Abstract
Objectives. We monitored the prevalence and patterns of use of the array of tobacco products available to young adults, who are at risk for initiation and progression to established tobacco use.
Methods. We used data from waves 1 to 3 of GfK’s KnowledgePanel (2011–2012), a nationally representative cohort of young adults aged 18 to 34 years (n = 2144). We examined prevalence and patterns of tobacco product use over time, associated demographics, and state-level tobacco policy. We used multivariable logistic regression to determine predictors of initiation of cigarettes as well as noncombustible and other combustible products.
Results. The prevalence of ever tobacco use rose from 57.28% at wave 1 to 67.43% at wave 3. Use of multiple products was the most common pattern (66.39% of tobacco users by wave 3). Predictors of initiation differed by product type and included age, race/ethnicity, policy, and use of other tobacco products.
Conclusions. Tobacco use is high among young adults and many are using multiple products. Efforts to implement policy and educate young adults about the risks associated with new and emerging products are critical to prevent increased initiation of tobacco use.
The 2009 Family Smoking Prevention and Tobacco Control Act (FSPTCA)1 gave the US Food and Drug Administration (FDA) the authority to regulate tobacco and marked a new beginning in the field of tobacco control. A number of restrictions arising from the FSPTCA have already taken place, including banning characterizing flavors in cigarettes and tobacco brand sponsorships of sports and entertainment events.1 At the same time, there has been increased use of noncigarette combustible products, such as cigars,2 and the introduction of several noncombustible tobacco products (NCPs), including e-cigarettes and snus. A spokesman for Phillip Morris recently called NCPs their “most exciting growth opportunity,”3(p5) and several tobacco companies have endorsed their development for harm reduction.4–6 However, rather than promoting these products as a substitution for cigarettes, they are often advertised as an extension of major cigarette brands7 and promoted for use alongside cigarettes.8,9 These new products—as well as other noncigarette combustible products such as cigars and hookahs—also may be especially attractive to young adults.10–12
Since the 1998 Master Settlement Agreement put forth restrictions on tobacco industry marketing to youths,13 young adults have become an increasingly important target audience.14,15 This is evident in the data, as approximately 20% of young adults still smoke cigarettes,16 and recent national data show that young adults aged 18 to 25 years had the highest prevalence of current tobacco use (40.8%) compared with youths (ages 12–17 years) or adults (ages 26 years and older).17 Use of certain noncigarette products are particularly high; for example, data from a nationally representative sample found that 30% of young adults aged 18 to 34 years had ever used cigars; 26% had used little cigars, cigarillos, or bidis; and 18% had used a hookah.18 This last finding corroborates other studies reporting high rates of hookah use among young adults,19–23 reaching past 30-day rates of 8.5% among college-age students.19 Young adults were also more likely than older adults to have ever used snus,10 and expressed interest in experimenting with the product.24 e-Cigarette use has also been reported to be inversely related to age,25,26 with rates in adults decreasing by 2% to 3% with every year of increased age.11
Tobacco use is the number one cause of preventable death in America27 and results in almost 5.2 million years of potential life lost each year.28 In light of the increasingly diversified tobacco product marketplace, and the introduction of products that currently fall outside FDA regulatory authority, it is critical to monitor tobacco product use among young adults. This includes not only current, regular use but also experimentation. To decrease the prevalence and improve long-term health outcomes, focus should be extended to understanding to what extent consumers are using multiple tobacco products (poly-use). Little is known about patterns of poly-use, but studies suggest that this is becoming increasingly prevalent among young adults. For example, a recent study found that 30% of young adult current tobacco users had ever used multiple tobacco products,18 which corroborates findings in other samples.29,30 Another nationally representative study reported the highest prevalence of poly-use among those aged 18 to 24 years compared with those aged 25 years and older.17 This has implications as poly-users may have higher rates of nicotine addiction,7,31 may have more withdrawal symptoms during quit attempts,31 and may be less likely to quit.32
Young adulthood marks a critical developmental period,33 one that often coincides with the both the initiation34–36 and establishment of regular tobacco use.35–37 Prevention efforts directed at this audience will depend on a precise understanding of the degree to which young adults are using different tobacco products. To address this, we examined the prevalence and patterns of tobacco product use over time, demographic correlates of use, and predictors of initiation among a large, nationally representative cohort of young adults. Data will help inform program planners and policymakers on the impact of new and alternative tobacco product use among young adults.
