Abstract
Accumulating evidence suggests that non-daily smokers have higher disease and mortality risks than never smokers. Yet, the accuracy of self-reported non-daily cigarette smoking is poorly understood.
We examined the concordance between self-reported non-daily smoking and serum cotinine in 18,835 adult participants (20 years or older) of the 2007 to 2014 National Health and Nutrition Examination Surveys, in comparison with daily smokers and nonsmokers. We also analyzed concentrations of the urinary biomarker 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanol (NNAL) by smoking status.
In the study sample, 77.8% (14,660) reported currently not smoking (nonsmokers), 18.3% (3,446) smoked every day (daily smokers), and 3.9% (729) smoked on some days of the past month (non-daily smokers). Just 2.1% of nonsmokers had cotinine concentrations in the active smoking range (>10 ng/mL), compared with 70.4% of non-daily and 98.8% of daily smokers. Non-daily smokers reported smoking a median of 24 cigarettes per month [interquartile range (IQR) = 9–60] and had substantially higher concentrations of NNAL (median = 72.5; IQR = 14.8–211.0 pg/mL) than nonsmokers (median = 0.4; IQR = 0.4–2.1 pg/mL), although lower than daily smokers (median = 294.0; IQR = 148.0–542.0 pg/mL). Among non-daily smokers, concentrations of cotinine and NNAL were positively correlated with days and cigarettes smoked per month (P < 0.001).
We observed excellent concordance between self-reported non-daily cigarette smoking and concentrations of serum cotinine.
These results provide evidence for the validity of self-reported non-daily smoking and indicate that non-daily smokers are exposed to substantial concentrations of carcinogenic nitrosamines regardless of the low number of cigarettes they smoke per month.
Introduction
In the last decade, the proportion of non-daily cigarette smokers, also called occasional or some day smokers, has grown worldwide. In the United States, 25.4% of current smokers smoke on a non-daily basis (1) and this proportion is even higher in India (45.0%), Panama (58.6%), Mexico (61.9%), and Ecuador 68.2% (2). In addition, the emergence of dual-users of cigarettes and other tobacco products, such as e-cigarettes, suggest that the proportion of non-daily cigarette smokers may continue to increase in the future. Although some people may consider this low-level of smoking to be harmless, a growing body of literature suggests that non-daily smokers have higher mortality risks than never smokers (3–8) and a median life-expectancy that is 5 years shorter (7). Within these studies, there is evidence of elevated mortality risk among those who smoked as low as six to ten cigarettes per month and higher risks with larger number of cigarettes smoked per month (8). Furthermore, most of these non-daily smokers started smoking at a young age (7, 8), implying a long exposure to thousands of toxicants present in cigarette smoke (9).
Cotinine is the main nicotine metabolite and the most widely used biomarker of smoking exposure (10). Although self-reported smoking status is generally regarded as valid, prior studies report a higher discordance between cotinine measurements and self-reported smoking status among low-intensity smokers (11, 12). As cotinine has a relatively short half-life (8–30 hours), it may not be the best biomarker to validate non-daily smoking. Also, it is possible that non-daily smokers may not accurately report their smoking use. For instance, they could identify themselves as nonsmokers (because of their low cigarette consumption) or they may report non-daily but actually smoke every day.
Tobacco-specific nitrosamines (TNSA) are also specific biomarkers of tobacco use and are considered to be major contributors to smoking related cancers. The most studied are 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone (NNK) and N′-nitrosonornicotine (NNN), which induce tumors in nasal and oral cavities, esophagus, and lungs (13, 14). When NNK is metabolized, it is rapidly converted to 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanol (NNAL). With a half-life of 40 to 45 days, NNAL can be used as a longer-term biomarker of cigarette exposure (15) particularly for non-daily use.
Here, we leverage nationally representative data from the National Health and Nutrition Examination Survey (NHANES) to explore the distribution of two specific biomarkers of tobacco exposure: serum cotinine (shorter-term biomarker) and urinary NNAL (longer-term biomarker). The objectives of this study were (i) to evaluate the concordance between self-reported smoking status in the past 30 days (nonsmoker, non-daily, or daily smoker) and measurements of serum cotinine using previously characterized cutpoints (16, 17) in adult participants of NHANES (2007–2014), and (ii) to explore the distribution of serum cotinine and urinary NNAL concentrations among non-daily smokers according to self-reported days smoked in the past month, the number of cigarettes smoked in the past month, and how recently participants reported smoking in the past 5 days.
Materials and Methods
The study sample includes adult participants (ages 20 years or older) of the 2007 to 2008, 2009 to 2010, 2011 to 2012, and 2013 to 2014 NHANES cycles who provided information on cigarette use and had complete measurements of serum cotinine, urinary creatinine and urinary NNAL (Fig. 1). The tobacco component for NHANES collects data on history and current use of cigarettes (past 30 days) as well as the recent use (past 5 days) of cigarettes and other tobacco or nicotine products. These data are available from https://wwwn.cdc.gov/Nchs/Nhanes/. Briefly, from the 23,482 adults who answered the tobacco use questionnaire in the household interview, 12,900 had never smoked at least 100 cigarettes in their entire life (never smokers). Among the remaining 10,567 ever smokers, 5,597 reported not smoking currently (former smokers), 4,085 smoked every day (daily smokers), and 882 smoked some days of the last month (non-daily smokers). We define daily and non-daily cigarette smoking based on use in the past 30 days. Our nonsmoking comparison group included former and never smokers, as both groups are predicted to have biochemical measures of cotinine in the nonsmoking range.
