Nonsteroidal anti-inflammatory drug therapy for neonatal calf diarrhea complex: Effects on calf performance

Todd, C. G.; Millman, S. T.; McKnight, D. R.; Duffield, T. F.; Leslie, K. E.

Body weight at weaning was evaluated using a GLM. Kaplan-Meier survival function estimates were used to estimate the median time to weaning for each treatment group. A Cox proportional hazards regression model was constructed to examine the effect of treatment on time to weaning, while controlling for the effects of significant covariates. The exact method was used to handle ties between calves that were weaned on the same day of age. Table 1 contains the predictor variables considered for inclusion in the multivariable models. Correlations between all predictor variables were examined to iden-tify highly collinear relationships. If any 2 predictor variables were identified as strongly collinear, the more biologically relevant predictor was chosen for inclusion in the model. The assumption of linearity between continuous predictor variables and each outcome was assessed by visual examination of smoothed scatter plots and the introduction of quadratic terms into the models. The full model was reduced using a manual backward elimination procedure. Predictor variables with P-values less than 0.05 were retained in the final model. Confounding was evaluated by removing individual predictor variables from the model and then determining the change in model coefficients. Any nonsignificant predictor variable causing greater than a 30% change in model coefficients on removal was considered a confounding variable and was included in the final model. Two-way interactions between MEL treatment and significant main effect predictor variables were evaluated by addition of product terms to the model. Statistically significant 2-way interaction terms (P < 0.05) were included in the final model. When significant interactions with treatment were identified, the data were stratified on the variable that was causing the effect modification and reanalyzed.