Assessing Sex, Gender Identity, Sexual Orientation, Race, Ethnicity, Socioeconomic Status, and Mental Health Concerns in Tobacco Use Disorder Treatment Research: Measurement Challenges and Recommendations From a Society for Research on Nicotine and Tobacco Pre-conference Workshop

Abstract

This paper reports on topics discussed at a Society for Research on Nicotine and Tobacco pre-conference workshop at the 2019 annual Society for Research on Nicotine and Tobacco meeting. The goal of the pre-conference workshop was to help develop a shared understanding of the importance of several tobacco-related priority groups in tobacco use disorder (TUD) treatment research and to highlight challenges in measurement related to these groups. The workshop focused on persons with minoritized sex, gender identity, and sexual orientation identities; persons with minoritized racial and ethnic backgrounds; persons with lower socioeconomic status (SES); and persons with mental health concerns. In addition to experiencing commercial tobacco-related health disparities, these groups are also underrepresented in tobacco research, including TUD treatment studies. Importantly, there is wide variation in how and whether researchers are identifying variation within these priority groups. Best practices for measuring and reporting sex, gender identity, sexual orientation, race, ethnicity, SES, and mental health concerns in TUD treatment research are needed. This paper provides information about measurement challenges when including these groups in TUD treatment research and specific recommendations about how to measure these groups and assess potential disparities in outcomes. The goal of this paper is to encourage TUD treatment researchers to use measurement best practices in these priority groups in an effort to conduct meaningful and equity-promoting research. Increasing the inclusion and visibility of these groups in TUD treatment research will help to move the field forward in decreasing tobacco-related health disparities.

Implications: Tobacco-related disparities exist for a number of priority groups including, among others, women, individuals with minoritized sexual and gender identities, individuals with minoritized racial and ethnic backgrounds, individuals with lower SES, and individuals with mental health concerns. Research on TUD treatments for many of these subgroups is lacking. Accurate assessment and consideration of these subgroups will provide needed information about efficacious and effective TUD treatments, about potential mediators and moderators, and for accurately describing study samples, all critical elements for reducing tobacco-related disparities, and improving diversity, equity, and inclusion in TUD treatment research.

Introduction

Tobacco use is the leading cause of preventable death and disease in the United States (US) and a leading cause globally.^1,2 Cigarettes are the most commonly used tobacco product throughout the world³ and among US adults⁴; however, rates of smoking and associated consequences differ for a number of priority groups. Groups who exhibit a higher prevalence of smoking cigarettes include sexual and gender minority (SGM) versus cisgender and heterosexual individuals, lower socioeconomic status (SES) versus higher SES individuals, and individuals with mental health concerns compared to those without mental health concerns (eg,^4-8). Furthermore, unlike in the general population, smoking prevalence has not declined over time for some of these groups (eg, people with mental health concerns).^9,10 While Black individuals and women have a lower smoking prevalence than White individuals and men, respectively, members of these two subgroups experience some areas of disproportionate smoking-related health consequences.^4,11,12 Consequently, it is critical that tobacco use disorder (TUD) treatments be efficacious and effective for these priority groups to reduce smoking inequities.

At the 2019 Annual Meeting of the Society for Research on Nicotine and Tobacco (SRNT), the SRNT Treatment Research Network sponsored a pre-conference workshop titled “An update to the basics: Current approaches for measuring and understanding key individual differences and cessation outcomes.” The goal of the pre-conference workshop was to help develop a shared understanding of the importance of several tobacco-related priority groups in TUD treatment research and to highlight challenges in measurement related to these groups. For example, some subgroups (eg, SGM individuals) have been underrepresented in research because there is no attempt to identify the subgroups via assessment or the assessments used do not accurately reflect the full range of subgroups.¹³ Importantly, terminology used to describe some subgroups has changed over time (eg, terms related to SGM status). As part of the workshop, talks provided recommendations for conducting TUD treatment research, with particular attention to measurement challenges related to sex, gender identity, sexual orientation, race, ethnicity, SES, and mental health concerns. While there are additional important priority groups (eg, those from rural areas, those from low- and middle-income countries, those with comorbid medical conditions such as HIV), the aim of this paper is to translate the information from the SRNT pre-conference workshop into a written resource. It should be noted that the information discussed here is based primarily, but not exclusively, on US studies as global differences were beyond the scope of the pre-conference workshop and this summary, although recommendations relevant for the larger global research field are included where possible.

For each of the four sections below [(1) sex, gender, and sexual orientation, (2) race and ethnicity, (3) SES, (4) mental health concerns], we first define key terms (Definitions) and then provide information about assessing subgroups (Assessment). Finally, each section provides group-specific recommendations (Recommendations). In the final section of the paper, we discuss overall recommendations and conclusions. Developing a common language and methods for assessment may improve the generalizability and reach of TUD treatment research findings and our ability to integrate such research into clinical practice, including tailoring treatment effectively among a number of tobacco-related priority groups.

