Related Experiment Videos
A model to estimate the optimal sample size for microbiological surveys
S F Altekruse1, F Elvinger, Y Wang
1Center for Veterinary Medicine, Food and Drug Administration, Rockville, Maryland, USA. sean.altekruse@fsis.usda.gov
Applied and Environmental Microbiology
|October 9, 2003
Summary
Laboratory managers can now estimate optimal microbiological sample sizes using a new Bayesian statistical model. This approach prevents biased results from under-sampling and conserves resources by avoiding over-sampling, ensuring accurate bacterial identification.
Area of Science:
- Microbiology
- Statistical Modeling
- Bioinformatics
Background:
- Estimating sample size for microbiological surveys presents challenges for laboratory managers, risking biased inferences or wasted resources.
- Accurate bacterial identification relies on appropriate sample sizes to avoid under- or over-sampling.
- Current methods lack a robust framework for optimizing sample size in diverse microbiological contexts.
Purpose of the Study:
- To present a statistical model for estimating optimal sample size for accurate bacterial subtype identification in microbiological surveys.
- To provide a practical tool for laboratory managers to improve resource allocation and inferential accuracy.
- To demonstrate the model's application in real-world scenarios for bacterial colony sampling.
Main Methods:
- Development of a Bayesian inference model combining prior scientific knowledge with observed survey data.
- Inputting a prior distribution of strains per specimen from an informed microbiologist and survey data on strains per specimen.
- Generating an updated probability distribution of strains per specimen to estimate the probability of observing all present strains based on sampled colonies.
Main Results:
- The model provides an updated probability distribution, enabling estimation of the likelihood of detecting all bacterial strains present.
- Two scenarios illustrate the model's use: estimating sample size for identifying Campylobacter types on broiler carcasses and Salmonella enterica serotype Enteritidis phage types in poultry fecal swabs.
- The model's precision in sample size estimation is expected to increase with the incorporation of ongoing survey data.
Conclusions:
- The presented Bayesian statistical model offers a robust method for determining optimal sample sizes in microbiological surveys.
- This approach enhances the accuracy of bacterial subtype identification while optimizing laboratory resource utilization.
- Continuous incorporation of new data into the model promises increasingly precise and reliable sample size estimations for future surveys.