Table 5.
Accuracy of NLST versus the HUNT Lung Cancer Model for lung cancer (LC) diagnosis within 6 years, using the same number of screenings as NLST in the CONOR ever-smokers with complete data. As compared with NLST criteria, our model's criteria identified 103 vs 69 out of 270 cases showing an improved sensitivity (38.14% vs 25.6%, P = 0.0216) and positive predictive value (4.95% vs 3.3%, P < 0.000001), with the same specificity (95.61% vs 95.5%, P = 0.7321) and similar negative predictive value (99.6% vs. 99.5%, P = 0.95374).
Criteriaa | Participants with LC (N) | Participants without LC (N) | Participants total (N) | Predictive value |
---|---|---|---|---|
NLST criteria | 270 | 45,117 | 45,387 | |
Criteria positive | 69 TP (3·3%) | 2012 FP (96·7%) | 2081 | PPV 3·3% |
Criteria negative | 201 FN (0·5%) | 43,105 TN (99·5%) | 43,306 | NPV 99·5% |
Sensitivity | 25·6% | |||
Specificity | 95·5% | |||
HUNT Lung Cancer Model criteria | ||||
Criteria positive | 103 TP (4·95%) | 1978 FP (95·05%) | 2081 | PPV 4·95% |
Criteria negative | 167 FN (0·4%) | 43,139 TN (99·6%) | 43,306 | NPV 99·6% |
Sensitivity | 38·14% | |||
Specificity | 95·61% |
FN = false negative; FP = false positive; NPV = negative predictive value; PPV = positive predictive value; TN = true negative; TP = true positive.
NLST criteria for study entry included a history of cigarette smoking of at least 30 pack-years, age between 55 and 74 years and, for former smokers, cessation within the previous 15 years.