A few notes about the language used throughout this paper: (1) “Tobacco use” refers to commercial tobacco use and not ceremonial tobacco use. (2) A diagnosis of a TUD includes a range of symptoms including difficulty reducing or stopping the use of tobacco and withdrawal symptoms.¹⁴ Treatment studies may include a range of inclusion criteria; some include TUD while other studies include other tobacco-related criteria (eg, number of cigarettes per day). This paper will refer to “TUD treatment” for consistency. (3) The language used below to refer to demographic groups is based on the language from the individual articles being described. Demographic groups may be referred to by different terms across studies (eg, African-American vs. Black, sex vs. gender) even when the terms are not necessarily interchangeable.

Sex, Gender Identity, and Sexual Orientation

Definitions

One’s sex is often assigned at birth by medical providers and/or parents based on external genitalia and refers to biological differences (eg, chromosome differences); whereas, gender is a social construct with gender identity consisting of an individual’s concept of themselves.¹⁵ Labels for one’s sex at birth may include male, female, and intersex. Labels for gender identity may include man, woman, or labels related to nonbinary gender identity (eg, genderqueer, gender nonconforming, agender, gender fluid).¹⁶ The relationship between sex at birth and gender identity can be described as cisgender (ie, sex assigned at birth does not differ from gender identity) or transgender (ie, sex assigned at birth differs from gender identity).¹⁶

Sexual orientation encompasses one’s sexual attraction, behavior, and identity.^15,17Sexual attraction may be to the same or other sexes/genders to varying degrees, and sexual behavior can similarly include partners from the same sex/gender or other sexes/genders as well as no sexual partners. While sexual attraction and sexual behavior are aspects of sexual orientation, sexual identity is the most common way that one thinks about and self-describes their sexual orientation. Labels for one’s sexual orientation include gay or lesbian (ie, attracted to same-sex/gender partners), heterosexual (ie, attracted to opposite-sex/gender partners), bisexual (ie, attracted to both male and female partners), pansexual (ie, attraction includes partners outside of binary conceptualizations of sex or gender), and asexual (ie, no attraction or no strong attraction to partners).

SGM is a commonly used umbrella term in research that encompasses individuals with any minoritized gender identity and/or sexual orientation. However, gender and sexual identity are distinct and should not be conflated when measured.¹⁸

Assessment

In the treatment literature, there have been longstanding issues with sex and gender assessment and reporting. These terms are often erroneously used interchangeably, and many studies erroneously assess sex and gender identity as binary characteristics¹⁹ when both are more continuous in nature (eg, intersex, nonbinary gender).

The most frequently cited guides for assessing gender identity and sexual orientation come from the Gender Identity in US Surveillance (GenIUSS) group¹⁵ and Williams Institute’s Sexual Minority Assessment Research Team (SMART).²⁰ Regarding gender identity, the GenIUSS group recommends a two-step approach; first assessing sex assigned at birth, on the original birth certificate, then assessing current gender identity. SMART recommends assessing sexual orientation within the following three domains: sexual attraction, sexual behavior, and sexual identity. These domains should be assessed separately because they are not always congruent—for example, individuals who report engaging in same-sex sexual behavior may identify as heterosexual. How researchers operationalize sexual orientation in smoking treatment research can affect their findings if attraction, behavior, and identity have different relations with outcomes of interest. Further, some domains may not capture information about sexual orientation (eg, for adolescents not yet engaging in sexual behavior, sexual behavior would not be a good representation of sexual orientation). Information from all three domains should be assessed and presented in papers.

Given the need for brevity where SGM status is not a major focus of the study, there are practical suggestions for implementing these guidelines broadly in TUD treatment research. See Figure 1 for domains and sample question wording,²¹ which are based on the GenIUSS and SMART recommendations. For additional guidance, NIH’s Sexual and Gender Minority Research Office maintains a repository of resources (https://dpcpsi.nih.gov/sgmro/measurement).

Figure 1.

Recommended response and answer terminology for measuring dimensions of sexual orientation and gender. At minimum, a measure of sexual orientation and a measure of gender should be included as standard demographic items across research modalities. Note, however, that research questions involving the sexual and gender minority (SGM) community may require more advanced questions and/or response options to allow more specificity. For example, the response options used here may be too binary for some SGM community members. “Male sexual partners” could be interpreted as “partners who identify as men,” “partners assigned male at birth,” or a combination of the above. Measures should be chosen and adapted to suit the research question and to be appropriate for the target population. Reproduced with permission from Dermody SS, Heffner JL, Hinds JT, et al. We are in this together: promoting health equity, diversity, and inclusion in tobacco research for sexual and gender minority populations. Nicotine Tob Res. 2020;22(12):2276–2279.

