A predictive model with repeated measures analysis of Staphylococcus aureus growth data

• Published in: Food microbiology 2003-04, Vol.20 (2), p.139-147
• Main Authors: McCann, T.L., Eifert, J.D., Gennings, C., Schilling, M.W., Carter, W.H.
• Format: Article
• Language: English
• Subjects: Bacteriology; Biological and medical sciences; Food industries; Food microbiology; Fundamental and applied biological sciences. Psychology; Growth; Growth, nutrition, cell differenciation; Microbiology; Predictive model; Repeated measures; Staphylococcus aureus
• ISSN: 0740-0020; 1095-9998
• DOI: 10.1016/S0740-0020(02)00141-7

Food microbiologists usually assume that multiple growth measurements collected from the same broth culture flask are independent of one another. However, these data should be considered repeated measures data since multiple measurements of a response variable are taken on the same experimental unit over time. These data are likely to be correlated, especially if the observation times are not randomized. A reparameterized Gompertz model was developed to predict the growth of Staphylococcus aureus 196E as a function of temperature, pH, and NaCl concentration, over time. Since the growth responses were measured repeatedly from each experimental unit (broth in flask), the need for a repeated measures approach to modeling is suggested. This approach was compared to the traditional approach of assuming independence among growth measurements. Each model (dependence or independence) adequately predicted the growth response of S. aureus 196E over the range of the growth factor levels tested. However, the repeated measures approach, where there are random components as well as an exponential spatial covariance structure, proved to be more appropriate than the independence model, where there are no random components and the covariance structure is the simple structure. Additionally, the likelihood ratio test provided further evidence that the dependence model was best suited for these data.