Table 4.
Contrast | Difference in AUC (95% CI) | P-value | IDI (95% CI) | P-value | NRI (95% CI) | P-value |
---|---|---|---|---|---|---|
Training set | ||||||
Model1–Model7 | 0 | – | <−0.01 (<−0.01, <0.01) | 0.170 | −0.04 (−0.44, 0.35) | 0.840 |
Model2–Model7 | −0.05 (−0.10, <0.01) | 0.053 | −0.10 (−0.15, −0.04) | <0.001 | −0.69 (−1.06, −0.32) | <0.001 |
Model3–Model7 | −0.06 (−0.12, −0.01) | 0.021 | −0.11 (−0.16, −0.05) | <0.001 | −0.61 (−0.99, −0.24) | 0.002 |
Model4–Model7 | −0.06 (−0.12, 0.01) | 0.096 | −0.09 (−0.16, −0.02) | 0.011 | −0.41 (−0.80, −0.02) | 0.043 |
Model5–Model7 | −0.07 (−0.13, −0.01) | 0.027 | −0.12 (−0.18, −0.05) | <0.001 | −0.57 (−0.95, −0.19) | 0.005 |
Model6–Model7 | −0.01 (−0.05, 0.02) | 0.447 | −0.03 (−0.07, 0.02) | 0.228 | −0.29 (−0.68, 0.11) | 0.157 |
Testing set | ||||||
Model1–Model6 | 0 | – | <−0.01 (<−0.01, <−0.01) | <0.001 | −1.10 (−1.61, 0.59) | <0.001 |
Model2–Model6 | −0.04 (−0.16, 0.09) | 0.560 | −0.09 (−0.21, 0.03) | 0.134 | −0.50 (−1.09, 0.09) | 0.114 |
Model3–Model6 | −0.08 (−0.18, 0.03) | 0.161 | −0.11 (−0.22, <0.01) | 0.057 | −0.80 (−1.36, −0.24) | 0.011 |
Model4–Model6 | 0.03 (−0.08, 0.13) | 0.606 | 0.06 (−0.08, 0.20) | 0.394 | 0.60 (<0.01, 1.19) | 0.058 |
Model5–Model6 | −0.07 (−0.22, 0.08) | 0.367 | −0.14 (−0.29, 0.01) | 0.074 | −0.40 (−1.00, 0.20) | 0.206 |
Model6–Model7 | 0.03 (−0.03, 0.09) | 0.304 | 0.04 (<0.01, 0.16) | 0.033 | 1.00 (0.48, 1.52) | 0.002 |
Model 1: LA/total n-3 PUFAs + systolic BP + duration of diabetes; Model 2: GLA/total n-3 PUFAs + systolic BP + duration of diabetes; Model 3: EDA/total n-3 PUFAs + systolic BP + duration of diabetes; Model 4: DGLA/total n-3 PUFAs + systolic BP + duration of diabetes; Model 5: AA/total n-3 PUFAs + systolic BP + duration of diabetes; Model 6: the 1st PC + the 2nd PC + systolic BP + duration of diabetes; Model 7: Total n-6/n-3 PUFAs + systolic BP + duration of diabetes.
AUC, the area under the receiver-operating-characteristic (ROC) curve; IDI, the integrated discrimination improvement; NRI, the net reclassification improvement.