Correlates of Tobacco Use among Asian and Pacific Islander Youth and Young Adults in the U.S.: A Systematic Review of the Literature

Abstract

Rates of tobacco product use, including use of combustible and electronic cigarettes, remain high in youth and young adults within the U.S. Though a substantial body of work has examined risk factors associated with initiation and ongoing use of tobacco products, research on tobacco use and associated correlates among Asian and Pacific Islanders (APIs) has been more limited despite high rates in select API subgroups. This systematic review outlines recent research (2010–2020) on correlates of tobacco use in APIs ages 9–29. To better understand determinants of tobacco use and identify gaps in the literature, we framed correlates based on the National Institute on Minority Health and Health Disparities (NIMHD) multidimensional research framework. Database and author-focused searches were conducted, followed by article abstract and full-text reviews, much of which were guided by a discrete set of inclusionary and exclusionary criteria related to tobacco use and youth/young adults. A total of 24 articles were included in this review. The majority of articles focused on individual-level correlates, with a high number of studies demonstrating association between behavioral and sociocultural factors and tobacco product use behavior. Interpersonal factors also made up a substantial portion of the literature, commonly focusing on peer, family, and social norms. Gaps related to the NIMHD model are addressed. Future research should examine biological and community/societal factors associated with API smoking in order to better understand unique correlates in this population and to inform tobacco prevention and intervention approaches.

Tobacco Use within the U.S.

Tobacco use is the leading preventable cause of mortality in the U.S., with approximately 480,000 smoking-related deaths from health conditions such as lung cancer, cardiovascular disease, and respiratory illness each year (Centers for Disease Control and Prevention [CDC], 2020a). These health outcomes are a consequence of a pattern of chronic smoking that often becomes habitual during youth and emerging adulthood (Chassin et al., 1996; U.S. Department of Health and Human Services [HHS], 2016). Although rates of smoking in the U.S. have decreased substantially over time, current data suggests that approximately 6% of adolescents and 13% of young adults (ages 18–24) currently smoke combustible cigarettes (CCs; CDC, 2020b; Creamer et al., 2019; Jamal et al., 2018). Additionally, the use of electronic cigarettes (e-cigarettes) has increased dramatically since their introduction to the market in the mid-2000s, with current estimates suggesting that up to 20% or more of adolescents and young adults have used e-cigarettes (i.e., vaping nicotine) within the past month (Schulenberg et al., 2020; Vallone et al., 2020; Wang et al., 2020).

E-cigarettes have become increasingly popular over the last decade and include both disposable and reusable devices that utilize a liquid cartridge to inhale an aerosol (“vaping”) that commonly includes nicotine (derived from tobacco), other chemicals, and/or flavorings. A burgeoning market of e-cigarette products and devices, such as vape pens, cig-a-likes, e-hookahs, and other mod- or pod-based systems (e.g., JUUL), are widely available online and in the community at various retail establishments (e.g., vape shops, grocery stores), with considerable marketing focused on youth and young adults (Collins et al., 2019). Despite the fact no e-cigarette product is currently approved by the FDA to modify tobacco risk (U.S. Food and Drug Administration, 2019), many e-cigarette products are often marketed as a safer alternative to CCs (Bhalerao et al., 2019; McCausland et al., 2019; Pokhrel, Fagan, et al., 2015; Pokhrel et al., 2019). Alarmingly, up to 70% of young adult daily or occasional e-cigarette users have never smoked CCs (Boyle et al., 2019). Hence, although e-cigarette use is common among CC smokers, it is likely that among young people, e-cigarette use is widely prevalent among those who never smoked CCs or were never regular CC smokers. This is a significant problem; had it not been for the presence of e-cigarettes, these young people might not have been exposed to toxicants and nicotine. Recent research has highlighted youths’ use of e-cigarettes and increased risk for contracting breathing and lung-related abnormalities, such as asthma or COVID-19 (Gaiha et al., 2020; Schweitzer et al., 2017). Further, a recent outbreak of e-cigarette or vaping product use-associated lung injury (EVALI) in predominantly young adults points to the lack of safety in a poorly regulated e-cigarette market (Ghinai et al., 2020).

Tobacco- and E-cigarette-Related Racial/Ethnic Disparities

Although Asian/Pacific Islander (API) groups represent one of the fastest growing demographic groups in the U.S., they represent one of the least studied groups in the context of tobacco use prevention and cessation in the U.S. (Romero & Pulvers, 2013). In the U.S., persons of Chinese, Indian, Filipino, Vietnamese, Korean and Japanese descent make up approximately 85% of all Asian Americans (Pew Research Center, 2019). Native Hawaiian and Pacific Islanders (NHPIs) include the indigenous people of the Pacific Islands, such as Native Hawaiians (NHs), Chamorros, Samoans, Tongans, and other groups within the U.S. Affiliated Pacific Islands (Chuukese, Marshallese). Due to a history of discrimination, persecution, marginalization, and/or cultural colonization, factors influencing tobacco use among APIs in the U.S. may differ from those of other racial/ethnic groups (McElfish et al., 2019). The combination of the “model minority” stereotype perpetuated towards Asian groups and lower overall substance abuse rates compared to other ethnic groups has contributed to the lack of awareness and prevention programs available to the public (Fang et al., 2011).

When assessed as a single group, current CC prevalence among API adults is approximately 12% in the U.S. (Mukherjea et al., 2014). However, select API groups such as Filipinos, Koreans, and NHPIs demonstrate elevated CC smoking rates between 14–20% (Mukherjea et al., 2014). Large, national studies rarely disaggregate smoking data by API subgroup, masking tobacco use variations across these subgroups. In Hawai‘i (HI) and California (CA), the states with the largest proportion of API subgroups, smoking-related disparities are common. For example, in HI, though 12.3% of all adults report current smoking, the smoking prevalence is highest for NHPIs at 19.7% (CDC, 2015). The distribution of e-cigarette prevalence trends in similar directions, with the highest rates found in NHPIs compared to persons of all other racial/ethnic groups (Narcisse et al., 2020). Subica et al. (2020) found a substantially higher rate of past month e-cigarette use (39%) among Samoan and Marshallese young adults living in NHPI communities in CA and Arkansas (AR) and data from a Hawaiian sample of adults (Seto et al., 2016) showed that NHs had the highest rates of e-cigarette experimentation compared to participants from other racial/ethnic groups. These data are consistent with the fact that despite immense progress in reducing smoking prevalence across the U.S. as a whole, the burden of tobacco use is now concentrated among marginalized populations (Healton & Nelson, 2004).