In addition to these questions, participants were also asked about their use of cigarettes, as well as any product containing nicotine (pipes, cigars, smokeless tobacco, hookahs, e-cigarette, nicotine patches, gum, or any other nicotine products) in the past 5 days, on the day of their blood and urine collection at the medical examination center. At this time, they also reported on how many of the past 5 days they smoked, the number of cigarettes smoked on those days, and when they smoked their last cigarette (the day of the examination, the day before, or 3–5 days before). We used the past 5-day variables to exclude 86 cigarette smokers who also consumed any other tobacco or nicotine products and 1,111 nonsmokers who reported using any tobacco or nicotine products. We also excluded 3,432 (14.6%) participants who lacked either cotinine, NNAL, or creatinine measurements. Thus, our analysis consisted of 18,835 adults ages 20 years or older with complete measurements of serum cotinine, urinary creatinine, and NNAL.
The number of cigarettes smoked per month is derived by multiplying the number of days participants smoked in the last month by the number of cigarettes they reported smoking on those days. As smokers tended to report round numbers for their cigarette use (e.g., 5, 10), we created five categories for days smoked in the last month (1–5, 6–10, 11–15, 16–20, >21) and four categories for the total number of cigarettes per month (<10, 11–30, 31–60, >60) for non-daily smokers.
Laboratory measurements
Serum cotinine was measured by isotope dilution high-performance liquid chromatography/atmospheric pressure chemical ionization tandem mass spectrometry (ID HPLC-APCI MS/MS). Briefly, serum samples were basified and spiked with methyl-D3 cotinine as an internal standard. Cotinine was isolated from serum using supported liquid extraction, and the concentrated organic extracts were injected into a C18 HPLC column. The eluant was monitored by APCI-MS/MS, and the m/z 80 daughter ion from the m/z 177 quasi-molecular ion was quantitated for cotinine. Additional ions for the internal standard and for confirmation were also monitored. Cotinine concentrations were derived from the ratio of native to labeled cotinine in the sample, by comparison to a standard curve.
NNAL was quantified using liquid chromatography linked to tandem mass spectrometry (LC/MS-MS). The urine sample was fortified with a NNAL-13C6 internal standard, and then hydrolyzed using β-glucuronidase in incubations for at least 24 hours. The samples were then extracted and cleaned up on a specially-designed solid-phase molecularly-imprinted polymer (MIP) column, after which the analyte was eluted and analyzed by LC/MS-MS, monitoring the m/z 210→180 native, and m/z 216→186 internal standard transition ions. NNAL concentrations were derived from the ratio of the integrated peaks of native to labeled ions by comparison to a standard calibration curve, as described previously (18, 19).
NNAL measurements were adjusted by urine creatinine to account for urinary dilution (20). Creatinine analysis was performed by a combination of ion selective electrode (ISE) and phothometric determination. Each urine sample was prediluted by the analyzer in one of the reaction cells and after a series of enzymatic steps the ending color product was measured by a photometric lector at 546 nm. There were no changes on lab site or method for creatinine measurements throughout the different NHANES cycles. However, in 2013 to 2014 there was a replacement of the clinical analyzer Roche/Hitachi Mod P used to measure creatinine until 2011 to 2012 by the Roche/Hitachi Cobas 6000 (21). A detailed description of this laboratory methodology can be found on the NHANES website (https://wwwn.cdc.gov/Nchs/Nhanes/2013-2014/ALB_CR_H.htm).
The limit of detection (LOD) using these methods is 0.015 ng/mL for serum cotinine, 0.6 pg/mL for NNAL, and 5 mg/dL for urinary creatinine. Those participants with measurements below these limits were assigned the value of the LOD divided by the square root of two (0.011 ng/mL, 0.4 pg/mL, and 4 mg/dL, respectively; refs. 22, 23). To monitor the quality of the analyses performed by the contract laboratories, the National Center for Health Statistics developed and distributed a quality control protocol which outline the use of Westgard rules when running NHANES specimens (24). Overall, the coefficient of variation (CV) for each assay was less than 7%.
Statistical analysis
The NHANES survey uses a complex, stratified, multistage probability cluster sampling to produce estimates representative of the U.S. population. Weights indicating the probability of selection of each individual are available for each cycle. However, to produce combined estimates for the analytic variables it is necessary to calculate the 8-year weights by multiplying the 2-year weights by one-fourth. More details can be found in the NHANES analytic guidelines (25, 26). To calculate the overall percentage of participants exposed to detectable concentrations of cotinine and NNAL we created a binary variable using the concentration and LOD for each biomarker, then we applied the survey sampling weights to calculate proportions and 95% confidence intervals (CI). We compared measurements of serum cotinine and NNAL between nonsmokers, non-daily, and daily smokers, using the median and the interquartile range (IQR) of each biomarker overall and by stratum of sex and race/ethnicity. We obtained estimates for non-Hispanic (NH) Blacks, NH Whites and Hispanics as well as the subgroup of Mexican-Americans. However, we did not examine the “Other” category since the definition used for this group varied over the NHANES cycles included in our analysis (https://wwwn.cdc.gov/Nchs/Nhanes/2013-2014/DEMO_H.htm#RIDRETH1).
