Serum cotinine verification of self-reported smoking status among adults eligible for lung cancer screening in the 1999-2018 National Health and Nutrition Examination Survey.

BACKGROUND:Self-reported smoking behavior is widely used in identifying the eligibility for lung cancer screening (LCS). In clinical trials, self-reported smoking status was shown to be a reliable measure, while its effectiveness outside of trial settings remains to be clarified. We aimed to verify self-reported smoking status with serum cotinine levels among LCS-eligible adults from the National Health and Nutrition Examination Survey (NHANES). METHODS:From the combined 1999-2018 NHANES data, we selected adults (aged ≥18 years) with complete data on serum cotinine and smoking behavior, and determined their LCS eligibility according to the United States Preventive Services Task Force recommendation. Using established race/ethnicity specific cotinine cutpoints as the gold standard, we verified self-reported current smoking status with five measures: sensitivity (Se), specificity (Sp), positive predictive value (PPV), negative predictive value (NPV), and reliability Kappa. We also compared these performance measures in subgroups stratified by sex, race/ethnicity, education, family income to poverty ratio, health insurance, and secondhand smoking exposure (SHS) at home. All analyses took into account the complex survey design. RESULTS:Approximately 9.3% (n = 2335, equivalent to 8.82 million weighted population) of all adults who currently smoke or formerly smoked were eligible for LCS. The prevalence of current smoking was 52.6% and 60.8% based on self-report and cotinine, respectively. The Se and Sp were 86.4% (95%CI: 83.9%-88.9%) and 99.7% (95%CI: 99.4%-100%), respectively; PPV and NPV were 99.8% (95%CI: 99.6%-100%) and 82.6% (95%CI: 79.4%-85.7%), respectively; and Kappa was 0.83 (95%CI: 0.80-0.86). The reliability Kappa was higher among females than males (0.87 (95%CI: 0.82-0.93)) vs 0.80 (95%CI: 0.77-0.84), the lowest among non-Hispanic white (0.82 (95%CI: 0.78-0.86)), and higher among those with SHS (0.72 (95%CI: 0.63-0.80) vs (0.68 (95%CI: 0.61-0.76)). CONCLUSION:Self-reported smoking status is reasonably reliable among adults with high risk for developing lung cancer in the general population.