Tobacco Product Use Patterns, and Nicotine and Tobacco-Specific Nitrosamine Exposure: NHANES 1999–2012

The prevalence of current cigarette smoking among adults in the United States has declined in recent years from 23.9% in 2005 to 16.7% in 2015 (1). Meanwhile, dual- and poly-tobacco use has become a public health concern given an increasing popularity of noncigarette tobacco products (e.g., cigars, hookah, electronic nicotine delivery systems). Multiple tobacco product use is also prevalent among adult tobacco users. For example, among U.S. adult tobacco users in 2010, 8.6% of cigarette smokers, 50.3% of cigar users, 54.8% of chewing tobacco users, and 42.5% of snuff users reported using at least one additional tobacco product every day or some days (2).

Two limitations characterize current literature on dual- and poly-tobacco use. First, researchers have defined dual- and/or poly-tobacco use as consumption of any two or more tobacco products (2, 3). This definition neglects the relative difference in health risks associated with nicotine delivery mechanisms that fall on a risk continuum with noncombustible products being the relatively less harmful and cigarettes being the most harmful tobacco products. Noncigarette combustible products generally are also less harmful than cigarette largely due to the lower frequency of product use compared with cigarettes (4). To overcome this limitation, we developed the Tobacco Product Use Patterns (T-PUPs) model to classify tobacco product use, incorporating the risk continuum framework of tobacco product categories (cigarettes; noncigarette combustibles, e.g., cigars, pipes, hookah; noncombustibles, e.g., chewing tobacco, snuff) and the number of product categories used (one, two, or three). This results in seven mutually exclusive categories: cigarettes only, noncigarette combustibles only, noncombustibles only, dual noncigarette combustibles and noncombustibles (i.e., using at least one noncigarette combustible and at least one noncombustible), dual cigarettes and noncombustibles (i.e., using cigarettes and at least one noncombustible), dual cigarettes and noncigarette combustibles (i.e., using cigarettes and at least one noncigarette combustible), and poly-tobacco use (i.e., using cigarettes and at least one noncigarette combustible and one noncombustible). We applied this model to youth tobacco users participated in the National Youth Tobacco Survey (5) and discerned differences between T-TUPs in demographic characteristics smoking-related beliefs and exposure to tobacco advertising. To date, this model has not been applied to U.S. adult tobacco users.

Second, the literature is lacking documentation on intensity of product consumption among dual- and poly-tobacco users compared with single-product category users, as well as likely differences in exposure to key biomarkers of nicotine dependence and cancer risks such as cotinine and tobacco-specific nitrosamines. To overcome these limitations in the literature, we applied the T-TUPs model to a national representative sample of U.S. adult tobacco users who participated in the National Health and Nutrition Examination Survey (NHANES). We then examined how levels of exposure to nicotine and tobacco-specific nitrosamines among dual- and poly-tobacco product category users differ from those of single tobacco product category users.

NHANES is a nationally representative sample of noninstitutionalized civilian U.S. population obtained through a complex multistage sampling design. The National Center for Health Statistics has conducted the study annually since 1999. Participants completed home interviews about their health. They completed an additional survey at a Mobile Examination Center (MEC) where they provided information on recent tobacco use and received a medical examination during which biospecimens (e.g., blood and urine) were collected. Additional information on NHANES is available online (6). We pooled NHANES data collected between 1999 and 2012 to maximize the sample size for each T-PUP. The National Institutes of Health Office of Human Subjects Research Protections determined the analysis to be exempted from institutional review board review. This was a secondary data analysis on de-identified data and therefore exempted from obtaining informed written consent from the participants.

We classified participants into seven mutually exclusive categories based on the T-PUPs model (5). While preserving the relative placement of tobacco products along a risk continuum, the T-PUPs model is parsimonious in classifying users of any number of tobacco products based on products nicotine delivery mechanisms. These characteristics make the T-PUPs an appropriate model for population-level surveillance. Tobacco use was determined based on number of days on which participants used cigarettes, pipes, cigars, chewing tobacco, and snuff in the past 5 days [i.e., “During the past five days (including today), on how many days did you (smoke cigarettes/smoke cigars/smoke pipes/use chewing tobacco/use snuff)?”]. Intensity of consumption was assessed for cigarettes, cigars, and pipes using the question “During the past five days (including today), on the days you smoked (cigarettes/cigars/pipes), how many (cigarettes/cigars/pipes) did you smoke each day?” We classified participants who reported using cigarettes exclusively in the past 5 days as cigarette-only users. Similarly, we classified participants who reported using exclusively cigars and/or pipe as noncigarette combustible-only users, and those who reported using exclusively chewing tobacco and/or snuff as noncombustibles-only users. Participants who reported using products from any two combinations of product categories (i.e., cigarettes, noncigarette combustibles, noncombustibles) were classified as dual-category users: dual noncigarette combustible and noncombustible users, dual cigarette and noncombustible users, and dual cigarette and noncigarette combustible users. Finally, participants who reported using products from all three product categories were classified as poly-tobacco users.

