Impact of Hispanic Ethnicity on Adolescent Tobacco Use Estimates in the United States

Abstract

Introduction:

National estimates of tobacco use are reported by racial category after excluding all individuals who identify as Hispanic (e.g., non-Hispanic [NH], Black, and NH-White). In this study, we assess the impact of excluding Hispanics from racial categories on national estimates of youth tobacco use in the United States.

Aims and Methods:

We pooled four years of Youth Risk Behavioral Surveillance Survey (YRBSS) data (2015–2021). Participants were n = 60 327 high school students (9–12th grade). We report the prevalence of past 30-day (current) use of cigarettes, cigars, smokeless tobacco, and electronic cigarettes, stratified by racial category: (1) American Indian/Alaskan Native (AI/AN); (2) Asian; (3) Black; (4) Native Hawaiian/Other Pacific Islander (NHOPI); (5) White; (6) multiracial; and (7) unspecific. Within racial category, we compared the prevalence for each product between NH and Hispanic youth; comparisons controlled for sex, grade, and year.

Results:

Descriptively, the majority of AI/ANs (75.6%) and Pacific Islanders (58.9%) identified as Hispanics, as did a minority of multiracial (20.6%), White (14.6%), Black (8.2%), and Asian (7.9%) youth. Prevalence estimates were significantly greater for all four tobacco products among Hispanic-Asian and Hispanic-Black youth, relative to their NH counterparts. Conversely, tobacco use was lower among Hispanic-White youth relative to NH-Whites.

Conclusions:

Systematically removing Hispanics from classification within each racial category results in an underestimation of tobacco use among Asian and Black youth, an overestimation of tobacco use among White youth, and unstable estimates among AI/AN and NHOPI youth. These findings challenge the utility of the “race/ethnicity” variable in the study of youth in the United States.

Implications:

Findings highlight the need to reconsider the operationalization of “race/ethnicity” which currently excludes Hispanics from each racial category. Hispanic ethnicity appears to function as a “within-group” difference; the study of these within-group differences may provide unique insights into tobacco use disparities.

Introduction

Hispanics are a large, young, and growing population in the United States. As of 2022, there are ~62 million Hispanics in the United States and Hispanics are the largest demographic in the two most populous states (California and Texas). Approximately 65% of Hispanics in the United States are under the age of 40 and, from 2000 to 2020, ~18% of the cumulative population growth in the United States were Hispanics under the age of 18.^1,2 These demographic indicators show that the future of public health research, practice, and policy must consider Hispanics in order to have meaningful impact on the population level. In the United States, tobacco use is the leading cause of preventable death,³ and thus investigating patterns of use among Hispanics is essential to understanding health outcomes among this population.^4–10

Hispanics are a racially diverse population; however, this racial diversity is not accounted for in most epidemiological studies.¹¹ For example, annual prevalence estimates of youth tobacco use reported by the Centers for Disease Control and Prevention (CDC) note that “Hispanic persons could be of any race (White, Black or African American, Asian, AI/AN, Native Hawaiian/Other Pacific Islanders [NHOPI], or multiracial).”⁸ Typically, researchers group race and ethnicity into a single construct (i.e., “race/ethnicity”) and generate four mutually exclusive categories: (1) Hispanic of any race(s); (2) non-Hispanic (NH), White (NHW); (3) NH-Black (NHB); and (4) NH-“Other” (NHO), which is comprised of Asian, American Indian/Alaskan Native (AI/AN), NHOPI, and multiple racial identities.¹² In effect, affirmative responses to Hispanic ethnicity “supersede any race response[s].”¹³ As a result, this approach to coding “race/ethnicity” provides no ability to investigate Hispanic ethnicity as a within-group determinant of health for any racial category (e.g., Hispanic-Black relative to NH-Black).¹⁴

Per the Minority Health Research (MHR) framework,¹⁵ study of health disparities must consider between-group differences (e.g., White relative to Black) and within-group differences (e.g., Asian female relative to Asian male). More recently, studies have explored the overlap of race and Hispanic ethnicity using the MHR framework.¹⁶ A recent analysis of data from high school students in Maryland found that ~44% of Hispanics reported no race at all, 11% identified as Black, ~23.5% identified as White, and 5% identified as multiple racial categories.¹⁶ Results from Webb et al. (2022) reveal substantial within-group variance in substance use (i.e., cannabis, alcohol, and tobacco) that was not observed using the standard, four-group categorization applied in health research. Put another way, youth substance use prevalence estimates reported in national studies are likely to be inaccurate due to the high degree of heterogeneity in their response to questions about race and ethnicity.

