Non-stochastic sampling error in quantal analyses for Campylobacter species on poultry products

Using primers and fluorescent probes specific for the most common food-borne Campylobacter species (Campylobacter jejuni and Campylobacter coli), we developed a multiplex, most probable number (MPN) assay using quantitative PCR (qPCR) as the determinant for binomial detection: i.e., number of p positive pathogen growth responses out of n = 6 observations each of 4 mL (V) per dilution. Working with media washes of thrice frozen-thawed chicken pieces which had been spiked with known levels of C. jejuni and C. coli, we found that about 20% of the experiments had a significant amount of error in the form of either greater than 25% MPN calculation error (Δε) and/or a low apparent recovery rate (R less than 1 = MPN observed ÷ CFU spiked). Assuming such errors were exacerbated by an excessively small n, we examined computer-generated MPN enumeration data from the standpoint of stochastic sampling error (Δ) and found that such binomial-based assays behaved identically to Poisson-based methods (e.g., counting data) except that fewer technical replicates (n) appeared to be required for the same number of cells per test volume (μ). This result implies that the qPCR detection-based MPN protocol discussed herein should accurately enumerate a test population with a μ ≥ 1 using n = 6 observations per dilution. For our protocol, this equates to ≥ 8 cells per 400-500 g of sampled product. Based on this analysis, the error rate we saw in spiked experiments (where μ >> 1) implied a non-stochastic source. In other experiments we present evidence that this source was, at least in part, related to the cell concentration step (i.e., centrifugation). We also demonstrate that the error rate lessened (from ~38% to ~13%) at lower Campylobacter levels (μ ≤ 40) as would most likely exist in nature. Using this protocol, we were able to quantify 14 to 1,226 MPN per 450 g of naturally contaminated chicken for skinless pieces and 11 to 244 MPN per 450 g for wings, breasts, legs, and thighs (skin on) whereupon about 50% of the 29 samples tested negative for both species. Four of these chicken wash samples did have substantially lower Campylobacter levels (1 to 6 MPN per 450 g) which might be better enumerated using a larger n. However, we established that the limit of quantification of this protocol diminishes for n > 6 because one is ever more diluting the sample, or lessening V, to achieve the requisite n.