Skip to main content
. 2019 Sep 7;34(11):1055–1074. doi: 10.1007/s10654-019-00555-w

Table 4.

Outcomes of the two-category models for inferring smoking history (years of cessation time) in former smokers from blood based on 13 CpGs

Former < 5 year versus Former ≥ 5 year cessation time Former < 10 year versus Former ≥ 10 year cessation time Former < 15 year versus Former ≥ 15 year cessation time
Model building data set (N = 1332) External validation Model building data set (N = 1332) External validation Model building data set (N = 1332) External validation
Model building Fivefold cross-validation KORA (N = 652) Model building Fivefold cross-validation KORA (N = 652) Model building Fivefold cross-validation KORA (N = 652)
Accuracya (95% CI) ± SD 0.725 (0.700, 0.749) 0.715 ± 0.020 0.830 (0.799, 0.858) 0.730 (0.705, 0.753) 0.721 ± 0.029 0.799 (0.766, 0.829) 0.732 (0.707, 0.756) 0.718 ± 0.016 0.759
Specificity 0.715 0.691 ± 0.090 0.494 0.694 0.682 ± 0.063 0.471 0.663 0.644 ± 0.033 0.449
Sensitivity 0.727 0.718 ± 0.026 0.879 0.740 0.733 ± 0.026 0.900 0.767 0.756 ± 0.015 0.902
AUC 0.793 0.774 ± 0.024 0.760 0.778 0.766 ± 0.033 0.764 0.779 0.767 ± 0.020 0.754

Cross-validation analysis results are presented as mean ± standard deviation

AUC Area under the curve

aProportion accurately inferred smoking habits, 95% confidence interval (CI)