Similar patterns of racial heterogeneity in Hispanics have been found in national studies of adults (Study of Latinos [SOL]).¹⁷ In fact, the SOL methodology states: “A high proportion of participants (40% based in waves 1 and 2) did not respond to the question about race. [The race] variable should be avoided or used under very limited circumstances.” The SOL recommendation to disregard racial diversity conflicts with recommendations from leading health authorities. Specifically, several reports by the US Surgeon General¹⁸ and the CDC^19,20 have noted the importance of measuring and addressing racial and ethnic disparities in tobacco use. Thus, research must consider the racial diversity of the Hispanic population in order to explore health disparities among Hispanic individuals.

Study Aims and Hypotheses

The primary aim of this study is to examine differences in tobacco use by Hispanic ethnicity across racial categories. For this study, racial categories are: (1) AI/AN; (2) Asian; (3) Black; (4) NHOPIs; (5) White; (6) “multiracial”; and (7) no response (i.e., missing) among a nationally representative sample of high school students. We conduct these analyses using nationally representative Youth Risk Behavioral Surveillance Survey (YRBSS) data from 2015 to 2021. The outcomes of this study are past 30-day (current) use of four (4) tobacco products: (1) combustible cigarettes; (2) combustible cigars; (3) smokeless tobacco; and (4) electronic cigarettes (e-cigarettes). Data are presented by year and pooled across all years. Findings from this study may inform the ongoing discussion of how race and ethnicity data should be collected and presented, as noted by the OMB¹³ and the American Medical Association (AMA).²¹

Methods

Study Sample

We pooled 4 years (2015–2021) of cross-sectional data from the Youth Risk Behavior Surveillance Survey (YRBSS). The YRBSS uses a three-stage cluster sample design to produce a nationally representative sample of 9–12th grade students in the United States. Data are collected biennially, in odd number years. A total of n = 61 298 high school students completed the YRBSS from 2015 to 2021. We excluded participants with incomplete data on any study variable with the exception of race, for which nonresponse was considered informed missing and categorized as “non-response.”

For each analysis, participants with missing data on the outcome (e.g., cigarette smoking) were removed from that analysis. However, missing data on one outcome (e.g., cigarettes) was not considered an exclusionary criterion for analysis of another outcome (e.g., e-cigarettes), in order to be consistent with the reporting of national figures.⁸ This approach resulted in variations in the analytic sample for each product: n = 57 126 for cigarettes; n = 55 547 for e-cigarettes; n = 56 877 for cigars; and n = 57 092 for smokeless tobacco.

Measure

Ethnicity

Participants were asked, “Are you Hispanic or Latino?” with categories of response “yes” and “no.” Approximately 1.2% of participants had missing data on this question and were excluded from the analysis.

Race

Participants were asked: “What is your race? (Select one or more responses).” The possible responses were: (1) AI/AN; (2) Asian; (3) Black; (4) NHOPIs; and (5) White. Participants who selected multiple racial groups were categorized as (6) “multiracial.” For this study, participants who chose no racial identity were classified as (7) “no response.”

Tobacco Use

Participants were asked to self-report the use of four (4) tobacco products in the past 30 days: (1) combustible cigarettes; (2) cigar products; (3) smokeless tobacco; and (4) electronic cigarettes. The use of each tobacco product was assessed independently. Participants were asked the following root question with product-specific changes: “During the past 30 days, on how many days did you [smoke cigarettes]/[smoke cigars, cigarillos, or little cigars?]/[use chewing tobacco, snuff, dip, snus, or dissolvable tobacco]/[use an electronic vapor product]?” The question and response options were uniform for all four questions across all years of data collection.