To evaluate the concordance between self-reported daily and non-daily smoking and the serum cotinine measurements, we used the previously characterized overall cutoff point of 10 ng/mL (16) to classify participants as nonsmokers (≤10 ng/mL) or active smokers (>10 ng/mL). In addition, we conducted a sensitivity analysis using specific race/ethnicity cutpoints for NH Blacks (6 ng/mL), NH Whites (5 ng/mL), and Mexican-Americans (1 ng/mL) as has also been recommended (17).
The agreement between self-report and the serum cotinine measurements was evaluated using Cohen's κ statistic. Spearman's correlation coefficients were computed to examine the correlation between biomarkers concentrations and both days and cigarettes smoked per month. In addition, we fitted linear regression models with both untransformed and log-transformed values for serum cotinine, NNAL, and NNAL adjusted by creatinine (NNAL-CR) to evaluate the statistical significance of possible differences between groups as well as examine trends across categories. All the statistical tests were conducted using the svy commands of STATA version 16, which perform estimates appropriately weighted to account for the complex survey design (27).
Results
The median age of the 18,835 adult participants was 49 years (IQR = 35–63). In study sample, 77.8% (14,660) reported currently not smoking (nonsmokers), 18.3% (3,446) smoked every day (daily smokers), and 3.9% (729) smoked on some days of the past month (non-daily smokers). Per 30 day-month, non-daily smokers reported smoking a median of 10 days and 24 cigarettes (IQR = 9–60). Conversely, daily smokers smoked a median of 300 cigarettes per month (10 cigarettes per day). The highest consumption of cigarettes per month among non-daily smokers was found in NH Blacks (median, 40), those ages 60 years or more (median, 30), and those with a high school education (median, 30). Among daily smokers, those with higher consumption were NH Whites (median, 450) and those ages 40 to 49 (median, 390). Mexican-Americans reported smoking the fewest cigarettes per month for both non-daily (median, 16) and daily (median, 180) smokers (Supplementary Table S1).
Overall, serum cotinine was estimated to be detectable in 72.1% (95% CI, 70.2%–73.9%) of the individuals in the 2007 to 2014 non-institutionalized adult U.S. population, with cotinine detected in 62.0% of nonsmokers, 96.8% of non-daily smokers, and 100% of daily smokers. For NNAL, 59.0% (95% CI, 56.7%–61.2%) of adult U.S. population had concentrations in urine above the limit of detection, 44.4% of nonsmokers, 95.0% of non-daily smokers, and 99.8% of daily smokers.
Serum cotinine, NNAL and NNAL-CR measurements were lowest among nonsmokers (cotinine, median = 0.0, IQR = 0.0–0.1 ng/mL; NNAL, median = 0.4, IQR = 0.4–2.1 pg/mL; and NNAL-CR, median = 0.9, IQR = 0.4–2.2 ng/g), intermediate among non-daily smokers (cotinine, median = 47.3, IQR = 7.1–148.0 ng/mL; NNAL, median = 72.5, IQR = 14.8–211.0 pg/mL; and NNAL-CR, median = 62.4, IQR = 13.9–158.5 ng/g), and highest among daily smokers (cotinine, median = 244.0, IQR = 153.0–338.0 ng/mL; NNAL, median = 294.0, IQR = 148.0–542.0 pg/mL; and NNAL-CR, median = 297.2, IQR = 158.3–514.3 ng/g); P < 0.001 for all (Table 1).
. | . | Proportion with detectable serum cotinine (≥0.015 ng/mL) . | Serum cotinine (ng/mL) . | Proportion with detectable urinary NNAL (≥0.6 pg/mL) . | NNAL (pg/mL) . | NNAL-CR (ng/g) . | ||
---|---|---|---|---|---|---|---|---|
. | N . | %a . | 95% CI . | Median (IQR) . | %a . | 95% CI . | Median (IQR) . | Median (IQR) . |
Nonsmokersb | ||||||||
Overall | 14,660 | 62.0 | 59.8–64.2 | 0.0 (0.0, 0.1) | 44.4 | 41.8–47.1 | 0.4 (0.4, 2.1) | 0.9 (0.4, 2.2) |
Sex | ||||||||
Male | 6,590 | 65.9 | 63.5–68.1 | 0.0 (0.0, 0.1) | 48.3 | 45.3–51.3 | 0.6 (0.4, 2.5) | 0.8 (0.4, 2.1) |
Female | 8,070 | 58.9 | 56.5–61.2 | 0.0 (0.0, 0.1) | 41.3 | 38.5–44.1 | 0.4 (0.4, 1.7) | 1.0 (0.5, 2.3) |
