Tobacco Taxation and Its Prospective Impact on Disparities in Smoking Initiation and Progression among Young Adults

Abstract

Purpose:

Limited research exists on tobacco taxes and cigarette smoking initiation and progression, particularly across different sociodemographic groups in young adulthood. This project examines how cigarette pack price in late adolescence prospectively relates to smoking initiation and progression by age 21, focusing on differences across demographics.

Methods:

Data are from the longitudinal Monitoring the Future (MTF) project (2001–2017). MTF examines drug use behaviors with nationally representative samples of 12th graders annually. Subsamples of 12th graders are followed longitudinally. We examined past 30-day cigarette smoking among baseline never smokers (N=9,232) and daily smoking among youth who were not daily cigarette smokers at baseline (N=15,141). Using logistic regression, we examined state-level cigarette pack price at modal age 18 and smoking at follow-up ages 19–20; we used interaction terms to assess differences across sociodemographic groups (by gender, race/ethnicity, and parental education).

Results:

For each dollar increase in price at baseline, the odds of initiation by age 19/20 were reduced by 12% (AOR= 0.88; 95% CI=0.78, 0.99) and the odds of progression to daily smoking were reduced by 16% (AOR= 0.84; 95% CI=0.76, 0.92). After adjusting for multiple testing, for both outcomes there were no statistically significant interactions between price and demographics.

Conclusions:

Cigarette prices in late adolescence were associated with a prospective reduction in cigarette smoking initiation and progression among young adults, with limited differences across sociodemographic characteristics. Higher cigarette prices can prevent smoking initiation and progression; however, complementary interventions are needed to reduce initiation and progression among subgroups disproportionately affected by tobacco.

Despite declines in cigarette smoking among youth,^1,2 recent trends indicate smoking initiation and progression to regular smoking increasingly occurs in young adulthood.^3–7 Cigarette smoking in young adulthood leads to continued use in adulthood,⁸ and preventing escalation and entrenchment of smoking in young adulthood is critical to reducing long-term health effects of smoking.⁷ Shedding light on smoking initiation and rapid escalation to regular smoking among young adults will help to prevent uptake.^3,5,7 Additionally, tobacco-related sociodemographic differences have emerged as public health concerns.⁹ Males are more likely to smoke and initiate at younger ages,^1,9,10 and they are more at risk of rapid escalation to heavy smoking in young adulthood relative to females.⁴ Non-Hispanic White youth are more likely to smoke and escalate to regular smoking in young adulthood compared to non-Hispanic Black youth and young adults, who tend to initiate smoking at later ages. Consequently, disparities exist in adolescence but additional disparities also emerge in adulthood.^1,4,8,9,11 Finally, people of low socioeconomic status (SES), and adolescents and young adults with lower parental educational, are more likely to initiate smoking at younger ages, be current smokers, and rapidly progress to regular smoking than their more affluent counterparts.^4,8,9,12

Research consistently indicates tobacco taxation is effective for reducing cigarette smoking.^13,14 Yet few tobacco policy evaluations directly address prospective effects during the critical age range from adolescence into young adulthood and across different subgroups.^9,15–17 There are no studies to our knowledge that address how the context of tobacco taxes in adolescence relates to cigarette smoking initiation and escalation in young adulthood. This is an important gap because contexts in adolescence impact behaviors in young adulthood,^18,19 and research is needed on ways to prevent smoking onset among young adults.^7,20 Cigarette taxation is effective for reducing adult smoking prevalence,^21,22 but recent evidence shows that taxes may not reduce gaps in smoking between vulnerable groups and the general population.^15,17,21 In short, limited studies on cigarette prices conducted since 2000 have not examined initiation, have devoted less attention to smoking escalation in young adulthood, are not from national samples, and have not examined cigarette smoking initiation and progression differences across subgroups.^4,7,20 Studies have also relied on data collected before the 1997 Master Settlement Agreement (MSA), which explicitly prohibited industry targeting of youth. This is a critical gap because young adults became increasingly important for marketing after the MSA.⁷

