Determining the microbiological criteria for lot rejection from the performance objective or food safety objective

Use of a Poisson-gamma model to assess the performance of the EC process hygiene criterion for Enterobacteriaceae on Irish sheep carcasses

“…Consequently, in the development of microbiological criteria/ sampling plans, the variability of the quality of the end product has been as a norm assumed more or less constant (Smelt & Quadt,1990), or said otherwise, independent of the level of contamination of the batch (Whiting et al, 2006). Recently, in an AOAC report concerning statistical process control for microbial load in foods, it is stated that the correlation between means and variances often observed for plate counts data can be removed by logarithmic transformation (AOAC, 2005).…”

“…Regulation EC No 1441/2007(Anonymous, 2007 lays down microbiological criteria defining the acceptability of a batch of foodstuff or a process, based on the absence, presence or number of microorganisms tested per batch. Historically, in the development or evaluation of acceptance sampling plans by attributes (Dahms, 2004;Dahms & Hildebrandt, 1998;Hildebrandt, Bohmer, & Dahms, 1995;Legan, Vandeven, Dahms, & Cole, 2001;Van Schothorst, Zwietering, Ross, Buchanan, & Cole, 2009;Whiting et al, 2006) and by variables (Kilsby, Aspinall, & Baird-Parker, 1979;Malcolm, 1984;Smelt & Quadt, 1990), two simplifying assumptions have been always made: that the true concentration of microorganisms is lognormally distributed within the batch, and that the variance of the samples is the same for a low or highly contaminated lot. Legan et al (2001) provided a basic method for assessing the performance of two-class and three-class attributes sampling plans translating the information about the proportion of units that are defective into an estimate of the microbial concentration in the batch.…”

“…Legan et al (2001) provided a basic method for assessing the performance of two-class and three-class attributes sampling plans translating the information about the proportion of units that are defective into an estimate of the microbial concentration in the batch. The assumption of constant within-batch variance was made to facilitate the assessment of the performance of a sampling plan as distributions move right or left on the x (contamination level) axis with different mean levels of contamination in an operating characteristic curve (Dahms, 2004;Dahms & Hildebrandt, 1998;Legan et al, 2001;Hildebrandt et al, 1995;Van Schothorst et al, 2009;Whiting et al, 2006). Van Schothorst et al (2009) kept the within-batch lognormal assumption but considered the situations in which microbiological testing of samples involves enrichment and absence/presence testing rather than enumeration.…”

Proposal of sampling protocols to verify possible performance objectives for Campylobacter species control in Italian broiler batches

“…I dati di prevalenza sulle carcasse sono stati analizzati mediante distribuzione Beta e il criterio di performance (PO) è stato fissato al 50° percentile di tale distribuzione. Utilizzando l'approccio suggerito da Whiting et al (2006), il numero di campioni da analizzare per determinare almeno un campione positivo segue la seguente equazione:…”

“…I dati di prevalenza sulle carcasse sono stati analizzati mediante distribuzione Beta e il criterio di performance (PO) è stato fissato al 50° percentile di tale distribuzione. Utilizzando l'approccio suggerito da Whiting et al (2006), il numero di campioni da analizzare per determinare almeno un campione positivo segue la seguente equazione:dove N s è il numero di campioni da analizzare, p R è la probabilità di rigettare il campione (95% CL), e P è la prevalenza attesa, stimata attraverso l'analisi di distribuzione. Per il calcolo di N s è stata effettuata una simulazione MonteCarlo nel ModelRisk v4.3 (Vose Consulting, Ghent, Belgium) con 10.000 interazioni.…”