Metabolic Syndrome and Its Components Predict the Risk of Type 2 Diabetes Mellitus in the Mainland Chinese: A 3-Year Cohort Study

Table 4

Three prediction models for T2DM using components of MetS.

Model	AUC (95% CI)	Cut-off value	Sensitivity (%)	Specificity (%)	PPV (%)	NPV (%)
Male
MFP	0.781 (0.762, 0.800)	−2.038	73	68	26	94
Full	0.779 (0.760, 0.799)	−2.071	74	67	26	94
Stepwise	0.779 (0.759, 0.799)	−2.064	74	67	26	94
Female
MFP	0.861 (0.837, 0.885)	−2.925	91	69	18	99
Full	0.861 (0.838, 0.885)	−3.050	90	70	18	99
Stepwise	0.860 (0.836, 0.883)	−3.067	88	69	18	99

For males: MFP model: −1.92837-0.17487(age/100)⁻² + 1.17798(PG2h ≥ 7.8 = 1) + 0.90866(T ≥ 1.7 = 1) + 0.97866(FBG ≥ 6.1 = 1) + 0.26414(hypertension = 1). Full model: −4.63421 + 0.03516age e + 0.23344(hypertension = 1) + 0.10789(overweight = 1) + 1.20251(PG2h ≥ 7.8 = 1) + 0.96130(FBG ≥ 6.1 = 1) + 0.95317(TG ≥ 1.7 = 1) + 0.00978(HDLc < 0.9 = 1). Stepwise model: −4.57730 + 0.03431age + 0.28763(hypertension = 1) + 1.20539(PG2h ≥ 7.8 = 1) + 0.96970(FBG ≥ 6.1 = 1) + 0.98210(TG ≥ 1.7 = 1). For females: MFP model: 0.98518–1.94843(age/100)⁻¹ + 1.12166(PG2h ≥ 7.8 = 1) + 1.01383(TG ≥ 1.7 = 1) + 0.82966(FBG ≥ 6.1 = 1). Full model: −6.82459 + 0.07260age −0.07548(hypertension = 1) + 0.21320(overweight = 1) + 0.82601(FBG ≥ 6.1 = 1) + 1.14685(PG2h ≥ 7.8 = 1) −0.38866(HDLc <1.0 = 1) + 1.09116(TG ≥ 1.7 = 1). Stepwise model: −6.81323+ 0.07236age + 0.90655(FBG ≥ 6.1 = 1) + 1.17439(PG2h ≥ 7.8 = 1) + 1.12417(TG ≥ 1.7 = 1). AUC: area under curve. Using bootstrap resampling (times = 500).