Statistical Analyses

All descriptive and analytic figures were weighted to adjust for school and student nonresponse and to make the data nationally representative to high school students across all years. First, we report the percent of each racial category that identifies as Hispanic among all participants (Figure 1) and then report the racial distribution of all Hispanics (Figure 2). Second, we conduct a series of logistic regressions to compare differences in the use of each tobacco product by Hispanic ethnicity (relative to NH ethnicity), among each racial category. Thus, we will make the following comparisons for each tobacco product: (1) NH-AI/AN relative to Hispanic-AI/AN; (2) NH-Asian relative to Hispanic-Asian; (3) NH-Black relative to Hispanic-Black; (4) NH-NHOPI relative to Hispanic-NHOPI; (5) NH-White relative to Hispanic-White; (6) NH-Multiracial relative to Hispanic-Multiracial; and (7) NH-Missing to Hispanic-Missing. All models controlled for sex, grade, and year. Analyses were conducted in Stata 18.0 (College Station, Texas).

Figure 1.

Percent Hispanic by race category among high school students in the United States, 2015–2021.Youth Risk Behavior Surveillance Survey (YRBSS).AIAN is American Indian or Alaskan Native.NHOPI is Native Hawaiian/Other Pacific Islander.Multi-races is anyone who selected more than one racial category.

Figure 2.

Racial identity among Hispanics from 2015 to 2021.

Results

Ethnicity and Race

As seen in Figure 1, approximately 23.7% of high school students identified as Hispanic from 2015 to 2021. Most AI/ANs (75.6%) and NHOPIs (58.9%) identified as Hispanic. Similarly, 20.6% of multiracial youth, 14.6% of White youth, 8.2% of Black youth, and 7.9% of Asian youth identified as Hispanic. Furthermore, approximately 10.4% of the high school students from 2015 to 2021 did not report a racial identity; nearly all of whom (93.3%) identified as Hispanic. As seen in Figure 2, the most common racial identity among Hispanics was “none” (40.1%), followed by White (37.2%), American Indian (7.7%), multiracial (5.5%), and Black (4.8%).

Tobacco Use Prevalence

As seen in Table 1, Hispanic youth (of all races) had a lower prevalence of cigarette smoking (5.9% versus 7.6%), cigar smoking (5.8% versus 6.9%), and smokeless tobacco (3.2% versus 5.0%), relative to NH youth of all races. E-cigarette use was 21.9% for both Hispanic and NH youth.

Table 1.

Tobacco Use by Hispanic Ethnicity and Racial Category among High School Students in the United States, 2015–2021 (n = 59 618)