Data on serum cotinine were available for the 1999–2012 survey years, whereas data on urinary total NNAL were available for 2007–2012. Cotinine is a primary metabolite of nicotine and is highly sensitive and specific to tobacco use (7). Serum cotinine has a relatively longer half-life than other biomarkers (e.g., carbon monoxide) of 16 hours in the general population on average, and therefore is also a recommended biomarker to assess nicotine intake over time and severity of nicotine dependence (7). An isotope dilution-high-performance liquid chromatography/atmospheric pressure chemical ionization-tandem mass spectrometry process was used to measure serum cotinine. Total 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanol (i.e., NNAL), a tobacco-specific nitrosamine and a known lung carcinogen, has an estimated half-life between 10 and 18 days (8). A liquid chromatography linked to tandem mass spectrometry approach was used to measure urinary total NNAL. Urinary total NNAL readings were divided by urinary creatinine levels and then log transformed to normalize the data. A detailed description of laboratory methods employed to ascertain levels of serum cotinine and urinary NNAL in NHANES are published (9).

Information on gender, age, race/ethnicity, and education attainment was collected. We categorized participants into five racial/ethnic groups: Mexican American, other Hispanic, non-Hispanic white, non-Hispanic black, and other race (including multirace). We created three categories for education attainment: less than high school graduate, high school graduate or graduate education diploma, and more than high school graduate.

We limited our analyses to tobacco users age 18 years or above because those participants have legal access to tobacco products. The total sample size for our analysis was 9,158, of which 9,157 participants had serum cotinine measurements and 3,864 had urinary total NNAL measurements. We conducted bivariate analysis to examine correlates of T-PUPs. We used general linear models to compare consumption of single tobacco products based on number of days and quantity used on the day used across T-PUPs that include the use of such product. For example, number of days smoking cigarettes and number of cigarettes smoked on the day smoked during the past 5 days were assessed across cigarette-only, dual cigarette and noncigarette combustible, dual cigarette and noncombustible, and poly-tobacco users. In these analyses, users of the single-category T-PUPs (i.e., cigarette only, noncigarette combustible only, and noncombustible only) were set as references. Last, we used general linear models to compare levels of serum cotinine and urinary total NNAL between dual- and poly-tobacco T-PUPs (i.e., dual noncigarette combustibles and noncombustibles, dual cigarettes and noncombustibles, dual cigarettes and noncigarette combustibles, and poly-tobacco) and their respective single-category noncigarette T-PUP (i.e., cigarettes only, noncigarette combustibles only, and noncombustibles only). For example, we used noncigarette combustible users as a reference category when examining cotinine and NNAL levels among dual noncigarette combustible and noncombustible, dual cigarette and noncigarette combustible, and poly-tobacco users. Similarly, we used noncombustible users as a reference for dual noncigarette combustible and noncombustible, dual cigarette and noncigarette combustible, and poly-tobacco users. General linear models were conducted using PROC SURVEYREG adjusted for age, gender, race/ethnicity, and education. Although body mass index (BMI) was included as a covariate in previous reports (9, 10), we found that adjusting for BMI did not substantially change our findings. We therefore did not control for BMI for model parsimony. Significance threshold was set to 0.05. We used SAS version 9.4 (SAS Institute) to conduct the analyses. All analyses were weighted to account for sampling design and representativeness of population using the MEC weights, masked variance pseudo-primary sampling unit, and masked variance pseudo-stratum.