National Institute on Minority Health and Health Disparities (NIMHD) Research Framework

To better understand complex factors that contribute to tobacco use initiation and maintenance in APIs, it is important to examine a broad range of social, environmental, and cultural determinants of health (Alvidrez et al., 2019). The multidimensional NIMHD framework, which is partly based on socioecological theory (Bronfenbrenner, 1977), outlines the importance of nested and changing levels of the environment (e.g., microsystem, mesosystem, exosystem, and macrosystem) that influence human development and behavior, with a significant focus on the role of social interaction. Within the model, complex domains of influence (Biological, Behavioral, Physical/Built Environment, Sociocultural Environment, Healthcare System) interact with varying levels of influence (Individual, Interpersonal, Community, Societal) to impact minority health and health disparities (see Figure 1; NIMHD, 2017). Though a number of components of the model have been tested for various health behaviors and conditions, the majority of past work focuses on single cells of influence, predominantly within the individual level (Alvidrez et al., 2019). As a result, the knowledge base of various health behaviors may be incomplete. As far as we are aware, no past studies have applied the NIMHD framework to tobacco use behavior, in general or in relation to API subgroups.

Figure 1.

The National Institute on Minority Health and Health Disparities Research Framework (NIMHD, 2017)

Current Study

To our knowledge, the most recent literature review of smoking-related correlates among API groups was conducted by Kim et al. (2008), prior to the widespread use of e-cigarettes. Although similar correlates of CC smoking were found in API and non-API youth (ages 11–18), some factors (e.g., advertising) had not been studied adequately. In addition, Kim et al. emphasized the need for more work on gender differences and in U.S. geographic regions outside of HI and CA. To provide an update of the literature and evaluate the research base through the lens of the NIMHD multidimensional model, we conducted a systematic review of published, peer-reviewed articles since 2010. Our study seeks to identify some of the current gaps in the literature to further understand correlates of tobacco use in API populations across the U.S. In this review, we sought to outline the factors associated with use of tobacco and e-cigarette use in API youth and young adults in order to guide future work and better tailor prevention and intervention programs in API communities.

Method

Literature Search

Figure 2 illustrates the PRISMA flowchart of the process of identifying articles for this review. In Step 1 of our search, three of the study co-authors (DJ, KM, and MR) systematically examined the peer-reviewed published literature using two primary methods – (1) a computerized search of online databases using specific keywords, and (2) a focused author search. For the computerized database search, we used PubMed and PsycNet, the latter of which includes both PsycArticles and PsycInfo. In each database, the primary terms “youth,” “Pacific Islander,” “Asian American,” and “smoking,” were used, filtering articles published between 2010 and 2020. To expand our search perimeter, we included derivatives of smoking (e.g., “vaping,” “vape,” “e-cigarette”), of Asian Americans (e.g., “Chinese American,” “Japanese American,” etc.), and removed “youth.” During these initial searches, the online databases yielded 6,792 articles. For the focused author search, we used a University search engine that mined 234 individual databases for 15 authors that were highly published in the areas of tobacco use and Asian Americans or Pacific Islanders (see Table 1). These 15 authors had a minimum of three first-authored publications in these substantive areas. The focused author search yielded a total of 942 articles. Duplicate articles and articles that were clearly outside the scope of our literature review (i.e., articles that did not focus on Asian Americans and/or Pacific Islanders, tobacco use, and youth/young adults) were eliminated from the computerized database search (6,652 of 6,792 articles, or 98% of the articles) and focused author search (878 of 942 articles, or 93% of the articles). After eliminating overlapping articles between the two search methods, there were 134 unique articles for potential inclusion in this review.

Figure 2.

PRISMA Diagram of Study Identification and Selection

Table 1.

Focused Author Searches: 2010–2020

Author	Number of Articles
Huh, Jimi	53
Leventhal, Adam	68
Unger, Jennifer B.	128
Cassel, Kevin	10
Pokhrel, Pallav	50
Subica, Andrew M.	28
Tanjasiri, Sora Park	15
Willis, Thomas A.	30
Clark, Trenette T.	35
Hofstette, Richard C.	17
Maxwell, Annette E.	52
Zhu, Shu-Hong	60
Kim, Seo-Ryung	146
Rogers, Christopher J.	4
Moon, Sung Seek	246
Total	942

In Step 2 of our search, the same three co-authors read and evaluated all 134 abstracts based on the inclusionary and exclusionary criteria for this study. Articles were included in this review if the following criteria were met: (1) at least 50% of the study sample included APIs residing in the U.S. and/or U.S. territories (e.g., Guam); (2) the study participants were between the ages of 9–29 years and/or had a mean age under 29 years; (3) the article exclusively focused on tobacco products and their correlates (e.g., social context, such as peer influences); (4) the article used an observational or empirically driven approach; and (5) the article was published between 2010 and 2020. Articles were excluded from this review if: (1) the article described a sample that did not include at least 50% Asian Americans or Pacific Islanders; (2) the article focused on tobacco use with other substances, but did not specifically examine tobacco-use correlates; (3) the article primarily focused on testing an intervention (e.g., counter-marketing or health communication interventions); (4) the article was non-empirical in nature (e.g., descriptive or theoretical); (5) the article was not peer reviewed; and (6) the article was published prior to 2010. Based on our assessment in Step 2, the number of articles was reduced to 68.

In Step 3, the entire research team screened and validated the full text of the 68 remaining articles using the inclusionary and exclusionary criteria. This process was guided by three senior co-authors with expertise in tobacco and substance use correlates with Asian Americans and/or Pacific Islanders. In this step, the team collectively examined the Methods and Results sections of each of the articles more closely, and discussed how to resolve ambiguities related to percentages of APIs in the sample and the ages of the study participants. In many of these cases, ambiguities were resolved through team discussion and consensus on whether the study met the inclusionary and exclusionary criteria. Due to our interest in biopsychosocial variables and the possibility that those using multiple substances might have different patterns and reasons for smoking, we excluded studies primarily focused on alcohol or other substance use correlates, as well as alcohol and other substance use correlates that were collected as part of eligible studies. This process eliminated 44 additional articles, reducing the final number of articles to 24.