Recommendations

Assessment of sexual and gender identity is important for understanding TUD treatment response. The low representation in research of intersex participants and participants with minoritized gender identities may be partly due to limited assessment options (eg, not including intersex or nonbinary as response options). Assessment should include careful and inclusive measurement of the range of sexual and gender identities, using the most recent best-practice guides (see Refs. ^{15, 20} and Figure 1). Individual countries have developed their own definition and assessment guidelines and these documents should be consulted for country-specific information (eg, Australia, https://www.abs.gov.au/statistics/standards/standard-sex-gender-variations-sex-characteristics-and-sexual-orientation-variable/latest-release; United Kingdom, www.ons.gov.uk/economy/environmentalaccounts/articles/whatisthedifferencebetweensexandgender/2019-02-21).

As noted above, it is recommended that sexual orientation be assessed as three separate domains (ie, sexual attraction, sexual behavior, sexual identity). Further, sexual orientation may develop or change over time, and changing, versus stable, sexual orientations are associated with greater risk of tobacco use.²² When disseminating research, it is essential to clearly state how variables were operationalized. For instance, because sex and gender identity have often (mistakenly) been used interchangeably, it is unclear to what extent biological factors related to one’s sex versus social factors related to one’s gender identity differentially relate to TUD treatment processes and outcomes. This lack of clarity can be addressed by: (1) clearly stating how constructs were assessed, and (2) consistently using the correct terminology when describing research findings and implications. Guidance on using non-biased language when discussing sex, gender identity, and sexual orientation in scientific writing (eg, “sexual orientation” rather than “sexual preference”) is provided by the American Psychological Association.^16,17 Adherence to up-to-date best practices facilitates cross-study comparisons as well as aggregation of data, which is sometimes needed given the low numbers of SGM individuals included in any one study.

Understanding and assessing sexual and gender identity is also critical for the protection and respectful treatment of research participants.²³ For example, clinical studies, especially those administering medication, often include procedures to protect pregnant women or women of childbearing potential. As SGM individuals become better represented in research, it is important to properly assess whether or not individuals are of childbearing potential, irrespective of their gender or sexual identity, if becoming pregnant is an exclusion criterion for a study. Interviewing may also be useful for assessing SGM individuals’ preferred language related to sex and gender identity.

Race and Ethnicity

Definitions

Race is a sociopolitical construct used to categorize individuals into social groups.²⁴ Racial terms originated from an unscientific and racist belief that certain population groups are superior to others based on phenotypic genetic expression or place of origin.²⁵ Ethnicity is defined as a group of people that identify with each other based on shared ancestry.²⁴

Assessment

Measurement of race and ethnicity is shaped by historical and current social and political structures.²⁶ Changing demographics and social contexts have resulted in changes in the way race and ethnicity have been measured over time.²⁶ The way in which individuals self-identify can also be dynamic. Researchers have questioned the purpose of measuring race and ethnicity for these reasons and because racial/ethnic categories continue to be incorrectly used to infer biological differences across groups.²⁶

There are several guidelines published on the measurement of race and ethnicity in social and behavioral sciences research.^24,26,27 In addition, the Phenotype and eXposures project (PhenX), led by RTI International with funding from the National Human Genome Research Institute, has developed consensus measures for the examination of race and ethnicity in tobacco control research (https://www.phenxtoolkit.org).²⁸ PhenX follows the US Department of Health and Human Services recommendation to use separate questions to assess race and ethnicity,²⁹ with questions about ethnicity coming before questions about race.²⁹ This method, however, is known to result in high rates of nonresponse among those who identify as Hispanic/Latino.³⁰

In TUD treatment research, the details of questions used to gather participants’ racial/ethnic background, the national background of Asian American, Hispanic, or Latino participants, and the tribe of American Indian participants are not consistently included in published reports. There are studies where ethnicity is not reported along with race, and participants who are White have been used as the reference group without a clear rationale.

Recommendations

Due to high nonresponse rates among Hispanic/Latino individuals when using separate questions to assess race and ethnicity, a single self-reported race and ethnicity question is recommended (see Table 1).³⁰ In addition, research from the US Census Bureau³⁰ indicates that the single-item format better reflects racial/ethnic self-identification. This recommendation deviates from the current federal guidelines, which may raise concern about the comparability of measures across studies. However, valid and useful measurement of race and ethnicity should supersede this concern, and federal guidelines often change. We recognize, however, that some studies may be required to use separate questions to assess race and ethnicity to meet reporting requirements by national funding agencies. Regardless, respondents should be permitted to indicate all racial/ethnic groups with which they identify (eg, “select all that apply”) as opposed to requiring selection of a single category.³⁰ The latter forces individuals who are multiracial to select only one race (unless there is a “multiracial” category, which has its own limitations),³¹ precluding an assessment of the growing multiracial population.

Table 1.

