Identification of Chicken Component Samples Containing Salmonella Concentrations Greater Than or Equal to 1 CFU/g

Introduction

On May 1, 2024, the United States Department of Agriculture Food Safety and Inspection Service (USDA-FSIS) published a Final Determination in the Federal Register that beginning on May 1, 2025, not ready-to-eat (NRTE) breaded stuffed chicken products that contain Salmonella at levels of 1 colony forming unit per gram (hereafter, “1 CFU/g”) or higher are adulterated (USDA-FSIS, 2024c). The Final Determination states that USDA-FSIS will enforce the adulteration threshold by testing 1-lb verification samples of the raw incoming chicken components used to produce NRTE breaded stuffed chicken products (USDA-FSIS, 2024c). The USDA-FSIS Microbiology Laboratory Guidebook (MLG) Chapter 4.15, Table 3, indicates that a 325-gram portion of the verification sample will be suspended in 1,625 mL of buffered peptone water (BPW) (USDA-FSIS, 2024a). For the purposes of this report, we have named this sample-BPW mixture the “Salmonella Screening Suspension.” According to MLG Chapter 4.15, the Salmonella Screening Suspension is incubated at 35°C for 20 to 24 h producing the “Salmonella Screening Enrichment.” Then, an aliquot of Salmonella Screening Enrichment is tested with the Neogen Molecular Detection Assay 2 Salmonella (MDA2-Salmonella) system. Verification samples with a positive MDA2-Salmonella outcome are declared “Salmonella Screen Positive.”

The Final Determination specified that Salmonella Screen Positive verification samples will be analyzed to determine if the Salmonella concentration meets or exceeds the 1 CFU/g adulteration threshold. The Final Determination indicated that this “adulteration assessment” may be performed with either the BioMerieux GENE-UP QUANT Salmonella Test Kit or a most probable number (MPN) quantification (USDA-FSIS, 2024c). The current BioMerieux 1 CFU/g lower limit of quantification GENE-UP QUANT protocol consists of a 4-h enrichment of a Salmonella Screening Suspension, then a genomic DNA extraction involving centrifugation, followed by a proprietary quantitative PCR (qPCR) protocol (Biomerieux, 2023). The use of MPN quantification as the adulteration assessment would require at least 9 overnight enrichments for each sample followed by Salmonella detection for each of the ≥9 enrichments.

Alternatively, the recently described Poisson Limit One Tube (PiLOT) (Schmidt et al., 2024) method could provide an adulteration assessment result within 2 hours of a Salmonella Screen Positive result. Schmidt et al. (2024) demonstrated the PiLOT threshold approach performed as well or better than 2 commercial quantification methods for assessing whether a sample met or exceeded 10 CFU/mL of Salmonella in inoculated turkey wing rinses. The optimal Salmonella quantification or threshold method depends on the testing application as well as the relative importance of technical, financial, and time resource constraints. Because the PiLOT method uses the Poisson probability distribution, the threshold Salmonella level can be set to user specifications. However, a PiLOT protocol with a theoretical 95% probability of detecting as positive a sample containing the threshold level of 1 CFU/g (a PiLOT-95 protocol) has a 50% probability of detecting as positive a sample containing 0.2 CFU/g Salmonella (Figure 1). This PiLOT-95 protocol entails an enrichment of an aliquot of sample suspension followed by a test of the enrichment for the presence of Salmonella. A modification of the PiLOT method involving enriching 3 aliquots of sample suspension theoretically reduces the potential for false positive outcomes. We named the modification the “Confidently Above Threshold” (CAT) method. A CAT protocol with a theoretical 80% probability of detecting as positive a sample containing 1 CFU/g (a CAT-80 protocol) has a 50% probability of detecting a positive sample containing 0.6 CFU/g Salmonella (Figure 1). Thus, the PiLOT threshold approach may be well suited for detecting samples with Salmonella levels ≥1 CFU/g and a 3-tube version theoretically optimizes the approach with a refined balance between false positive and false negative results. The objective of this study was to empirically demonstrate that compared with the PiLOT-95 protocol, the CAT-80 protocol reduces false adulteration (False Positive outcome) results while minimally impacting false not adulterated (False Negative outcome) results. This was achieved using raw chicken component samples inoculated with Salmonella strains at levels near 1 CFU/g.

Figure 1.

Adulteration outcome probabilities (P_Adult) for PiLOT-95 and CAT-80 protocols were performed with raw chicken component samples harboring Salmonella at concentrations between 0.01 and 2.0 CFU/g. The dark red and dark blue lines correspond to a 325-g raw poultry sample suspended in 1,625 mL of BPW. The light red and blue shaded areas correspond to the MLG 4.15 raw poultry sample weight (325 ± 32.5 g) and BPW volume (1,625 ± 32.5 mL) tolerances.

Materials and Methods

Results

Across the 80 inoculated raw chicken component samples reference Salmonella quantities ($Q_{\mathit{ref}}$) ranged from < 0.1 to 52.5 CFU/g (Data File A6). Sens_PiLOT-95 and Sens_CAT-80 were both 1.00 because all 40 raw chicken component samples with $Q_{\mathit{ref}}$ ≥ 1.0 CFU/g (including 24 samples with $Q_{\mathit{ref}}$ values between 1.0 and 1.9 CFU/g) were accurately identified as adulterated by PiLOT-95 and CAT-80 (Table 1). As expected, the Spec_CAT-80 and Acc_CAT-80 values were higher than Spec_PiLOT-95 and Acc_PiLOT-95 values (Table 1). As predicted by P_Adult values, the fraction of the 12 inoculated raw chicken component samples with $Q_{\mathit{ref}}$ values between 0.5 and 0.9 CFU/g falsely declared adulterated were lower for the CAT-80 protocol (Table 2). PiLOT-95 falsely declared adulterated 32.1% of the 28 samples with $Q_{\mathit{ref}}$ values <0.5 compared to 3.6% for CAT-80 (Table 2).