Using follow-up data from the 2001–2017 Monitoring the Future (MTF) project,^23,24 this project examines how tobacco taxes (i.e., cigarette prices during late adolescence) relate to cigarette smoking initiation and progression after age 18 among a national sample of young adults aged 19/20. We focus on ages before 21 because policies before this age have the best potential to disrupt smoking initiation and progression in young adulthood, as initiation and changes in smoking trajectories are most likely to occur before this age.^4,6 While smoking trajectories are likely to be established before 21, cigarette smoking initiation and progression can still occur for many young adults before age 25.^4,7,25 We also examine cigarette use up to 2017 because e-cigarette use has increased dramatically across the U.S. after 2017,² and we aimed to exclude any confounding role of e-cigarette use. Finally, we examine whether the relationship between cigarette prices in late adolescence and smoking initiation and progression in young adulthood differs by gender, race/ethnicity, and parental education.

Methods

Data and Sample

The MTF study is a national, repeated cross-sectional and prospective study of drug use with nationally representative samples of 12th graders from the 48 contiguous states recruited annually.^2,23,24 Subsamples of approximately 2,450 students are selected from each 12th-grade sample to complete follow-up surveys, with oversampling of substance users.²⁴ We used data on cigarette smoking behavior, demographics, state of residence, and school zip code in 12^th grade, with baseline data from 2000 to 2016. Primary analyses examined the first follow-up at modal ages 19/20. Data were collected in 2000–2017. Weights were used to account for attrition, oversampling of substance users, and the complex, stratified survey design of the MTF study.^6,26

For cigarette smoking initiation analyses, we restricted our sample to baseline never smokers (no smoking reported in lifetime).⁵ We limited the sample to participants who were not daily smokers at baseline (never [60.6%], previous/not current [28.3%], and current non-daily smokers [11.1%]) to examine smoking progression.⁴ Initiation analyses included 52.5% of the baseline sample (unweighted N=39,936), and progression analyses included 86.8% of the baseline sample (unweighted N=39,889). We used list-wise deletion for missing data on covariates. For primary analyses, the unweighted sample size at modal age 19/20 for initiation was 12,928 and 15,141 for progression. At modal age 19/20, 47.2% and 49.8% of the baseline sample was lost due to attrition for initiation and progression analyses, respectively. No variable had more than 5% missing data after accounting for attrition, and less than 9% of the sample was removed due to missing data. The University of Minnesota and University of Michigan Institutional Review Boards approved this project.

Cigarette Smoking Outcomes

At modal age 19/20, we examined past 30-day smoking (0=no vs. 1=smoked at least one cigarette) and daily smoking (0=none or nondaily smoking vs. 1=smoked at least 1 cigarette per day).^4,7

Cigarette Pack Price

Using Tax Burden on Tobacco data, tobacco price was based on average cost per cigarette pack, in constant dollars (adjusted to 2017 dollar value), by state.²⁷ Because MTF surveys are collected in the spring and average pack prices are measured on November 1, we used an average before and after each respondent’s survey. We use cigarette prices instead of taxes because taxes tend to only be effective if cigarette prices increase in response to taxes.^13,15,28

Demographic Characteristics

Demographics included self-reported sex (herein gender; male=1, female=0), race/ethnicity (non-Hispanic White, non-Hispanic Black, Hispanic, non-Hispanic Asian, and non-Hispanic Other), and parental education (highest education of either parent: some high school or less, high school/GED, some college, or college degree or more).