Race/ethnicityc | ||||||||
Hispanic | 4,082 | 61.8 | 58.7–64.9 | 0.0 (0.0, 0.0) | 48.7 | 45.0–52.4 | 0.4 (0.4, 1.6) | 0.8 (0.4, 1.8) |
Mexican-American | 2,459 | 61.5 | 57.2–65.7 | 0.0 (0.0, 0.0) | 46.7 | 42.0–51.4 | 0.4 (0.4, 1.5) | 0.8 (0.4, 1.7) |
NH White | 6,134 | 58.1 | 55.2–61.0 | 0.0 (0.0, 0.1) | 41.7 | 38.2–45.2 | 0.4 (0.4, 2.0) | 0.9 (0.4, 2.2) |
NH Black | 2,763 | 80.8 | 78.9–82.6 | 0.0 (0.0, 0.2) | 60.1 | 56.0–64.0 | 1.1 (0.4, 4.4) | 1.0 (0.4, 3.2) |
Non-daily smokersb,d | ||||||||
Overall | 729 | 96.8 | 94.0–98.3 | 47.3 (7.1, 148.0) | 95.0 | 90.2–97.5 | 72.5 (14.8, 211.0) | 62.4 (13.9, 158.5) |
Sex | ||||||||
Male | 444 | 97.3 | 94.8–98.6 | 40.3 (5.9, 139.5) | 95.2 | 89.5–97.8 | 67.1 (14.3, 208.5) | 53.0 (11.6, 137.7) |
Female | 285 | 96.1 | 90.7–98.4 | 51.1 (8.9, 156.0) | 94.7 | 89.3–97.5 | 85.8 (15.2, 216.0) | 79.2 (18.3, 191.2) |
Race/ethnicityc | ||||||||
Hispanic | 266 | 96.3 | 92.6–98.2 | 14.5 (1.1, 62.3) | 95.6 | 93.0–97.3 | 29.5 (8.0, 107.0) | 25.5 (6.2, 96.8) |
Mexican-American | 178 | 95.9 | 90.9–98.2 | 12.7 (0.7, 48.5) | 96.1 | 92.2–98.1 | 24.9 (7.1, 75.3) | 21.9 (5.7, 78.9) |
NH White | 210 | 95.7 | 90.3–98.1 | 52.9 (10.0, 167.0) | 93.1 | 83.0–97.4 | 101.0 (19.5, 216.0) | 82.9 (18.2, 214.7) |
NH Black | 182 | 100.0 | 100.0–100.0 | 131.0 (51.3, 246.0) | 100.0 | 100.0–100.0 | 177.5 (78.3, 335.0) | 111.5 (59.1, 197.7) |
Daily smokersb | ||||||||
Overall | 3,446 | 100.0 | 100.0–100.0 | 244.0 (153.0, 338.0) | 99.8 | 99.5–99.9 | 294.0 (148.0, 542.0) | 297.2 (158.3, 514.3) |
Sex | ||||||||
Male | 1,920 | 100.0 | 100.0–100.0 | 249.0 (154.0, 342.0) | 99.8 | 99.5–99.9 | 295.0 (146.0, 543.0) | 258.5 (139.7, 458.7) |
Female | 1,526 | 100.0 | 100.0–100.0 | 238.0 (152.0, 334.0) | 99.8 | 99.3–99.9 | 293.0 (149.0, 541.0) | 345.4 (189.9, 597.9) |
Race/ethnicityc | ||||||||
Hispanic | 562 | 100.0 | 100.0–100.0 | 161.0 (86.7, 263.0) | 99.6 | 98.0–99.9 | 229.0 (101.0, 437.0) | 220.0 (102.4, 393.5) |
Mexican-American | 309 | 100.0 | 100.0–100.0 | 133.0 (77.3, 228.0) | 99.4 | 95.9–99.9 | 197.0 (88.7, 432.0) | 206.8 (89.2, 385.0) |
NH White | 1,823 | 100.0 | 100.0–100.0 | 253.0 (174.0, 340.0) | 99.9 | 99.5–100.0 | 345.0 (170.0, 606.0) | 397.7 (227.2, 645.9) |
NH Black | 826 | 100.0 | 100.0–100.0 | 277.5 (179.0, 381.0) | 99.5 | 98.5–99.9 | 279.0 (161.0, 518.0) | 215.9 (134.7, 333.9) |
. | . | Proportion with detectable serum cotinine (≥0.015 ng/mL) . | Serum cotinine (ng/mL) . | Proportion with detectable urinary NNAL (≥0.6 pg/mL) . | NNAL (pg/mL) . | NNAL-CR (ng/g) . | ||
---|---|---|---|---|---|---|---|---|
. | N . | %a . | 95% CI . | Median (IQR) . | %a . | 95% CI . | Median (IQR) . | Median (IQR) . |
Nonsmokersb | ||||||||
Overall | 14,660 | 62.0 | 59.8–64.2 | 0.0 (0.0, 0.1) | 44.4 | 41.8–47.1 | 0.4 (0.4, 2.1) | 0.9 (0.4, 2.2) |
Sex | ||||||||
Male | 6,590 | 65.9 | 63.5–68.1 | 0.0 (0.0, 0.1) | 48.3 | 45.3–51.3 | 0.6 (0.4, 2.5) | 0.8 (0.4, 2.1) |
Female | 8,070 | 58.9 | 56.5–61.2 | 0.0 (0.0, 0.1) | 41.3 | 38.5–44.1 | 0.4 (0.4, 1.7) | 1.0 (0.5, 2.3) |
Race/ethnicityc | ||||||||
Hispanic | 4,082 | 61.8 | 58.7–64.9 | 0.0 (0.0, 0.0) | 48.7 | 45.0–52.4 | 0.4 (0.4, 1.6) | 0.8 (0.4, 1.8) |
Mexican-American | 2,459 | 61.5 | 57.2–65.7 | 0.0 (0.0, 0.0) | 46.7 | 42.0–51.4 | 0.4 (0.4, 1.5) | 0.8 (0.4, 1.7) |
NH White | 6,134 | 58.1 | 55.2–61.0 | 0.0 (0.0, 0.1) | 41.7 | 38.2–45.2 | 0.4 (0.4, 2.0) | 0.9 (0.4, 2.2) |
NH Black | 2,763 | 80.8 | 78.9–82.6 | 0.0 (0.0, 0.2) | 60.1 | 56.0–64.0 | 1.1 (0.4, 4.4) | 1.0 (0.4, 3.2) |
Non-daily smokersb,d | ||||||||
Overall | 729 | 96.8 | 94.0–98.3 | 47.3 (7.1, 148.0) | 95.0 | 90.2–97.5 | 72.5 (14.8, 211.0) | 62.4 (13.9, 158.5) |
Sex | ||||||||
Male | 444 | 97.3 | 94.8–98.6 | 40.3 (5.9, 139.5) | 95.2 | 89.5–97.8 | 67.1 (14.3, 208.5) | 53.0 (11.6, 137.7) |
Female | 285 | 96.1 | 90.7–98.4 | 51.1 (8.9, 156.0) | 94.7 | 89.3–97.5 | 85.8 (15.2, 216.0) | 79.2 (18.3, 191.2) |
Race/ethnicityc | ||||||||