We used data from a nationally representative sample of young adults, aged 18 to 34 years, designed to monitor the prevalence and patterns of tobacco use among young adults over time. The sample was drawn from GfK’s KnowledgePanel, an online panel of adults aged 18 years and older, recruited via address-based sampling (http://www.knowledgenetworks.com/knpanel/index.html). The probability-based random sampling method provides statistically valid representation of the US population and maintains representation of the offline population by providing households without Internet access with a free netbook computer and Internet service.
The baseline survey (wave 1) was fielded in July 2011 and subsequent surveys have been fielded at 6-month intervals: wave 2 (January 2012) and wave 3 (July 2012). We used data from the 2144 young adults who completed the first 3 waves of data collection and had complete information on tobacco use at all waves. This last requirement caused the exclusion of 38 respondents: 3 with missing information on tobacco use at all 3 waves, 13 at 2 of the 3 waves, and the remainder at 1 of the 3 waves. Further details can be found elsewhere.18
Ever tobacco use was assessed at each wave with the question “which, if any, of the following tobacco or nicotine products have you ever used or tried?” Respondents were instructed to select among the following products: cigarettes; cigars; little cigars, cigarillos, or bidis; hookah; pipe; e-cigarettes; dip or snuff; chewing tobacco; snus; and dissolvables. Brand information was included for little cigars, cigarillos, and bidis; e-cigarettes; dip and snuff; chewing tobacco; snus; and dissolvables, because previous research has shown that it increases validity of estimates.38,39 Once a respondent endorsed ever using a tobacco product, they were labeled as an ever user of that product at all subsequent waves. For simplicity, ever use will be referred to as “use” from this point forward. Current use was originally examined as well, but the proportion of individuals endorsing ever use at wave 1 and then subsequently “recanting” use was deemed too high (> 10% for some products) for this analysis.
We defined tobacco use subgroups in the following manner: “Non-user”—respondents with no reported tobacco use; “cigarette-only”— respondents who only endorsed use of cigarettes; “NCP-only”— respondents who endorsed ever use of an NCP only (e.g., e-cigarettes, snus, dissolvables, chewing tobacco, dip or snuff); “other combustible-only”—respondents who endorsed use of a noncigarette combustible product only (e.g., hookah, pipe, cigars, little cigars, cigarillos, bidis); and “poly-user”—respondents who reported use of more than 1 type of tobacco product.
This includes respondents reporting use of cigarettes and NCPs, cigarettes and other combustibles, other combustibles and NCPs, or use of all 3 product types. Because of the complexity of the data, we did not refer to individuals reporting use of more than 1 product type (e.g., use of 2 NCPs, such as snus and e-cigarettes) as poly-users in this study.
Current use of marijuana was assessed at wave 1 by asking how often, if ever, respondents currently used marijuana. Response categories included every day, some days, and not at all. We dichotomized responses into either “no current use” or “current use.” This was included in analysis because of the known association between marijuana and use of other combustible products, specifically cigars.40
Variables were from wave 1 and included age (18–24 years and 25–34 years), gender, race/ethnicity (non-Hispanic White, non-Hispanic Black, non-Hispanic other, Hispanic), educational attainment (< high school, high school, some college), and self-identified financial situation (do not meet basic expenses, just meet basic expenses, meet needs with a little left, live comfortably).
Variables included state smoking prevalence,41,42 total tax per cigarette pack (state plus federal),43 state-level per capita tobacco control expenditures rounded to the nearest cent (F. J. Chaloupka, written communication, 2011), and level of state clean indoor air legislations measured as percentage of state population covered as of 201344 across all US states and the District of Columbia. We treated all policy variables as continuous.
We performed all analyses with Stata IC version 12.1 (StataCorp LP, College Station, TX) by using SURVEY commands. We used poststratification weights set to wave 1 benchmarks to offset any nonresponse or noncoverage bias.
Univariate analyses produced weighted prevalence estimates for use of each product described previously. We also calculated patterns of ever tobacco use (nonuser of tobacco, NCP-only user, other combustible–only user, cigarette-only user, poly-user) at each wave. Because of the tobacco-use definition (see “Measures” section), wave-3 estimates represent the total tobacco use over the study period. For example, a wave-3 poly-user represents anyone who had endorsed the use of at least 2 different types of tobacco products (e.g., an NCP and cigarettes) at any point during the study period. This represents respondents who endorsed the use of multiple products at 1 wave (e.g., wave 1) as well as those who endorsed the use of multiple products across different waves (e.g., cigarettes at wave 1 and e-cigarettes at wave 2).
