Dynamic glycemic exposure and trajectory profiling for incident diabetic kidney disease: A longitudinal cohort study

Table 3 Predictive performance of baseline glycated hemoglobin vs time-weighted average glycated hemoglobin for incident diabetic kidney disease, median (interquartile range)

Evaluation metric	Baseline HbA1c	TWA-HbA1c1	P value	IDI (95%CI)	P value	NRI (95%CI)	P value
Overall C-index	0.651 (0.613, 0.689)	0.724 (0.687, 0.760)	0.007
Time-dependent AUC
3 years	0.617 (0.516, 0.718)	0.692 (0.595, 0.790)	< 0.001	0.021 (0.011, 0.032)	< 0.001	0.268 (0.129, 0.407)	< 0.001
5 years	0.662 (0.601, 0.724)	0.730 (0.672, 0.788)	< 0.001	0.037 (0.012, 0.062)	< 0.001	0.297 (0.014, 0.580)	< 0.001
7 years	0.653 (0.602, 0.704)	0.726 (0.677, 0.775)	< 0.001	0.058 (0.017, 0.099)	< 0.001	0.301 (0.046, 0.557)	< 0.001

¹Baseline glycated hemoglobin refers to the single measurement at enrollment, whereas time-weighted average-glycated hemoglobin was incorporated into the models as a time-varying covariate. The P values indicate the statistical significance of the improvement in predictive performance when time-weighted average-glycated hemoglobin is used compared to baseline glycated hemoglobin.

HbA1c: Glycated hemoglobin; TWA-HbA1c: Time-weighted average glycated hemoglobin; CI: Confidence interval; IDI: Integrated discrimination improvement; NRI: Net reclassification improvement; AUC: Area under the receiver operating characteristic curve.