Covariates

We included census region, and state percent of: non-Hispanic Black, Hispanic, and living below the poverty line. Four census regions were included as dummy variables (Northeast, Midwest, South, and West). Percent Black and Hispanic populations for each state were calculated using Survey of Epidemiology and End Results (SEER) population estimates.²⁹ To control for tobacco policy environment, we included percentage of residents covered by smoke-free workplace and hospitality laws at the county level using data from American Nonsmokers’ Rights Foundation.³⁰

Analytic Strategy

We conducted multiple weighted logistic regression models. We first conducted logistic regressions to examine cigarette smoking initiation and progression, controlling for covariates. We then examined multiplicative interaction terms between price and gender, race/ethnicity, and parental education in separate models, adjusting for all covariates used in the main regression models. For parental education and race/ethnicity, we examined interactions by using joint F-tests. We further tested interaction effects on the additive scale by predicting marginal probabilities with covariates at their measured values (average marginal effects, AME).³¹ For all significance tests for interactions, we adjusted for multiple testing by using Benjamini-Hochberg tests (false discovery rate of 5%).³² Analyses were conducted in Stata v.15 and accounted for attrition and the MTF sample design.

We conducted multiple sensitivity analyses. We conducted logistic regression models for smoking progression by including baseline smoking history as a covariate and effect modifier for pack price. We conducted separate models that examined effect modification of pack price based on baseline year and a dichotomous metric of pre- vs. post-2009 to examine differences before and after the federal tax increase included in the Tobacco Control Act. We tested whether the relationship between pack price and smoking initiation and progression at follow-up depended on base year via multiplicative interactions. We examined models that included individual characteristics at follow-up, adjusting for the following additional covariates: highest degree earned, and full-time student and employment status. Since cigarette smoking initiation and progression still occurs for many young adults before age 25, we conducted analyses that examined follow-ups at modal ages 21/22 and 23/24, which included main effects and interaction terms; we also ran additional analyses that excluded any smokers or daily smokers from the previous wave for initiation and progression, respectively. To account for the fact that 14% of students reported living in a different state at modal age 19/20 and that within-state pack price could change from baseline to follow-up, we created a measure that captured change in pack price between baseline and follow-up and adjusted for pack price change. Finally, we also conducted ordinal regression analyses that used an ordered smoking outcome at modal age 19/20 (0=no smoking, 1=nondaily smoking, and 2=daily smoking).

Results

Table 1 presents descriptive statistics for initiation and progression samples at modal age 19/20. Average pack price was $5.68 (initiation sample) and $5.56 (progression sample; ranges of $3.52 to $10.63). Difference in average pack price across base years is presented in Supplemental Figure 1. At modal age 19/20, cigarette use was 4.5% and daily smoking was 5.2% for initiation and progression samples, respectively.

Table 1.

Weighted Descriptive Statistics for Smoking Initiation and Progression Analyses

	Initiation Sample		Progression Sample
Variables	% or Mean (SE)
Baseline cigarette pack price*	5.68	(0.02)	5.56	(0.01)
Past 30-day smoking	4.5%	(0.23)	~	~
Daily smoking	~	~	5.2%	(0.19)
Sex
Male	54.9%	(0.57)	53.7%	(0.45)
Female	45.1%	(0.57)	46.3%	(0.45)
Race/ethnicity
Non-Hispanic White	64.3%	(0.58)	65.3%	(0.46)
Non-Hispanic Black	15.0%	(0.49)	13.3%	(0.37)
Hispanic	13.5%	(0.42)	14.7%	(0.35)
Asian	5.3%	(0.25)	4.4%	(0.18)
Other	1.9%	(0.15)	2.3%	(0.13)
Parental education
Less than high school	6.7%	(0.31)	7.7%	(0.27)
High school degree	18.4%	(0.46)	19.4%	(0.37)
Some college	19.1%	(0.45)	19.8%	(0.36)
College degree or more	55.8%	(0.57)	53.1%	(0.45)
Census region
Northeast	18.9%	(0.43)	18.5%	(0.34)
Midwest	25.0%	(0.47)	25.2%	(0.37)
South	35.4%	(0.56)	36.0%	(0.44)
West	20.7%	(0.45)	20.3%	(0.36)
% workplaces covered by smoke-free laws	42.0%	(0.53)	38.1%	(0.42)
% hospitality venues covered by smoke-free laws	52.0%	(0.54)	47.9%	(0.44)
% of state that is Black	13.1%	(0.09)	13.1%	(0.08)
% of state that is Hispanic	14.7%	(0.14)	14.4%	(0.11)
% of state below poverty	13.4%	(0.03)	13.4%	(0.03)