We carried out bivariate analyses to examine associations between wave-3 tobacco-use patterns and both demographic characteristics and state-level tobacco control policy variables. We made statistical comparisons in reference to 2 groups: (1) nonuser and (2) cigarette-only smokers. We used a second-order (Satterthwaite) Rao–Scott χ2 test to adjust for the survey design and converted the χ2 statistic into an F-statistic with noninteger degrees of freedom.45,46 We set statistical significance at P < .05.
We used multivariable logistic regression models to estimate the influence of the independent variables on initiating use of 3 separate product types: (1) NCPs, (2) other combustibles, and (3) cigarettes. We ran separate models for each of the 3 dependent variables. All demographic and policy variables previously described were included as covariates. In addition, we included an indicator variable reflecting use of other tobacco products (apart from the outcome) at any wave before or concurrent to initiation. This variable was specific to each model. For example, in the model in which NCP initiation is the outcome, this covariate reflects cigarette use, other combustible use, or both. For the model in which other combustibles are the outcome, this reflects cigarette use, NCP use, or both. For the model in which initiation of cigarettes is the outcome, this reflects NCP use, other combustible use, or both.
Ever use of any tobacco product was endorsed by 57.28% of the sample at wave 1 and increased by 10 percentage points over a 1-year period, to reach a rate of 67.43% (Table 1). The wave-1 prevalence of any combustible product use was 56.67%, and any NCP use was 14.61%. By wave 3, rates of ever use were highest for cigarettes (57.65% of the sample); cigars (39.75%); little cigars, cigarillos, or bidis (32.91%); and hookahs (21.77%). The products that showed the greatest percentage change in use over time were cigars (10.20% increase in use); little cigars, cigarillos, or bidis (9.08%); and cigarettes (8.69%).
TABLE 1— Prevalence of Tobacco Product Use Among US Young Adults (Weighted): GfK’s KnowledgePanel, 2011–2012
| Product | Wave 1, Freqa (%) | Wave 2, Freqa (%) | Wave 3, Freqa (%) | Wave 1–Wave 3, % Change |
| Cigarettes | 1042 (48.96) | 1172 (55.07) | 1227 (57.65) | 8.69 |
| Other combustible products | ||||
| Cigars | 629 (29.55) | 782 (36.74) | 846 (39.75) | 10.20 |
| Pipe | 183 (8.59) | 238 (11.18) | 273 (12.82) | 4.23 |
| Little cigars, cigarillos, or bidis | 507 (23.83) | 694 (32.61) | 700 (32.91) | 9.08 |
| Hookah | 331 (15.55) | 425 (19.97) | 463 (21.77) | 6.22 |
| Noncombustible products | ||||
| Electronic cigarettes | 107 (5.03) | 169 (7.92) | 219 (10.31) | 5.28 |
| Chewing tobacco, dip, or snuff | 227 (10.64) | 287 (13.49) | 325 (15.25) | 4.61 |
| Snus | 99 (4.65) | 133 (6.26) | 148 (6.96) | 2.31 |
| Dissolvables | 16 (0.74) | 21 (0.97) | 27 (1.26) | 0.52 |
| Any noncombustible | 311 (14.61) | 410 (19.25) | 468 (22.00) | 7.40 |
| Any other combustible | 855 (40.17) | 1034 (48.59) | 1095 (51.45) | 11.28 |
| Any combustibleb | 1206 (56.67) | 1369 (64.33) | 1422 (66.82) | 10.15 |
| Any tobacco use | 1219 (57.28) | 1385 (65.08) | 1435 (67.43) | 10.15 |
Note. Use is defined as “ever use” of a tobacco product.
a This represents a weighted frequency.
b This represents use of cigarettes as well as other combustible products.