	Currenta cigarette		Currenta cigar		Currenta smokeless tobacco		Currenta electronic cigarettes
	Non-Hispanic	Hispanic	Non-Hispanic	Hispanic	Non-Hispanic	Hispanic	Non-Hispanic	Hispanic
All youth	7.2% (6.4–8.0)		6.6% (6.0–7.4)		4.6% (4.0–5.2)		21.9% (20.5–23.4)
Current use (95% CI)	7.6% (6.7–8.5)	5.9% (5.1–6.8)	6.9% (6.2–7.7)	5.8% (5.0–6.8)	5.0% (4.4–5.6)	3.2% (2.7–3.7)	21.9% (20.4–23.5)	21.9% (20.0–23.9)
	p < .001		p = .022		p < .001		p = .948
AI/ANb	9.1% (7.4–11.1)		8.8% (6.9–11.1)		6.2% (4.7–8.2)		24.8% (21.3–28.6)
Current use (95% CI)	13.1% (9.5–18.0)	8.0% (6.1–10.5)	13.8% (8.9–20.9)	7.2% (5.4–9.6)	9.6% (6.4–14.3)	5.1% (3.6–7.2)	32.1% (26.3–38.5)	22.2% (18.3–26.8)
	p = .024		p = .018		p = .012		p = .006
Asian	3.9% (2.7–5.5)		3.0% (2.1–4.3)		2.4% (1.7–3.4)		10.7% (8.8–13.0)
Current use (95% CI)	2.8% (1.8–4.3)	14.3% (8.6–22.9)	1.9% (1.2–3.0)	18.5% (10.1–23.9)	1.4% (0.9–2.2)	13.5% (8.1–21.6)	9.2% (7.4–11.4)	31.0% (21.9–41.8)
	p < .001		p < .001		p < .001		p < .001
Black	4.2% (3.3–6.9)		7.2% (6.1–8.4)		3.2% (2.6–3.8)		16.0% (14.6–17.6)
Current use (95% CI)	4.0% (3.3–4.9)	7.6% (5.5–10.5)	6.9% (5.9–8.2)	9.2% (6.9–12.3)	3.0% (2.5–3.7)	5.0% (3.4–7.4)	15.0% (13.5–16.6)	26.4% (22.4–30.7)
	p < .001		p = .054		p = .011		p < .001
NHOPIc	4.7% (3.2–6.9)		6.2% (4.6–8.3)		4.9% (3.2–7.4)		22.5% (18.3–27.4)
Current use (95% CI)	5.4% (3.2–8.7)	4.3% (2.5–7.2)	6.7% (3.9–11.3)	5.8% (3.9–8.7)	5.7% (2.9–11.2)	4.3% (2.7–6.8)	20.7% (14.7–28.3)	23.8% (18.9–29.5)
	p = .512		p = .699		p = .465		p = .459
White	8.3% (7.3–9.4)		6.9% (6.1–7.7)		5.5% (4.7–6.3)		24.2% (22.6–26.0)
Current use (95% CI)	8.8% (7.7–9.9)	5.5% (4.6–6.5)	7.1% (6.3–8.0)	5.3% (4.4–6.5)	5.8% (5.0–6.8)	3.2% (2.5–4.0)	24.6% (22.8–26.5)	22.2% (19.9–24.6)
	p < .001		p = .003		p < .001		p = .062
2 + Races	8.4% (7.0–10.0)		7.9% (6.6–9.4)		4.0% (3.1–5.1)		22.5% (19.9–25.3)
Current use (95% CI)	8.3% (6.9–10.0)	8.6% (6.1–12.2)	7.6% (6.3–9.2)	9.2% (6.6–12.7)	4.0% (3.1–5.1)	3.5% (2.2–5.6)	21.4% (18.7–24.3)	26.4% (21.9–31.5)
	p = .851		p = .274		p = .615		p = .034
Missing	5.1% (4.2–6.2)		5.1% (4.2–6.3)		2.5% (2.0–3.1)		19.7% (17.5–22.1)
Current use (95% CI)	4.6% (2.4–8.4)	5.2% (4.2–6.3)	7.8% (4.4–13.5)	4.9% (3.9–6.1)	8.5% (4.0–17.3)	2.0% (1.6–2.6)	18.6% (11.1–29.4)	19.9% (17.6–22.4)
	p = .706		p = .126		p < .001		p = .787

^aAny use of corresponding product within the past 30 days.

^bAI/AN is American Indian or Alaskan Native

^cBivariate chi-square test comparing prevalence between Hispanic and non-Hispanic youth within stratified sample.

As seen in Table 2, Asian youth who identified as Hispanic had greater odds of reporting e-cigarette use (aOR: 6.15; 95% CI: 2.98 to 12.69), cigarette smoking (aOR: 4.85; 95% CI: 2.90 to 8.10), cigar/cigarillos use (aOR: 9.75; 95% CI: 4.94 to 19.23), and smokeless tobacco use (aOR: 10.96; 95% CI: 4.95 to 24.26), relative to their NH counterparts, controlling for sex, grade, and year. Similarly, Black youth who identified as Hispanic had greater odds of reporting current e-cigarette use (aOR: 2.31; 95% CI: 1.57 to 3.41), cigarette smoking (aOR: 2.02; 95% CI: 1.59 to 2.55), cigar/cigarillos use (aOR: 1.52; 95% CI: 1.13 to 2.04), and smokeless tobacco use (aOR: 1.81; 95% CI: 1.19 to 2.76), relative to their NH counterparts, controlling for sex, grade, and year.

Table 2.

Differences in Past 30-Day Tobacco Use, by Hispanic Ethnicity, Stratified by Racial Category (YRBSS, 2015–2021; N = 59 618)

