Evaluation and Validation of Various Sampling Plans for the Detection of Pathogenic or Indicator Microorganisms on Pre-harvest Leafy Greens

Abstract

The consumption of leafy greens increased over the last few decades due to health concerns. However, leafy green vegetables are highly susceptible to microbial contamination. The pre-harvest sampling and testing are highly important to ensure safety of leafy greens. Z-pattern sampling scheme is extensively used currently. However, the scientific rationale and performance attributes of these sampling plans are unclear in relation to detection of both indicator microorganisms and pathogens. The overall goal of this study is to evaluate and validate various sampling plans for the detection of pathogenic bacteria on pre-harvest leafy greens and find the optimal sampling plan. Computer simulations and field trials were performed to compare the effectiveness of various sampling plans, including simple random sampling, stratified random sampling, Z-pattern sampling, “samples of opportunity” sampling and iterative Bayesian sampling. Studies showed that Z-pattern sampling plan had larger variability than random sampling plan and stratified sampling plan when the contamination sites were randomly distributed, although the mean detection probabilities of these three sampling plans were the same. Samples of opportunity sampling performed better than random sampling plan and stratified sampling plan when the contamination sites were non-randomly distributed, such as flooded field, field with animal house nearby or field with power line above. And iterative Bayesian sampling was suggested when the number of samples is limited. A what-if sampling strategy would be made to get a more efficient detection of pathogenic bacteria for the industry and government farms. This study provides the scientific and mathematical rationale for various sampling plans and allows leafy green growers to make informed decisions regarding strategies for optimizing pre-harvest microbiological testing programs.