Notes. Unweighted N for initiation analyses=9,232; unweighted N for progression analyses=15,141. Attrition weights were used to generate estimates, which account for the complex survey design of the Monitoring the Future study, oversampling of drug users, and attrition; standard errors are in parentheses; all estimates are at baseline except for the smoking outcome measures; pairwise deletion was used for missing data. All estimates presented in the table are from baseline and correspond with the sample size for the first follow up (modal age 19/20).

^*This metric is in $, adjusted for inflation. ~ indicates that particular variable was not included in analyses.

Cigarette Initiation

Main Effects

Table 2 presents logistic regression models for smoking at modal age 19/20 among baseline never smokers. Each one dollar increase in pack price at baseline was associated with a 12% reduction in the odds of smoking initiation, adjusting for covariates (AOR=0.88; 95% CI=0.78, 0.99). Males had higher odds of initiating smoking than females (AOR=1.55; 95% CI=1.27, 1.90), and non-Hispanic Black young adults had lower odds of initiation relative to non-Hispanic White young adults (AOR=0.60; 95% CI=0.39, 0.91). Baseline never smokers in the South, compared to the Northeast, had lower odds of initiation (AOR=0.60; 95% CI=0.39, 0.92).

Table 2.

Weighted Logistic Regression Results for Smoking Initiation and Progression at Follow-up among Monitoring the Future Longitudinal Sample (Baseline Years 2000–2016)

	Follow-up 1 (Modal age 19/20)
	Initiation Results		Progression Results
Variables	AOR	(95% CI)	AOR	(95% CI)
Cigarette pack price (in $ adjusted for inflation)	0.88	(0.78, 0.99)	0.84	(0.76, 0.92)
Male (vs. female)	1.55	(1.27, 1.90)	1.34	(1.15, 1.56)
Race/ethnicity (vs. non-Hispanic White)
Non-Hispanic Black	0.60	(0.39, 0.91)	0.51	(0.36, 0.73)
Hispanic	0.71	(0.46, 1.09)	0.62	(0.44, 0.86)
Asian	0.59	(0.35, 1.01)	0.52	(0.32, 0.84)
Other	0.90	(0.44, 1.84)	1.51	(1.01, 2.27)
Parental education (vs. less than high school)
High school graduate	1.11	(0.61, 2.01)	1.38	(0.90, 2.10)
Some college	1.10	(0.61, 1.97)	1.19	(0.78, 1.83)
College degree or more	1.04	(0.60, 1.83)	0.92	(0.61, 1.38)
Census region (vs. Northeast)
Midwest	0.97	(0.69, 1.39)	0.81	(0.64, 1.04)
South	0.60	(0.39, 0.92)	0.65	(0.48, 0.87)
West	0.99	(0.65, 1.51)	0.66	(0.48, 0.92)
% workplaces covered by smoke-free laws	0.79	(0.55, 1.15)	1.00	(0.73, 1.37)
% hospitality venues covered by smoke-free laws	1.03	(0.70, 1.51)	0.70	(0.51, 0.97)
% of state that is Black	5.06	(0.67, 38.05)	2.23	(0.52, 9.61)
% of state that is Hispanic	0.88	(0.24, 3.29)	1.97	(0.71, 5.45)
% of state below poverty	29.98	(0.41, 2206.65)	0.09	(0.01, 2.26)

Notes. Unweighted N for initiation model=9,232; unweighted N for progression model=15,141; 95% confidence intervals are in parentheses; administrative price, and county and state-level controls are measured at baseline; % workplaces and hospitality venues covered by smoke-free laws are measured at the county level. Bolded odds ratios indicate statistical significance (p<.05). Initiation analyses examined past 30-day smoking among baseline never smokers; progression analyses examined daily smoking among baseline never, previous, and nondaily smokers.