Hispanic | 266 | 96.3 | 92.6–98.2 | 14.5 (1.1, 62.3) | 95.6 | 93.0–97.3 | 29.5 (8.0, 107.0) | 25.5 (6.2, 96.8) |
Mexican-American | 178 | 95.9 | 90.9–98.2 | 12.7 (0.7, 48.5) | 96.1 | 92.2–98.1 | 24.9 (7.1, 75.3) | 21.9 (5.7, 78.9) |
NH White | 210 | 95.7 | 90.3–98.1 | 52.9 (10.0, 167.0) | 93.1 | 83.0–97.4 | 101.0 (19.5, 216.0) | 82.9 (18.2, 214.7) |
NH Black | 182 | 100.0 | 100.0–100.0 | 131.0 (51.3, 246.0) | 100.0 | 100.0–100.0 | 177.5 (78.3, 335.0) | 111.5 (59.1, 197.7) |
Daily smokersb | ||||||||
Overall | 3,446 | 100.0 | 100.0–100.0 | 244.0 (153.0, 338.0) | 99.8 | 99.5–99.9 | 294.0 (148.0, 542.0) | 297.2 (158.3, 514.3) |
Sex | ||||||||
Male | 1,920 | 100.0 | 100.0–100.0 | 249.0 (154.0, 342.0) | 99.8 | 99.5–99.9 | 295.0 (146.0, 543.0) | 258.5 (139.7, 458.7) |
Female | 1,526 | 100.0 | 100.0–100.0 | 238.0 (152.0, 334.0) | 99.8 | 99.3–99.9 | 293.0 (149.0, 541.0) | 345.4 (189.9, 597.9) |
Race/ethnicityc | ||||||||
Hispanic | 562 | 100.0 | 100.0–100.0 | 161.0 (86.7, 263.0) | 99.6 | 98.0–99.9 | 229.0 (101.0, 437.0) | 220.0 (102.4, 393.5) |
Mexican-American | 309 | 100.0 | 100.0–100.0 | 133.0 (77.3, 228.0) | 99.4 | 95.9–99.9 | 197.0 (88.7, 432.0) | 206.8 (89.2, 385.0) |
NH White | 1,823 | 100.0 | 100.0–100.0 | 253.0 (174.0, 340.0) | 99.9 | 99.5–100.0 | 345.0 (170.0, 606.0) | 397.7 (227.2, 645.9) |
NH Black | 826 | 100.0 | 100.0–100.0 | 277.5 (179.0, 381.0) | 99.5 | 98.5–99.9 | 279.0 (161.0, 518.0) | 215.9 (134.7, 333.9) |
aPercentage of individuals in the 2007–2014 noninstitutionalized adult U.S. population with concentrations of serum cotinine and urinary NNAL above the limit of detection (0.015 ng/mL and 0.6 pg/mL, respectively).
bNonsmokers include self-reported never smokers and former smokers (>30 days since quit); daily smokers and non-daily smokers were defined based on use in the past 30 days.
cThe number of participants in analyses stratified by race/ethnicity do not sum to the total because we are not presenting results for “Other” group separately as the definition of this group varied over time. In addition, Mexican-American is a subgroup of the larger Hispanic category.
dConcentrations of serum cotinine, urinary NNAL, and NNAL-CR were different in non-daily smokers compared with nonsmokers (P < 0.001) and daily smokers (P < 0.001) overall and across strata of sex and race/ethnicity using linear regression.
Similar patterns of cotinine, NNAL, and NNAL-CR concentrations were observed across strata of sex and race/ethnicity (Table 1). Yet, concentrations within each smoking group varied by race/ethnicity. Among non-daily smokers, the highest median cotinine and NNAL concentrations were found in NH Blacks (cotinine: 131.0 ng/mL; NNAL: 177.5 pg/mL) followed by NH Whites (cotinine: 52.9 ng/mL; NNAL: 101.0 pg/mL) and Hispanics (cotinine: 14.5 ng/mL; NNAL: 29.5 pg/mL). Within Hispanics, even lower levels were found in the Mexican-American group for both biomarkers (cotinine: 12.7 ng/mL; NNAL: 24.9 pg/mL). Nevertheless, a smaller difference was observed between NH Whites and NH Blacks in median concentrations of NNAL-CR (NH Blacks: 111.5 ng/g, NH Whites: 82.9 ng/g). The same pattern of higher concentrations of cotinine in NH Blacks followed by NH Whites, Hispanics, and Mexican-Americans was observed among daily smokers. However, NNAL and NNAL-CR concentrations were higher in NH Whites (NNAL median: 345.0 pg/mL; NNAL-CR median: 397.7 ng/g) compared with NH Blacks (NNAL median: 279.0 pg/mL; NNAL-CR median: 215.9 ng/g) and Hispanics (NNAL median: 229.0 pg/mL; NNAL-CR median 220.0 ng/g). As observed in non-daily smokers, the lowest concentrations of cotinine, NNAL and NNAL-CR were found in Mexican-Americans daily smokers (cotinine median: 133.0 ng/mL, NNAL median: 197.0 pg/mL, NNAL-CR median: 206.8 ng/g). Significant differences in cotinine, NNAL, and NNAL-CR levels by race/ethnicity among non-daily smokers persisted in linear regression models after adjustment for cigarettes smoked per month (Supplementary Table S2).
