California’s Tobacco 21 Minimum Sales Age Law and Adolescents’ Tobacco and Nicotine Use: Differential Associations Among Racial and Ethnic Groups

Abstract

Objective

A California, USA, law raised the minimum tobacco sales age to 21 (T21) on June 9, 2016. We investigated whether T21 was associated with reductions adolescents’ use of tobacco cigarettes, smokeless tobacco, and electronic cigarettes and whether these associations differed across racial and ethnic groups.

Methods

Secondary analyses of data from 2,956,054 7^th, 9^th, and 11^th grade students who participated in the California Healthy Kids Survey from 2010-11 to 2017-18.

Results

Multilevel mixed effects logistic regression analyses showed that T21 was associated with reduced prevalence of lifetime smokeless tobacco and e-cigarette use and past month smokeless tobacco use in the overall student population. T21 was associated with increases in prevalence of past month e-cigarette use. Moderation analyses indicated differences by racial and ethnic groups. Notably, T21 was associated with reductions in lifetime and past-30-day use of all tobacco and nicotine products among Latinx youth. The findings were more mixed for other racial and ethnic groups. Slopes analyses indicated that T21 was associated with accelerated downward trends for 30-day cigarette and smokeless use; moderated trends for lifetime cigarette smoking such that downward slopes became less steep; and reversed downward trends for e-cigarette use. Changes in slopes varied across racial and ethnic groups.

Conclusions

Our findings highlight the importance of understanding the complex associations that T21 and other tobacco control policies have with the use of different tobacco and nicotine products among racial and ethnic groups. Future research should investigate mechanisms underlying these differences to inform tobacco control efforts.

INTRODUCTION

Although cigarette smoking and smokeless tobacco use by adolescents have declined over the past several decades,[1] they remain serious health issues. The mean age of initiation of smoking among adults who have ever smoked is 15.3 years and the mean age of beginning daily smoking is 18.2 years.[2] Only 13% of adult smokers report that they began smoking after the age of 18 years. The rapid rise in e-cigarette use among adolescents raises additional health concerns and may increase the likelihood of cigarette smoking.[3, 4] Preventing initiation of tobacco and nicotine use by adolescents is a public health priority.

Minimum legal sales age 21 policies for tobacco

Increasing the minimum legal sales age (MLSA) for tobacco and nicotine products to 21 years [5-7] has been recommended to reduce adolescents’ access to tobacco and nicotine products by making purchase more difficult. These policies may also decrease availability of these products through social sources by making it more difficult for peers to obtain and share them. T21 policies may also denormalize tobacco and nicotine use[8] or increase beliefs about the risks associated with use.

T21 laws and adolescents’ tobacco and nicotine use

Studies have investigated the associations of T21 laws with adolescents’ smoking behaviors. In Needham, Massachusetts significant reductions were observed in adolescents’ smoking prevalence and self-reported purchases of cigarettes relative to neighboring communities without a T21 law.[9] Similar reductions in adolescents’ smoking rates were found in a study of community-level T21 policies across the U.S.[10] Greater decreases in recent smoking and established smoking have been found among young adult smokers in communities implementing T21 laws.[11] In contrast, New York City’s T21 law was not associated with significant reductions in adolescents’ tobacco use compared to control sites.[12] Possible explanations were the already strict tobacco control policies in New York City (e.g., high taxes, tobacco retailer licensing requirements) and the lack of retailer compliance with new T21 law. A majority of students living in cities or localities of the state that implemented T21 policies in Kansas between November 2015 and August 2018 were unaware of the minimum legal age for tobacco sales and use.[13] Tobacco non-users were less likely to know of the law than were users, although those who were aware of the T21 law were less likely to report that they intended use tobacco. To our knowledge, no studies have investigated associations of T21 laws with smokeless tobacco or e-cigarette use, nor examined possible differential associations of T21 laws with tobacco use among adolescents across racial and ethnic groups.

T21 laws and health disparities

T21 laws may differentially affect young people from different racial and ethnic backgrounds. Attempts to denormalize tobacco use may further stigmatize marginalized groups, leading to smaller impacts or even reactance.[14-16] Differences in public support for raising the MLSA for tobacco to 21 years might undermine the effectiveness of T21 policies in some communities, though support for T21 is consistently high across racial and ethnic groups.[17] In some studies, including in California, greater support for T21 laws has been found among African American and Latinx,[18-20] suggesting that T21 laws are likely to be as effective, if not more so, in these communities. T21 laws may lack support from tobacco retailers, reducing their willingness to comply with the law. For example, despite broad community support, some evidence suggests that New York City’s T21 reduced retailer compliance.[21] In contrast, a study in California found that a great majority of tobacco retailers (99%) were aware of T21, a majority (61%) supported it, and sales to underage decoys decreased after its implementation.[19] Tobacco compliance rates, however, are lower in neighborhoods with higher percentages of African Americans and Latinx, suggesting T21 laws may be less effective for youth in these neighborhoods.[22] Relatedly, exposure to tobacco outlets and marketing, including point-of-sale advertising, may be greater for young people living in low SES and minority neighborhoods,[23-26] which also may undermine T21 laws and increase risk for tobacco and nicotine use by these youth.

