Establishing the change in antibiotic resistance of Enterococcus faecium strains isolated from Dutch broilers by...

In this study, we investigated the change in the resistance of Enterococcus faecium strains isolated from Dutch broilers against erythromycin and virginiamycin in 1998, 1999 and 2001 by logistic regression analysis and survival analysis. The E. faecium strains were isolated from caecal samples that had been randomly collected from six slaughterhouses. Moreover, between the sample collection in 1998 and the sample collection in 1999, virginiamycin and the macrolide antibiotics (of which erythromycin is a member) have been banned in The Netherlands from use in broiler feeds as growth promoter. In the logistic regression analysis we used the internationally accepted cut-off values to determine whether bacteria were resistant or not. In the survival analysis, inhibition of bacterial growth was the event and time to event was replaced by concentration of antibiotic to event. As a consequence, changes in the growth of bacteria can be tested over an entire range of concentrations and no cut-off value for resistance has to be determined. We performed the survival analysis by use of a Cox logistic model with an odds ratio (OR) for the increase of the odds of the basic hazard rate as outcome. Both the logistic regression and the survival analyses showed that resistance to erythromycin and virginiamycin decreased during the study period. In the logistic regression model the ORs associated with the fraction of bacteria inhibited by the antibiotics in 2001 as compared to 1998 were 3.76 (2.57-5.49) for erythromycin and 11.65 (7.68-17.66) for virginiamycin. The corresponding ORs from the survival analysis were lower; 2.88 (2.21-3.76) and 2.11 (1.80-2.49), respectively. The reason for the differences between the ORs of the survival analysis and the logistic regression analysis is probably because most changes in resistance included the cut-off value and logistic regression specifically examines those changes.