Most Probable Number (MPN) tables.
- To introduce precise calculation methods and Bayesian statistics for improved MPN estimation and uncertainty quantification.
Main Methods:
- Utilized recently developed exact MPN calculation methods.
- Applied Bayesian statistics to derive credible intervals for MPN estimates.
- Investigated diffuse prior and empirical Bayes procedures with a Poisson prior distribution.
Main Results:
- Exact MPN calculation methods resolve issues related to approximate procedures and rounding conventions.
- Bayesian statistics provide a robust framework for quantifying uncertainty in MPN estimates, particularly for credible intervals.
- Empirical Bayes procedures with a Poisson prior yielded narrower credible interval widths compared to other methods.
Conclusions:
- Exact MPN calculations and standardized rounding improve table consistency.
- Bayesian approaches are essential for accurate credible interval determination in MPN analysis.
- The proposed Bayesian methods enhance the precision of microorganism concentration estimates in microbiological assays.