Purpose of this study

California implemented a T21 law in June 2016. We investigated the associations of this law with adolescent prevalence of cigarette smoking, smokeless tobacco use, and e-cigarette use using data from the California Healthy Kids Survey (CHKS) 2010-11 through 2017-2018. We undertook supplemental analyses to investigate whether T21 was associated with changes in trends (slopes). We considered differential associations of T21 with tobacco and nicotine use and trends in use among specific racial and ethnic groups (African Americans, Asian Americans, American Indian/Alaska Natives, Native Hawaiian/Pacific Islanders, non-Latinx Whites, and Latinx). Considering previous research and social-ecological models of substance use,[27] we hypothesized that the implementation of T21 in California would be associated with overall reductions in the prevalence of cigarette smoking, smokeless tobacco use, and e-cigarette use. We expected T21 to moderate the secular trends in tobacco and nicotine use such that downward trends would be accelerated. We made no specific hypotheses about racial and ethnic differences, but considered these questions to be exploratory.

METHOD

Study sample

California Health Kids Survey.

We analyzed data from a statewide sample of 7^th, 9^th, and 11^th graders from public schools in California that participated in the California Healthy Kids Survey (CHKS) from 2010-11 to 2017-18. Data collection is staggered such that it occurs in approximately half of schools each school year, and in any given school in alternate years.[28] Minimum CHKS requirements include participation by students in 7th, 9th, and 11th grades with a targeted response rate of 70% using passive or active parental consent. Based on CDE data, we estimate for the 2017-2018 and 2018-2019 survey years that 71% of schools with 7^th, 9^th, and 11^th grades participated. Using a random sample of high schools from 48 school districts in 19 counties across several years we found the average response rates were 73% for 9^th graders and 68% for 11^th graders, with an overall rate of 71%. Additional details on the CHKS methodology are in Appendix 1.

Sample characteristics.

The study sample includes 2,956,054 students who provided complete data on tobacco and nicotine use and other study variables. Because the survey did not include items about e-cigarette use until 2013, fewer students are included in those analyses (N = 2,069,487). Of the study sample, 51% were female, 49% were Latinx, 22% identified as non-Latinx White, 1% as American Indian/Alaska Native, 2% as Native Hawaiian/Pacific Islander, 11% as Asian American, 4% as Black or African American, and 12% as more than one race or unknown race/ethnicity. About 34% of students in the study sample were in 7^th grade, 35% in 9^th grade, and 31% in 11^th grade.

Measures

Tobacco and nicotine use.

Students were asked how many times they had smoked a whole cigarette, used smokeless tobacco, and used electronic cigarettes [2013-14 onward] in their lifetime, each with 6 responses from “0 times” to “7 or more times”. They were also asked how many days they smoked or used each product in the past 30 days, with 6 responses from “0 days” to “20 to 30 days”. Because of the skewed distribution of the responses, we dichotomized these measures (1 = any lifetime/past month use).

Demographics.

Students were asked their sex, grade, race, and ethnicity. Sex was coded as a dummy variable (1 = Female). Grade was coded into two dummy variables, with 7^th grade as the reference group. Race/ethnicity was coded as a series of dummy variables, with non-Latinx White as the reference group.

Analysis plan

The primary analyses comprised mixed-effects logistic regressions to account for nesting of students within schools. The school-level ICCs were 0.38 and 0.29 for lifetime and past 30-day smoking, respectively; 0.23 and 0.17 for smokeless tobacco use, and 0.25 and 0.21 for e-cigarette use. The models included a linear term for survey year to control for secular trends, which allowed us to determine whether post-T21 prevalence of tobacco and nicotine use differed from what would be expected given the trends in use prior to T21 [29], a common approach in interrupted time series studies evaluating the effects of policies or public health interventions [29, 30]. A dummy variable was included to represent the implementation of T21, coded as 0 for 2010-2015 and as 1 for 2016-2018. Grade, sex, and race/ethnicity were entered as covariates. Overall associations of T21 with tobacco and nicotine use behaviors in the general adolescent population were first examined (Model 1). To test for differential associations of T21 with prevalence of tobacco and nicotine use, the race/ethnicity x T21 interactions were examined (Model 2). Because non-Latinx White students were the reference group, the overall T21 effect in Model 2 represents the association for that group and the interaction terms test the differences between that association and those of the other racial/ethnic groups. When significant interactions were found, an additional model simultaneously estimated T21 effects for each racial and ethnic group, controlling for sex, grade, and survey year.