Categorization of Studies to Domains and Levels of Influence

We categorized correlates to the five NIMHD Domains of Influence—Biological, Behavioral, Physical/Built Environment, Sociocultural Environmental, and Health Care System—and four Levels of Influence—Individual, Interpersonal, Community, and Societal (see Figure 1). This process entailed four steps—(1) outlining all individual correlates investigated in the included studies, (2) categorizing correlates within each level and domain of the NIMHD model, (3) developing subcategories of correlates, based on overarching themes across groups of correlates, and (4) refining and validating the categorization and subcategorization of correlates. Steps 2 and 3 used multiple independent coders to sort and categorize correlates within each study. Step 4 entailed an iterative process among co-authors, in order to best capture the dimensions of each tobacco correlate. The final structure of the categorization and subcategorization of correlates was based on consensus among co-authors.

Within the NIMHD Domains of Influence, Biological correlates include a wide range of genetic or other biological vulnerabilities, mechanisms, or exposures, whereas Behavioral correlates focus on actions/behaviors, responses, and functioning. Correlates of the Physical/Built Environment include an individual’s surrounding environment, while Sociocultural Environment correlates comprise social and cultural traditions and norms. Lastly, correlates of the Health Care System center on aspects of health services and options. Within the NIMHD Levels of Influence, Individual influences include factors about the person, while Interpersonal factors include social interactions. Community-level factors focus on environmental factors within the local community, while Societal-level influences include larger, system-level factors, such as at the state, regional, country, or global level. Figure 1 provides examples of types of correlates under each domain and level of influence.

Results

All studies included in the review are outlined in Table 2. Results are organized below by characteristics of the studies reviewed and the levels of influence addressed by the studies as they pertain to the NIMHD Model. Figure 3 summarizes findings organized by domains of influence/correlate specified by the Model. We relied on original work by NIMHD (2017; Figure 1) and Alvidrez et al. (2019) and our collective expertise to place correlates into particular domains/levels of influence. Of note, Table 2 includes all potential correlates examined in each study, whereas Figure 3 solely lists variables that have been found to be significantly associated with smoking behavior.

Table 2.