Among non-daily smokers, both biomarkers were correlated with self-reported days smoked per month (cotinine ρ = 0.49; NNAL ρ = 0.45; P < 0.001) and number of cigarettes smoked per month (cotinine ρ = 0.47; NNAL ρ = 0.47; P < 0.001). Serum cotinine concentrations were significantly higher with more days smoked per month (e.g., serum cotinine increased by 38.5 ng/mL from 1–5 to 6–10 days; Ptrend < 0.001) and with more cigarettes smoked per month (Ptrend < 0.001; Figs. 2 and 3). A similar positive trend was observed for NNAL-CR and days smoked per month (e.g., NNAL-CR increased by 51.0 ng/g from 1–5 to 6–10 days; Ptrend < 0.001) as well as the number of cigarettes smoked per month (Ptrend < 0.001; Figs. 2 and 3). Higher concentrations of both biomarkers were found among those participants who smoked the day of the sample collection (today) and were lower among participants reporting smoking the day before (yesterday) and 3 to 5 days before (Ptrend < 0.001; Fig. 4).
There was a strong concordance (79% of agreement, κ = 0.48; P < 0.001) between self-reported smoking status and serum cotinine. Just 2.1% of nonsmokers had cotinine concentrations in the active smoking range (>10 ng/mL), compared with 70.4% of non-daily smokers and 98.8% of daily smokers (Table 2). Also, the prevalence of current smoking was 21.4% (95% CI, 20.2–22.6) based on self-report and 21.8% (95% CI, 20.7–23.1) according to cotinine concentrations. For NH Blacks, NH Whites, and Mexican-Americans, the agreement obtained with the race/ethnicity-specific cutoff points was comparable with the overall cutoff point to classify active smoking. However, among non-daily smokers, using specific cutoff points improved the concordance between the self-report and serum cotinine concentrations (Supplementary Table S3).
. | . | Serum cotinine ≤10 ng/mL . | . | Serum cotinine >10 ng/mL . | ||
---|---|---|---|---|---|---|
. | N . | %a . | 95% CI . | N . | %a . | 95% CI . |
Nonsmokersb | ||||||
Overall | 14,327 | 97.9 | 97.6–98.2 | 333 | 2.1 | 1.8–2.4 |
Sex | ||||||
Male | 6,407 | 97.4 | 96.9–97.9 | 183 | 2.6 | 2.1–3.1 |
Female | 7,920 | 98.3 | 97.9–98.7 | 150 | 1.7 | 1.3–2.1 |
Race/ethnicityc | ||||||
Hispanic | 4,039 | 98.7 | 98.1–99.1 | 43 | 1.3 | 0.9–1.9 |
Mexican-American | 2,434 | 98.8 | 98.1–99.3 | 25 | 1.2 | 0.7–1.9 |
NH White | 6,003 | 98.1 | 97.7–98.5 | 131 | 1.9 | 1.5–2.3 |
NH Black | 2,634 | 95.0 | 94.0–95.9 | 129 | 5.0 | 4.1–6.0 |
Non-daily smokersb | ||||||
Overall | 206 | 29.6 | 25.1–34.5 | 523 | 70.4 | 65.5–74.9 |
Sex | ||||||
Male | 132 | 31.9 | 26.4–38.0 | 312 | 68.1 | 62.0–73.6 |
Female | 74 | 26.4 | 20.4–33.3 | 211 | 73.6 | 66.7–79.6 |
Race/ethnicityc | ||||||
Hispanic | 114 | 43.1 | 36.2–50.3 | 152 | 56.9 | 49.7–63.8 |
Mexican-American | 81 | 44.2 | 36.2–52.5 | 97 | 55.8 | 47.5–63.8 |
NH White | 53 | 28.1 | 20.9–36.8 | 157 | 71.9 | 63.2–79.1 |
NH Black | 12 | 7.0 | 3.3–14.2 | 170 | 93.0 | 85.8–96.7 |
Daily smokersb | ||||||
Overall | 40 | 1.2 | 0.8–1.8 | 3,406 | 98.8 | 98.2–99.2 |
Sex | ||||||
Male | 19 | 1.0 | 0.5–1.9 | 1,901 | 99.0 | 98.1–99.5 |
Female | 21 | 1.4 | 0.8–2.2 | 1,505 | 98.6 | 97.8–99.2 |
Race/ethnicityc | ||||||
Hispanic | 20 | 4.5 | 2.7–7.4 | 542 | 95.5 | 92.6–97.3 |
Mexican-American | 17 | 7.4 | 4.3–12.4 | 292 | 92.6 | 87.6–95.7 |
NH White | 12 | 0.8 | 0.4–1.5 | 1,811 | 99.2 | 98.5–99.6 |
NH Blackd | – | – | – | 823 | 99.6 | 98.8–99.9 |
. | . | Serum cotinine ≤10 ng/mL . | . | Serum cotinine >10 ng/mL . | ||
---|---|---|---|---|---|---|
. | N . | %a . | 95% CI . | N . | %a . | 95% CI . |
Nonsmokersb | ||||||
Overall | 14,327 | 97.9 | 97.6–98.2 | 333 | 2.1 | 1.8–2.4 |
Sex | ||||||
Male | 6,407 | 97.4 | 96.9–97.9 | 183 | 2.6 | 2.1–3.1 |
Female | 7,920 | 98.3 | 97.9–98.7 | 150 | 1.7 | 1.3–2.1 |
Race/ethnicityc | ||||||
Hispanic | 4,039 | 98.7 | 98.1–99.1 | 43 | 1.3 | 0.9–1.9 |
Mexican-American | 2,434 | 98.8 | 98.1–99.3 | 25 | 1.2 | 0.7–1.9 |
NH White | 6,003 | 98.1 | 97.7–98.5 | 131 | 1.9 | 1.5–2.3 |
NH Black | 2,634 | 95.0 | 94.0–95.9 | 129 | 5.0 | 4.1–6.0 |
Non-daily smokersb | ||||||
Overall | 206 | 29.6 | 25.1–34.5 | 523 | 70.4 | 65.5–74.9 |
Sex | ||||||
Male | 132 | 31.9 | 26.4–38.0 | 312 | 68.1 | 62.0–73.6 |
Female | 74 | 26.4 | 20.4–33.3 | 211 | 73.6 | 66.7–79.6 |