The changing patterns of tobacco use over time are displayed in Figure 1. The most prominent group at wave 1 was nonusers of tobacco products, which represented 42.70% of young adults. However, by wave 3 this percentage had decreased to only 32.56% as the former nonusers reported using tobacco. Coinciding with the decrease in former nonusers was a sharp rise of poly-users, which started as the 2nd-most prominent tobacco group at wave 1 (34.10%) and rose to be the most prominent group by wave 3 (44.77%). Among poly-users, the majority (76.18%) reported ever poly-use at baseline, and the remainder endorsed use of multiple tobacco products at subsequent waves (waves 2 or 3). The majority of poly-users endorsed use of cigarettes as one of their products (97.91%). Few reported initiation of different products across different waves; for example, only 2.51% of tobacco users first reported use of cigarettes only and subsequently reported use of NCPs. This transition to poly-use, however, accounts for the slight decrease in the cigarette-only group over time (15.31% at wave 1 to 13.84% at wave 3). There were few changes in the prevalence of other combustible-only and NCP-only users over time as the transition to poly-use from respondents in these categories at earlier waves was balanced by individuals initiating use of these products.
FIGURE 1— Longitudinal patterns of tobacco use among a nationally representative sample of young adults (weighted): GfK’s KnowledgePanel, 2011–2012.
To understand more about the tobacco use patterns identified in Figure 1, we examined the demographic characteristics of young adults according to their tobacco use pattern by the end of the study period (wave 3). We excluded NCP-only users because of their small sample size (n = 14). However, they showed signs of differing from the other tobacco-use groups: the majority were males (75.60%), were non-Hispanic White (70.78%), and had at least some college education (97.54%), and a high percentage were current marijuana users (25.23%).
The remaining comparisons are shown in Table 2. Cigarette-only smokers were more likely than nonusers to be male (68.97% vs 51.23%), to be in the “older” (25–35 years) age group (67.55% vs 51.72%), to have a high-school education or less (42.91% vs 31.31%), to be a current marijuana user (4.73% vs 1.27%), and to live in a state with a higher smoking prevalence (mean prevalence = 18.83% vs 17.53%). Other-combustible-only users were more likely than nonusers to currently use marijuana (5.81% vs 1.27%) and to be Hispanic (32.52% vs 21.49%). Other-combustible-only users were also more likely than cigarette-only smokers to be Hispanic (32.52% vs 14.16%) and were also more likely to be younger (18–24 years; 46.21% vs 32.45%), female (45.37% vs 31.03%), and at least meet basic expenses (95.64% vs 88.21%).
TABLE 2— Demographic Characteristics of Young Adults by Smoking Pattern Over Time (Weighted): GfK’s KnowledgePanel, 2011–2012
| Definitions of Tobacco Use Patternsa at the End of Wave 3 | |||||
| Characteristic | Total Sample (n = 2144), % or Mean (SD) | Nonuser (n = 725), % or Mean (SD) | Cigarette-Only (n = 261), % or Mean (SD) | Other Combustible–Only (n = 181), % or Mean (SD) | Poly-Userb (n = 963), % or Mean (SD) |
| Gender | |||||
| Male | 50.16 | 51.23 | 68.97* | 54.63** | 54.12** |
| Female | 49.84 | 48.77 | 31.03* | 45.37** | 45.88** |
| Age, y | |||||
| 18–24 | 41.85 | 48.28 | 32.45* | 46.21** | 39.16* |
| 25–34 | 58.15 | 51.72 | 67.55* | 53.79** | 60.84* |
| Race/ethnicity | |||||
| Non-Hispanic White | 59.74 | 52.38 | 59.18 | 51.72 | 66.58* |
| Non-Hispanic Black | 13.08 | 16.86 | 16.73 | 10.37 | 9.60* |
| Non-Hispanic other | 7.45 | 9.27 | 9.94 | 5.40 | 5.83 |
| Hispanic | 19.74 | 21.49 | 14.16 | 32.52*,** | 17.99 |
| Education | |||||
| < high school | 12.39 | 12.47 | 15.10 | 13.78 | 11.41 |
| High school | 20.34 | 18.84 | 27.81 | 18.59 | 19.69 |
| ≥ some college | 67.27 | 68.69 | 57.09* | 67.64 | 68.90** |
| Perceived financial situation | |||||
| Do not meet basic expenses | 8.54 | 8.36 | 11.77 | 4.36** | 8.31 |
| Just meet basic expenses | 29.16 | 26.75 | 29.21 | 30.63 | 30.78 |
| Meet needs with a little left | 38.69 | 37.46 | 32.44 | 43.22 | 40.71 |
| Live comfortably | 23.60 | 27.43 | 26.56 | 21.79 | 20.21* |
| Marijuana use | |||||
| No current use | 91.35 | 98.73 | 95.27* | 94.19* | 84.44*,** |
| Current marijuana | 8.65 | 1.27 | 4.73* | 5.81* | 15.56*,** |
| State-level variables | |||||
| Smoking prevalence | 17.99 (3.30) | 17.53 (3.40) | 18.83 (3.39)* | 17.14 (3.00)** | 18.22 (3.17)* |
| Total tax (state + federal) | 2.48 (1.02) | 2.52 (1.06) | 2.60 (1.17) | 2.32 (0.89)** | 2.44 (0.96) |
| Per capita funding | 2.04 (1.64) | 2.10 (1.69) | 2.00 (1.50) | 2.13 (1.69) | 2.00 (1.63) |
| Clean indoor air laws | 81.35 (33.51) | 81.92 (32.97) | 77.56 (36.70) | 87.67 (29.07) | 81.09 (33.41) |
a Because of the small sample size (n = 14), noncombustible product–only users were excluded from the table.
