Table 4.

Multiple Logistic Regression Results for Ileal Disease Location.

Fctors			P-values
Age at diagnosis (>16yrs)			0.6809
Gender			0.3071
NOD2 risk allele carriage			0.1408
Cohort			0.2269
Smoking			0.0044
Elevated GM-CSF Ab			0.0108
Perianal location			0.0029
Smoking* Elevated GM-CSF Ab			0.0023
Cohort*Smoking			0.0010
Cohort*Smoking* Elevated GM-CSF Ab			0.0048
Factors	By Smoking	By Cohort	OR (95%CI)
GM-CSF Ab Elevation Yes vs. No	Yes	Brisbane	0.19 (0.04–0.81)
		NIDDK	2.67 (0.88–8.06)
	No	Brisbane	2.27 (1.21–4.26)
		NIDDK	2.25 (1.21–4.25)

A multiple logistic regression model was fitted with Ileal CD location as the response variable. Main effects for the variables age at diagnosis, gender, NOD2 allele, Cohort, Smoking, elevated GM-CSF Ab (≥ 5mcg/mL), and Perianal location were included in the model automatically, while interaction terms between these main effects were included only if significant after backward elimination. Table 4 displays the p-values for all fitted terms in the final model, as well as the estimated odds ratios for association of Ileal CD location with Elevated GM-CSF-Ab. The odds ratio estimates were broken down by smoking and cohort as the significant association was defined by interactions of elevated GM-CSF-Ab with smoking and cohort.