Sample Items to Assess Race, Ethnicity, and Socioeconomic Status

	Sample Measures
Race and ethnicity	Which categories describe you?
	Select all that apply
	White
	Hispanic, Latino, or Spanish
	Black or African American
	Asian
	American Indian or Alaska Native
	Middle Eastern or North African
	Native Hawaiian or other Pacific Islander
	Some other race, ethnicity or origin
Socioeconomic status	Is your annual household income from all sources…
	Less than $10,000
	Less than $15,000 ($10,000 to less than $15,000)
	Less than $20,000 ($15,000 to less than $20,000)
	Less than $25,000 ($20,000 to less than $25,000)
	Less than $35,000 ($25,000 to less than $35,000)
	Less than $50,000 ($35,000 to less than $50,000)
	Less than $75,000 ($50,000 to less than $75,000)
	$75,000 or more
	What is the highest grade or year of school you completed?
	Never attended school or only attended kindergarten
	Grades 1 through 8 (elementary)
	Grades 9 through 11 (some high school)
	Grade 12 or GED (high school graduate)
	College 1 year to 3 years (some college or technical school)
	College 4 years or more (college graduate)
	Are you currently…?
	Employed for wages
	Self-employed
	Out of work for 1 year or more
	Out of work for less than 1 year
	A Homemaker
	A Student
	Retired or
	Unable to work

The following items come from surveys conducted in the United States. The race and ethnicity item comes from a United States Census Bureau survey that compared different question formats and response options to assess race and ethnicity.³⁰ It is not a standard item used in United States Census Bureau assessments. The socioeconomic status items come from the Behavioral Risk Factor Surveillance Survey. These survey questions are sample items and may need to be adapted for use in certain study populations or updated to reflect demographic changes. Studies conducted in other countries, for example, should use race/ethnicity and socioeconomic status measures relevant to their population.

According to the US Office of Management and Budget, the minimum racial and ethnic categories to be collected when assessing race and ethnicity are the following: White; Black or African American; American Indian or Alaska Native; Asian; Native Hawaiian or Other Pacific Islander; and Hispanic or Latino or not Hispanic or Latino.²⁹ These standard categories may not be sufficient, however, to comprehensively capture racial/ethnic background. In addition, individuals may identify in ways that are distinct from standard categories (eg, Latinx). Therefore, studies may need to supplement standard categories with others that are of particular relevance to the population or outcomes being studied. For example, in the US, individuals with Middle Eastern or North African ancestry are instructed to select ‘White’ ³²; however, many such individuals do not self-identify as White and would prefer a Middle Eastern or North African race category.³³ Also, there are country-specific racial and ethnic minorities with a higher prevalence of tobacco use that may warrant assessment (eg, Maori in New Zealand,³⁴ Ta Oi and Bru Van Kieu in Vietnam,³⁵ Jung Po in China³⁶). When appropriate, use terms/categories that are more accurate and specific (“Korean American” vs. “Asian American”). Avoid terms that may be offensive or “othering” (eg, “nonwhite”).²⁴ If comparing racial/ethnic groups, describe the rationale used to determine the reference group (eg, group at highest risk).

Measurement of race and ethnicity in TUD treatment research is important for monitoring racial/ethnic disparities in treatment outcomes. Research is needed that also assesses the sociopolitical factors that contribute to these disparities. Understanding and addressing the causes of racial/ethnic disparities in TUD treatment outcomes—for example, racism—is critical for advancing equity. Naming and assessing the underlying power systems and structures that produce racial/ethnic disparities in treatment outcomes may also help to overcome faulty assumptions that differences in outcomes are biologically driven. If race/ethnicity is used as a proxy for other constructs, provide a rationale for this decision.

Categorizing individuals by race and ethnicity may provide a false sense of homogeneity within racial/ethnic groups. Although some studies may not have sufficient resources or access to a large sample, when possible conduct sampling so there is adequate sample size to examine heterogeneity within racial/ethnic groups. TUD treatments that are designed for particular racial or ethnic groups may have limited effectiveness if they assume homogeneity within racial/ethnic groups. Manley³⁷ notes that an individual’s racial classification provides no information about an individual’s cultural experiences. Examining differences in treatment outcomes within racial/ethnic groups may provide insight into targets for intervention.

Socioeconomic Status

Definition

SES is defined as a relative position in a hierarchical social structure, based on access to or control over wealth or power.³⁸

Assessment

Several publications provide guidance on the measurement of SES in health-related research and discuss strengths and limitations of different approaches.^38,39 TUD treatment researchers have employed highly variable definitions and measurement of SES. Although self-reported income and education are commonly used measures of US SES, US studies operationalize SES differently. For example, low SES may be defined by some based on an income relative to the federal poverty level and by others as an income below a certain threshold (eg, ≤$10 000).^40,41

Other measures of SES used in TUD disorder treatment research include occupation, employment, insurance coverage, financial strain, and composite indices that combine multiple SES measures. Some studies report a sample is socioeconomically disadvantaged based on participants’ enrollment in publicly subsidized healthcare programs, attendance at safety-net hospitals, or based on residence in subsidized housing. These definitions coincide with recruitment location.

