Table 2

AUROC for predicting the risk of adverse outcomes in survival models with clinical and sociodemographic variables with and without simple count and severity-weighted scores (training data set)

All-cause mortalityAny-cause hospitalizationCV
hospitalization
Diabetes-related hospitalizationClustered CV/diabetes-related hospitalizationHypoglycemia hospitalizationCV procedures
Model 10.75280.62310.70330.62490.63290.69970.6665
Model 2
CU score
(SWU score)
0.7570
(0.7556)
0.6298
(0.6299)
0.7281
(0.7169)
0.6324
(0.6323)
0.6445
(0.6433)
0.7052
(0.7056)
0.6858
(0.6778)
Model 2
C10 score
(SW10 score)
0.7563
(0.7553)
0.6288
(0.6291)
0.7267
(0.7162)
0.6312
(0.6315)
0.6427
(0.6422)
0.7042
(0.7047)
0.6850
(0.6773)
Model 2
C5 score
(SW5 score)
0.7553
(0.7548)
0.6274
(0.6280)
0.7226
(0.7138)
0.6297
(0.6302)
0.6404
(0.6404)
0.7026
(0.7036)
0.6833
(0.6759)
Model 30.75290.62300.70570.62540.63290.71040.6723
Model 4
CU score
(SWU score)
0.7570
(0.7554)
0.6293
(0.6293)
0.7298
(0.7185)
0.6327
(0.6325)
0.6442
(0.6431)
0.7162
(0.7163)
0.6903
(0.6825)
Model 4
C10 score
(SW10 score)
0.7563
(0.7552)
0.6283
(0.6286)
0.7288
(0.7179)
0.6316
(0.6317)
0.6425
(0.6420)
0.7151
(0.7155)
0.6901
(0.6823)
Model 4
C5 score
(SW5 score)
0.7553
(0.7547)
0.6270
(0.6276)
0.7247
(0.7156)
0.6302
(0.6306)
0.6404
(0.6403)
0.7136
(0.7144)
0.6885
(0.6809)
  • Model 1: age, gender, deprivation, and ethnicity.

  • Model 2: model 1+ severity score.

  • Model 3: age, gender, deprivation, ethnicity, and HbA1c (glycated hemoglobin).

  • Model 4: model 3 + severity score.

  • Simple count (C) score measured using unlimited (CU), 10-year (C10), and 5-year (C5) look-back windows.

  • Severity-weighted (SW) score measured using unlimited (SWU), 10-year (SW10), and 5-year (SW5) look-back windows.

  • AUROC, area under the receiver operating characteristics curve; CV, cardiovascular.