Table 3. Multivariate linear regression for erythropoietin.

	Men (n = 3,395)			Women (n = 3,382)			Interaction P-value
Variable	β*	95% Confidence interval	P-value	β*	95% Confidence interval	P-value
Age (per 10 years)	0.048	0.022 to 0.074	<0.001	-0.008	-0.050 to 0.031	0.692	0.132
Waist circumference (cm)	0.006	0.003 to 0.008	<0.001	0.007	0.005 to 0.010	<0.001	0.294
Premenopausal	-	-	-	0.010	-0.069 to 0.088	0.813	-
UAE (mg/24h)†	0.010	-0.008 to 0.028	0.262	0.016	-0.007 to 0.040	0.179	0.328
Hs-C-reactive protein (mg/L)†	0.008	-0.009 to 0.024	0.371	0.010	-0.006 to 0.026	0.236	0.419
eGFR (per 10mL/min/1.73m2)	0.009	-0.008 to 0.025	0.319	-0.005	-0.021 to 0.012	0.587	0.713
[Hemoglobin (g/dL)– 14.5]2	0.063	0.053 to 0.074	<0.001	0.079	-0.072 to 0.085	<0.001	0.023
Glucose (mmol/L)	0.038	0.017 to 0.058	<0.001	0.029	0.005 to 0.053	0.017	0.866
Cholesterol (mmol/L)	-0.055	-0.078 to -0.032	<0.001	-0.041	-0.066 to -0.016	0.001	0.608
Smoking or quit <1 year	-0.078	-0.131 to -0.024	0.004	-0.058	-0.112 to -0.003	0.038	0.529
Systolic blood pressure (per 5mmHg)	0.001	-0.007 to 0.009	0.893	0.014	0.006 to 0.022	0.001	0.058
Heart rate (per 5b.p.m.)	-0.008	-0.020 to 0.004	0.168	-0.005	-0.018 to 0.008	0.454	0.367
Myocardial infarction	0.055	-0.050 to 0.160	0.303	0.114	-0.089 to 0.317	0.271	0.756
Stroke	0.063	-0.193 to 0.320	0.628	-0.107	-0.384 to 0.171	0.451	0.414
Venous thromboembolism	0.358	0.066 to 0.650	0.016	-0.041	-0.381 to 0.298	0.811	0.085

* The dependent variable erythropoietin was double log-transformed before included in the model. Therefore, a β of 1 should be interpreted as a doubling for each unit rise of the independent variable.

^† Both the dependent and independent variables were double log-transformed before included in the model. Therefore, a β of 1 should be interpreted as a doubling for each doubling of the independent variable.

UAE = Urinary Albumin Excretion, eGFR = estimated Glomerular Filtration Rate.