Evaluation of a hybrid in-field sampling method for the detection of pathogenic bacteria through consideration of a priori knowledge of factors related to non-random contamination

Abstract

Pre-harvest testing is increasingly used to enhance the microbial safety of fresh produce. Traditional sampling assumes that sample collectors have no information on potential contamination sources. Knowledge of such factors could potentially increase the effectiveness of pre-harvest sampling programs. Simulation modeling and field validation trials were used to evaluate a hybrid “Samples of Opportunity” (SOO) sampling method that included a portion of the samples based on the sampler's knowledge of risk factors in pre-harvest produce fields. Relative effectiveness of SOO sampling was compared with three traditional sampling methods. These evaluations were based on three non-random contamination scenarios. The mean detection probability of SOO is 96% higher than traditional sampling methods (p < 0.001). However, if the site of actual contamination is offset from assumed area of contamination, the detection probability of SOO sampling drops, and becomes similar or even worse than that achieved by the other sampling methods. Preliminary field validation trials indicated indeed that SOO performed better than the other three sampling methods. This study provides a mathematical approach for evaluating the effectiveness of four pre-harvest sampling methods, and suggests that having a priori knowledge of the contamination source in the field would improve effectiveness of sampling, particularly if done using a standardized protocol.