Skip to main content
. Author manuscript; available in PMC: 2013 Jul 25.
Published in final edited form as: Ann Intern Med. 2012 Aug 21;157(4):242–250. doi: 10.7326/0003-4819-157-4-201208210-00004

Table 3.

Sensitivity, Specificity, Net Benefit, Net Gain, and AUC of the LLP Risk Model for Specified Risk Thresholds in Each of 3 Populations

Variable and Measure Study Population
EUELC Harvard LLPC
Absolute risk threshold*
  2.5%

    Sensitivity, % 55.2 64.7 74.3

    Specificity, % 69.8 73.1 67.4

    Accuracy, % 65.2 68.1 67.8

    Net benefit 0.0190 0.0345 0.0261

    Net gain, % 1.1 0.9 0.53

  5.0%
    Sensitivity, % 35.7 49.4 57.4

    Specificity, % 84.3 87.1 81.1

    Accuracy, % 69.1 64.7 79.8

    Net benefit 0.0127 0.0249 0.0171

    Net gain, % 3.0 2.52 2.3
  10.0%

    Sensitivity, % 14.2 25.3 27.9

    Specificity, % 94.3 95.7 92.5

    Accuracy, % 69.2 53.8 89.0

    Net benefit 0.0055 0.0134 0.0060

    Net gain, % 7.65 6.87 6.78

Discriminative performance
  AUC (95% CI) 0.67 (0.64–0.69) 0.76 (0.75–0.78) 0.82 (0.80–0.85)

AUC = area under the receiver-operating characteristic curve; EUELC = European Early Lung Cancer; = Liverpool Lung Project; = Liverpool Lung Project prospective cohort.

*

Absolute risk for lung cancer at which screening is initiated.

Gain in net benefit for LLP risk model over a screen-all strategy.