Intervention steps for controlling E. coli contamination in wheat flours

Abstract

The frequency of recalls involving Shiga toxin-producing E. coli (STEC) – contaminated wheat flours has increased in recent years. Such events have led to the need for food safety intervention strategies for wheat flours. This is an overlooked area of food safety research in flour milling. To better understand the issue, the initial objective was to characterize the transfer of E. coli into the mill fractions and equipment through consecutive milling steps. To evaluate this, 40 batches (900 g / batch) of wheat were prepared and split into inoculated (n = 20) and non-inoculated (n = 20) groups. Inoculated wheat (~3 log CFU/g) was prepared by tempering (16% moisture, 24 h) wheat using an E. coli cocktail (ATCC 1427, 1429, 1430, and 1431) while non-inoculated wheat was tempered with sterile distilled water. The inoculated wheat batches were milled first (20 batches) followed by the non-inoculated batches (20 batches) using a Chopin lab-scale roller mill (2 break, 1 sizing, and 3 reduction rolls). E. coli counts of the mill surfaces ( log CFU/ 100 cm²), flour (break, sizing, reduction, and straight-grade flour), and non-flour (bran, fine bran, shorts, rough middling, and fine middling) fractions (log CFU/g) were enumerated after milling. E. coli counts for each flour fraction were then plotted against the cumulative wheat quantity milled (kg). A best-fit model describing the E. coli counts trend for each flour fraction was then selected. Results show that E. coli counts were higher (P ≤ 0.05) on the rolls (break and reduction rolls), feeders, hoppers, and break sifter surfaces after the inoculated wheat milling. In the initial milling batch, the bran, fine bran, rough middling fractions (2.3 – 2.7 log CFU/g) had higher (P ≤ 0.05) E. coli counts than all the flour fractions (1 – 1.9 log CFU/g). E. coli counts of all flour (3.4 – 3.9 log CFU/g) and non-flour (3.7 – 4.2 log CFU/g) fractions from the last inoculated wheat batch milled were not different (P > 0.05). For the non-inoculated milling sample, E. coli surface counts declined after milling (0.4 – 1.9 log CFU/ 100 cm²). E. coli counts (log CFU/g) of all flour and non-flour fractions declined to levels below detection limit (< 1 log CFU/g) as the quantity of wheat milled increased. Break, sizing, reduction, and straight-grade flours (n = 20 / fraction) obtained from non-inoculated wheat were positive for E. coli. The model describing the E. coli transfer to flour fractions (break, sizing, reduction, and straight-grade) as a function of wheat quantity milled (kg) had good fit (P < 0.001; se < 0.43). Results from this study could provide insights to mill managers on how E. coli contaminates equipment and mill fractions produced in wheat milling operations. The second objective evaluated the efficacy of adding sodium bisulfate (SBS) to the tempering water to reduce STEC O121 and O26 loads of wheat before milling. The minimum inhibitory concentration (MIC) of SBS was evaluated. Wheat samples were inoculated (~6 log CFU/g) with E. coli O121 (ATCC 2219) or O26 (ATCC 2196) and tempered (17%, 24 h) with increasing SBS concentrations (0, 0.5, 0.75, 1.0, 1.25, and 1.5% SBS). Tempering concentrations that resulted in ≥ 3 log reductions were evaluated for their effects on flour quality. The MIC of SBS against both E. coli serotypes was 0.32% w/v. After tempering, the 1.25 (O121 – 3.5 logs; O26 – 3.2 logs) and 1.5% SBS (> 4 logs for O121 and O26) resulted in wheat STEC load reductions (P ≤ 0.05). The concentrations of 0, 1.25, and 1.5% SBS were used for subsequent flour quality evaluations. Wheat flours produced with 1.25 and 1.5% SBS were more acidic (pH = 5.41 – 5.60; P ≤ 0.05) and had comparable composition, size, pasting, and baking quality to the control treatment. Overall, wheat grains and milling equipment were observed to be viable routes of E. coli contamination of wheat flours and SBS tempering was a feasible intervention step for controlling STEC contamination in wheat prior to milling.