Skip to main content
. 2018 Aug 16;17:188. doi: 10.1186/s12944-018-0842-1

Table 3.

The AUCs’ performance for distinguishing subjects with isolated low HDL-C phenotype in gender-based analysis

Models AUC (95%CI) p value Cut-off value Specificity Sensitivity
Male subjects
 miR-222-3p 0.671 (0.566, 0.730) 0.002 0.474 0.667 0.622
 miR-221-3p/miR-222-3p ratio 0.731 (0.626, 0.836) < 0.001 0.564 0.800 0.622
 Clinical model 0.759 (0.659, 0.859) < 0.001 0.630 0.867 0.556
 Clinical model + miR-222-3pa 0.813 (0.721, 0.904) < 0.001 0.522 0.800 0.778
 Clinical model + miR-221-3p/miR-222-3p ratioa 0.851 (0.770, 0.933)c < 0.001 0.430 0.756 0.889
Female subjects
 miR-222-3p 0.623 (0.501, 0.745) 0.026 0.600 0.878 0.417
 miR-221-3p/miR-222-3p ratio 0.624 (0.503, 0.745) 0.025 0.566 0.780 0.512
 Clinical model 0.808 (0.714, 0.901) < 0.001 0.548 0.854 0.674
 Clinical model + miR-222-3pb 0.847 (0.761, 0.934) < 0.001 0.421 0.780 0.837
 Clinical model + miR-221-3p/miR-222-3p ratiob 0.841 (0.756, 0.926) < 0.001 0.556 0.853 0.721

AUC area under the curve. Best threshold of Cut-off value were estimated using youden method

a, b The AUC, sensitivities and specificities of the combination model of either miR-222-3p or miR-221-3p/miR-222-3p ratio and clinical indexes were compared with that of the predictive clinical model for male and female subjects, respectively

cp < 0.05