Table 5.

Mixed logistic regression on the binomial outcome that a negative herd would test positive the following year for antibodies against influenza A(H1N1)pdm09 using the haemagglutination inhibition test

Number of observations (positives)	Fixed effects*	OR	95% CI	P value
	Year of test (n = 5)
174 (46)	2010	1
406 (104)	2011	1·30	0·55–3·05	0·553
358 (121)	2012	5·72	2·31–14·17	<0·001
389 (128)	2013	6·40	2·43–16·83	<0·001
	Production class (n = 4)
258 (64)	Fattening herd	1
157 (43)	Nucleus herd	6·26	1·12–34·86	0·036
155 (56)	Multiplier herd	47·95	7·75–296·83	<0·001
757 (236)	Conventional herd	9·75	2·85–33·41	<0·001
	Herd size in three quantiles (pigs)
399 (89)	Small (<350)	1
405 (130)	Medium (350–665)	6·31	2·00–19·87	0·002
469 (165)	Large (>665)	9·24	2·78–30·72	<0·001
	Constant	0·0004	0·00003–0·00612	<0·001
	Random effects†
1327 (399)	County (n = 17)
	Var(constant)	2·72	1·01–7·35
1327 (399)	County>Herd id (n = 621)
	Var (constant)	19·72	10·34–37·61

OR, Odds ratio; CI, confidence interval.

^*The three categorical fixed effects were (1) year of test, (2) production class, and (3) herd size.

^†County and herd ID were included as random effects to account for non-fixed efects associated with county and the individual herd.