	American Indian/Alaskan native (n = 1,694)	Asian (n = 3,004)	Black (n = 9,702)	Native Hawaiian or Other Pacific Islander (n = 859)	White (n = 34,330)	Multiple racial categories (n = 4,017)	No race selected (n = 6,721)
	Combustible cigarettes
Hispanic ethnicity
No	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)
Yes	0.63* (0.44–0.90)	4.85*** (2.90–8.10)	2.02*** (1.59–2.55)	1.04 (0.67–1.62)	0.82** (0.71–0.94)	1.30* (1.00–1.68)	1.19 (0.73–1.94)
	Large cigars and cigarillos
Hispanic ethnicity
No	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)
Yes	0.55 (0.30–1.01)	9.75*** (4.94–19.23)	1.52** (1.13–2.04)	1.11 (0.52–2.37)	0.77** (0.63–0.94)	1.38 (0.94–2.02)	0.58 (0.31–1.08)
	Smokeless tobacco products
Hispanic ethnicity
No	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)
Yes	0.56* (0.31–0.99)	10.96*** (4.95–24.26)	1.81** (1.19–2.76)	0.98 (0.42–2.28)	0.57*** (0.43–0.74)	1.01 (0.61–1.69)	0.23*** (0.10–0.51)
	Electronic cigarettes
Hispanic ethnicity
No	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)
Yes	0.60* (0.37–0.98)	6.15*** (2.98–12.69)	2.31*** (1.57–3.41)	0.96 (0.47–1.98)	0.63*** (0.52–0.77)	1.14 (0.76–1.69)	1.09 (0.67–2.12)

^***p < .001;

^**p < .01;

^*p < .05.

All models controlled for sex (male referent), grade (9th referent), and year (2015 as referent).

Youth Risk Behavior Surveillance Survey (YRBSS).

AIAN is American Indian or Alaskan Native.

NHOPI is Native Hawaiian/Other Pacific Islander.

Multi-race is anyone who selected more than one racial category.

White youth who identified as Hispanic had lower odds of reporting e-cigarette use (aOR: 0.63; 95% CI: 0.52 to 0.77), cigarette smoking (aOR: 0.82; 95% CI: 0.71 to 0.94), cigar smoking (aOR: 0.77; 95% CI: 0.63 to 0.94), and smokeless tobacco (aOR: 0.57; 95% CI: 0.43 to 0.74) than those who identified as NH, controlling for sex, grade, and year. A similar relationship was observed among AI/ANs; however, this association may be limited by the proportion of AI/AN youth who identified as Hispanic (75.6%). AI/AN youth who identified as Hispanic had lower odds of reporting current electronic cigarette use (aOR: 0.60; 95% CI: 0.37 to 0.98), cigarette smoking (aOR: 0.63; 95% CI: 0.44 to 0.90), and smokeless tobacco (aOR: 0.59; 95% CI: 0.31 to 0.99) than those who identified as NH, controlling for sex, grade, and year. There was no significant difference in cigar smoking by Hispanic ethnicity (aOR: 0.55; 95% CI: 0.30 to 1.01; p = .053) among AI/AN, controlling for sex, grade, and year.

Among multiracial youth, Hispanic ethnicity was associated with greater odds of cigarette smoking (aOR: 1.30; 95% CI: 1.00 to 1.68). Among youth who did not report any racial category, Hispanic ethnicity was associated with lower odds of smokeless tobacco (aOR: 0.23; 95% CI: 0.10 to 0.51). There were no statistical differences in the use of any tobacco products by Hispanic ethnicity among NHOPI youth.

Discussion

Our study found that tobacco use prevalence differed by Hispanic ethnicity for several racial categories. Notably, we found that Asian and Black youth who identified as Hispanic were more likely to use all four tobacco products, relative to NH counterparts. In practical terms, this means that national prevalence estimates of tobacco use likely underestimate the true prevalence due to systematically removing Hispanics who have higher use rates among Asian and Black youth. Inversely, White youth who identified as Hispanic were less likely to use all four tobacco products, relative to their NH counterparts; this means that national prevalence estimates of tobacco use likely overestimate the true prevalence due to systematically removing Hispanics who have lower use rates within White youth. Our findings further support recent calls for reconsidering the operationalization of race and ethnicity in the surveillance of health behaviors, particularly tobacco use.^22,23

The primary implication of this study is that Hispanic ethnicity should be treated as a within-group variable for all racial groups, rather than as a single, homogeneous category. In the standard reporting procedures of national surveillance data, a participant’s response(s) to the racial category question are not considered if they identify as Hispanic¹³; in effect, Hispanic ethnicity serves as an exclusionary criterion for inclusion within any racial category. Our descriptive data show that Hispanics account for a sizable proportion of participants within each racial category. Our analytic results show that the standard approach to coding Hispanic ethnicity results in biased prevalence estimates; these findings in a national sample are similar to what was observed by Webb et al. (2021) in a state-level analysis of YRBSS data.¹⁶ Furthermore, the use of Hispanic ethnicity to supersede response(s) to the question on race runs counter to National Institutes of Health (NIH) standards for human subject inclusion and enrollment. Notably, the NIH Inclusion Enrollment Report requires research conducted with human subjects to report the number of Hispanic and NH participants expected to be recruited within each racial category (e.g., AI/AN; White). Indeed, the NIH Inc[lusion Enrollment Report template offers a framework for how to report Hispanic ethnicity as a within-group variable; our Supplementary Table S1 offers an example of this using the NIH Inclusion Enrollment Report template using data from 2021.

