Table 3.

Unadjusted and multivariable-adjusted logistic regression analyses of the combined exposures for PC risk.

Exposure		Group		Univariable		Multivariable
		PC	HC	OR (95% CI)	p	OR (95% CI)	p a
Low DII	with FBG <126 mg/dL	10	134	1 (Ref)		1 (Ref)
	with FBG ≥126 mg/dL	7	16	5.86 (1.96~17.6)	0.001	7.50 (2.31~24.4)	0.0008
High DII	with FBG <126 mg/dL	16	119	1.80 (0.79~4.12)	0.16	1.76 (0.64~4.80)	0.27
	with FBG ≥126 mg/dL	21	10	28.1 (10.5~75.7)	<0.0001	32.5 (9.85~106.9)	<0.0001
p for trend					<0.0001		<0.0001
RERI/S					21.5/4.79		24.2/4.34
p for interaction					0.01		0.04
Low DII	with non-smoker	8	82	1 (Ref)		1 (Ref)	0.39
	with ex/current smoker	6	69	1.34 (0.49~3.65)	0.57	1.67 (0.51~5.46)	0.13
High DII	with non-smoker	18	76	2.43 (1.00~2.77)	0.05	2.20 (0.79~6.16)	0.02
	with ex/current smoker	20	53	3.87 (1.59~9.42)	0.002	3.88 (1.25~12.0)	0.01
p for trend					0.0009		0.01
RERI/S					1.10/1.63		1.01/1.54
p for interaction					0.58		0.76

ORs and 95% CIs were evaluated to examine the relationship between the DII score and FBG levels or smoking status for PC risk using a multivariate logistic regression analysis. The DII scores were divided into two ranges based on the median value: high and low. ^a p-Values were adjusted for sex, age, BMI, smoking status, and energy intake. The p-values for trends were computed by treating the values as continuous variables, evaluating the continuous scale, and allocating median values to every quantile. The combined effect was assessed through additive interactions using RERI/S and the p-values for multiplicative interactions from dichotomized exposures. Values of RERI > 0 and S > 1 indicate a synergistic interaction between the two factors. Abbreviations: PC, pancreatic cancer; HC, healthy control; OR, odds ratio; CI, confidence interval; DII, dietary inflammatory index; FBG, fasting blood glucose; Ref, reference value; RERI, relative excess risk due to interaction; S, synergy index.