Interaction Results

Interactions are shown in Table 3. Multiplicative interaction results were not statistically significant for gender and race/ethnicity (gender, p=.83; race/ethnicity, p=0.31); additive interactions were not statistically significant (gender, p=0.69; race/ethnicity, p=0.22). The joint F-test showed the parental education and pack price interaction was not statistically significant (p=0.24). The global test for the additive interaction between parental education and pack price was nonsignificant (p=0.21). The relationships for the covariates not included in the interaction terms were similar to the relationships found for the covariates in the main effect models.

Table 3.

Partial and Interaction Effects for Initiation and Progression at Follow-up in Monitoring the Future Longitudinal Sample (Baseline Years 2000–2016)

	Follow-up 1 (Modal age 19/20)
	Initiation Results		Progression Results
Variables	AOR	(95% CI)	AOR	(95% CI)
Model 1
Cigarette pack price (in $ adjusted for inflation)	0.87	(0.75, 1.00)	0.77	(0.69, 0.86)
Male (vs. female)	1.41	(0.58, 3.45)	0.60	(0.30, 1.19)
Cigarette pack price X Sex	1.02	(0.87, 1.19)	1.16	(1.02, 1.32)
Significance test for interactions		p-value		p-value
Multiplicative		0.83		0.06
Additive		0.69		0.15
Model 2
Cigarette pack price (in $ adjusted for inflation)	0.88	(0.78, 1.00)	0.84	(0.75, 0.93)
Race/ethnicity (vs. non-Hispanic White)
Non-Hispanic Black	0.22	(0.04, 1.21)	0.18	(0.07, 0.51)
Hispanic	6.68	(0.64, 69.34)	0.72	(0.21, 2.45)
Asian	0.58	(0.12, 2.87)	0.58	(0.09, 3.52)
Other	10.67	(0.08, 1470.54)	14.00	(0.98, 200.00)
Cigarette pack price X Race/ethnicity
X Non-Hispanic Black	1.19	(0.89, 1.60)	1.22	(1.01, 1.47)
X Hispanic	0.66	(0.42, 1.03)	0.97	(0.78, 1.21)
X Asian	1.00	(0.77, 1.32)	0.98	(0.71, 1.36)
X Other	0.62	(0.24, 1.63)	0.64	(0.37, 1.09)
Overall significance test for interactions		p-value		p-value
Multiplicative		0.31		0.17
Additive		0.33		0.05
Model 3
Cigarette pack price (in $ adjusted for inflation)	0.58	(0.41, 0.82)	0.58	(0.38, 0.89)
Parental education (vs. less than high school)
High school graduate	0.14	(0.02, 1.20)	0.21	(0.03, 1.66)
Some college	0.18	(0.02, 1.54)	0.23	(0.03, 1.95)
College degree or more	0.10	(0.01, 0.66)	0.11	(0.01, 0.89)
Cigarette pack price X Parental education
High school graduate	1.49	(1.00, 2.21)	1.46	(0.94, 2.25)
Some college	1.41	(0.96, 2.09)	1.39	(0.89, 2.17)
College degree or more	1.58	(1.11, 2.24)	1.50	(0.98, 2.30)
Overall significance test for interactions		p-value		p-value
Multiplicative		0.24		0.27
Additive		0.33		0.15

Notes. Unweighted N for initiation model=9,232; unweighted N for progression model=15,141; 95% confidence intervals are in parentheses; administrative price, and county and state-level controls are measured at baseline; all models also include all covariates included in Table 2 (not shown): sex, race/ethnicity, parental education, census region, % workplaces and hospitality venues covered by smoke-free laws are measured at the county level, % of state that is Black, Hispanic, and below poverty. Model 1 includes only the interaction for sex, Model 2 only includes the interaction for race/ethncitiy, and Model 3 only includes the interaction for parental education. Bolded odds ratios indicate statistical significance (p<.05). Initiation analyses examined past 30-day smoking among baseline never smokers; progression analyses examined daily smoking among baseline never, previous, and nondaily smokers. Interaction p-values based on Benjamini-Hochberg tests.