Race/ethnicityc | ||||||
Hispanic | 114 | 43.1 | 36.2–50.3 | 152 | 56.9 | 49.7–63.8 |
Mexican-American | 81 | 44.2 | 36.2–52.5 | 97 | 55.8 | 47.5–63.8 |
NH White | 53 | 28.1 | 20.9–36.8 | 157 | 71.9 | 63.2–79.1 |
NH Black | 12 | 7.0 | 3.3–14.2 | 170 | 93.0 | 85.8–96.7 |
Daily smokersb | ||||||
Overall | 40 | 1.2 | 0.8–1.8 | 3,406 | 98.8 | 98.2–99.2 |
Sex | ||||||
Male | 19 | 1.0 | 0.5–1.9 | 1,901 | 99.0 | 98.1–99.5 |
Female | 21 | 1.4 | 0.8–2.2 | 1,505 | 98.6 | 97.8–99.2 |
Race/ethnicityc | ||||||
Hispanic | 20 | 4.5 | 2.7–7.4 | 542 | 95.5 | 92.6–97.3 |
Mexican-American | 17 | 7.4 | 4.3–12.4 | 292 | 92.6 | 87.6–95.7 |
NH White | 12 | 0.8 | 0.4–1.5 | 1,811 | 99.2 | 98.5–99.6 |
NH Blackd | – | – | – | 823 | 99.6 | 98.8–99.9 |
aPercentages were weighted for the US population. Agreement 79%, κ = 0.48, P < 0.001.
bNonsmokers include self-reported never smokers and former smokers (>30 days since quit); daily smokers and non-daily smokers were defined based on use in the past 30 days.
cThe number of participants in analyses stratified by race/ethnicity do not sum to the total because we are not presenting results for “Other” group separately as the definition of this group varied over time. In addition, Mexican-American is a subgroup of the larger Hispanic category.
dData not reported due to a small number of participants (n < 10).
We additionally examined the proportion of non-daily and daily smokers above two cutoff-points of NNAL previously used to classify active smokers from second-hand smoke exposure (28, 29). We found that 74.2% of non-daily smokers and 98.8% of daily smokers had urinary NNAL concentrations above 14.4 pg/mL (29). By using the cutoff point of 47.3 pg/mL instead (28), the proportion was reduced to 57.6% for non-daily smokers and 94.2% for daily smokers. These results were similar to those found for cotinine (70.4% and 98.8% among non-daily and daily smokers, respectively).
Finally, the nonsmoking category included both never smokers and former smokers (>1 month since quit). Serum cotinine concentrations were modestly higher among former smokers (median = 0.03, IQR = 0.01–0.09 ng/mL) compared with never smokers (median = 0.02, IQR = 0.01–0.06 ng/mL, P < 0.001). However, both groups were in the nonsmoking range and comparable in terms of the proportion below the 10 ng/mL cutpoint (95.9% vs. 98.7%, respectively). Similarly, we found modestly higher concentrations of NNAL and NNAL-CR among former smokers (median NNAL = 0.6, IQR = 0.4–3.0 pg/mL; median NNAL-CR = 1.0, IQR = 0.4–2.9 ng/g; P < 0.001) compared with never smokers (median NNAL = 0.4, IQR = 0.4–1.8 pg/mL; median NNAL-CR = 0.8, IQR = 0.4–2.0 ng/g; P < 0.001).
Discussion
Non-daily smokers had substantially higher concentrations of both serum cotinine and urinary NNAL than nonsmokers, yet lower concentrations of each biomarker than daily smokers. Similar patterns have been found in prior epidemiologic studies where disease and mortality risks among non-daily smokers are higher than nonsmokers but also lower than daily smokers (6–8). Furthermore, results from two Asian cohort studies indicate specific association of NNAL concentrations in the range we observed among non-daily smokers in our current study with increased risk of lung cancer. In that study, the median of NNAL-CR in the Singapore and Shangai cohorts were 46.0 ng/g (range = 4.2–952.0 ng/g) and 186.2 ng/g (range = 20.9–734.4 ng/g) respectively, and smokers with NNAL concentrations in the highest tertile had two times higher lung cancer risk than participants in the lowest tertile (30). Thus, our results provide evidence as to the validity of self-reported non-daily smoking and suggest that previously observed associations between non-daily smoking and mortality are real rather than a result of misclassified smoking behaviors. Our current study also supports results suggesting a dose–response relationship among non-daily smokers (8), as we observed that both cotinine and NNAL concentrations were positively correlated with higher number of days smoked per month and cigarettes smoked per month, and fewer days since last smoked. In summary, our study provides further evidence that there is no safe level of cigarette smoking.
