Table 4.

Concordance probability of different multivariable models including serum uromodulin to predict all-cause mortality, kidney failure, and major cardiovascular events

Outcome	Events	Univariable	Model 1a	Model 2b	Model 3c
All-cause mortality	335/5143
Concordance probability for event within first year		0.674	0.622	0.605	0.646
Concordance probability for event within 4 years		0.573	0.589	0.582	0.555
Kidney failure	229/5143
Concordance probability for event within first year		0.744	0.796	0.699	0.706
Concordance probability for event within 4 years		0.575	0.667	0.769	0.740
MACE	417/5143
Concordance probability for event within first year		0.567	0.677	0.647	0.659
Concordance probability for event within 4 years		0.528	0.551	0.547	0.575

Concordance probability indicates the discriminatory power and predictive ability of statistical models for respective outcomes. It estimates the probability of concordance between predicted and observed outcomes. MACE, major cardiovascular event—defined as a composite of fatal cardiovascular event, nonfatal myocardial infarction, nonfatal stroke, or incident peripheral vascular disease.

^aAdjusted for age, sex, and body mass index.

^bModel 1 plus eGFR and urinary albumin-creatinine ratio.

^cModel 2 plus prevalent cardiovascular disease, diabetes and hypertension at baseline, systolic BP, diastolic BP, serum high and LDL concentration, serum C-reactive protein and phosphorus concentration, prescription of diuretics, and lipid- and BP-lowering medication.