Overall and T-PUP–specific weighted sample characteristics appear in Table 1. The most common T-PUP was cigarettes only (82.7%), followed by noncombustibles only (7.4%), noncigarette combustibles only (6.8%), dual cigarettes and noncombustibles (1.4%), dual cigarettes and noncigarette combustibles (1.4%), dual noncigarette combustibles and noncombustibles (0.2%), and poly-tobacco (0.1%). Demographic characteristics differed by T-PUPs. For example, men comprised 53.9% of cigarette-only users and >86% of all remaining T-PUPs (P < 0.0001). Tobacco users ages 18 to 25 years comprised >32% of all dual- and poly-tobacco T-PUPs, and <19% of cigarette-only, noncigarette combustible only, and non–combustible-only users (P < 0.0001). Noncigarette combustible and dual cigarette and noncigarette combustible users had the highest proportion of non-Hispanic black (>20%), whereas dual cigarette and noncombustible users had the highest proportion of non-Hispanic white (91.7%). Finally, noncigarette combustible users had the highest proportion of attaining high school education or above (57.7%) than users of other T-PUPs (P < 0.0001).

Data on tobacco product consumption by T-PUPs appear in Table 2. Compared with cigarette-only users, dual cigarette and noncombustible users smoked cigarettes on 0.6 fewer days and 3.3 fewer cigarettes on the day smoked in the past 5 days (P < 0.05). However, days smoked in the past 5 days and cigarettes smoked on the day smoked did not differ between cigarette-only, dual cigarette and noncigarette combustible, and poly-tobacco users. Compared with noncigarette combustible-only users, dual noncigarette combustible and noncombustible users smoked 3.6 more cigars on the day they smoked cigars, whereas poly-tobacco users smoked 0.7 fewer cigars on the day they smoked cigars (P < 0.01). Finally, compared with noncombustible-only users, dual noncigarette combustible and noncombustible, dual cigarettes and noncombustible, and poly-tobacco users consumed chewing tobacco and snuff on fewer days during the past 5 days (P < 0.01).

Mean serum cotinine and urinary total NNAL levels by T-PUPs appear in Table 3. Compared with cigarette-only users, adjusted mean serum cotinine level among dual cigarette and noncombustible users was 62.2 ng/mL higher (P < 0.05). Compared with noncigarette combustible-only users, dual cigarette and noncigarette combustible users, and dual noncigarette combustible and noncombustible users had higher adjusted mean levels of serum cotinine (93.1 and 60.3 ng/mL higher, respectively; P < 0.05). Compared with non–combustible-only users, adjusted mean serum cotinine level among dual cigarette and noncombustible users was 45.2 ng/mL higher (P < 0.05). Regarding urinary total NNAL, data from dual noncigarette combustible and noncombustible users (n = 8) and poly-tobacco users (n = 7) were not presented due to small number of participants in these T-PUPs. Compared with cigarette-only users, dual cigarette and noncigarette combustible users, and dual cigarette and noncombustible users had higher adjusted mean levels of urinary total NNAL (54.4 and 259.6 ng/mL higher, respectively). Compared with noncigarette combustible-only users, dual cigarette and noncigarette combustible users, and dual noncigarette combustible and noncombustible users had higher adjusted mean levels of urinary total NNAL (161.1 and 199.5 ng/mL higher, respectively).

This study is one of the first to compare tobacco product consumption and exposure to nicotine and tobacco-specific nitrosamines of dual-product category users and poly-tobacco product category users against single-product category users, while including different types of tobacco products, including pipe, cigars, and smokeless tobacco that have not been simultaneously studied previously. We found that dual- and poly-tobacco product category use does not necessarily translate to lower consumption of individual products. On the one hand, for example, compared with cigarette-only users, dual cigarette and noncombustible users reported smoking fewer days and fewer cigarettes; similarly, compared with noncombustible-only users, dual noncigarette combustible and noncombustible users, dual cigarette and noncombustible users, and poly-tobacco users reported using noncombustibles on fewer days. On the other hand, for example, compared with cigarette-only users, dual cigarette and noncigarette combustible users and poly-tobacco users showed no difference in cigarette consumption, in agreement with a previous study showing that cigarette consumption did not differ by smokeless tobacco use consumption among men (11). These findings suggest that some dual- and poly-tobacco product category users may have higher exposure to harmful and potentially harmful constituents of tobacco products, e.g., acetaldehyde, arsenic, and cadmium. In addition, the lower consumption of noncombustible tobacco products associated with dual- and poly-tobacco users than noncombustible-only users may not yield net health benefits, because this could be offset by addition of cigarettes and noncigarette combustible products that are probably more detrimental to health.

