Validity of self-reported smoking status, NHANES III

Objectives: (1) To describe the age-, race-, and gender-specific validity of self-reported non-smoking status using cotinine-determined smoking status as the gold standard. (2) To quantify the association between age-specific misclassification of smoking status and race-gender categories. Methods: 15,182 adults 18+ years of age were identified in the National Health and Nutrition Examination Survey III. Misclassification of smoking status was defined as cotinine-determined smokers self-reported as non-smokers. The main outcome measure was the age-specific odds ratios (OR) and 95% confidence interval (CI) for the association between misclassification of smoking status and race-gender categories. The ORs were calculated using multiple logistic regression modeling, taking into account the complex survey design and sample weights. Results: The overall misclassification of cotinine-determined smokers who self-reported as non-smokers was 8.0% with the proportion increasing from 5.8% of 18-34 year olds to 24.5% of 75+ year olds. Misclassification of smoking status ranged from 0.0% for 75+ year old Mexican-American (MA) females to 67.8% for 75+ year old Non-Hispanic Black (NHB) females. Among 18-34 year olds, NHB and MA females were more likely to be misclassified (OR=3.62; 95%CI: 2.03-6.47 and OR=6.76; 3.16-14.49, respectively) than Non-Hispanic White (NHW) females; MA females were more likely to be misclassified (OR=2.34; 1.10-4.98) than MA males; NHB and MA males were more likely to be misclassified (OR=3.44; 1.55-7.65 and OR=2,70; 1.30-5.62, respectively) than NHW males. Among 75+ year olds, NHB females were more likely to be misclassified (OR=8.90; 2.07-38.25) than NHW females; and NHB females were more likely to be misclassified (OR=21.43; 4.28-107.23) than NHB males. Conclusions: Validity of self-reported smoking was associated with race-gender for adults in age-specific categories. This misclassification of cotinine-determined smokers based on self-reported non-smoking status suggests caution in interpreting such survey data. Research Support: NIH/NIDCR 1K08DE016031-01A1, 7K08DE016031-02