Discussion

Testing methods designed to produce binary outputs regarding a threshold bacterial concentration in meat samples are a relatively novel development in food safety. Proposed titles for these approaches include “threshold testing methods,” “limit testing methods,” and “qualitative threshold methods” (National Advisory Committee on Microbiological Criteria in Foods, 2024; Schmidt et al., 2024). This study supported the hypothesis that the CAT-80 threshold testing protocol reduces the occurrence of false adulterated (false positive) outcomes compared with the PiLOT-95 threshold testing protocol (Tables 1 and 2). The mathematical foundation of the CAT method was the Poisson probability distribution with the assumption that the probability of >0 Salmonella cells in each of the 3 CAT Aliquots were independent events (Eq. 3). The 3 CAT Aliquots were not sensu stricto statistically independent, but a satisfactory and feasible alternative statistical model was not apparent. The alignment of the empirical results with the CAT P_Adult values (Table 2) demonstrated that Eq. 3 was an acceptable mathematical model for the CAT method.

This research provides proof of concept for the CAT threshold method, which is compatible with the current FSIS verification sample testing method for raw poultry products (Supplemental Figure 1). The FSIS MLG 4.15 raw poultry verification sample testing method includes tolerances for portion size (325 ± 32.5 g) and BPW volume (1,625 ± 32.5 mL) used to prepare the Salmonella Screening Suspension (USDA-FSIS, 2024a). The lowest P_Adult values occur when the smallest portion size tolerated (g = 292.5 g) is suspended in the largest BPW volume tolerated (f = 1657.5 mL). The highest P_Adult values occur when the largest portion size tolerated (g = 357.5 g) is suspended in the smallest BPW volume tolerated (f = 1,592.5 mL) (Figure 1). The impacts of the Salmonella Screening Suspension tolerances on P_Adult values for PiLOT-95 and CAT-80 are depicted in Figure 1. We conclude that the CAT-80 threshold protocol is robust to these tolerances because P_Adult is reduced by only 0.05 at the 1.0 CFU/g threshold when 292.5 g is suspended 1,657.5 mL (Figure 1).

The CAT-80 threshold protocol was selected for the empirical proof-of-concept experiments described in this manuscript based on one possible balance between Sens and Spec. Prospective users, including but not limited to USDA-FSIS, industry, and commercial testing laboratories, may target different Sens and Spec balances. The mathematical foundation of the CAT method enables each user to select threshold and P_Adult values and remove the appropriate volumes of sample suspension for the 3 CAT Aliquots. For example, Eq. 3 demonstrates that 95% detection probability (CAT-95 protocol; P_PCO = 0.95) of Salmonella at the 1 CFU/g threshold (T = 1) in a 325-g sample (g = 325) suspended in 1,625 mL (f = 1,625) is achieved with 3 CAT Aliquots of 25 mL ($v_{CAT}$ = 25). Eq. 4 was employed to graph the P_Adult tolerance ranges for CAT-80 ($v_{CAT}$ = 16 mL), CAT-90 ($v_{CAT}$ = 20 mL), CAT-95 ($v_{CAT}$ = 25 mL), and CAT-99 ($v_{CAT}$ = 32 mL) protocols for a 325 ± 32.5 g sample suspended in 1,625 ± 32.5 mL at a 1 CFU/g threshold (Figure 2).

Figure 2.

Adulteration outcome probabilities (P_Adult) for CAT-99 (green), CAT-95 (purple), CAT-90 (orange), and CAT-80 (blue) protocols performed with raw chicken component samples harboring Salmonella at concentrations between 0.01 and 2.0 CFU/g. Lines correspond to a 325-g raw poultry sample suspended in 1,625 mL of BPW. The shaded areas correspond to the MLG 4.15 raw poultry sample weight (325 ± 32.5 g) and BPW volume (1,625 ± 32.5 mL) tolerances.

Conclusions

Ease of use is a key strength of the CAT method. Essentially, the CAT method consists of transferring sample suspension to 3 separate tubes, incubating these tubes, and then determining the presence or absence of Salmonella in each of these 3 enrichments. Theoretically, any Salmonella molecular or immunological detection system is compatible with the CAT method and analogous methods such as PiLOT. For example, in the present study, the Neogen MDA2 system was used to detect Salmonella in PiLOT and CAT enrichments while the Hygiena BAX Real-Time PCR system was used to detect Salmonella during the study initially describing PiLOT methods (Schmidt et al., 2024). Another key strength of the CAT method is the ability to accommodate user-defined thresholds across many different sample types. Given the novelty of threshold methods, additional proof-of-concept and equivalence studies are required across various matrices, bacterial targets, and thresholds. However, applying threshold testing concepts such as PiLOT and CAT to the 10 CFU/g and 10 CFU/mL Salmonella thresholds, which are a component of recent FSIS Salmonella proposals (USDA-FSIS, 2024b), is a critically important research area that we plan to address. PiLOT and CAT methods are also amenable to shorter enrichment times or other alternatives to overnight enrichment. Studies focused on more rapid identification of meat and poultry samples exceeding specific Salmonella thresholds are planned.

Competing Interests

The authors declare no conflicts of interest.

Author Contribution Statement

John W. Schmidt: conceptualization, methodology, formal analysis, investigation, resources, and data curation; original draft writing, review, and editing; and visualization, supervision, project administration, and funding acquisition. Weifan Wu: investigation and data curation. Dayna Harhay: investigation and resources. Tommy L. Wheeler: writing, review, editing, supervision, resources, and funding acquisition.