Herd and cow characteristics affecting the odds of veterinary treatment for disease – a multilevel analysis

Table 4 Explanatory variables^a significantly associated with veterinary treatment (yes = 1/no = 0), given a disease event (two-level logistic model) for young animals using different estimating algorithms.

		LL^b		PQL ^c		MCMC ^d
Variables	Categories	Estimate	SE	Estimate	SE	Estimate	SE
Fixed part:
Intercept		-4.51	0.76	-4.71	0.82	-5.01	0.84
Age	< 2	0		0		0
	2 to 15	0.62	0.58	0.68	0.61	0.73	0.62
	> 15	2.02	0.71	2.20	0.70	2.36	0.76
Another animal with an event at the same date	Yes	1.30	0.48	1.38	0.50	1.47	0.52
	No	0		0		0
Disease complex	Lameness disorders	2.43	0.81	2.40	0.85	2.55	0.88
	Other disorders	1.80	0.63	1.84	0.67	1.95	0.68
	Cough	1.48	0.72	1.49	0.77	1.60	0.79
	Gastro-intestinal disorders	0		0		0
Good record-keeping ability	Yes	0		0		0
	No	2.34	0.95	2.45	1.08	2.64	1.18
Random part:
Herd		1.39	0.89	2.02	0.96	2.81	1.86

^a Study month was identified as a confounder and was therefore also included in the model, although not statistically significant and thus not presented in the table.
^b Log likelihood.
^c Second-order penalized quasi-likelihood (PQL) estimates with restricted iterative generalised square algorithm.
^d Markov-chain Monte Carlo (MCMC) estimates.

ISSN: 1751-0147