Progression to Regular Cigarette Smoking

Main Effects

Daily smoking results at modal age 19/20 are also presented in Table 2. Each one dollar increase in pack price at baseline was associated with 16% reduction in the odds of daily smoking, adjusting for covariates (AOR=0.84, 95% CI=0.76, 0.92). Males had higher odds of progression to regular smoking compared to females (AOR=1.34, 95% CI=1.15, 1.56). Relative to non-Hispanic White young adults, non-Hispanic Black (AOR=0.51, 95% CI=0.36, 0.73), Hispanic (AOR=0.62, 95% CI=0.44, 0.86), and non-Hispanic Asian (AOR=0.52, 95% CI=0.32, 0.84) young adults had lower odds of progression to regular smoking, but other races/ethnicities had higher odds of progression (AOR=1.51, 95% CI=1.01, 2.27). Individuals in the South and West census regions compared to the Northeast had lower odds of progression (respectively, AOR=0.65, 95% CI=0.48, 0.87; AOR=0.66, 95% CI=0.48, 0.92). As the percent of residents covered by smoke-free hospitality venues increased, the odds of regular smoking progression decreased (AOR=0.70, 95% CI=0.51, 0.97).

Interaction Results

Interactions for smoking progression are shown in Table 3. The multiplicative interaction between baseline pack price and gender was statistically significant, but it was not statistically significant after adjusting for multiple testing (p=.058). Higher pack price was more strongly associated with lower daily smoking at follow-up for females compared to males (Figure 1). The additive interaction for gender was not statistically significant (p=.15). The multiplicative race/ethnicity interaction was not statistically significant (p=0.17). The joint test for the additive interaction for race/ethnicity was not statistically significant (p=.05), even though the individual difference between non-Hispanic White young adults compared to non-Hispanic Black young adults was significant (p<.01). Daily smoking progression was higher for non-Hispanic White young adults at lower levels of price, but similar to other races/ethnicities at higher levels of price. The joint F-test was not statistically significant (p=0.27) for the multiplicative interaction for parental education. The global test for the additive interaction for parental education was not statistically significant (p=0.15); yet the individual AME of pack price was stronger for the parental education of less than a high school degree compared to college or more (p=.049). Again, no major differences in effect sizes or statistical significance were found for covariates relative to main effect results.

Figure 1.

Predicted Probability of Daily Smoking at Follow-Up According to Cigarette Pack Price by Gender

Note: high and low measures of price per pack in dollars were based on the 10^th and 90^th percentile of the tax measure’s distribution.

Sensitivity Analyses

The relationship between pack price and daily smoking at follow-up remained relatively unchanged in terms of effect size and statistical significance (AOR=0.90; 95% CI=0.82, 0.99), adjusting for baseline smoking history. Relative to never smokers, previous and non-daily smokers had higher odds of daily smoking at follow-up (respectively, AOR=5.12; 95% CI=4.09, 6.41; AOR=17.11; 95% CI=13.67, 21.42). The multiplicative interaction between baseline pack price and smoking history showed the relationship between higher prices and lower odds of daily smoking was stronger for never smokers compared to non-daily smokers (AOR=1.23; 95% CI=1.02, 1.48). The interaction between the binary cohort year (before vs. after 2009) and pack price was not statistically significant for either initiation or progression. The interaction between each base year and pack price was not statistically significant, indicating that the relationship between pack price and smoking initiation and progression did not change over time. Results were unchanged after adjusting for follow-up education, student status, and employment. The relationship between baseline pack price and smoking initiation at modal ages 21/22 and 23/24 among baseline never smokers was not statistically significant (Supplemental Table 1). All multiplicative and additive interactions for smoking initiation at modal ages 21/22 and 23/24 were not statistically significant (Supplemental Table 2). A dollar increase in price per pack at baseline was associated with a 14% decrease in the odds of daily smoking at modal age 21/22 (AOR=0.86, 95% CI=0.79, 0.95), and a 12% reduction at modal age 23/24 (AOR=0.88, 95% CI=0.79, 0.90), adjusting for covariates (Supplemental Table 1). Multiplicative and additive interactions were not statistically significant for smoking progression at modal ages 21/22 or 23/24 (Supplemental Table 2).