For the supplemental analyses testing changes in trends, an initial model (Model 1) included the T21 x survey year interaction term to test whether T21 was associated with changes in the slopes for the overall population. The dummy variable representing implementation of T21, race/ethnicity, grade, and sex were also included as predictors. Model 2 added the 3-way race/ethnicity x T21 x survey year interaction terms to test whether T21 was differentially associated with changes in trends across the racial and ethnic subgroups. Model 2 also included all relevant lower-level 2-way interactions. When there were significant 3-way interactions, we disaggregated the effects by racial/ethnic subgroup.

RESULTS

Overall, 11% students reported any lifetime cigarette smoking and 6% reported smoking at least one cigarette in the past month. In all, 5% of students reported any lifetime smokeless tobacco use and 3% reported using it in the past month. Use of e-cigarettes was highest, with 21% of students reporting any lifetime use and 9% reporting past month use.

Cigarette smoking

Prevalence of lifetime cigarette smoking.

Associations of T21 with prevalence of cigarette smoking are summarized in Table 1. Controlling for sex, grade, race, ethnicity, and survey year, T21 was not significantly associated (OR = 1.01) with overall changes in lifetime prevalence of cigarette smoking in the general population of students (Model 1). Racial and ethnic differences in this association were observed (Model 2). Disaggregating the T21 effects by race and ethnicity (Table 2) showed that T21 was significantly associated with increases in lifetime smoking prevalence among non-Latinx White (OR = 1.17), American Indian/Alaska Native (OR = 1.15), Asian American (OR = 1.05), and multi-racial students (OR = 1.04). These increases were significantly smaller for Asian American and multi-racial adolescents than for non-Latinx Whites. Non-significant interactions indicated that the associations for American Indians/Alaska Natives and non-Latinx Whites should be considered equivalent. The association for African American students was non-significant. T21 was associated with a reduction in lifetime smoking prevalence among Latinx students (OR = 0.94).

Table 1.

Results of Mixed Effects Logistic Regressions Testing Associations Between California’s T21 Law and Prevalence of Cigarette Smoking (OR, 95% CI)

	Lifetime Smoking		Past 30-Day Smoking
Predictor	Model 1	Model 2	Model 1	Model 2
T21	1.01 (1.00, 1.03)	1.17 (1.14, 1.20)**	0.99 (0.97, 1.01)	1.08 (1.05, 1.12)**
Survey year	0.82 (0.81, 0.82)**	0.82 (0.81, 0.82)**	0.82 (0.81, 0.82)**	0.82 (0.81, 0.82)**
9th Grade1	3.36 (3.21, 3.51)**	3.36 (3.21, 3.51)**	2.17 (2.06, 2.27)**	2.16 (2.06, 2.27)**
11th Grade1	5.93 (5.68, 6.20)**	5.94 (5.68, 6.21)**	3.29 (3.13, 3.45)**	3.29 (3.13, 3.45)**
Female	0.77 (0.76, 0.78)**	0.77 (0.76, 0.78)**	0.72 (0.71, 0.73)**	0.72 (0.71, 0.73)**
American Indian/Alaska Native2	1.62 (1.57, 1.68)**	1.63 (1.57, 1.69)**	1.70 (1.63, 1.78)**	1.68 (1.61, 1.77)**
Native Hawaiian/Pacific Islander2	1.12 (1.09, 1.15)**	1.14 (1.11, 1.18)**	1.12 (1.08, 1.17)**	1.13 (1.08, 1.18)**
Asian American2	0.47 (0.46, 0.48)**	0.48 (0.47, 0.49)**	0.49 (0.48, 0.50)**	0.49 (0.48, 0.51)**
African American2	0.89 (0.87, 0.91)**	0.92 (0.89, 0.92)**	1.08 (1.05, 1.11)**	1.10 (1.07, 1.14)**
Multi-racial/unknown2	1.10 (1.10, 1.12)**	1.12 (1.11, 1.14)**	1.08 (1.06, 1.10)**	1.09 (1.10, 1.11)**
Latinx2	1.14 (1.13, 1.16)**	1.19 (1.17, 1.20)**	1.07 (1.05, 1.09)**	1.09 (1.07, 1.11)**
T21 x American Indian/Alaska Native	–	0.99 (0.90, 1.08)	–	1.07 (0.95, 1.20)
T21 x Native Hawaiian/Pacific Islander	–	0.90 (0.83, 0.98)*	–	0.98 (0.86, 1.10)
T21 x Asian American	–	0.90 (0.86, 0.95)**	–	0.96 (0.89, 1.03)
T21 x African American	–	0.82 (0.77, 0.87)**	–	0.89 (0.82, 0.97)*
T21 x Multi-racial/unknown	–	0.89 (0.85, 0.92)**	–	0.92 (0.87, 0.97)*
T21 x Latinx	–	0.81 (0.79, 0.83)**	–	0.87 (0.84, 0.90)**