Studies included in the Systematic Literature Review

Study	Age Range (M age)	Race/Ethnicity	Region	Study Design (N)	Tobacco Type(s)	Correlate(s)	Description	Major Finding(s)
Cerrada, Unger, et al. (2016)	18–25 (21.1)	AA (KA)	S CA	QUAN (475)	CC	Gender, ethnicity, father’s smoking status, perceived smoking prevalence, acculturation	Examined whether predictors of perceived smoking prevalence differed by ethnicity and gender and whether smoking intensity was associated with perceived smoking prevalence	Current smoking status was associated with perceived smoking prevalence for most ethnic groups. Light smokers differed from non-smokers in terms of certain smoking prevalence estimates.
Cerrada, Ra, et al. (2016)	18–25 (22.4)	AA (KA)	N/A, likely S CA	QUAN (78)	CC	Social context, presence of friends, location, activities, food/beverage consumption, perceived stress, positive and negative affect, anhedonia, craving	Assessed contextual factors associated with CC use in real-time	Smoking instances (vs not smoking) were associated with being outside, with Korean friends, socializing, feeling stressed, exercising, and craving CCs.
Huh et al. (2013) Huh, Shin, et al. (2014)	18–25 (20.8) 18–24 (21.2)	AA (KA) AA (KA)	S CA N/A, likely S CA	QUAL (67) QUAN (22)	CC CC	Culture, gender, access/availability, social context Social context, craving, positive and negative affect, perceived stress	Examined the cultural meaning and attitudes related to smoking and cessation with current and non-smokers Examined contextual antecedents to smoking among KA smokers in real-time	Smoking CCs was seen as accepted and accessible in Korean culture; natural for KA men; comfortable for KA women. Negative affect and social context were associated with increased likelihood
Huh, Paul Thing, et al. (2014)	18–25 (20.8, 21.1)	AA (KA)	S CA	MIX (67, 475)	CC	Perceived smoking prevalence, social context	Examined acculturation and normative beliefs as risk factors for smoking	Participants overestimated smoking rates among KAs and in Korean-owned businesses.
Huh & Leventhal (2016)	18–25 (22.4)	AA (KA)	N/A, likely S CA	QUAN (78)	E-CIG, CC	Gender, craving, nicotine dependence	Examined whether within-subject variation in E-CIG use was associated with lower within-subject variation in CC frequency and craving; assessed whether gender and nicotine dependence would moderate these relationships	No relationship was found between within-subject variation in E-CIG use and within-subject variation in CC frequency or craving. Females who had higher craving were more likely to use E-CIGs.
Lee et al. (2013)	N/A (youth)	AA	N CA	MIX (93, 73)	CC	Tobacco advertising exposure, point-of-sale (POS) advertising, availability, accessibility, types of products	Descriptively examined environmental influences on tobacco use	Participants in different study phases reported high exposure to tobacco ads and other advertising (e.g., POS). Products were widely available and accessible.
Lee et al. (2019)	12–19 (16.2)	PI	N/A, likely S CA	QUAN (284)	Tobacco (smoking in general)	Smoking attitudes, expectancies	Examined the mechanisms, including smoking, through which biculturalism impacted overall health	Smoking attitudes and expectancies were associated with past 30-day smoking.
Maglalang et al. (2016)	18–25 (21.0)	API	N CA	QUAN (501)	E-CIG	Awareness of E-CIGs from various sources, perceived E-CIG harm, demographics, flavors	Study explored associations between E-CIG use and awareness and perceived risks	E-CIG use was associated with awareness of E-CIGs through peers, low perceived risk, and particular ethnic backgrounds.
Pokhrel & Herzog (2014)	18+ college sample (27.5)	NH	HI	QUAN (128)	CC	Discrimination, historical trauma, alcohol/marijuana concurrent use	Studied historical trauma, discrimination, and tobacco and other substance use among NHs	Findings suggested an indirect path to higher substance use through perceived discrimination and a direct path between historical trauma and lower substance use.
Pokhrel et al. (2014)	18–40 (23.5)	API	HI	QUAN (307)	E-CIG	Outcome expectancies, race/ethnicity, age	Examined participant characteristics and E-CIG outcome expectancies and their association with E-CIG use and susceptibility	Current and past 30-day CC smoking was positively associated with positive expectancies. Inverse relationships were found between negative expectancies and current and past 30-day smoking. Older age and Filipino background were associated with E-CIG use.
Pokhrel,…Fagan (2015)	18–35 (25.1)	API	HI	QUAL (62)	E-CIG	Reasons for liking and not liking E-CIGs	Explored the reasons ENDs users liked and disliked E-CIGs	Four major themes emerged: perceptions that E-CIGs are safer than CCs, benefits related to vaping, flavor advantages, and use for recreational purposes.
Pokhrel,…Regmi, & Fagan (2015)	18–35 (25.1)	API	HI	QUAL (62)	E-CIG, CC	Contexts when CCs and E-CIGs are used	Explored the contexts surrounding CC and E-CIG use in dual users	CCs and E-CIGs were used during select activities (e.g., before/after a meal). CCs were used for craving or stimulation, in certain locations (e.g., being outdoors), and with other substances. E-CIGs were used when CCs were unavailable or unable to be used.
Pokhrel et al. (2018)	18–25 (20.9)	AA	HI	QUAN (470)	E-CIG	Social media E-CIG exposure, outcome expectancies	Tested whether/how social media E-CIG exposure was associated with E-CIG use and outcome expectancies	Social media E-CIG exposure was associated with current E-CIG use; select positive outcome expectancies mediated the relationship.
Pokhrel et al. (2020)	18–25 (21.2)	API	HI	QUAN (2401)	CC	Physical activity	Examined the association between CC and E-CIG use with physical activity cross-sectionally and prospectively	Higher moderate physical activity was associated with reduced CC use six months later. Higher physical activity of all intensities was associated with increased E-CIG use at follow-up.
Rosario-Sim et al. (2013)	16–19 (17.2)	AA	NY	QUAN (328)	Tobacco (smoking in general)	Social context (peer groups and physical environment), peer pressure, tobacco friendly spaces, meta-motivational states	Explored meta-motivational states and environments during first smoking events	Youth who smoked the first time tended to be in a paratelic state (i.e., aroused, present-oriented), and in environments permissive toward smoking, with peers, but not adults, present.
Shih et al. (2015)	N/A (11.5)	AA	S CA	QUAN (953)	CC	Ethnicity, outcome expectancies, resistance self-efficacy, close adult substance use, perceived prevalence among peers	Study examined ethnic differences in smoking (and other substance use) initiation and risk factors associated with lifetime use and initiation	Rates of CC initiation were highest for Korean and Filipino youth. Subgroup differences were found for negative expectancies, perceived peer use, and close adult cigarette use.
Subica et al. (2020)	18–30 (23.6)	NHPI	S CA, AR	QUAN (143)	E-CIG, CC	Outcome expectancies, gender, race/ethnicity	Explored risk factors associated with E-CIG use	Positive outcome, but not negative, expectancies, were associated with current E-CIG use. Men were more likely to report CC use and Samoans were more likely to report ever using E-CIGs and reported higher positive and negative expectancies.
Tanjasiri et al. (2013)	15–25 (N/A)	API	S CA	MIX (111)	Tobacco (smoking in general)	Proximity to pro-tobacco locations, community locations, perceived neighborhood safety, positive smoking attitudes, gender, spending money	Examined environmental influences of smoking	Study revealed that particular community locations, pro-tobacco locations, perceiving one’s neighborhood as unsafe, and positive smoking attitudes are associated with tobacco use.
Wills, Sargent, Knight, et al. (2016)	N/A (14.7)	API	HI	QUAN (2309)	E-CIG, CC	Age, ethnicity, parental education, support, and monitoring, parent- adolescent conflict, academic and social competence, sensation seeking, rebelliousness, willingness to use CC, smoking expectancies, peer smoker affiliation, prototypes of smokers	Examined the relation between E-CIG use, willingness to smoke CCs, and social-cognitive factors that predict smoking CCs	Those who had used E-CIGs had greater willingness to smoke a CC. This relation was partly mediated through positive expectancies about smoking. Parent-adolescent conflict and parental monitoring also predicted willingness to smoke CC. Willingness was associated with future CC onset.
Wills, Sargent, Gibbons, et al. (2016)	N/A (14.7)	API	HI	QUAN (1136)	E-CIG, CC	E-CIG use, propensity to smoke (including rebelliousness, parental support, and willingness to smoke), ethnicity	Studied whether E-CIG use onset differs for youth who are at low- vs. high-risk of smoking	Findings suggested that E-CIG use was a risk factor for future CC use among those who had never smoked previously. Caucasian, Filipino, NH, and those from Other backgrounds were more likely to begin smoking than AA.
Wills et al. (2017)	14–16 (14.7)	API	HI	QUAN (2338)	E-CIG, CC	CC smoking onset, gender, ethnicity, age, rebelliousness, sensation seeking, parental support and monitoring, parental education	Examined relationship between E-CIG use and future CC smoking, as well as predictors of future E-CIG uptake	E-CIG use was a risk factor for future CC use. Caucasians and NHs were at higher risk for E-CIG uptake.
Yang et al. (2013)	11–18+ (N/A)	API	CA	QUAN (1287, 5024)	CC	Positive and negative attitudes about smoking, perceived prevalence of peer smoking, peer approval or disapproval of smoking, academic performance, truancy, perceived harm, ethnicity, gender	Assessed psychosocial correlates of past 30-day CC use	Demonstrated ethnic and gender variation in smoking. Attitudes toward CCs, perceived harm, perceived prevalence of peer smoking, friend disapproval, truancy, and academic performance were all correlated with current CC use.
Yu et al. (2010)	9–21 (14.1)	AA	U.S.	QUAN (1368)	CC	Refusal of CCs, truancy, perceived smoking safety, awareness of secondhand smoking harm, home smoking restrictions, exposure and receptivity to tobacco advertising, positive smoking images, sex, income, age, family smoking	Examined intra- and inter-personal correlates of smoking status	Experimental, occasional, and past-month smoking were predicted by different combinations of demographic, knowledge (e.g., awareness of second-hand smoking harm) and attitudinal (e.g., perceived smoking safety) variables.
Wills et al. (2017)	14–16 (14.7)	API	HI	QUAN (2338)	E-CIG, CC	CC smoking onset, gender, ethnicity, age, rebelliousness, sensation seeking, parental support and monitoring, parental education	Examined relationship between E-CIG use and future CC smoking, as well as predictors of future E-CIG uptake	E-CIG use was a risk factor for future CC use. Caucasians and NHs were at higher risk for E-CIG uptake.
Yang et al. (2013)	11–18+ (N/A)	API	CA	QUAN (1287, 5024)	CC	Positive and negative attitudes about smoking, perceived prevalence of peer smoking, peer approval or disapproval of smoking, academic performance, truancy, perceived harm, ethnicity, gender	Assessed psychosocial correlates of past 30-day CC use	Demonstrated ethnic and gender variation in smoking. Attitudes toward CCs, perceived harm, perceived prevalence of peer smoking, friend disapproval, truancy, and academic performance were all correlated with current CC use.
Yu et al. (2010)	9–21 (14.1)	AA	U.S.	QUAN (1368)	CC	Refusal of CCs, truancy, perceived smoking safety, awareness of secondhand smoking harm, home smoking restrictions, exposure and receptivity to tobacco advertising, positive smoking images, sex, income, age, family smoking	Examined intra- and inter-personal correlates of smoking status	Experimental, occasional, and past-month smoking were predicted by different combinations of demographic, knowledge (e.g., awareness of second-hand smoking harm) and attitudinal (e.g., perceived smoking safety) variables.