Recommendations

There is no single best indicator of SES that can be used in all TUD treatment research. Rather, we recommend that measurement of SES be guided by its relevance to the study population and study questions and outcomes. When the SES of study participants is central to a study, researchers should explain how this demographic characteristic plays an important role.²⁴

Commonly used measures of SES such as education, income, and occupation reflect access to different types of power and social benefits.³⁸ A study focused on examining the impact of economic resources on TUD treatment outcomes, for example, may be better suited using annual household income as a measure of SES as opposed to an individual’s occupation because income may more directly assess economic resources. If measures of income and education are similarly appropriate for the study, education is recommended. There is often a high rate of nonresponse for questions about income.⁴²

The life stage of the study population should also inform the SES measure used.⁴³ For example, an occupation-based indicator of SES may be less relevant for the elderly in the US because a minority of the elderly remain in the workforce.⁴⁴ Although occupation-based indicators of SES are commonly used in studies of adults conducted in European countries these measures have been critiqued.^38,39,45 Occupational classes often consist of a range of occupations with varying income, prestige, and benefits. Also, these measures do not consider potential heterogeneity in benefits that may arise from employment within occupational categories due to race/ethnicity or gender.³⁸ See Table 1 for sample items to assess SES.

Mental Health Concerns

Definitions

We will refer to individuals with “mental health concerns” to refer broadly to individuals with psychiatric diagnoses or self-reported psychiatric symptoms but not to individuals with alcohol and other nontobacco substance use disorders. “Transdiagnostic assessment measures” refers to measures of emotional vulnerabilities that are applicable across multiple diagnostic categories.

Assessment (see Supplemental Table 1)

Mental Health Variables Among Individuals Who Smoke

Unless there is a specific hypothesis about mental health or the sample was selected for mental health diagnoses or symptoms, few studies report mental health data. Treatment researchers can assess mental health status through a variety of procedures and choices should be guided by the study’s inclusion/exclusion criteria, methods, and hypotheses. Options include clinician-administered diagnostic interviews, self-report psychological symptom screeners and measures of distress, and transdiagnostic assessment measures.

Diagnostic interviews are the most comprehensive assessments, providing treatment researchers with formal psychiatric diagnoses. The Structured Clinical Interview for DSM-5 (SCID-5) is the gold-standard diagnostic assessment,⁴⁶ and the research version (SCID-5-RV) is flexible and customizable, allowing researchers to examine diagnoses of interest consistent with research needs.⁴⁷ The SCID-5 can take a considerable amount of time to complete, is proprietary and carries a fee for use, and is intended for use by a trained mental health professional, limiting its utility in some settings. The World Health Organization World Mental Health Composite International Diagnostic Interview (WHO WMH-CIDI), based on ICD-10 and DSM-IV diagnoses, is another gold-standard assessment which can be administered by trained lay interviewers and professionals.⁴⁸ The WHO WMH-CIDI is available in many languages and is in the public domain; however, training at a certified training site is required and carries a fee. Similarly, the Mini International Neuropsychiatric Interview (MINI) assesses common psychiatric diagnoses.⁴⁹ The MINI is intended to be a briefer assessment and can be administered by research staff; however, this is also a proprietary measure and there is a fee associated with its use. The SCID-5, WHO WMH-CIDI, and MINI have been translated for use by non-English speakers. Two newer semi-structured interviews include the Diagnostic Assessment Research Tool (DART)⁵⁰ and the Diagnostic Interview for Anxiety, Mood, and Obsessive-Compulsive and Related Neuropsychiatric Disorders (DIAMOND).⁵¹ The DIAMOND is intended for use by mental health professionals, can be administered during a briefer timeframe and is free for researchers who complete required training. Like the DIAMOND, the DART is free to use and briefer than the SCID and WMH-CIDI. Unlike the DIAMOND, there is no official training requirement before its use—though the authors state that it is intended to be used by trained professionals who are knowledgeable about diagnostic assessment and psychopathology. Each of the diagnostic assessments demonstrates good reliability and validity^49,52–54 and high specificity and sensitivity.^49,52 Because it is so new, only preliminary data on the DART has been published⁵⁰; however, most modules appear to have strong construct, convergent, and discriminant validity and high rates of inter-rater agreement.

Psychological symptom screeners and measures of distress can be used to assess disorder-specific symptoms (eg, anxiety-related symptoms) or provide a more general summary of psychological distress or impairment. The PRIME-MD Patient Health Questionnaire (PHQ) is used to assess current symptoms, and, importantly, the severity of symptoms, associated with depression, anxiety, eating pathology, and somatoform disorder.⁵⁵ The PHQ has excellent validity and high sensitivity and specificity⁵⁵ with respect to diagnoses and researchers can also use subsections of the PHQ to assess diagnoses of interest. The most commonly used are those assessing generalized anxiety disorder symptoms (GAD-7) and depression symptoms (PHQ-9), each of which demonstrates good validity and reliability and high specificity and sensitivity.^56,57 It is important to have a process in place for quickly identifying, and addressing, suicidal ideation when using a measure such as the PHQ-9 which assesses for thoughts of death or of hurting oneself. Finally, the Kessler Psychological Distress Scale (K6), a brief self-report measure, can be used to screen for serious psychological distress and impairment.⁵⁸ The K6 has been demonstrated to have high concordance with CIDI diagnoses of serious mental illness in several countries.⁵⁹