¹Reference category is 7^th grade.

²Reference category is non-Latinx White.

^*p < 0.05

^**p < 0.001

Table 2.

Results of Mixed Effects Logistic Regressions Testing Associations of California’s T21 Law with Prevalence of Tobacco and Nicotine Use by Race and Ethnicity¹

	Lifetime Smoking		Past 30-Day Smoking
Racial/Ethnic Group	OR	95% CI	OR	95% CI
Non-Latinx White	1.17***	1.14, 1.20	1.08***	1.05, 1.12
American Indian/Alaska Native	1.15***	1.06, 1.26	1.16**	1.04, 1.29
Native Hawaiian/Pacific Islander	1.06	0.97, 1.15	1.06	0.94, 1.19
Asian American	1.05**	1.00, 1.10	1.04	0.97, 1.11
African American	0.96	0.90, 1.01	0.97	0.89, 1.04
Multi-Racial/Unknown	1.04**	1.00, 1.07	1.00	0.95, 1.05
Latinx	0.94***	0.92, 0.96	0.94***	0.92, 0.97
	Lifetime Smokeless Use		Past Month Smokeless Use
Non-Latinx White	1.00	0.97, 1.03	0.95*	0.91, 1.00
American Indian/Alaska Native	1.01	0.92, 1.11	0.94	0.83, 1.07
Native Hawaiian/Pacific Islander	0.99	0.89, 1.11	1.02	0.88, 1.17
Asian American	0.95	0.89, 1.02	0.98	0.90, 1.07
African American	0.92*	0.86, 0.99	0.89**	0.82, 0.97
Multi-Racial/Unknown	0.98	0.94, 1.02	0.93*	0.88, 0.99
Latinx	0.96**	0.94, 0.99	0.92***	0.89, 0.95
	Lifetime E-Cigarette Use		Past Month E-Cigarette Use
Non-Latinx White	0.85***	0.83, 0.87	1.43***	1.38, 1.47
American Indian/Alaska Native	0.76***	0.71, 0.82	1.06	0.97, 1.17
Native Hawaiian/Pacific Islander	0.76***	0.71, 0.80	1.11**	1.03, 1.21
Asian American	0.78***	0.75, 0.80	1.17***	1.12, 1.23
African American	0.68***	0.65, 0.71	0.95	0.89, 1.00
Multi-Racial/Unknown	0.74***	0.72, 0.76	1.14***	1.10, 1.18
Latinx	0.67***	0.66, 0.68	0.96**	0.94, 0.99

^*p < 0.05.

^**p < .01.

^***p < 0.001.

¹All models controlled for survey year, grade in school, race, ethnicity, and sex. The individual group effects of T21 were estimated using models with the race/ethnicity x T21 product terms, but no main effect for T21.

Trends in lifetime smoking.

Trend analyses for cigarette smoking are summarized in supplemental Table S1 in Appendix 1. For lifetime smoking, the overall effect for survey year (secular trend) was negative (OR = 0.82, b = −0.20), indicating a reduction in prevalence of cigarette smoking over time. However, the interaction between T21 and survey year (Model 1) was significant and positive (OR = 1.02), indicating that the downward trend was attenuated following T21 in the overall population, although it remained negative (b = −0.18). None of the 3-way interactions were statistically significant (Model 2), indicating that T21 was not differentially associated with changes in the slopes across the racial/ethnic subgroups.

Prevalence of past 30-day cigarette smoking.

A similar pattern was found for past 30-day cigarette smoking (Table 1). There was no overall association between T21 and monthly smoking (Model 1), but significant interactions (Model 2) indicated differences among the racial and ethnic groups. Disaggregating the effects (Table 2), T21 was associated with increases in the prevalence of past month smoking among non-Latinx White (OR = 1.08) and American Indian/Alaska Native students (OR = 1.16). A significant decrease in the odds of past 30-day smoking associated with T21 was found for Latinx students (OR = 0.94). The associations of T21 with past 30-day cigarette smoking for Native Hawaiian/Pacific Islanders (OR = 1.06) and Asian Americans (OR = 1.04) were similar to that for non-Latinx Whites. The associations for African American and multi-racial students were also non-significant.

