P3-71 Comparison of Predictive Models for Growth of Shiga Toxin-producing Escherichia coli (STEC) in Ground Beef

Wednesday, August 6, 2014
Exhibit Hall D (Indiana Convention Center)
Harshavardhan Thippareddi, University of Nebraska-Lincoln, Lincoln, NE
Introduction: Foodborne illness caused by Shiga toxin-producing E. coli (STECs) has been a major concern. While predictive models for the growth of E. coli O157 and other E. coli are available, they have not been validated for growth of non-O157 STEC serogroups. 

Purpose: The objective of the research was to evaluate the performance of Combase model (for E. coli O157:H7) and Ratkowsky square-root model recommended by Huang for the prediction of the growth of STECs in ground beef (Huang model). 

Methods: Irradiated ground beef (fat, 73/27; and lean, 93/7) was inoculated with ca. 2 CFU/g of five-strain cocktails of O157:H7, O26, O45, O103, O104, O111, O121, O145, and six-serovar STEC cocktails from USDA-ARS and ATCC, respectively. Inoculated ground beef was packaged in vacuum bags, placed in programmable water baths to follow time-varying temperature profiles (sinusoidal; low or high temperature profiles) for a total of 300h and 25h, respectively. Growth data were collected and fitted into the models described by Huang and Combase. Model performance indicators (Root mean squared error, RMSE; bias factor, BF; and accuracy factor, AF) were calculated to evaluate the performance of these models. 

Results: The Huang model underestimated the growth of all STECs for both temperature profiles; the average RMSE for fat and lean meat were 1.87 and 2.14 at low temperature, and 2.39 and 2.42 at high temperature profiles. Combase model over-predicted the growth of all STECs with a mean RMSE of 1.44 and 1.60 log CFU/g for fat and lean ground beef, respectively. Besides, Combase model cannot predict the growth of E. coli O157:H7 at temperatures below 10°C where growth of STEC can occur.

Significance: None of the two models precisely predicted the growth of STECs in ground beef. However, Combase could be a conservative option given the limited availability of predictive models for STECs.