Note.

Race/Ethnicity: AA = Asian American, API = Asian/Pacific Islander, KA = Korean American, NH = Native Hawaiian, NHPI = Native Hawaiian/Pacific Islander.

Study Design: MIX = Mixed Methods, QUAL = Qualitative, QUAN = Quantitative

Locations: S = Southern, N = Northern, CA = California, HI = Hawaii, AR = Arkansas, NY = New York

Tobacco Type: E-CIG = Electronic Cigarette, CC = Combustible Cigarette

Other: N/A = not available

Figure 3.

Correlates of Tobacco Use for API Youth and Young Adults within the NIMHD Model

Note: Number of studies in parentheses

Study Characteristics

A total of 24 articles met criteria for our search (Table 2). Detailed information from each study related to participants’ age and race/ethnicity, as well as the region of the country where data was collected, the type of research methodology used, and the tobacco type, are noted in Table 2. Of the articles, 6 (25%) focused exclusively on Korean Americans, 3 (12.5%) examined Asian Americans, 12 (50%) APIs, and 3 (12.5%) focused specifically on NHPIs. The majority of studies examined participants residing in either HI (n = 9; 37.5%) or CA (n =12; 50%). Articles included quantitative (n = 18; 75%), qualitative (n = 3; 12.5%), and mixed methods (n = 3; 12.5%) studies. Of the quantitative studies, 10 were cross-sectional, four were prospective or longitudinal, and four utilized intensive longitudinal modeling/ecological momentary assessment (EMA). Papers addressed correlates of CCs (n = 11), e-cigarettes (n = 4), or a combination of CCs, e-cigarettes, and smokeless forms of tobacco (n = 9), including as related to initiation, recent or current use (e.g., in the past year or month), and smoking intensity (number of cigarettes).

Individual-Level Factors

No studies from our search examined Biological or Health Care System correlates. Studies examining Behavioral factors corresponded to three primary categories – (1) attitudinal/motivational factors, (2) dispositional factors, and (3) other factors. Attitudinal and motivational factors included cognitive, motivational, attitudes/beliefs, and perceptions about smoking. Often assessed interchangeably, attitudes or expectancies about smoking include both positive and negative perceptions or beliefs about possible outcomes (e.g., feeling relaxed or becoming addicted) that may occur if engaging in smoking behavior (Brandon et al., 1999). Positive images of smoking focus on beliefs that smoking makes young people look cool or fit in with their peers (Yu et al., 2010). Studies focused on dispositional correlates included affective and personality factors. Other factors included any variables that could not be classified into the first two categories.

Studies examining attitudinal/motivational factors comprised 52% of the total evidence from Behavioral studies. They focused on correlates such as positive or negative attitudes or outcome expectancies related to smoking (Lee et al., 2019; Yu et al., 2010; Yang et al., 2013; Tanjasiri et al., 2013; Shih et al., 2015; Pokhrel et al., 2018; Pokhrel et al., 2014; Subica et al., 2020). Specifically, positive attitudes, expectancies, or images of smoking have been shown to be positively associated (Lee et al., 2019; Subica et al., 2020; Tanjasiri et al., 2013; Yang et al., 2013; Yu et al., 2010; Pokhrel et al., 2014), while negative attitudes and expectancies have been inversely associated (Pokhrel et al., 2014; Yang et al., 2013), with current and lifetime smoking among APIs. In addition, studies in this category also focused on correlates of current and lifetime smoking, such as craving (Cerrada, Ra, et al., 2016; Huh, Shin, et al., 2014; Pokhrel, Herzog, Muranaka, Regmi, & Fagan, 2015), willingness to smoke (Wills, Sargent, Knight, et al., 2016), meta-motivational states (Rosario-Sim et al., 2013), and motives for using/reasons for liking e-cigarettes (Pokhrel, Herzog, Muranaka, & Fagan, 2015; Pokhrel, Herzog, Muranaka, Regmi, & Fagan, 2015; Maglalang et al., 2016) in APIs. For example, meta-motivational states were examined in one study (Rosario-Sim et al., 2013), suggesting that first-time smoking in Asian American adolescents reflected a present-oriented, as opposed to goal- or future-oriented, state. A subset of studies examined either perceptions of smoking harm or health benefits of e-cigarettes in APIs (Maglalang et al., 2016; Pokhrel, Herzog, Muranaka, & Fagan, 2015; Yang et al., 2013; Yu et al., 2010; Wills et al., 2017). Additional studies examined dispositional factors associated with current smoking or smoking onset, including negative affect (Huh, Shin, et al., 2014), perceived stress (Cerrada, Ra, et al., 2016, Huh, Shin, et al., 2014), and rebelliousness (Rosario-Sim et al., 2013; Wills, Sargent, Gibbons, et al., 2016; Wills et al., 2017) in APIs, making up 16% of the evidence from behavioral studies. Finally, there were a number of Behavioral studies that were unique, examining behavioral correlates of cigarette and e-cigarette use (including experimental, occasional, and regular use) such as truancy in API adolescents (Yang et al., 2013; Yu et al., 2010) and physical activity (Pokhrel et al., 2020; Pokhrel, Herzog, Muranaka, Regmi, & Fagan, 2015) in API young adults. These comprised 32% of the evidence for Behavioral studies. For example, Pokhrel et al. (2020) found that higher physical activity (of varying intensities) among API young adults was associated with reduced e-cigarette and CC use cross-sectionally. However, prospective associations showed that higher physical activity of all intensities at baseline was associated with increased e-cigarette use six months later.