Trends in past 30-day cigarette smoking.

Overall, the effect for survey year was negative (OR = 0.82, b = −0.20) indicating a downward trend in 30-day smoking over the years of the study (Model 1, Table S1). There was also a significant negative T21 x survey year interaction (OR = 0.97). That is, the downward trend in 30-day smoking was accelerated somewhat post-T21 (b = −0.23). None of the race/ethnicity x T21 x school year interactions were significant (Model 2), suggesting the change in trend was similar across the racial and ethnic subgroups.

Smokeless tobacco use

Prevalence of lifetime smokeless tobacco use.

T21 was associated with a small, but statistically significant, reduction (OR = 0.97) in prevalence for the overall population, after controlling for demographics and the secular trend (Model 1, Table 3). Moderation analyses (Model 2) indicated racial ethnic differences in this association -- T21 was significantly associated with reductions in use among African American (OR = 0.92) and Latinx students (OR = 0.96). No significant associations between T21 and lifetime smokeless tobacco use were found for the other subgroups.

Table 3.

Results of Mixed Effects Logistic Regressions Testing Associations Between California’s T21 Law and Prevalence of Smokeless Tobacco Use (OR, 95% CI)

	Lifetime Smokeless Use		Past Month Smokeless Use
Predictor	Model 1	Model 2	Model 1	Model 2
T21	0.97 (0.95, 1.00)*	1.00 (0.97, 1.03)	0.93 (0.91, 0.96)**	0.95 (0.91, 1.00)*
Survey year	0.88 (0.88, 0.89)**	0.88 (0.88, 0.89)**	0.88 (0.88, 0.89)**	0.88 (0.88, 0.89)**
9th grade1	1.86 (1.78, 1.94)**	1.86 (1.78, 1.94)**	1.54 (1.47, 1.61)**	154 (1.47, 1.61)**
11th grade1	2.95 (2.82, 3.08)**	2.95 (2.83, 3.08)**	1.99 (1.90, 2.08)**	1.99 (1.90, 2.08)**
Female	0.45 (0.44, 0.45)**	0.45 (0.44, 0.45)**	0.44 (0.43, 0.45)**	0.44 (0.43, 0.45)**
American Indian/Alaska Native2	1.65 (1.59, 1.72)**	1.65 (1.58, 1.72)**	1.98 (1.88, 2.08)**	1.98 (1.88, 2.10)**
Native Hawaiian/Pacific Islander2	0.98 (0.94, 1.03)	0.99 (0.94, 1.03)	1.20 (1.14, 1.27)**	1.19 (1.12, 1.27)**
Asian American2	0.40 (0.39, 0.41)**	0.40 (0.39, 0.42)**	0.50 (0.48, 0.52)**	0.49 (0.47, 0.52)**
African American2	1.04 (1.01, 1.07)*	1.05 (1.02, 1.09)*	1.54 (1.48, 1.59)**	1.55 (1.49, 1.62)**
Multi-racial/unknown2	0.95 (0.93, 0.97)**	0.96 (0.93, 0.98)**	1.02 (1.00, 1.05)	1.03 (1.00, 1.06)
Latinx	0.93 (0.91, 0.94)**	0.93 (0.92, 0.95)**	1.07 (1.05, 1.10)**	1.08 (1.06, 1.11)**
T21 x American Indian/Alaska Native	–	1.01 (0.91, 1.11)	–	0.98 (0.86, 1.12)
T21 x Native Hawaiian/Pacific Islander	–	0.99 (0.89, 1.11)	–	1.07 (0.92, 1.24)
T21 x Asian American	–	0.95 (0.88, 1.02)	–	1.03 (0.94, 1.13)
T21 x African American	–	0.93 (0.86, 1.00)*	–	0.94 (0.86, 1.03)
T21 x Multi-racial/unknown	–	0.98 (0.93, 1.03)	–	0.98 (0.91, 1.05)
T21 x Latinx	–	0.96 (0.93, 0.99)*	–	0.96 (0.92, 1.01)

¹Reference category is 7^th grade.

²Reference category is non-Latinx White.

^*p < 0.05

^**p < 0.001

Trends in lifetime smokeless tobacco use.