Some psychological symptoms are common across multiple diagnoses (ie, are “transdiagnostic”); recent evidence suggests that transdiagnostic vulnerabilities, such as distress intolerance, anxiety sensitivity, and anhedonia, may serve as a link between mood and emotional disorders and smoking. Researchers can use transdiagnostic vulnerabilities assessments to guide TUD treatment and help explain differential treatment responses (eg, high vs. low anxiety sensitivity) and mechanisms of change during treatment (ie, smoking reduction via increasing distress tolerance). The Distress Tolerance Scale (DTS), a widely used measure of distress tolerance, measures an individual’s perceived ability to withstand distress. The factor structure of the DTS has been validated in people who smoke cigarettes and demonstrates good internal consistency.⁶⁰ The Anxiety Sensitivity Index-3 (ASI-3), which examines an individual’s focus on anxiety-related symptoms and perceived harm from consequences of anxiety-related symptoms, is also validated in people who smoke cigarettes⁶¹ and has high internal consistency. Finally, measures of anhedonia can be used to examine an individual’s inability to experience pleasure (ie, consummatory anhedonia) or an individual’s expected experience of pleasure from typically pleasant activities (ie, anticipatory anhedonia). Importantly, anticipatory anhedonia has recently been posited as a potential nicotine withdrawal symptom.⁶² Measures to assess anhedonia include the Snaith Hamilton Pleasure Scale (SHAPS) and the Temporal Experience of Pleasure Scale (TEPS), each of which demonstrates strong psychometric properties.^63,64 Each of the transdiagnostic measures described above are available in the public domain.

Tobacco Use Among Individuals With Mental Health Concerns

An additional concern related to TUD research is the effect of mental health concerns on reporting patterns of key tobacco-related outcomes and the extent to which those outcomes (and existing cutoffs) are valid for these groups. There are several aspects of tobacco use, dependence, and cessation that are likely influenced by mental health symptoms. For example, in some of the authors’ data of adults with serious mental illness who smoke cigarettes, baseline assessments of nicotine withdrawal (ie, assessed while still smoking) looked similar to withdrawal symptoms reported by general population adults who smoke cigarettes after 24-hours of abstinence (ie, actually experiencing withdrawal).⁶⁵ Commonly used measures of nicotine withdrawal may not adequately differentiate between true withdrawal symptoms and mental health symptoms; specifically, affective components of withdrawal (eg, anxiety, depressed mood, and irritability); and are likely to over-estimate true withdrawal symptoms. This suggests that treatment researchers should assess withdrawal at baseline (even though research participants are still smoking) and examine change scores from baseline to follow-up as a better indicator of withdrawal. On the other hand, the most commonly used measure of cigarette dependence, the Fagerström Test of Cigarette Dependence (FTCD)^66,67 may underestimate dependence in some patients with psychiatric symptoms.⁶⁸ For example, one of the six FTCD items asks “How soon after waking do you smoke your first cigarette?” Given the association between sleep disturbance and mental health symptoms,⁶⁹ study participants may endorse waiting a long time before smoking their first cigarette of the morning because they smoked in the middle of the night, rather than because of low dependence.

Recommendations

Given that those with mental health concerns are more likely to smoke cigarettes and less likely to quit, it is imperative that tobacco treatment researchers assess mental health concerns in their studies. An important caveat (not unique to mental health assessment or psychologists, but especially relevant here) is to follow the guidelines of the American Psychological Association suggesting that “Psychologists who conduct psychological testing, assessment, and evaluation strive to practice with cultural competence.” (p. 18, American Psychological Association⁷⁰) and that “Psychologists who conduct psychological testing, assessment, and evaluation aspire to ensure awareness of individual differences, various forms of biases or potential biases, cultural attitudes, population appropriate norms, and potential misuse of data.” (p. 19, American Psychological Association⁷⁰).

Clinician-administered diagnostic interviews are the gold standard for the assessment of specific psychiatric diagnoses in treatment research focusing on tobacco users with a mental health condition. Benefits of clinician-administered interviews include more accurate diagnosis. However, the cost, training requirements, and time to administer may limit their use in certain contexts (eg, survey research) and these factors, along with inclusion/exclusion criteria and study hypotheses, should guide researchers’ measurement choices.

It may be useful to pair the semi-structured clinical interview with relevant symptom severity measures to quantify the severity of the symptoms, rather than simply confirming a specific diagnosis. This may provide opportunities to examine more nuanced relationships between psychiatric symptoms and cessation and examine the influence of a range of symptoms including those that are less severe or otherwise do not satisfy diagnostic criteria. Furthermore, tobacco treatment researchers who are not focusing on a particular set of disorders may be more interested in including symptom severity measures (eg, PHQ-9, GAD-7) to evaluate symptoms that commonly co-occur with tobacco use without using more cost- and time-consuming semi-structured interviews.