Within the Individual Level of Influence, the remainder of the studies were categorized within the Sociocultural Environment and Physical/Built Environment domain. Studies within the Sociocultural Environment domain focused mainly on examining sociodemographic correlates and their influence on tobacco use, such as age (Pokhrel et al., 2014; Wills et al., 2017; Yu et al., 2010), sex/gender (Cerrada, Unger, et al., 2016; Huh & Leventhal, 2016; Huh et al., 2013; Maglalang et al., 2016; Subica et al., 2020; Yang et al., 2013; Yu et al., 2010), race/ethnicity (Cerrada, Unger, et al., 2016; Maglalang et al., 2016; Pokhrel et al., 2014; Shih et al., 2015; Subica et al., 2020; Wills, Sargent, Gibbons, et al., 2016; Wills et al., 2017; Yang et al., 2013), and income (Yu et al., 2010). These comprised 85% of the evidence within the Sociocultural Environment domain. For example, a study with Korean American smokers found that female gender moderated the relation between craving and CC use, suggesting some potential differences in influences for male versus female young adults (Huh & Leventhal, 2016). The remaining two studies examining Sociocultural Environmental domains focused on acculturation and its relation to perceived smoking rates (Huh, Paul Thing, et al., 2014) and CC smoking (Huh et al., 2013). Lastly, only two studies contributed to the Physical/Built Environment domain and both examined the context of being outdoors (Cerrada, Ra, et al., 2016; Pokhrel, Herzog, Muranaka, Regmi, & Fagan, 2015). Cerrada, Ra, et al. found that being outside was associated with smoking instances among Korean American young adults, as assessed using EMA. In a qualitative study, Pokhrel, Herzog, Muranaka, Regmi, and Fagan examined the context of CC vs e-cigarettes use, with CC use being common outdoors, while e-cigarettes were used frequently indoors.

Interpersonal-Level Factors

Correlates of smoking in youth and young adults across ethnic groups have included a focus on interpersonal and social factors at different levels (Amin et al., 2020). The API tobacco studies in this review focused on correlates related to Interpersonal-Level Factors within two primary categories – (1) family factors, and (2) peer factors. Family factors were within the Physical/Built Environment domain, while peer factors cut across the Physical/Built Environment and Sociocultural Environment domains. Family factors comprised 60% of the evidence in the Physical/Built Environment domain, and focused on correlates of smoking onset, experimentation, and willingness to smoke, such as parental or household smoking (Yu et al., 2010; Shih et al., 2015), parental support and monitoring (Wills, Sargent, Knight, et al., 2016; Wills, Sargent, Gibbons, et al., 2016; Rosario-Sim et al., 2013; Wills et al., 2017), parent-adolescent conflict (Wills, Sargent, Knight, et al., 2016), and parent education level (Wills et al., 2017) in API middle-schoolers and adolescents. Lower parental education and support were related to initiation of e-cigarette use (Wills et al., 2017). In a complex structural equation model examining the association between e-cigarette use and willingness to smoke in adolescent API non-smokers, those who had used e-cigarettes reported greater willingness to smoke CCs (Wills, Sargent, Knight, et al., 2016). The only other factors that had direct effects on smoking willingness were greater parent-adolescent conflict and decreased parental monitoring.

Peer factors comprised 40% of the evidence in the Physical/Built Environment domain and 89% of the evidence in the Sociocultural Environment domain. These studies focused on adolescent and young adult smoking and smoking onset correlates such as social context/presence of peers (Cerrada, Ra, et al., 2016; Huh, Shin, et al., 2014; Rosario-Sim et al., 2013; Huh et al., 2013) in Korean and Asian American adolescents and young adults; normative attitudes among API adolescents and young adults (Huh, Paul Thing, et al., 2014; Yang et al., 2013; Shih et al., 2015; Cerrada, Unger, et al., 2016); and peer pressure, refusal, and influence (Huh et al., 2013; Rosario-Sim et al., 2013; Yu et al., 2010; Maglalang et al., 2016) in API adolescents and young adults. For example, two studies found that socializing with Korean friends were related to in-the-moment decisions to smoke among Korean American emerging adults (Cerrada, Ra, et al., 2016; Huh, Shin, et al., 2014). Further, API youth have reported learning about e-cigarettes from their peers and this has been associated with initial use (Maglalang et al., 2016). Studies with API youth and young adults have focused on both descriptive and injunctive norms, such as the overestimation and/or misperceptions of smoking rates for Korean Americans (Huh, Paul Thing, et al., 2014). Only one study (Pokhrel & Herzog, 2014) provided evidence of the association between greater perceived ethnic discrimination, assessed via self-reported frequency of everyday experiences of unfair treatment, and increased cigarette use in NH young adults, which was classified under the Sociocultural level.

