Table C.4.

Potential sources of uncertainty identified in the assessment of the minimum HPP requirements for the control of pathogens in raw milk and raw colostrum to achieve an equivalent efficacy to thermal pasteurisation

Source or location of the uncertainty	Nature or cause of the uncertainty	Impact of the uncertainty on the conclusions (e.g. over/underestimation)
Data used to fit the models (number of studies available per pathogen, number of strains used in eligible studies)	Representativeness of the variability in the efficacy of HPP against the relevant pathogens in different types of milk. Log10 reduction data were all based on experimental data subjected to error, variability and uncertainty. In many studies mean log10 reduction of few replicates are reported, missing the variability (lower and higher values than that collected).	The impact of this uncertainty is estimated to be low‐moderate as a high number of studies were included in the assessment covering a relatively wide range of experimental conditions, which resulted in a wide variability of the log10 reduction values. This uncertainty may result in an under‐ or over‐estimation of the HPP efficacy. However, based on model prediction, this source may have contributed more to the underestimation of the HPP efficacy.
Model structure	Log‐linear or biphasic inactivation with two linear parts is assumed to describe the inactivation trends in response to P and t. Non‐linear trends may also exist that could not be adequately captured by fitting all data by a global model	This uncertainty source may lead to over‐ or under‐estimation of HPP efficacy
Relevant factors determining the efficacy of HPP processing: microbiological factors	The physiological state of cells, which is highly dependent on culture preparation (e.g. exponential vs. stationary phase cells, or stress adapted cells) may influence the efficacy of HPP. The resistance of the strain and its physiological state are very hard to elucidate under real industrial conditions due to the variability of the strains that can contaminate the food processing environment and the food products, as well as the variety of stresses that these strains can be exposed to. This factor is particularly relevant when the pathogenic bacteria contaminate raw materials (e.g. raw milk) and are exposed to heat during manufacturing, which may confer resistance to the cell and reduce the efficacy of HPP compared to what could be observed if the pathogen would contaminate the final product during post‐processing conditions. The effect of the level of contamination before HPP is experimentally assessed through the inoculum concentration, which may be a limiting factor to quantify the number of log10 reductions caused by the HPP treatment if the contamination level is not high enough or the HPP causes a reduction of the microorganisms below the quantification limit. This adds uncertainty about the actual magnitude of the inactivation or the occurrence of pseudo‐tails within a range of concentrations of 2.4 log10 units, between the detection limit 1 cell/25 g (i.e. −1.4 log10 CFU/g) up to just below the quantification limit (i.e. 1 log10 CFU/g). Analytical methods (culture media) to count L. monocytogenes after HPP, can make the recovery of the sublethally injured cells difficult, leading to an overestimation of the HPP efficacy. Many data used in the present assessment used selective agar (e.g. PALCAM, Oxford and more recently ALOA). The overestimation is less relevant since ALOA became of extensive use, as pressure injured L. monocytogenes are able to grow when directly plated on ALOA without previous repair incubation period (Jantzen et al., 2006).	This uncertainty source may lead to over‐ or under‐estimation of HPP efficacy. However, the impact of this uncertainty is estimated to be low‐moderate as a high number of studies were included in the assessment covering a relatively wide number of different strains at stationary phase, which represents the worst‐case scenario of HPP resistance. The majority of log10 reduction data has been obtained from high inoculum level, allowing a proper quantitative determination of the survivors. Though some studies have been used selective media for enumeration (that may fail to recover sublethally injured cells), in the case of Listeria monocytogenes the use by many studies of ALOA and ALOA‐like media is known to recover HPP injured cells (Jantzen et al., 2006; Morales et al., 2006). This uncertainty may result in an over‐estimation of the HPP efficacy.
Relevant factors determining the efficacy of HPP: Impact of fat content from different animal species (few inconclusive studies were available)	Fat could play a protective role on microorganisms against HPP	This source would lead to under‐estimation of HPP efficacy.
Relevant factors determining the efficacy of HPP: processing parameters (P, t, T)	Some extrinsic factors, mainly temperature during processing (once at the target pressure) is not reported during HPP cycle in many studies. Besides the uncertain compression heating, the impact of temperatures below 45°C has not been studied. Most records did not specify whether the CUT was included or not in the period during which, HPP inactivation was assessed. The compression stage may cause some inactivation, depending on the microorganism, the product and the pressure come‐up rate.	This uncertainty source may lead to over‐ or under‐estimation of HPP efficacy. When the considered duration of the HPP treatment considers also CUT, the assessed log10 reductions practically refer to shorted holding time and thus, the killing effect is underestimated.

ALOA: Agar Listeria Ottavani & Agosti; CUT: come‐up time; HPP: high‐pressure processing; P: pressure; t: time; T: temperature.