Revising a constrained 2-Class attributes sampling plan when laboratory methods are changed

The Food Safety and Inspection Service (FSIS) implemented a 2-class attributes sampling plan for Campylobacter (i.e., a performance standard) for comminuted chicken based on a 1 mL direct plating assay. The performance standard specified only one allowable positive out of 52 samples. This paper explores an alternative performance standard that is based on an assay with a lower limit of detection (i.e., 30 mL instead of 1 mL) but that continues to meet the same objectives as the original standard. A multi-objective optimization algorithm is applied to determine a new number of allowable Campylobacter positives based on a 30 mL enrichment assay, such that the new performance standard would meet the public health and regulatory cost objectives of the original direct plating standard. Two scenarios are considered. Scenario one assumes there was no change in pathogen occurrence since the original 1 mL performance standard was implemented. For Scenario one, the new number of allowable positives (five) that minimizes the objective function also reflects a naïve expectation based on the ratio of the mean prevalence levels from 30 mL and 1 mL data. Scenario two considers the implications of a change in pathogen occurrence since the original 1 mL performance standard was implemented. This latter approach demonstrates the complications of reconciling the dynamics of industry distributions while referencing performance standards derived from past data.