In Model 1 (Table S3, Appendix 1), there was a significant negative effect for survey year (OR = 0.88, b = −0.13) showing an overall decrease in lifetime smokeless tobacco use over time. The interaction between T21 and survey year was non-significant (OR = 0.98), suggesting no change in slope after T21. However, there was significant and positive three-way interaction (OR = 1.08) in Model 2 for Latinx students. Disaggregating this effect (Table S2) showed there was a significant acceleration of the downward slope for non-Latinx White students (b = −0.18), while the slope for Latinx students was unchanged after T21 and similar to the overall secular trend (b = −0.13). None of the other three-way interactions were significant, indicating that T21 was not associated with changes in the downward slopes for the other racial/ethnic subgroups.

Prevalence of past 30-day smokeless tobacco use.

Table 3 shows T21 was associated with reductions in past 30-day smokeless tobacco use (OR = 0.93) for the overall sample (Model 1). None of the T21 by race/ethnicity interactions were significant (Model 2), indicating that this association was statistically consistent across all subgroups of students. The overlap among the subgroup ORs can be seen in Table 2.

Trends in past 30-day smokeless tobacco use.

Model 1 (Table S3) showed a significant downward trend across survey years (OR = 0.88, b = −0.12) in past 30-day smokeless tobacco use for the overall population. The interaction between T21 and survey year was also significant and negative (OR = 0.94), indicating the downward trend accelerated after T21 (b = −0.18). A significant 3-way interaction (Model 2) indicated that the change in trend was moderated for Latinx students (OR = 1.10). Further analyses (Table S2) showed that T21 was unrelated to changes in the slope for these students, although it remained negative (b = −0.14). No other three-way interactions were significant, suggesting that the downward acceleration in slopes was similar across all other racial/ethnic subgroups and consistent with the population effect.

E-cigarette use

Prevalence of lifetime e-cigarette use.

T21 was associated with a substantial reduction (OR = 0.72) in lifetime e-cigarette use overall (Table 4). The moderation analysis (Model 2) indicated that the associations were stronger for all racial/ethnic groups relative to non-Latinx Whites. Nonetheless, the associations were consistently negative and robust, with ORs ranging from 0.76 for Latinx students to 0.85 for non-Latinx White students.

Table 4.

Results of Mixed Effects Logistic Regressions Testing Associations Between California’s T21 Law and Prevalence of E-Cigarette Use (OR, 95% CI)

	Lifetime E-Cigarette Use		Past Month E-Cigarette Use
Predictor	Model 1	Model 2	Model 1	Model 2
T21	0.72 (0.71, 0.73)**	0.85 (0.83, 0.87)**	1.09 (1.07, 1.11)**	1.43 (1.39, 1.47)**
Survey year	1.00 (1.00, 1.01)	1.00 (1.00, 1.01)	0.87 (0.86, 0.88)**	0.87 (0.86, 0.87)**
9th grade1	2.66 (2.57, 2.74)**	2.66 (2.57, 2.75)**	2.13 (2.05, 2.22)**	2.13 (2.10, 2.22)**
11th grade1	4.20 (4.10, 4.34)**	4.21 (4.07, 4.35)**	2.84 (2.73, 2.95)**	2.84 (2.73, 2.95)**
Female	0.90 (0.90, 0.91)**	0.90 (0.90, 0.91)**	0.84 (0.83, 0.84)**	0.84 (0.83, 0.84)**
American Indian/Alaska Native2	1.27 (1.23, 1.32)**	1.33 (1.27, 1.39)**	1.38 (1.32, 1.45)**	1.55 (1.46, 1.63)**
Native Hawaiian/Pacific Islander2	1.24 (1.20, 1.28)**	1.29 (1.25, 1.35)**	1.20 (1.15, 1.25)**	1.32 (1.26, 1.38)**
Asian American2	0.48 (0.47, 0.49)**	0.50 (0.49, 0.51)**	0.45 (0.44, 0.46)**	0.49 (0.47, 0.51)**
African American2	0.92 (0.90, 0.94)**	1.01 (0.97, 1.03)	0.90 (0.87, 0.92)**	1.04 (1.01, 1.08)*
Multi-racial/unknown2	1.13 (1.11, 1.14)**	1.18 (1.16, 1.21)**	1.05 (1.04, 1.07)**	1.15 (1.13, 1.18)**
Latinx	1.31 (1.30, 1.32)**	1.43 (1.42, 1.45)**	1.17 (1.15, 1.18)**	1.35 (1.33, 1.37)**
T21 x American Indian/Alaska Native	–	0.89 (0.83, 0.96)**	–	0.75 (0.68, 0.82)**
T21 x Native Hawaiian/Pacific Islander	–	0.89 (0.84, 0.95)**	–	0.78 (0.72, 0.85)**
T21 x Asian American	–	0.91 (0.88, 0.94)**	–	0.82 (0.78, 0.86)**
T21 x African American	–	0.80 (0.77, 0.83)**	–	0.66 (0.62, 0.70)**
T21 x Multi-racial/unknown	–	0.87 (0.85, 0.90)**	–	0.80 (0.77, 0.83)**
T21 x Latinx	–	0.79 (0.77, 0.80)**	–	0.67 (0.66, 0.69)**

¹Reference category is 7^th grade.

