Determination of the Validation Frequency for Commercial UV Juice Processing Units

摘要

The CiderSure 3500 is one of the most commonly used UV juice processing units in the United States for the nonthermal processing of apple cider and fulfills the 5-log performance standard established by the federal juice HACCP regulation. However, the appropriate validation frequency of this machine's quartz tubes is currently unknown by juice processors and regulatory agencies. Presently, an annual validation is recommended by the manufacturer. Historical validation data from 1998 to 2013 of commercially used quartz tubes underwent comprehensive statistical analysis. A total of 400 tubes were validated one time, and 212 of those units were revalidated at least once over the evaluated time frame. Validations were performed at 14 mJ·cm~2 UV dose and under turbulent flow conditions. Every validation showed a greater than 5-log reduction of Escherichia coli ATCC 25922, a nonpathogenic surrogate for pathogenic E. coli O157:H7, in each of three replicates. For initial validations, a mixed-effect model with log reduction of E. coli as response was constructed (400 tubes analyzed in triplicate). The model showed that the year of analysis and the initial inoculum level significantly affected the log reduction of E. coli (P ＜ 0.0001), which on average was 7.0 ± 0.7. A quadratic relationship between the year of analysis and the response was found. Likewise, for revalidations (212 tubes analyzed in triplicate), the constructed random coefficient model showed that the year of analysis, quadratic effect of year of analysis, and initial inoculum level significantly affected the log reduction of E. coli (P ＜ 0.0001). For this model, the major source of variance was explained by the year of analysis. The models describe the UV reactor's performance over time and suggest that a validation frequency of every 3 years would be conservatively adequate during the first 8 years of quartz tube use. After that, due to the reported quadratic trend, yearly validation would be recommended.