Alternatively, tobacco treatment researchers may wish to focus on transdiagnostic factors that are relevant across a wide range of psychiatric disorders and which are implicated in tobacco initiation, maintenance, and relapse (eg, DTS,⁶⁰ ASI-3,⁶¹ SHAPS,⁶⁴ TEPS⁶³). These measures may help guide treatment delivery and serve as potential moderators and mediators of change.

Unless researchers are excluding potential participants, who have specific symptoms or disorders, it should be assumed that individuals with mental health symptoms are represented—and likely overrepresented—among participants in TUD treatment studies. Even if the main focus of a study is not related to mental health concerns, tobacco researchers should include measures that can characterize the sample with respect to mental health (eg, the very brief K6 to identify nonspecific psychological distress). By failing to assess mental health concerns, the field is missing the opportunity to fully describe participants, understand the context of tobacco-related variables (eg, withdrawal, dependence), examine the mediators and moderators that will help the field better understand why and for whom our interventions work, and make interventions more efficacious and thus improve treatment delivery. Given the importance of assessing mental health status and the accessibility of these assessments (see Supplemental Table 1), widespread adoption by treatment researchers is attainable.

Overall Recommendations and Conclusions

Measure and Report the Assessment of Tobacco-Related Priority Groups in TUD Treatment Research Studies

When planning and implementing studies, researchers should adopt best practices in their assessments to ensure subgroups are properly identified, assessed, and described. As there tend to be multiple appropriate approaches to assessing these priority groups, tobacco researchers should choose corresponding measures based on factors such as the research hypotheses, the population of interest, study burden (eg, shorter vs. longer measures), and availability of research funds (eg, proprietary vs. public domain measures). Language related to these priority groups can change over time, so it is important to incorporate updates in language and assessment recommendations. For example, in a 2020 report,¹³ the US National Academies of Sciences, Engineering, and Medicine recommended that the Federal Interagency Working Group on Improving Measurement of Sexual Orientation and Gender Identity in Federal Surveys be reconvened to develop new standards for assessment that could be used across the country. Even if a study does not intend to examine these subgroups in hypothesis testing when disseminating results, the assessment of individual difference variables should be clearly explained (eg, listing all response options for a variable in the Methods section or as supplemental information when using non-standardized measures or measures that have been adapted for the purpose of the study) and sample characteristics should be described comprehensively (eg, in a demographics table and/or the Sample Characteristics section of the Results). Providing this descriptive information can help readers better judge the generalizability of the findings.

Moderators and Mediators

Various quantitative tools exist that can further our understanding of priority groups in TUD treatment research. To help describe differential treatment response between demographic groups, moderation analyses (ie, interaction analyses) can compare intervention effectiveness between subgroups. At the outset in grant proposals and study protocols, it is important to adequately power the study to examine outcomes across demographic subgroups. When a priori power analyses were not conducted, effect sizes and confidence intervals could be presented separately by subgroups of interest and labeled as exploratory, which could be included in meta-analyses or serve as the basis for future adequately powered research.

Research that identifies underlying mechanisms that explain group differences in the efficacy and effectiveness of TUD treatment is needed. While the priority groups we describe have differential smoking and treatment outcomes, these differences may be due to factors related to systems of privilege and oppression (eg, racism). See Refs. ^71,72 for information about macro-level factors (eg, access to and experiences with healthcare, geographic segregation, discrimination) and Ref. ⁷³ for a detailed framework for assessing social determinants of health. Pearson et al.⁷¹ also provide some recommendations on the assessment of macro-level variables in tobacco research. By evaluating these mechanisms directly using mediation analyses, for example, it is possible to identify modifiable factors that could be targeted in interventions to improve treatment efficacy, effectiveness, and health equity.

Intersectionality

Intersectionality has a long history rooted in Black feminist scholarship and activism, first recognizing that racism and sexism are interconnected.^74–76 More recently, intersectionality, which recognizes that systems of privilege and oppression are interconnected and produce overlapping experiences of oppression and disadvantage for people with intersecting minoritized identities (based on social categories such as race, class, gender identity, and sexual orientation), has been discussed as a useful framework for understanding and intervening upon public health problems and disparities.^77–79 Some studies of smoking behavior and TUD treatment have begun to examine disparities among individuals whose identities intersect across multiple minoritized social groups.^80–82 However, the tobacco treatment field is in the early stages of applying and using intersectionality; measuring and analyzing multiple intersecting identities is complex and current methods may not be reflective of a truly intersectional approach.⁸³ The application of intersectionality as a theoretical framework that centers the voices and lived experiences of those who have been minoritized due to historical and interlocking macro-level systems of power^84,85 may help practitioners, researchers, and policymakers better understand and intervene upon disparities in tobacco use and TUD treatment. See Etherington et al.⁸⁶ for a detailed example of how an intersectionality lens can be incorporated into intervention development research.