²Reference category is non-Latinx White.

^*p < 0.05

^**p < 0.001

Trends in lifetime prevalence of e-cigarette use.

Model 1 (Table S4) showed no systematic trend for lifetime e-cigarette use across the survey years (OR = 1.00). However, the T21 by survey year interaction was significant (OR = 1.06) indicating that the slope increased somewhat after T21 (b = 0.05) for the overall population, suggesting there was an increase in the rate of growth in e-cigarette use. Moderation analyses (Model 2) found significant negative 3-way interactions for American Indian/Alaska Native (OR = 0.82), Asian American (OR = 0.93), African American (OR = 0.75), multi-racial or unknown race (OR = 0.89), and Latinx (OR = 0.84) students. Disaggregating these effects (Table S2) showed that the slope for African American (b = −0.10) became negative and the downward slope for American Indian/Alaska Native became steeper (b = −0.08). The slope for non-Latinx White students (b = 0.17) became substantially more positive after T21. More modest upward increases in slopes were observed for the other racial and ethnic groups.

Prevalence of past month e-cigarette use.

T21 was associated with increases in past month e-cigarette use in the overall population of students (OR = 1.09). The moderation model, however, indicated significant differences in this association among the racial and ethnic groups. Disaggregating the interactions (Table 2) showed the T21 was associated with increases in past 30-day e-cigarette use among non-Latinx White (OR = 1.43), Native Hawaiian/Pacific Islander (OR = 1.11), Asian American (OR = 1.17), and multi-racial (OR = 1.14) students. In contrast, it was associated with decreases in past month e-cigarette use among Latinx students (OR = 0.96). There were no significant associations for American Indian/Alaska Native (OR = 1.06) or African American (OR = 0.95) students.

Trends in past month e-cigarette use.

Model 1 (Table S4) showed that, overall, there was a decrease in past 30-day e-cigarette use across survey years (OR = 0.80, b = −0.22). However, the T21 x survey year interaction indicated that there was a significant change in the slope after T21 (OR = 1.81), with it becoming positive (b = 0.38). Significant 3-way interactions (Model 2) indicated that this increase was somewhat less pronounced for African American (b = 0.09) and Latinx (b = 0.33) students and more pronounced for non-Latinx White students (b = 0.47).

DISCUSSION

We found that California’s T21 law was associated with reductions in prevalence of lifetime and past month smokeless tobacco use and lifetime e-cigarette use in the general student population, contributing to overall reductions in use and harms associated with these products. We found no overall associations, however, with lifetime and past 30-day cigarette smoking and a positive association with past month e-cigarette use. The lack of overall associations with lifetime and past month cigarette smoking and lifetime smokeless tobacco use may reflect the low rates of use of these products among California adolescents before T21 was implemented. Further reductions may be difficult to attain. The associations of T21 with reduced prevalence of lifetime e-cigarette use, but increased past month e-cigarette use may reflect differential effects of the law on established e-cigarette users as opposed to new e-cigarette smokers. Regular users may be less deterred by T21 or may have established sources and can more easily circumvent the restrictions on sale. These findings, however, should be replicated in future research and possible mechanisms investigated.

The associations of T21 with prevalence of adolescents’ use of other tobacco and nicotine products varied by race and ethnicity. When considering racial and ethnic differences, we found reductions in lifetime prevalence e-cigarette use following T21 across all racial/ethnic groups. These reductions were greater for all racial/ethnic minority groups relative to non-Latinx Whites. Given the potential role of e-cigarettes in initiation to tobacco cigarette use[3, 4] and possible harms associated with e-cigarette use by young people,[31] this is an important finding and may suggest that T21 can help reduce racial/ethnic disparities in the use of e-cigarettes.

The most consistent finding was for Latinx students, who, unlike the other groups, showed significant reductions across all tobacco and nicotine use outcomes. This pattern of findings is consistent with research showing greater support for T21 in Latinx communities,[20] but is inconsistent with other research showing lower retailer compliance rates and greater tobacco marketing in these communities.[22, 23] It suggests that T21 may, in fact, be broadly contributing to reductions in smoking and nicotine use disparities for this group, which represents almost half of all students in the CHKS sample. The findings regarding T21 and Latinx youth, however, need to be replicated.