b Poly-users are defined as respondents who endorsed ever use of more than 1 type of tobacco product (cigarettes and noncombustible products, cigarettes and other combustibles, noncombustible products and other combustibles, or all 3) any time by wave 3.
*P < .05 in comparison with nonusers.
**P < .05 in comparison with cigarette-only users.
Poly-users were more likely than nonusers to be younger (39.16% vs 48.28%), White (66.58% vs 52.38%), and living in states with a higher smoking prevalence (mean prevalence = 18.22% vs 17.53%). Compared with cigarette-only smokers, poly-users were more likely to be female (45.88% vs 31.03%), to be White (66.58% vs 59.18%), and to currently use marijuana (15.56% vs 4.73%).
There were several predictors of initiating use of tobacco products (Table 3). Initiation of NCP use was more likely in individuals who were younger (odds ratio [OR] = 2.00; 95% confidence interval [CI] = 1.24, 3.25), and among those who had used other tobacco products at any wave (OR = 5.12; 95% CI = 2.91, 9.00). Initiation was less likely among non-Hispanic Blacks than non-Hispanic Whites (OR = 0.31; 95% CI = 0.12, 0.83). Initiation of other combustibles was more likely among Hispanics than non-Hispanic Whites (OR = 1.82; 95% CI = 1.06, 3.14), and in areas with higher smoking prevalence (OR = 1.08; 95% CI = 1.01, 1.14), and less likely in states with higher tax (OR = 0.79; 95% CI = 0.65, 0.96). Initiation of cigarettes was predicted by younger age (OR = 1.85; 95% CI = 1.18, 2.91), Hispanic ethnicity (OR = 1.82; 95% CI = 1.02, 3.25), and areas with high smoking prevalence (OR = 1.17; 95% CI = 1.09, 1.27).
TABLE 3— Results of Multivariate Analysis to Identify Predictors of Initiating Tobacco Use From Baseline to Wave 3: GfK’s KnowledgePanel, 2011–2012
| Use of NCPs | Use of Other Combustibles | Use of Cigarettes | ||||
| Variable | OR (95% CI) | P | OR (95% CI) | P | OR (95% CI) | P |
| Age, y | ||||||
| 18–24 | 2.00 (1.24, 3.25) | .005 | 0.82 (0.54, 1.26) | .362 | 1.85 (1.18, 2.91) | .007 |
| 25–34 (Ref) | 1.00 | 1.00 | 1.00 | |||
| Gender | ||||||
| Male (Ref) | 1.00 | 1.00 | 1.00 | |||
| Female | 0.95 (0.59, 1.52) | .82 | 1.48 (0.98, 2.24) | .06 | 1.20 (0.75, 1.91) | .454 |
| Race/ethnicity | ||||||
| Non-Hispanic White (Ref) | 1.00 | 1.00 | 1.00 | |||
| Non-Hispanic Black | 0.31 (0.12, 0.83) | .02 | 1.11 (0.47, 2.61) | .811 | 1.37 (0.67, 2.79) | .394 |
| Non-Hispanic other | 1.40 (0.51, 3.83) | .516 | 0.76 (0.29, 2.03) | .591 | 1.08 (0.34, 3.47) | .893 |
| Hispanic | 0.78 (0.36, 1.69) | .527 | 1.82 (1.06, 3.14) | .03 | 1.82 (1.02, 3.25) | .042 |
| Education | ||||||
| < high school | 0.80 (0.29, 2.18) | .657 | 0.75 (0.34, 1.64) | .474 | 1.56 (0.61, 4.01) | .352 |
| High school (Ref) | 1.00 | 1.00 | 1.00 | |||
| ≥ some college | 0.80 (0.44, 1.46) | .469 | 0.65 (0.40, 1.05) | .078 | 1.60 (0.88, 2.88) | .12 |
| Perceived financial situation | ||||||
| Do not meet basic expenses | 0.93 (0.33, 2.61) | .892 | 0.86 (0.37, 1.99) | .725 | 0.59 (0.19, 1.79) | .353 |
| Just meet basic expenses | 1.60 (0.80, 3.21) | .182 | 1.30 (0.73, 2.31) | .38 | 0.84 (0.44, 1.63) | .613 |
| Meet needs with a little left | 0.89 (0.45, 1.74) | .731 | 0.83 (0.47, 1.45) | .508 | 0.59 (0.32, 1.09) | .093 |
| Live comfortably (Ref) | 1.00 | 1.00 | 1.00 | |||
| Marijuana use | ||||||
| No current marijuana use | 1.68 (0.91, 3.10) | .099 | 1.42 (0.64, 3.12) | .387 | 0.27 (0.10, 0.73) | .01 |
| Current marijuana use (Ref) | 1.00 | 1.00 | 1.00 | |||
| State-level variables | ||||||
| Smoking prevalence | 1.02 (0.95, 1.11) | .553 | 1.08 (1.01, 1.14) | .017 | 1.17 (1.09, 1.27) | < .001 |