Community-Level Factors

Studies focused on Community-Level Factors in this review described correlates to API tobacco use within the Physical/Built Environment Domain. The correlates corresponded to two primary categories – (1) advertising, and (2) geography/convenience. Studies focused on advertising correlates to current and lifetime tobacco use comprised 44% of the Community-Level studies. These studies focused on exposure and receptivity to advertising/marketing among adolescent and young adult APIs (Lee et al., 2013; Maglalang et al., 2016; Yu et al., 2010), social media e-cigarette exposure among young adult APIs (Pokhrel et al., 2018), and point-of-sale advertising and marketing exposure in Southeast Asian American adolescents and young adults (Lee et al., 2013). Pokhrel et al. (2018), for example, found that exposure to e-cigarette advertisements through social media was associated with current e-cigarette use. This relationship was mediated by two types of positive outcome expectancies – positive smoking experiences (e.g., smoking with approval from family) and positive sensory experiences (e.g., smells good).

Studies focused on geography or convenience factors comprised 56% of the Community-Level studies. These studies included correlates of smoking onset and current use, such as availability and accessibility of tobacco products in Asian American adolescents and young adults (Lee et al., 2013; Huh et al., 2013; Rosario-Sim et al., 2013), pro-tobacco and tobacco-friendly locations in API adolescents and young adults (Huh et al., 2013; Rosario-Sim et al., 2013; Tanjasiri et al., 2013), tobacco presence in businesses or community locations in Asian American adolescents and young adults (Huh, Paul Thing, et al., 2014; Tanjasiri et al., 2013), and perceived neighborhood safety in API (Tanjasiri et al., 2013). For example, a mixed methods study (Tanjasiri et al., 2013) utilized geographic information system (GIS) data to code pro-tobacco spaces in the local community in Southern California. Findings indicated that API youth ages 15–25 who lived closer to pro-tobacco spaces were more likely to report past 30-day smoking.

Societal-Level Factors

Evidence related to Societal-Level Influences of health behavior, such as broad state, regional, country, or global factors, have been less commonly studied with APIs (Alvidrez et al., 2019). In relation to smoking, this may include policies and laws designed to limit tobacco use, age restrictions related to tobacco products, taxes on tobacco products, smoke-free environments, and societal-level discrimination. In our review, we classified only one study as assessing a Societal-Level correlate. Pokhrel and Herzog (2014) examined the association between perceived ethnic discrimination, historical trauma (i.e., traumatic events experienced by one’s ancestors), and cigarette and other substance use among NH young adults. Historical trauma was related to higher substance use (including CCs, alcohol, and cannabis) through perceived discrimination as a mediator. In addition, a direct path between historical trauma and lower substance use was found.

Discussion

API populations have been understudied in relation to tobacco use correlates. The current systematic review aimed to summarize the recent literature on tobacco use influences in API youth and young adults and is the first study to apply the NIMHD model to ascertain the state of research concerning known correlates of tobacco use among API youth and young adults. By taking a multidimensional approach, we were able to evaluate evidence specific to particular levels and domains of smoking influence and identify gaps in the research base that require further study. Overall, we found that most evidence lies in the Individual Level of Influence within the Behavioral domain, followed by the Interpersonal Level, within the Environmental domains (i.e., Physical/Built and Sociocultural).

Individual-level factors such as smoking-related attitudes, motives, and positive and negative outcome expectancies have been the most widely studied in API youth and young adults. Similarly, interpersonal factors, including familial smoking, parent support and monitoring, as well as peer factors and higher normative beliefs, have been commonly studied as correlates. Findings concerning these intrapersonal and interpersonal risk and protective factors have generally supported the hypothesized relationships based on theories of health behaviors such as the Theory of Planned Behavior (TPB; Ajzen, 1985) and Social Cognitive Theory (SCT; Bandura, 1999). For example, consistent with research conducted in majority White populations (Amin et al., 2020; Kassel et al., 2003; Kinouani et al., 2019), studies in this review found that motives such as affect regulation and interpersonal factors, including peer and family influence, may play key roles in shaping adolescent smoking initiation and maintenance among API. Family influences may be particularly influential for adolescent APIs, who are more likely to live in multigenerational homes as compared to White adolescents (Ong et al., 2021), and may experience greater exposure to adult smoking.

Across all levels of influence, there were no studies focused on API youth and young adults that addressed the Biological domain. Clearly there is more research needed in this area. Epidemiologic research among adults indicate that NH smokers are at increased risk for cancer incidence and mortality, compared to other ethnic groups, even after accounting for the quantity of cigarettes smoked (Haiman et al., 2006). This line of research has examined ethnic differences in the genetic basis of nicotine metabolism among NH smokers and others, including Japanese-Americans. For example, Derby et al. (2008) found that slow metabolizer status is protective against lung cancer for Japanese Americans. Similar research among API youth and young adults is lacking. Future research may, for example, examine the genetic basis of disparities in smoking rates and dependence across API subgroups.

There are further areas for research and investigation of tobacco use correlates, based on our review. For example, few studies have examined smoking-related correlates that may be more relevant for diverse populations, such as discrimination, stigma, and historical trauma. Relationships between these variables are likely complex, as illustrated by Pokhrel & Herzog (2014), yet are important societal-level contributors to tobacco use, particularly for NHPIs. Recent increases in discrimination and assault directed towards Asian Americans following the COVID-19 pandemic may lead to increased negative health outcomes in APIs (Chen et al., 2020), making this a critical time to study the complex relation between discrimination, stress, and coping-related smoking behavior. Further, other areas of research in our review demonstrated conflicting findings, such as the impact of perceived stress and negative affect on the tobacco use of APIs (Cerrada, Ra, et al., 2016; Huh, Shin, et al., 2014). For example, Cerrada, Ra, et al. found that participants who experienced greater perceived stress were more likely to smoke; however, positive and negative affect were not associated with smoking instances. In comparison, Huh, Shin, et al. found that greater momentary negative affect, but not perceived stress, was related to smoking. Both studies examined relationships using EMA with Korean American young adults. Though variables were operationalized similarly, Cerrada, Ra, et al. had a larger sample and each study utilized different analyses and controls, possibly contributing to disparate findings.