Other racial groups showed more mixed associations of tobacco and nicotine use with T21. African American students showed reductions in lifetime and past 30-day smokeless tobacco use and in lifetime e-cigarette use, but no significant associations were found for lifetime or past 30-day cigarette use or past 30-day e-cigarette use. American Indian/Alaska Native students showed increases in lifetime and 30-day tobacco cigarette use, reductions in lifetime e-cigarette use, and no changes in smokeless tobacco use following T21. Importantly, T21 may have had adverse effects for some subgroups, such cigarette use among American Indians/Alaska Natives, contributing to health disparities. These unintended consequences may result from numerous factors, including cultural considerations, stigmatization and reactance,[14, 15] exposure to marketing,[23-26] or lack of enforcement of sales laws and compliance in disadvantaged neighborhoods.[22] Understanding why these differences occur across racial/ethnic groups and across tobacco and nicotine products is key to designing effective interventions to support T21. Culturally adapted educational campaigns for retailers,[32] mystery shop programs that reinforce refusal of sales to minors,[33, 34] and enforcement compliance checks that target retailers could be considered.

For the overall population, the downward trend in lifetime cigarette smoking was somewhat attenuated after T21, although it remained negative. For 30-day smoking and 30-day smokeless tobacco use, the downward trend was accelerated. In contrast, for lifetime and 30-day e-cigarette use the downward trends were reversed. This latter finding is troubling and suggests that T21 may not have deterred regular e-cigarette use among adolescents who were already users. T21 was differentially associated with changes in trends across racial and ethnic groups. While downward trends in smokeless tobacco use were accelerated among non-Latinx White students, no changes in trends following T21 were found for this behavior in some groups, notably Latinx students. For e-cigarette use, the observed increases in trends were most apparent for non-Latinx White and less so for African American students. These findings suggest that targeted interventions may be needed to reinforce T21 and slow the growth in use of some tobacco and nicotine products in some populations.

Limitations

Although this study provides mixed evidence about the potential of T21 to prevent or reduce tobacco use among adolescents of different racial and ethnic groups, some limitations must be considered. Although participation rates in CHKS are good, the students surveyed may not be representative of all adolescents in California; students who are frequently absent or not enrolled in public schools are not represented, nor are adolescents in other middle and high school grade levels. Moreover, the findings may not generalize to other states where adolescent tobacco use rates are higher than California. Because of the small number of tobacco and nicotine users, it was necessary to dichotomize the outcomes and it was not possible to consider associations of T21 with frequency of use. While concerns could be raised that the lifetime measures are inappropriate, these rates are meaningful in that fewer lifetime users in total (i.e., previous users + new users) would be expected following T21 if it reduced initiation. Unfortunately, the CHKS did not inquire about cigar smoking and we were unable to evaluate the association of T21 with this outcome. Given the popularity of cigars among some groups, notably African Americans,[35, 36] future research should address this issue. Finally, the data comprise successive cross-sectional surveys and some of the same students are likely included in more than one survey. We were not able to link individual data because the surveys were anonymous. This non-independence of observations, however, is taken into account in the analyses. Nonetheless, the lack of longitudinal data at the individual level and the absence of a control comparison site limit the extent to which causal conclusions can be made.

Conclusion

This study provides preliminary evidence that the implementation of T21 in California was associated with decreases in lifetime and past month smokeless tobacco use, and lifetime e-cigarette use in the general population. However, such reductions differed across racial and ethnic groups, with increases in the use of some tobacco and nicotine products for some groups after T21. Additional research on these disparities could increase our understanding of why they occurred and how they might be countered.

WHAT THIS PAPER ADDS.

• Raising the tobacco purchase age to 21 years (T21) is a recommended strategy to reduce adolescents’ tobacco and nicotine use.

• Limited research has investigated associations of T21 laws with adolescents’ use of products other than tobacco cigarettes or considered racial and ethnic differences in these associations.

• California’s T21 law was associated reductions in lifetime smokeless tobacco and e-cigarette use and past month smokeless tobacco use in the general student population, but with increases in past month e-cigarette use.

• Differential associations of California’s T21 law with tobacco and nicotine use were observed across racial and ethnic groups, with the most consistent reductions in tobacco and nicotine use found for Latinx youths.

• Future research should focus on identifying mechanisms underlying disparities associated with T21 laws to better inform prevention efforts.