Skip to main content
. Author manuscript; available in PMC: 2018 Jul 1.
Published in final edited form as: Kidney Int. 2017 Mar 17;92(1):238–247. doi: 10.1016/j.kint.2017.01.024

Table 5.

The comparison of Harrell’ C statistics of the adjusted Cox proportional hazard model and continuous net reclassification improvement (NRI) of baseline and enhanced model for predicting risk of death.

Models Harrell’C statistics Continuous NRI


C-statistic P value† Overall (%) with 95% CI
Base model: age, sex, race, comorbidities* 0.63 reference

Base model + low muscle mass, weakness, and slowness as dichotomous variables

 low muscle mass by height2 0.65 0.17 4.7 (−22.9, 29.2)
 low muscle mass by BW 0.63 0.56 −6.0 (−33.5, 27.2)
 low muscle mass by BSA 0.64 0.30 13.8 (−34.6, 35.0)
 low muscle mass by BMI 0.63 0.49 15.4 (−28.5, 42.4)
 weakness 0.66 0.56 33.7 (9.8, 62.7)
 slowness 0.68 0.004 50.5 (24.3, 73.0)

Base model + sarcopenia, defined as low muscle mass combined with weakness

 low muscle mass by height2 and weakness 0.65 0.31 −8.4 (−28.5, 20)
 low muscle mass by BW and weakness 0.64 0.28 2.8 (−32.1, 27.5)
 low muscle mass by BSA and weakness 0.65 0.41 14.3 (−25.5, 37.6)
 low muscle mass by BMI and weakness 0.64 0.18 1.7 (−26.7, 31.5)

Base model + sarcopenia, defined as low muscle mass combined with slowness

 low muscle mass by height2 and slowness 0.65 0.28 −7.1 (−28.1, 19.7)
 low muscle mass by BW and slowness 0.64 0.23 0.2 (−28.1, 32.0)
 low muscle mass by BSA and slowness 0.64 0.33 5.2 (−25.5, 31.1)
 low muscle mass by BMI and slowness 0.64 0.74 11.8 (−10.5, 39.0)

Base model + total muscle mass, handgrip strength, and gait speed as continuous variables

 muscle mass/height2 0.64 0.28 9.9 (−25.9, 34.2)
 muscle mass/BW 0.63 0.15 −28.5 (−32.7, 35.7)
 muscle mass/BSA 0.64 0.70 10.3 (−24.1, 38.6)
 muscle mass/BMI 0.65 0.17 −5.7 (−31.3, 31.6)
 handgrip strength 0.68 0.03 32.0 (6.8, 59.2)
 gait speed 0.66 0.01 18.8 (−12.5, 49.9)

BMI, body mass index; BSA, body surface area; b.s.CI, bootstrap confidence interval

*

Comorbidities including diabetes mellitus, congestive heart failure, and coronary artery disease.

p<0.05 consider significantly different between each model and the based model.