Few studies have examined advertising and marketing influences with API, including through social media and at vape shops, the latter which are increasingly present in local, disadvantaged communities (Berg et al., 2020; Huh et al., 2020). Evidence from our review shows that e-cigarette use may lead to future CC use in APIs who have never smoked, which is particularly concerning from a public health standpoint (Wills, Sargent, Knight, et al., 2016; Wills, Sargent, Gibbons, et al., 2016; Wills et al., 2017). Additional evidence demonstrates that API youth and young adults perceive e-cigarettes as healthier than CCs (Maglalang et al., 2016; Pokhrel, Herzog, Muranaka, & Fagan, 2015), which may influence their initiation. Future research is needed to better understand negative health outcomes related to e-cigarette use (e.g., EVALI) and how to translate preventive health information effectively to youth and young adults. Similarly, more work is needed to better understand the transition from e-cigarette to CC smoking, patterns of dual e-cigarette and CC use, and whether particular factors (e.g., efficiency of nicotine delivery, craving) are contributing to this transition. Lastly, dispositional and situational factors that can impact decision making related to smoking, such as positive and negative affect, anxiety, or impulsivity, have been poorly studied in API groups. Future research should attempt to clarify the relationship of these correlates to the tobacco use of API youth and young adults.

The use of social media among youth and young adults has become an integral part of their social interaction; yet, little is known about how social media platforms may influence smoking behavior. Substantial numbers of youth and young adults use social media platforms (Massey et al., 2021; Pew Research Center, 2018), raising the likelihood that smoking-related influence can occur through targeted advertising, social influencers, and exposure to peer/adult tobacco use and promotion. Only one study in our review examined social media exposure in APIs (Pokhrel et al., 2018). Future research is needed to better understand these contemporary social influences in order to inform regulatory decisions.

There has been recent emphasis on the need for innovative, multi-level research that can lead to a better understanding of intersecting influences on disease risk in health disparity populations (Alvidrez et al., 2019). The NIMHD has expressed an interest in research that hierarchically connects information at different levels of NIMHD’s research framework (NIMHD, 2017). One example of a multi-level study might be to use EMA to examine e-cigarette users’ decision-making processes and smoking behaviors emanating from social media exposure. This type of approach blends both individual-level and community-level influences of the NIMHD model within one study. In addition, evaluating smoking correlates in novel and objective ways can complement traditional methods of assessment, such as with EMA (e.g., rating real-time craving), digital biomarkers (e.g., heart rate captured through an activity tracker), or geospatial location (e.g., GPS location of high-risk smoking lapse contexts). The present study is a first step toward systematically examining tobacco use correlates for API youth and young adults, and provides a comprehensive framework to guide research within and across levels and domains within the NIMHD framework.

Implications for Prevention, Interventions, and Tobacco Regulation

It is important to have a clearer understanding of the correlates of smoking behavior in API youth and young adults in order to translate findings into prevention and intervention strategies. Based on our review, there is evidence that attitudinal, motivational, and descriptive norms may contribute to smoking initiation and current use in API youth and young adults, thus making these potentially valuable targets for prevention and intervention. Additional interpersonal factors, such as family smoking status, peer pressure, and the social nature of smoking (e.g., being in the presence of friends, often within the same cultural group), also play a large role in API smoking behavior. Past work with Korean American young adults suggests that smoking cessation interventions might benefit from a social support component to address peer influence (Cerrada et al., 2017). Cognitive and behavioral smoking cessation interventions, such as culturally and community-specific personalized normative feedback, should be further tested. Family-based preventive interventions have been shown to have moderate effectiveness in reducing smoking initiation and experimentation (Thomas et al., 2016). Ideally, these interventions would be grounded in the values, beliefs, and worldviews of API youth populations, in order to best meet the needs of the intended audience (Okamoto et al., 2018, 2019). Mobile interventions utilizing ecological momentary interventions or just-in-time-adaptive interventions (JITAI) may be particularly appealing to youth and young adult APIs and can address environmental risk factors in real-time (Nahum-Shani et al., 2018). A recent culturally-informed JITAI (Cerrada et al., 2017; Huh et al., 2020) was designed for Korean American young adults, targeting smoking-related contexts in-the-moment, when immediate support is needed. As emphasized by multiple NIH agencies (e.g., NCI, NIMHD, NIDA), there is a need to develop and test new multi-level interventions that target two or more levels of influence (Paskett et al., 2016).

Though there have been recent achievements related to tobacco regulation in the U.S. (HHS, 2016), additional work is needed. Evidence from our review suggests that API youth and young adults perceive low risk associated with e-cigarettes. Further regulation of e-cigarettes, such as advertising limits on social media and culturally resonant health warnings on product advertisements, may help reduce smoking. Increased public health messaging about e-cigarette smoking risk that targets API youth would be beneficial. Studies in this review also suggested that proximity to pro-tobacco influences (e.g., convenience shops) and greater access to and availability of tobacco products contributes to current smoking in APIs. Suggested tobacco control policies in API communities have included taking a multi-level approach (community, mainstream institutions, legislative, and corporate) to institute change (Asian Pacific Partners for Empowerment, Advocacy, and Leadership [APPEAL], 2012). Select APPEAL strategies include changing norms surrounding tobacco use (e.g., offering tobacco-free community events), ensuring that APIs have a voice related to mainstream tobacco control (e.g., tailoring tobacco cessation ads for APIs; placing anti-tobacco ads in Asian language newspapers), educating policy makers about the unique needs of API groups, and working to improve corporate-level change and compliance in API communities, such as working with API businesses to improve regulation of particular products (e.g., South Asian tobacco products such as bidis).

Limitations of the Study

There were some limitations to our systematic review. Although we engaged in a systematic and iterative categorization process among co-authors, our selection of particular variables into NIMHD categories was subjective; others may interpret classifications in a different manner. Because most included studies were from HI or CA, we cannot generalize findings to APIs in other U.S. regions. Though these states have the highest concentration of API groups, there is a need for greater representation from other U.S. regions. Lastly, we did not conduct a quality assessment of bias due to having studies with varying methodologies.

Conclusions

Evidence concerning the association between smoking behavior and individual, interpersonal, community, and societal correlates among API youth and young adults is growing, but still limited. We identified 24 articles that presented empirical evidence on a range of correlates of API smoking. Using the NIMHD model, our work suggests that evidence from Community/Societal domains of influence is lacking and we did not find any studies addressing biological-based correlates. Smoking remains the most preventable cause of morbidity and mortality in the U.S. (CDC, 2020a) and disproportionately impacts diverse communities (Herzog & Pokhrel, 2012). To better understand determinants of smoking in API, it will be critical to broaden the research base with API populations.