An abundance of research has recently called for changes to how health disparities are investigated and reported by race and ethnicity, including the Office of Management and Budget (OMB),¹³ the American Medical Association (AMA),²¹ and the American Psychological Association (APA).^24,25 One such recommendation by the APA is to provide information regarding missing data in terms of race, ethnicity, and culture.²⁵ As observed in this study, a plurality of Hispanic youth had missing data on race from 2015 to 2021; a similar rate of missingness was observed in the nationally representative SOL among adults and a state-level analysis of youth in Maryland.¹⁶ In 2023, the OMB recommended collecting data on race and ethnicity via a single item which would include Hispanic ethnicity as one of the selectable choices, along with each racial category. The approach outlined by the OMB would likely result in a substantial reduction in missing data as participants who skipped the race question but responded affirmatively to Hispanic ethnicity would be coded as “Hispanic, only.” As nearly 95% of participants in our study identified as Hispanic, the use of the OMB recommended question would result in less than 1% of participants with missingness on the “race and ethnicity” variable. However, independent of how race and ethnicity data are collected, our analytic results demonstrate the need to analyze the heterogeneity of these two variables in order to accurately estimate the prevalence of tobacco use subpopulations.

This study, along with others,^16,26 offers evidence for the need to reconsider how Hispanic ethnicity is conceptualized and operationalized in health research and broader social science. For example, our data suggest the need to update the definition of Hispanic to more accurately reflect Spanish colonization in regions of the United States (e.g., California, Texas, Florida) and South East Asia (e.g., Philippines).²⁷ To date, research on the diversity of Hispanics has focused largely on national origin, with an emphasis on Central and South America.^4–7 Per the OMB definition, Hispanic ethnicity includes being “…Spanish culture or origin.” As such, individuals of indigenous backgrounds within US regions colonized by Spain (i.e., southwest United States and Florida) would meet the definition of Hispanic, similar to those within Central and South America or the Philippines. This is supported by our data which shows that ~76% of AI/AN youth do identify as Hispanic. Similarly, the ~59% of NHOPI youth who identified as Hispanic demonstrate the need to update the definition of Hispanic to incorporate other regions (e.g., South East Asia) colonized by Spain. For example, individuals who identify as Filipino are classified as Asian¹³ but are often included within samples of Native Hawaiian and Pacific Islanders^28,29 and frequently identify as Hispanic.²⁷ In general, a more historically accurate reflection of colonization is needed to effectively define and measure Hispanic ethnicity, particularly as this population grows in the United States.

Our study has implications for MHR methods and, more specifically, the investigation of within-group differences across racial and ethnic groups. The systematic exclusion of Hispanics from each racial group ultimately generates ethnically stratified subsamples of each racial category (e.g., NH-White, NH-Black, and NH-Asian) which, as quantified in our analysis, results in a critical underestimation or overestimation of prevalence within each racial category. For example, Hispanic-Asian youth had the highest prevalence of using all four tobacco products, including e-cigarettes (31.0%), combustible cigarettes (14.3%), combustible cigars (18.5%), and smokeless tobacco (13.5%). Conversely, NH-Asians had the lowest prevalence of using all four tobacco products, including e-cigarettes (9.2%), combustible cigarettes (2.8%), combustible cigars (1.9%), and smokeless tobacco (1.4%). These ethnic differences in tobacco use prevalence among Asian youth are not observable when using the standard operationalization of Hispanics. Indeed, there is no ability to investigate Hispanic ethnicity as a within-group determinant of health for any racial category. It is plausible that the application of MHR methods, such as those presented in this study, may expand the understanding of health outcomes among Hispanics in the United States.

