TABLE 6.

Multivariable mixed-effects Poisson regression models showing associations between route and disease-specific antimicrobial use in turkey flocks (n = 334) and resistance to the number of antimicrobial classes in E. coli isolates (n = 1,317), 2016–2019.

AMUroute– and disease–specific1		Estimate	IRR (95% CI)	p-value	Random intercepts	Variances2 (std. dev)
Route	Disease
Feed	Late septicemia	0.336	1.398 (1.109–1.763)	0.004	Sampling network	(0.000)
Feed	Enteric diseases	0.400	1.492 (1.244–1.788)	<0.000	Year: Sampling network	0.033 (0.181)
Water	Enteric diseases	0.538	1.712 (1.300–2.254)	<0.000	Flock: Year: Sampling network	0.129 (0.360)
Intercept		−4.396	0.012 (0.010–0.015)	<0.000

¹AMU_{route–disease–specific} – Antimicrobial use in the model above is a binomial indicator of whether the flock used the antimicrobial via a given route for a specific disease syndrome. IRR, incidence rate ratio; CI, confidence interval.

²Four levels: isolates at the lowest, flocks at the second, years at the third, and veterinarians at the highest level. Level 1 variance was assumed to be 3.29 (Snijders and Bosker, 2011).