Removing any lifetime smokers from previous waves showed that the results did not change for smoking initiation. Removing any current daily smokers from previous waves rendered the association between baseline pack price and subsequent daily smoking nonsignificant at modal ages 21/22 and 23/24. Adjusting for change in pack price between baseline and modal age 19/20 did not change results for initiation or progression. Ordinal regression results indicated that pack price was similarly related to progression, supporting results found with the dichotomous measure of daily smoking.

Discussion

Cigarette prices in late adolescence were related to lower odds of cigarette smoking initiation and progression in young adulthood, indicating that price effects in late adolescence extend into young adulthood. Previous cross-sectional research on youth demonstrates that higher state pack prices are associated with lower smoking (and daily smoking) initiation among 8th graders, along with other smoking outcomes among 10^th and 12^th graders, such as lower smoking intensity.³³ We extend this research by demonstrating that pack prices in 12^th grade are related to smoking initiation and progression at age 19/20. Sensitivity analyses showed that the impact of pack price in adolescence did not extend beyond age 20 for initiation, but it was related to smoking progression up to age 24. This is important because smoking progression occurs for a sizable number of young adults through age 25.^4,6 However, this relationship between pack price and progression at later ages did not hold in all sensitivity analyses. Smoking initiation and progression to regular smoking occurs for the majority of smokers between adolescence and young adulthood.^6–8,34 Intervening during these ages, which are marked by the exploration of identities, roles, and life transitions³⁵, is key, and pack price in late adolescence is an intervention that can help to prevent smoking initiation, escalation of use, and nicotine dependence in young adulthood.

Tobacco-Related Disparities

Our findings generally support previous cross-sectional research that found no major differences across gender, race/ethnicity, or parental education in terms of pack price and cigarette smoking initiation and progression outcomes.³³

Disparities Based on Gender

Previous cross-sectional research on tobacco taxes have provided mixed results for the impact of cigarette prices on cigarette smoking according to gender among youth and adults.^36,37 For example, while Fleischer et al.³³ found no differential relationships based on gender, other research shows females have a larger price response compared to males.³⁸ State-level research demonstrates adult females are more price sensitive than males³⁶; whereas, repeated cross-sectional research shows youth males are more price sensitive price.³⁷ However, this research did not address smoking progression or young adulthood. A limited amount of longitudinal research on young adults shows that price has a consistent impact on smoking initiation for females but not males²⁸; this research did not address smoking progression. We found no major differences across gender in terms of sensitivity to pack price (for both smoking initiation and progression in young adulthood). Even though multiplicative interactions for smoking progression indicated a stronger response to price for females, we could not make definitive conclusions about this interaction after accounting for multiple testing. Additional interventions are needed to disrupt progression to regular smoking among males during the critical period of young adulthood, as males are more likely than females to rapidly progress to regular smoking.⁷

Disparities Based on Race/Ethnicity

Cigarette prices in late adolescence were not differentially associated with initiation according to race/ethnicity, corroborating previous cross-sectional research.³³ Marked disparities in prevalence and initiation rates from adolescence to young adulthood exist for certain racial/ethnic groups⁹ and studies find that tobacco taxation can be effective in all racial/ethnic groups and may even reduce disparities associated with race/ethnicity in older age groups.³⁹ However, our results indicate that the prospective relationship between pack price in late adolescence may not impact these disparities in young adulthood. The association between pack price and smoking progression also did not differ by race/ethnicity, and we could not definitively conclude that the impact of price was stronger in certain racial/ethnic subgroups. In sum, our findings suggest that other policies with higher potential to reduce racial/ethnic disparities in smoking initiation and progression (e.g., menthol bans) are needed in order to reach racial/ethnic groups most affected by tobacco use during adulthood.⁴⁰

