Table 4.

Cross-sectional and longitudinal relationships between CRP (quartile) and HOMA-IR (continuous)

	Insulin resistance: beta-coefficienta per quartile of year 15 CRP
	Cross-sectional associationb					Longitudinal associationc
Adjustments	1st	2nd	3rd	4th	p value for trend	1st	2nd	3rd	4th	p value for trend
Model 1d	Ref.	0.1234	0.2193	0.3514	<0.0001	Ref.	0.0330	0.0767	0.1140	<0.0001
Model 2e	Ref.	0.1227	0.2167	0.3423	<0.0001	Ref.	0.0353	0.0792	0.1186	<0.0001
Model 3f
BMI (kg/m2)	Ref.	0.0470	0.0833	0.0937	<0.0001	Ref.	0.0225	0.0515	0.0657	0.001
WC (cm)	Ref.	0.0283	0.0454	0.0575	0.001	Ref.	0.0167	0.0388	0.0516	0.01
Estimated % fat	Ref.	0.0247	0.0488	0.0783	<0.0001	Ref.	0.0152	0.0404	0.0594	0.002
Model 4g	Ref.	0.0230	0.0274	0.0557	0.004	Ref.	0.0181	0.0308	0.0499	0.02

^aThe beta coefficient is approximately the proportion increase in the geometric mean HOMA-IR compared with the lowest quartile of CRP

^bIn cross-sectional analyses, CRP and HOMA-IR were measured at year 15

^cIn longitudinal analyses, CRP was measured at year 15 and HOMA-IR was measured at year 20

^dModel 1: adjusting for age, sex, race and study centre. Year 15 HOMA-IR was additionally included in longitudinal analysis

^eModel 2: model 1 plus year 15 smoking status, physical activity, alcohol consumption and education

^fModel 3: model 2 plus body fat measurements (BMI, WC or estimated %fat in three separate models)

^gModel 4: model 2 plus WC, F2-isoprostanes and oxidised LDL

Ref. Reference