We observed the lowest levels of tobacco biomarkers in non-daily smokers among Hispanics and Mexican-Americans, which is consistent with their low cigarette consumption. In contrast, higher concentrations of cotinine and NNAL were found among NH Blacks, compared with NH Whites and Hispanics. Although NH Blacks smoked on more days of the month and higher number of cigarettes per month than NH Whites and Hispanics, differences persisted in linear regression models after adjustment for total number of cigarettes per month. Alternatively, differences may reflect cigarette smoking behavior (e.g., puff number and maximum volume) as well as other individual factors (hormonal, genetics) that modify the cotinine pharmacokinetics parameters. Racial/ethnic differences in nicotine and cotinine glucuronidation (31, 32), and to a lesser extent NNAL metabolism (33) have been noted in previous studies. In contrast, NNAL-CR concentrations were similar between NH Blacks and NH Whites. Creatinine levels were higher among NH blacks, likely affecting these comparisons. Indeed, a previous study conducted in NHANES 2007 to 2008 showed that creatinine concentration is the most influential nontobacco product variable in the TNSAs exposure assessment (20).
As NNAL has a longer half-life, we hypothesized that a greater proportion of self-reported non-daily smokers would have concentrations of NNAL within the range of active smokers than cotinine. Nevertheless, we found that the performance of urinary NNAL was similar to serum cotinine for detecting active non-daily cigarette smoking. These result suggest that even with a relatively short half-life, cotinine concentrations persist in the active smoking range among non-daily smokers for a number of days after smoking a cigarette. We observed additional data supporting this finding when we examined cotinine concentrations among participants reporting smoking cigarettes 3–5 days prior to urine collection (Fig. 4).
Strengths of our study include use of high-quality measurements of cotinine and NNAL conducted as part of the nationally representative NHANES surveys. Pooling different cycles of the survey (2007–2014) provided a sufficient sample size to evaluate the validity of self-reported smoking using the overall cutoff point for serum cotinine concentrations, as well as specific points for NH Blacks, NH Whites, and Mexican-Americans. Furthermore, the detailed questions on current cigarette use in the last month allowed us to analyze the distribution of these biomarkers according to the number of days smoked in the last month, and total number of cigarettes smoked per month. Likewise, using the last 5 days data, we were able to exclude those cigarette smokers who consumed any other tobacco products as well as nonsmokers who reported using any tobacco product. Limitations include that we cannot rule out the use of non-cigarette tobacco products beyond the 5 days before sample collection. In addition, e-cigarettes appeared in the U.S. market in 2007 (34), but were only assessed in NHANES starting in 2013 when the prevalence of recent use among adults 18 years or older was 2.6% (95% CI, 2.0–3.1; ref. 35). It is likely that the prevalence of e-cigarette use prior to 2013 was lower.
Nonresponse bias cannot be ruled out because the examination and interview response rates in NHANES have been declining over time (68.5% and 71.0% respectively in 2013–2014 compared with 75.4% and 78.4% in 2007–2008). This may have affected the validity of our estimates, especially in stratified analyses by race/ethnicity. Finally, the clinical analyzer used for urinary creatinine measurements changed for the 2013 to 2014 cycle, but any possible bias introduced by this change would be small relative to the large difference in NNAL-CR concentrations observed between nonsmokers, non-daily smokers and daily smokers. We noted similar differences between groups when NNAL was not adjusted for creatinine.
In summary, we observed excellent concordance between self-reported non-daily cigarette smoking and concentrations of serum cotinine among participants of the 2007 to 2014 NHANES surveys. Our results provide further evidence that non-daily cigarette smokers are exposed to high amounts of carcinogenic nitrosamines relative to nonsmokers, regardless of how few cigarettes they smoke per month. These results are complementary to the growing number of epidemiologic studies indicating substantial disease and mortality risks of non-daily smoking.
Authors' Disclosures
No disclosures were reported.
Disclaimer
The views and opinions expressed in this article are those of the authors only and do not necessarily represent the views, official policy or position of the U.S. Department of Health and Human Services or any of its affiliated institutions or agencies. Use of trade names is for identification only and does not imply endorsement by the CDC, the NCI, or the U.S. Department of Health and Human Services.
Authors' Contributions
D.S. Gutiérrez-Torres: Formal analysis, methodology, writing–original draft, writing–review and editing. L. Wang: Investigation, writing–review and editing. B.C. Blount: Investigation, writing–review and editing. B. Xia: Investigation, writing–review and editing. C.S. Sosnoff: Investigation, writing–review and editing. M.S. Shiels: Writing–review and editing. M. Inoue-Choi: Writing–review and editing. A. Etemadi: Writing–review and editing. N.D. Freedman: Conceptualization, supervision, writing–review and editing.
Acknowledgments
This study was supported by NCI, Division of Cancer Epidemiology and Genetics, Intramural Research Program.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.