Discussion of a biometrical model for the evaluation of feeding, age, and animal effects on transport properties of small intestinal mucosa
Lodemann, U.; Dahms, S.; Martens, H.; Arndt, G.
Deutsche Tierärztliche Wochenschrift 111(5): 209-212
2004
ISSN/ISBN: 0341-6593 PMID: 15233341 Document Number: 580458
This paper is concerned with the question in which ways study results can depend on the choice of the statistical model and factors included in this model. This is shown using example data of a study dealing with the effects of an Enterococcus faecium as probiotic in the diet of pigs. We focused on the effects on transport properties of pig jejunum. The experimental design was the following: the sows and piglets were randomly assigned to two different feeding groups. The control group was fed a conventional diet and the experimental group was additionally supplemented with a probiotic preparation of Enterococcus faecium NCIMB 10415. The animals were divided into four age groups. Two samples of five animals of each feeding and age group were taken and mounted into conventional Ussing chambers. Glucose transport rates were measured by changes in short-circuit current (Isc) of the pig jejunum epithelium. The appropriate reference base for evaluation of effects of feeding or age on Isc is the variation between animals which are submitted to identical conditions relating to these factors. To refer explicitly to this variation a random animal effect has to be included in the statistical model of variance analysis. Otherwise the variation between animals could be underestimated. With the example data set conclusions for the factor "feeding" would be different depending on whether a random animal effect is included in the model or not.