Longitudinal study of Clostridium difficile shedding in raccoons on swine farms and conservation areas in Ontario, Canada

Table 2 Univariable logistic regression model with random effects (site and animal) showing associations between Clostridium difficile status and raccoon age (adult or juvenile), sex, location type (farm or conservation area), and season (May to July or August to October)

Predictor	Sub-category (n = sample size)	% C. difficile positive (95 % CI)^a	Univariable models^b
			Odds ratio	95 % CI	Covariance parameter estimates (95 % CI)		P
					Site level	Animal level^d
Age^c	Juvenile (n = 83)	2.4 (0.3–8.4)	Referent
Age^c	Adult ( n = 360)	10.0 (7.1–13.6)	4.79	1.11–20.60	0.28 (0.04–1.8)		0.036
Location type	Conservation (n = 225)	7.1 (4.1–11.3)	Referent
Location type	Swine farm (n = 219)	10.0 (6.4–14.8)	1.42	0.57–3.59	0.21 (0.02–1.8)		0.448
Sex	Female (n = 245)	8.6 (5.4–12.8)	Referent
Sex	Male (n = 199)	8.5 (5.1–13.3)	0.93	0.46–1.88	0.27 (0.04–1.9)	0.14^e	0.831
Season	Aug. to Oct. (n = 165)	3.0 (1.0–6.9)	Referent
	May to July ( n = 279)	11.8 (8.3–16.2)	4.81	1.80–12.87	0.21 (0.06–1.8)		0.002

^aCI = confidence interval
^bSignificant differences are in bold
^cAge for 1 individual was unknown
^dThe random effect for animal was not included in the models for age, location type, and season because it explained 5.3 × 10^-31 to 4.2 × 10^-30 of the variance, and its removal from the model changed the coefficients little and did not change the model according to AIC and BIC
^eThe 95 % confidence interval was estimated to be between 3.6 × 10^-11 and 5.6 × 10⁸

ISSN: 1746-6148