| Total tax (state + federal) | 0.94 (0.77, 1.15) | .544 | 0.79 (0.65, 0.96) | .019 | 1.08 (0.84, 1.39) | .55 |
| Per capita funding | 0.97 (0.85, 1.10) | .598 | 1.00 (0.89, 1.12) | .971 | 0.92 (0.79, 1.07) | .306 |
| Clean indoor air laws | 1.00 (1.00, 1.01) | .304 | 1.01 (1.00, 1.07) | .06 | 1.01 (1.00, 1.02) | .101 |
| Other tobacco product usea | ||||||
| No use at any wave (Ref) | 1.00 | 1.00 | 1.00 | |||
| Use at any wave | 5.12 (2.91, 9.00) | < .001 | 1.10 (0.60, 2.01) | .753 | 0.98 (0.62, 1.55) | .93 |
Note. CI = confidence interval; NCP = noncombustible product; OR = odds ratio.
a For the model where NCP adoption is the outcome, this reflects cigarette use, other combustible use, or both. For the model where other combustibles are the outcome, this reflects cigarette use, NCP use, or both. For the final model, this reflects NCP use, other combustible use, or both.
Data show a high prevalence of tobacco use in this cohort across a broad spectrum of both combustible and noncombustible products. By the third wave, 67% of respondents had used at least 1 tobacco product, a growth of 10% points over the study period. This indicates at least a willingness to try tobacco products, which could increase vulnerability to more regular use. Although cigarette use was highest, cigar use was also high, which is consistent with other studies17,18 and signals a growing trend of cigar use. This is problematic as cigars are another combustible product with known adverse health consequences,47 but cigars are largely omitted from regulations set forth under the FSPTCA.
By the end of the study period, poly-use represented the most common pattern of ever tobacco use. This corroborates other studies that have reported high rates of poly-use,17,30 a pattern that may be particularly problematic as some studies indicate that it is associated with increased risk of tobacco-related disease48,49 and harm escalation,50 and may hinder cessation.7,32 The majority of poly-users first reported use of multiple tobacco product types at the same wave, suggesting that the sequence between initiating use of different tobacco products may not be long. Even if these young adults are not current users of multiple products, this willingness to experiment with various tobacco products, particularly in a short time frame, could result in an increased vulnerability to long-term use and addiction. For young adults vulnerable to poly-use, the challenge will be preventing use of that first tobacco product.
Younger age was associated with poly-use, which signals a need to invest in comprehensive tobacco prevention programs for youths and young adults. Furthermore, the gender breakdown is more evenly split among poly-users than among cigarette-only smokers. As many poly-users reported use of “other combustibles” as one product type, this may be related to the higher proportion of females using other combustible products, such as little cigars, cigarillos, or bidis and hookahs.51,52 This is reflected in the higher percentage of females in the other-combustible–only category as opposed to the cigarette-only category. Finally, poly-users had, by far, the highest prevalence of current marijuana use, as measured at wave 1. This suggests that, in agreement with other studies,53 poly-users are experimenting not only with tobacco but also with other substances. Understanding the broader characteristics of poly-users with respect to how they use other substances, such as alcohol and drugs, may help inform messaging and outreach tactics.