We found that dual- and poly-tobacco T-TUPs users had either equal or higher levels of serum cotinine and urinary total NNAL than single-product category T-PUPs users. These findings are supported by previous studies that reported higher levels of serum cotinine and urinary NNAL among dual cigarette and smokeless tobacco users compared with exclusive cigarette users (10–12). This could be due to the route of administration for noncombustible products that provides longer absorption time, or due to higher levels of nicotine and tobacco-specific nitrosamines in noncombustible products included in the analysis (13, 14). Our findings provide partial support to the argument that individuals with higher nicotine dependency may use tobacco products from multiple product categories to obtain nicotine from additional sources, perhaps especially when the environment does not allow the use of a specific product category (e.g., combustible users adding noncombustibles to circumvent smoke-free policies). However, finding on poly-tobacco T-PUP users showing statistically nondistinguishable level of serum cotinine from those of single-category T-PUPs users may suggest that tobacco dependence is not the only explanation for dual- and poly-tobacco product category use. Future studies need to further examine the role of nicotine dependency in dual- and poly-tobacco use patterns.

Our analyses have several limitations. First, NHANES did not have data on all tobacco products. Although electronic cigarettes, hookah, and snus became increasingly popular after 2010 (15, 16), we were unable to include these products in our analysis. Future studies need to include these tobacco products to confirm our findings. Second, individuals who did not use any tobacco products in the past 5 days were excluded from the analysis, and therefore our findings may not be generalizable to intermittent tobacco users. Third, we did not analyze exposure to all known harmful and potentially harmful constituents of tobacco products. Although our analysis on serum cotinine and urinary total NNAL may provide some indications on potential health risks associated with each T-PUP, to fully examine health risks associated with each T-PUP, we need to assess exposure to other tobacco product constituents that have known health risks and to directly examine disease risks using carefully conducted epidemiologic studies. The Population Assessment on Tobacco and Health Study includes a comprehensive list of biomarkers for tobacco product constituents, which can be used to validate our findings, and to perform a holistic comparison of health risks across T-PUPs. Fourth, prevalence of dual- and poly-tobacco T-PUPs was quite low in our current sample, which limited the statistical power to detect differences between T-PUPs and our ability to rank order all seven T-PUPs based on cotinine and NNAL levels. Studies with larger samples, especially those that purposefully sample dual- and poly-tobacco users and collect a broad range of biomarker data, are essential to understand tobacco consumption and health risks associated with all T-PUPs categories. Fifth, the cross-sectional nature of NHANES limited our ability to examine how changes from one T-PUP to another influence tobacco consumption, serum cotinine levels, and urinary NNAL levels. Finally, measures on smokeless tobacco consumption beyond the number of days used, as well as specific smokeless tobacco product use given the variability in constituent yields across these products (13), are needed to better quantify the intensity of smokeless tobacco use to enable more detail comparisons on noncombustible products consumption across T-PUPs.

In conclusion, tobacco product consumption, and nicotine and tobacco-specific nitrosamine exposure of dual- and poly-tobacco product category users somewhat differ from those of single-product category users as defined by the T-TUPs model. With an increasing availability of various tobacco products and prevalence of dual- and poly-tobacco use, research is needed to examine variations in tobacco product consumption and associated health risks with a comprehensive set of biomarkers that capture the physiologic underpinnings of tobacco product use patterns.

No potential conflicts of interest were disclosed.

Conception and design: K. Choi, S. El-Toukhy, D. Hatsukami

Development of methodology: K. Choi, D. Hatsukami

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): K. Choi, N.D. Freedman

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K. Choi, M. Sabado, N.D. Freedman

Writing, review, and/or revision of the manuscript: K. Choi, M. Sabado, S. El-Toukhy, E. Vogtmann, N.D. Freedman, D. Hatsukami

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): K. Choi, M. Sabado, E. Vogtmann, N.D. Freedman

Study supervision: K. Choi

Other [developer of the T-PUP (Tobacco Products Use Pattern) model upon which the study is based]: S. El-Toukhy

K. Choi, M. Sabado, and S. El-Toukhy's efforts were supported by the Division of Intramural Research, National Institute on Minority Health and Health Disparities. E. Vogtmann's and N.D. Freedman's efforts were supported by the Intramural Research Program of the National Cancer Institute. D. Hatsukami's effort was supported by a research grant funded by the National Cancer Institute (U19CA157345; PI: D. Hatsukami).

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.