Abstract: Ongoing regulatory changes are increasing the need for reliable process validation methods for pathogen reduction processes involving low-moisture products; however, the reliability of various validation methods has not been evaluated. Therefore, the objective was to quantify accuracy and repeatability of four validation methods (two biologically based and two based on time- temperature models) for thermal pasteurization of almonds. Almond kernels were inoculated with Salmonella Enteritidis phage type 30 or Enterococcus faecium (NRRL B-2354) at ~108 CFU/g, equilibrated to 0.24, 0.45, 0.58, or 0.78 water activity (aw), and then heated in a pilot-scale, moist-air impingement oven (dry bulb 121, 149, or 1778C; dew point ,33.0, 69.4, 81.6, or 90.68C; vair¼2.7 m/s) to a target lethality of ~4 log. Almond surface temperatures were measured in two ways, and those temperatures were used to calculate Salmonella inactivation using a traditional (D, z) model and a modified model accounting for process humidity. Among the process validation methods, both methods based on time-temperature models had better repeatability, with replication errors approximately half those of the surrogate (E. faecium). Additionally, the modified model yielded the lowest root mean squared error in predicting Salmonella inactivation (1.1 to 1.5 log CFU/g); in contrast, E. faecium yielded a root mean squared error of 1.2 to 1.6 log CFU/g, and the traditional model yielded an unacceptably high error (3.4 to 4.4 log CFU/g). Importantly, the surrogate and modified model both yielded lethality predictions that were statistically equivalent (a¼0.05) to actual Salmonella lethality. The results demonstrate the importance of methodology, aw, and process humidity when validating thermal pasteurization processes for low-moisture foods, which should help processors select and interpret validation methods to ensure product safety.
Citation: Jeong S, Marks BP, James MK. Comparing Thermal Process Validation Methods for Salmonella Inactivation on Almond Kernels. 2017;80(1):169-176. doi:10.4315/0362-028X.JFP-16-224.