The most striking predictor of NCP initiation was other tobacco product use. This may indicate that young adults are turning to NCPs to reduce harm associated with combustible tobacco products, or that they are simply more likely to supplement their combustible tobacco product use with NCPs (e.g., in situations when they cannot smoke). It may also reflect a tendency toward experimentation and may be related to the influence of peers and the social setting. Initiation of NCPs, however, was not predicted by educational level in any of the models, suggesting that it did not matter whether the young adults were in a college setting. From the analysis, it is unclear if those who initiate NCPs continue to use other tobacco products and, if so, how they used these products together. More research is necessary to clarify these issues.
Another predictor of tobacco use was race/ethnicity; Blacks were less likely than Whites to ever use NCPs and Hispanics were more likely to ever use both other combustibles and cigarettes. This may reflect cultural differences in attitudes about different products, differential marketing strategies of the industry, or other factors. Higher state-level smoking prevalence predicted initiation of both other combustibles and cigarettes and higher state-level tax was inversely related to initiation of other combustibles, verifying previous literature on the influence of the environment and policy in predicting tobacco use.54–58
The high level of poly-use identified in this study highlights the need to develop effective interventions to educate young adults about the harms associated with not only cigarettes but also with other tobacco products. This includes education about the ways in which poly-use may actually increase risk of tobacco-related disease. Furthermore, current screening protocols often focus on whether an individual smokes cigarettes, rather than whether they use other tobacco products. This may lead to an underestimation of tobacco use and cause health professionals to miss an opportunity to discuss cessation resources with their patients. Development and implementation of policies may also help curb the wide-ranging patterns of tobacco use in this vulnerable age group. This is especially important with emerging use of largely unregulated cigars and NCPs that have unknown long-term effects on population-level harm.
Several limitations of this study must be considered. First, assessment of tobacco product use is by self-report and may be subject to recall bias. Second, we defined tobacco use as ever use, which may not represent current or regular use. However, there is a need to understand experimentation among young adults, particularly as new products emerge in the market, as this often precedes more regular use. Third, we did not consider frequency of use. It is likely that results would differ if we had analyzed regular or heavy users separately from other more experimental users. Fourth, despite attempts to control for all relevant covariates, it is possible that there is some amount of unmeasured confounding. Fifth, the number of NCP-only users over the study period was small (n = 14) and, therefore, the demographic characteristics associated with that group may not be valid. Finally, we defined poly-use as ever use of more than 1 type of tobacco product and it does not necessarily imply concurrent use. In addition, this definition does not capture respondents who are using more than 1 tobacco product within a single tobacco product category (e.g., 2 NCPs). Results may differ if we had altered the definition to reflect either concurrent poly-use or use of any 2 products rather than product types.
Despite these limitations, this study presents critical data on tobacco use patterns over time in a large, nationally representative cohort of young adults. In the wake of FDA regulation of tobacco and an increasingly diversified tobacco product marketplace, the patterns of tobacco product use are changing. Continued rapid and rigorous surveillance of tobacco use is essential to develop evidence-based messaging and treatment strategies. Young adulthood is a particularly critical time period, marked by development transitions and associated with the establishment of life-long tobacco use. Reducing the burden of tobacco use will be contingent on continuing to monitor the evolving ways in which tobacco products are being used in this population so that appropriate policy can be implemented and interventions developed to improve population-level health.
Acknowledgments
This project has been funded in whole or in part with federal funds from the National Institute on Drug Abuse, National Institutes of Health, and the Food and Drug Administration, Department of Health and Human Services, under contract HHSN271201100027C.
The authors would like to acknowledge Cindy Tworek (Food and Drug Administration) and Genevieve Vullo (Kelly Government Solutions) for their participation in this project and contributions to the article.
Note. The views and opinions expressed in this presentation are those of the authors only and do not necessarily represent the views, official policy, or position of the US Department of Health and Human Services or any of its affiliated institutions or agencies.
Human Participant Protection
Approval for this study was granted by the Independent Investigational Review Board Inc.
