Table 3. Performances of the four models with different combinations of predictive variables in the derivation and validation cohorts.

Variables	Clinical model	Clinical-pathological model	Clinical-medication model	Full model
Age				0.998
Pathological grade		1.270*		1.300*
Cystatin C	2.728*	2.731*	2.305*	2.663*
eGFR	0.970*	0.972*	0.233	0.973
BNP	1.059*	1.057*	1.476	1.055*
Log urinary ACR (mg/g)	1.077*	1.066*	1.517*	1.066
RASB use			0.841*	0.768
C-statistics
Derivation cohort	0.864	0.865	0.842	0.866
Validation cohort	0.870	0.876	0.849	0.875
Hosmer-Lemeshow test P	0.558	0.909	0.418	0.623
AIC	332.73	332.53	354.44	336.17

Notes.

ACR

Albumin-to-creatinine ratio

eGFR

Estimated glomerular filtration rate

RASB

renin-angiotensin system blocker

AIC

Akaike information criterion

^*statistically significant at P < 0.05.