Toxoplasma gondii in small exotic felids from zoos in Europe and the Middle East: serological prevalence and risk factors

Table 5 Fixed effects in the final optimized generalized linear mixed models to determine potential risk factors for T. gondii-seropositivity in wild felids in human care. Modeling was performed assuming random effects by the variable “Zoo”

Model (AIC, model fit)	Category	Variable	Odds ratio (95% confidence interval)	z value	P-value
15 (276.8)		(Intercept)	16.82 (2.36–120)	2.818	0.005**
	Individual	Age	1.232 (1.13–1.34)	4.725	< 0.001***
	Food-related	Cattle: fresh (ref.)
		Cattle: fresh/frozen	0.507 (0.14–1.85)	− 1.028	0.304
		Cattle: frozen	0.143 (0.04–0.57)	− 2.756	0.006**
		Cattle: no	0.397 (0.12–1.37)	− 1.462	0.144
	Housing	Outdoor housing fenced in on all sides: no (ref.)
		Outdoor housing fenced in on all sides: yes	0.261 (0.08–0.88)	− 2.166	0.030*
	Housing	Mesh size more > 5 cm (ref.)
		Mesh size < 2 cm	0.547 (0.14–2.20)	− 0.849	0.396
		Mesh size 2–5 cm	0.317 (0.10–0.99)	− 1.982	0.047*
	Hygiene	Wearing gloves: no (ref.)
		Wearing gloves: yes	0.419 (0.20–0.90)	− 2.239	0.025*

Notes: Optimization of modeling was started with a full model [including all variables with a statistically significant effect (P < 0.1) in an initial bivariable generalized linear mixed model always including “Age” (years) in addition to the variable in question and proven independent by factor analysis] (Table 4). The Akaike information criterion (AIC) was used to characterize relative model quality. The full model had an AIC of 283.2. Optimization of the full model was done by a stepwise elimination of those variables that, if removed, did not cause an increase of AIC
Abbreviations: ref., reference
* P ≤ 0.05, ** P ≤ 0.01, *** P < 0.001

ISSN: 1756-3305