Recognizing Mistrust and Engaging Communities in TUD Treatment Research

Research procedures (eg, recruitment and retention, study visit procedures) should be equitable, inclusive, and affirming to all participants. Given the history and ongoing harm that healthcare systems, academic institutions, and researchers have perpetuated, it is also important for the public health field to “dismantle mistrust by acknowledging the role we have played in maintaining damaging power dynamics and perpetuating mistrust” (p. 193).⁸⁷ As such, community-engaged and community-based participatory research that focuses on the strengths of priority populations and works with (rather than on) communities through equitable partnerships to set and intervene upon health priorities through research, interventions, and policy is needed.^23,88 A community-based participatory research approach can be used to develop smoking cessation programs for priority groups (see Ref. ⁸⁹ for a review and Ref. ⁹⁰ for an example of using a community-based participatory research approach for tobacco and cancer control in an Asian-American community). See Refs. ^91,92 for more information about conducting community-based participatory research. Additionally, working with communities to adopt recruitment methods that help foster representation^93,94 and address practical barriers to participation is necessary. Finally, researchers must also fairly compensate communities and participants for their time and expertise.

Limitations

First, as mentioned in the Introduction, this paper includes a non-exhaustive list of priority groups experiencing tobacco-related disparities. Additional groups including other demographics (eg, age, rurality), people with medical co-morbidities (eg, HIV), and those with alcohol/substance use disorders should be considered in treatment research (eg, Refs. ^95–97). Second, we focused on cigarette smoking treatment; however, the use of other tobacco products can also differ between these groups.^5,98,99 Research on treatments for non-cigarette tobacco products and tobacco product polyuse by priority subgroups is also needed. Third, we do not provide an exhaustive list of valid approaches for assessing priority subgroups. For example, PhenX includes a range of assessments of demographics, social determinants of health, substance use, and mental health concerns and the RDoC (Research Domain Criteria Initiative; https://www.nimh.nih.gov/research/research-funded-by-nimh/rdoc/index.shtml) focuses on the measurement of mental health concerns and related constructs. Fourth, while this paper focuses primarily on US literature and considerations, and it was outside the scope of this paper to cover the full global literature, there are global differences in terms and guidelines used to describe subgroups (eg, racial and ethnic groups) and definitions of subgroups (eg, classifying SES) as well as other considerations that differ across countries (eg, stigma and legal issues related to SGM status). There are also global differences in the burden of tobacco (eg, increasing burden in low- and middle-income countries) and tobacco-related factors such as patterns of use of specific tobacco products.^100,101 These factors must all be considered by the research community.^102,103

Conclusions

There is a lack of research on TUD treatments for priority groups disproportionately affected by smoking and related disease. Accurately assessing and including these groups in research will provide needed information about efficacious, effective, and equitable treatments, a critical element to reducing tobacco-related disparities and advancing equity in tobacco control.

Supplementary Material

A Contributorship Form detailing each author’s specific involvement with this content, as well as any supplementary data, are available online at https://academic.oup.com/ntr.

Acknowledgments

The authors thank Drs. Leone Brose and Jan Blalock, the 2019 co-chairs of the Society for Research on Nicotine and Tobacco (SRNT) Treatment Research Network, for their assistance in organizing the pre-conference workshop; Dr. Monica Webb Hooper for her contributions to the pre-conference workshop talk on race, ethnicity, and SES and the initial proposal for this paper; and Drs. Leone Brose, Megan Piper, Anne Joseph, Neal Benowitz, and Natalie Walker for their contributions to the pre-conference workshop. We also thank Myndee Diamond for reading a draft of this paper, and members of the SRNT’s Board of Directors (Drs. Patricia Nez Henderson and Ben Toll) and Treatment Research Network Advisory Committee (Drs. Erika Bloom and Stuart Ferguson) for providing feedback on the paper proposal and on drafts of this manuscript.

Prior Presentations

Some data included in this manuscript were presented as a pre-conference workshop at the 2019 meeting of the Society for Research on Nicotine and Tobacco.

Funding

This work was supported by the National Institutes of Health (NIH)/National Institute on Drug Abuse (NIDA) [K01-DA040043, F31-DA052149, R33-DA041163, R34-DA050967], and the NIH/National Cancer Institute (NCI) [R21-CA236980, T32-CA128582]This paper is sponsored by the Society for Research on Nicotine and Tobacco (SRNT) Treatment Research Network (approval 11.27.2019). The NIH had no role in the design, analysis, interpretation, or publication of this study. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

Declaration of Interests

JLH has received research support from Pfizer. The other authors report no financial or other relationship relevant to the subject of this article.

Author notes

These authors made equal contributions to the paper and are listed in alphabetical order.