Building on the implications for MHR methods, this study also has implications for the development of tailored tobacco prevention interventions. For example, smokeless tobacco use and interventions for the same predominate among NH-White youth.³⁰ However, our findings show that the highest prevalence of smokeless tobacco use was among Asian (13.5%) youth who identified as Hispanic, followed by NH-AI/AN (9.6%). The high rate of smokeless tobacco use among Hispanic-Asian youth is reflected in recent studies of young people in Texas^31,32 and California³³; two states with large Hispanic and Asian populations. These findings showcase the need for more research and outreach to prevent tobacco use among diverse populations. Further study on tobacco use across racial and ethnic identities is needed to inform targeted interventions in the United States.

Our findings also indicate the need to explore health behaviors and outcomes among Hispanics beyond measures of nationality and acculturation (i.e., “healthy immigrant hypothesis”).^4–7 Instead, research should apply the same degree of rigor (supported by equitable investment) in exploring racial heterogeneity among Hispanics as has been conducted for NHs. As ~60% of Hispanics identified with at least one racial category, there is a considerable need to understand how these differing backgrounds correspond with health behaviors and outcomes. For example, cigarette smoking was 5.9% among all Hispanic youth but ranged from as high as 14.3% among Asian-Hispanics to 4.3% among NHOPI-Hispanics. Analyses like these should be replicated with relevant health outcomes during adolescence and/or adulthood to better understand the nuances of health behaviors and outcomes among Hispanic individuals.¹¹

This study has several limitations. First, the race and ethnicity variables used in this study are merely categorizations and do not necessarily reflect the racial and ethnic identity of each individual.³⁴ For example, individuals may identify as Hispanic for several reasons, including cultural identity, national origin, and/or household diversity (i.e., one parent identifies as Hispanic while the other does not).^27,35 However, the logic sequence (if Hispanic, then race is not considered) used to operationalize Hispanic ethnicity means that individuals categorized as Hispanic will have varied and diverse identities, even within each racial category. As such, our study cannot inform on nuances in racial identity among Hispanic youth, including the plurality (40.1%) of Hispanic youth who reported no racial category at all. Extensive qualitative and quantitative research is needed to better understand the nuances of racial identities among Hispanic youth as well as the implications on health surveillance, intervention, and evaluation. Second, our analyses included data collected in 2021 which may have resulted in bias due to several factors, including the COVID-19 pandemic as well as the enaction of Tobacco21 in late 2019.³⁶ However, our analytic approach did control for survey year as a covariate to account for variance across time. Third, these data are self-reported and thus subject to recall and response bias, particularly in assessing past 30-day tobacco use. Third, this study uses nationally representative data and, thus, our findings are not necessarily generalizable to specific regions within that sampling frame. Consequently, our study must be interpreted within the historical and cultural context of each specific region, including analysis of state-level YRBSS data. For example, our national figures show 75.6% of AI/ANs identified as Hispanic, but these figures vary considerably by state. Texas and New Mexico are the historical homes of many Native Americans who identify as Hispanic.³⁷ Conversely, neighboring state Oklahoma is home to the largest population of NH, AI/AN in the United States; one result of the American genocide known as the Trail of Tears, which forcibly relocated Native American tribes west to the region now known as Oklahoma (or “Red People” in Choctaw). Such historical factors must be considered when generalizing these nationally representative findings to specific regions.

Despite our limitations, this study offers valuable insights into the prevalence of tobacco use by race and ethnicity among nationally representative high school students from 2015 to 2021. Our methodology provides an example to investigate racial and ethnic differences across health behaviors. Similarly, these methods can be applied to any survey independently assessing race and ethnicity. The proposal makes a strong case for reformatting the assessment of race and ethnicity to improve the utility of this measure to detect and quantify the nuances of race and ethnicity. In addition, national datasets can make race and ethnicity variables available in all public-use files (e.g., NSDUH). Future research should examine correlates and determinants that explain the observed differences in tobacco use among Hispanics by race. In addition, while our study explores ethnic heterogeneity by racial category, future studies should examine racial heterogeneity among Hispanics to better understand the nuances of tobacco use across this large and growing population. The results presented in this study should be given substantial priority, even urgency, given the size and age of the Hispanic population in the United States.

Supplementary Material

Supplementary material is available at Nicotine and Tobacco Research online.

Data Availability

All data presented in this study are publicly available through the Centers for Disease Control and Prevention (CDC).