Table 3

Performance of two extended RUBY models in predicting AGT and compared with NCDRS at optimal cutoffs in the Better Health for Better Hong Kong_Hong Kong Family Diabetes Study cohort (n=1054)

ModelAUROC (95% CI)SensitivitySpecificityPPVNPVNNS
Original RUBY score0.698 (0.662 to 0.733)0.4850.7850.4240.8222.4
Extended RUBY model 10.756 (0.722 to 0.791)0.5880.7940.4840.8552.1
Extended RUBY model 20.778 (0.744 to 0.812)0.6350.8050.5160.8711.9
NCDRS0.648 (0.609 to 0.687)0.6030.6930.3890.8432.6
  • Optimal risk cut-offs: original RUBY score (12 out of 30), extended RUBY model 1 (risk threshold, 29.4%), and extended RUBY model 2 (risk threshold, 29.5%) for predicting AGT were determined by the Youden Index. NCDRS with cut-off points 27 for predicting pre-diabetes and diabetes based on ADA criteria. Extended RUBY models used corresponding absolute risk in percentage as the optimal risk cutoffs while NCDRS and original RUBY score provided risk score.

  • ADA, American Diabetes Association; AGT, abnormal glucose tolerance; AUROC, area under receiver operating characteristics; NCDRS, New Chinese Diabetes Risk Score; NNS, number needed to screen; NPV, negative predictive value; PPV, positive predictive value; RUBY, Risk-Understanding-By-Yourself.