High-density lipoprotein subclasses and cardiovascular disease and mortality in type 2 diabetes: analysis from the Hong Kong Diabetes Biobank

Table 3 Predictive performances of small HDL-P for CVD and all-cause mortality over RECODe model

	Incident CVD	All-cause mortality
RECODe model, C-statistic (95% CI)	0.760 (0.715, 0.805)	0.829 (0.789, 0.868)
+ Small HDL-P, C-statistic (95% CI)	0.770 (0.726, 0.813)	0.863 (0.826, 0.900)
Change in C-statistic (95% CI)	0.010 (− 0.001, 0.029)	0.034 (0.013, 0.055)
IDI for 5-year risk (95% CI)	0.017 (0.003, 0.038)	0.052 (0.014, 0.103)
Categorical NRI for 5-year risk (95% CI)
Case	− 0.041 (− 0.056, 0.087)	0.118 (− 0.026, 0.208)
Noncase	0.026 (− 0.009, 0.058)	0.038 (0.013, 0.067)
Overall	− 0.015 (− 0.047, 0.126)	0.156 (0.006, 0.252)
Continuous NRI for 5-year risk (95% CI)
Case	0.183 (0.052, 0.340)	0.224 (0.034, 0.421)
Non-case	0.099 (− 0.006, 0.205)	0.348 (0.163, 0.485)
Overall	0.282 (0.088, 0.486)	0.571 (0.246, 0.851)

95% CI estimated by 1000 times bootstrapping
Risk cutoffs for categorical NRI: 5% and 10%
RECODe model: age, sex, current smoking, systolic blood pressure, glycated hemoglobin, total cholesterol, high-density lipoprotein cholesterol, estimated glomerular filtration rate, ln (urinary albumin-creatinine ratio), antihypertensive drugs, statins and CVD history (for all-cause mortality)
CI confidence interval; CVD cardiovascular disease; HDL-P high-density lipoprotein particles; IDI integrated discrimination improvement; NRI net reclassification improvement

ISSN: 1475-2840