Mathematical Approaches To Estimating Lag-Phase Duration and Growth Rate for Predicting Growth of Salmonella Serovars, Escherichia coli O157:H7, and Staphylococcus aureus in Raw Beef, Bratwurst, and Poultry

Abstract

This study was done to optimize accuracy of predicting growth of Salmonella serovars, Escherichia coli O157:H7, and Staphylococcus aureus in temperature-abused raw beef, poultry, and bratwurst (with salt but without added nitrite). Four mathematical approaches were used with experimentally determined lag-phase duration (LPD) and growth rate (GR) values to develop 12 versions of THERM (Temperature History Evaluation for Raw Meats; http://www.meathaccp.wisc.edu/THERM/calc.aspx), a computer-based tool that calculates elapsing lag phase or growth that occurs in each entered time interval and sums the results of all intervals to predict growth. Each THERM version utilized LPD values calculated by linear interpolation, quadratic equation, piecewise linear regression, or exponential decay curve and GR values calculated by linear interpolation, quadratic equation, or piecewise linear regression. Each combination of mathematical approaches for LPD and GR calculations was defined as another THERM version. Time, temperature, and pathogen level (log CFU per gram) data were obtained from 26 inoculation experiments with ground beef, pork sausages, and poultry. Time and temperature data were entered into the 12 THERM versions to obtain pathogen growth. Predicted and experimental results were qualitatively described and compared (growth defined as >0.3-log increase) or quantitatively compared. The 12 THERM versions had qualitative accuracies of 81.4 to 88.6% across 70 combinations of product, pathogen, and experiment. Quantitative accuracies within ±0.3 log CFU were obtained for 51.4 to 67.2% of the experimental combinations; 82.9 to 88.6% of the quantitative predictions were accurate or fail-safe. Piecewise linear regression or linear interpolation for calculating LPD and GR yielded the most accurate THERM performance.