Skip to main content

Table 6 Validation statistics for each models in the THIN and QResearch® validation cohorts

From: Predicting the risk of Chronic Kidney Disease in Men and Women in England and Wales: prospective derivation and external validation of the QKidney®Scores

  

moderate-severe CKD

End Stage Kidney Failure

Women

THIN cohort

  
 

R2 statistic (%)

56.19 (55.67 to 56.70)

54.02 (51.67 to 56.37)

 

D statistic

2.32 (2.30 to 2.34)

2.22 (2.11 to 2.32)

 

ROC statistic

0.875 (0.872 to 0.877)

0.818 (0.803 to 0.833)

 

QResearch cohort

  
 

R2 statistic (%)

56.45 (55.40 to 57.50)

55.39 (0.52.59 to 58.18)

 

D statistic

2.33 (2.28 to 2.40)

2.28 (2.15 to 2.41)

 

ROC statistic

0.877 (0.873 to 0.880)

0.843 (0.825 to 0.860)

Men

THIN cohort

  
 

R2 statistic (%)

57.41 (54.56 to 60.27)

52.86 (50.55 to 55.17)

 

D statistic

2.38 (2.24 to 2.51)

2.17 (2.07 to 2.27)

 

ROC statistic

0.875 (0.873 to 0.878)

0.839 (0.827 to 0.850)

 

QResearch cohort

  
 

R2 statistic (%)

58.29 (55.31 to 61.26)

56.65 (53.94 to 59.35)

 

D statistic

2.42 (2.28 to 2.56)

2.34 (2.21 to 2.47)

 

ROC statistic

0.878 (0.874 to 0.882)

0.846 (0.829 to 0.862)

  1. Notes on understanding validation statistics:
  2. R2 statistic shows explained variation - higher values indicate more variation is explained
  3. ROC statistic is a measure of discrimination - higher values indicate better discrimination
  4. D statistic is a measure of discrimination - higher values indicate better discrimination and an increase of 0.1 or more over other risk prediction models is a good marker of improved prognostic separation