Disparities Based on Parental Education

The association between pack price and cigarette smoking initiation and progression did not differ by parental education, supporting a limited amount of previous cross-sectional research.³³ These results contradict other cross-sectional research that show low-SES adults and adolescents have a stronger response to tobacco prices compared to high-SES adults and youth;^17,38,41 however, these contradictory studies were not conducted during the same time period as the current study, and they did not examine national samples of young adults over time. There is limited research on tobacco prices and disparities among national samples of young adults over time to which we could compare our results. While our findings suggest potential differences across SES groups for smoking initiation at age 19/20, we could not draw definitive conclusions about statistical significance and strength of these differences. We also found no differential impact of pack price in late adolescence on progression to regular smoking in young adulthood across levels of parental education. Therefore, strategies that address SES-related disparities in smoking progression between ages 19 and 25 are especially needed,⁷ as low-SES groups are more likely to rapidly progress to regular smoking in young adulthood compared to high-SES groups.⁴

Implications for Practice

While taxes are effective for preventing initiation and progression in young adulthood, taxes should be implemented in conjunction with complementary programs in order improve the impact of taxes on tobacco-related disparities among young adults. Evidence-based interventions, if implemented with tobacco tax increases, could help to reduce disparities in young adulthood. Such interventions can address multilevel factors during adolescence and young adulthood,⁴³ which subsequently reduce smoking later in life.⁴³ These programs can be tailored to subpopulations, and implementing programs that are strategically tailored to priority populations in conjunction with tax increases could increase impact on smoking initiation and progression disparities.

Limitations

The sample consisted only of participants who were selected from MTF baseline data, which includes students attending school. Previous research indicates school-based surveys have fewer biases for measuring substance use than other types of surveys.⁴⁴ We examined smoking initiation over a time when the tobacco landscape changed considerably, particularly for youth. Policies were changing and new tobacco products (e-cigarettes) were being introduced into the market. Moreover, smokeless tobacco use among young adults had increased.^10,34 E-cigarette use started to increase among youth in 2014, yet e-cigarette use increased most dramatically at the end of the current study period,² potentially reducing confounding of any potential impact of e-cigarette use in the current analysis. Additionally, we found no differences over time in the relationship between pack price and smoking initiation and progression, indicating that trends in e-cigarette and smokeless tobacco use over time likely did not dramatically influence results. However, we acknowledge that these results might not hold in later years given increases in e-cigarette use since 2017.² Our ability to investigate how e-cigarette use at baseline was associated with cigarette use in the current study was limited due to data characteristics including sample size and available e-cigarette items within the longitudinal MTF sample. (With a limited sample size, inclusion of current e-cigarette use increased standard errors of the price coefficients, but otherwise did not substantially affect the results). We included controls for policies that are related to differences in tobacco policy contexts across states, which also accounted for dimensions of the changing tobacco landscape. We adjusted for local policies, but reliable tobacco tax data at the local level is not available over time.

Conclusion

Higher cigarette prices during late adolescence are prospectively associated with lower odds of cigarette smoking initiation and progression to regular smoking in early young adulthood, which is a critical age for smoking behavior. The impact of higher prices on health equity is mixed. We found no definitive differences in the prospective relationship between pack price and smoking initiation and progression according to gender, race/ethnicity, or parental education. Tobacco tax increases should be implemented in conjunction with tailored prevention programs geared toward reducing smoking among young adults who are disproportionately affected by tobacco.

Supplementary Material

1

Supplemental Figure 1. Average Pack Price Across Base Years

Implications and Contribution.

Cigarette smoking initiation and progression to regular smoking increasingly occurs in young adulthood. In a national sample, higher cigarette prices in 12th grade help to prevent smoking initiation and progression in young adulthood; however, complementary policies and programs are needed to increase the impact of tobacco taxes on tobacco-related disparities.