Conceptual and methodological issues for research on tobacco-related health disparities


  • All authors have no conflicts of interests.

This supplemental issue of Addiction is devoted to research on tobacco-related health disparities, with an emphasis on some of the conceptual and methodological issues underpinning this research. The papers are products of the Tobacco Research Network on Disparities (TReND),* funded by the National Cancer Institute and the American Legacy Foundation [1]. We have organized the papers in this supplemental issue into three sections.

The first section includes five papers that begin with a comprehensive overview of the major constructs (disparities, inequities, inequalities, diversity), the knowledge and evidence base on tobacco-related health disparities, and a review of the evidence on disparities in the United States by poverty, education and occupation [2]. The second paper is equally broad, but focuses on a comprehensive model of the developmental life-span and the tobacco addiction cycle with some emphasis on treatment [3]. The third paper is an overview of the elements of a biopsychosocial model of tobacco-related disparities for racially classified social groups [4]. The fourth paper in this section asks about the role of genetics in tobacco-related disparities among racially classified social groups and provides a series of questions for the field to consider [5]. Finally, the fifth paper addresses some of the conceptual issues with an interesting and unique analysis of the potential confounding influences of socio-economic status and race on traditionally reported results of health disparities in smoking [6].

The second section of this supplemental issue includes two empirical papers that focus on the reliability and validity of brief measures (a four-item scale and an eight-item scale) of sensation seeking, a personality construct assumed to be driven biologically and to exist on four dimensions [7,8] (experience seeking, thrill and adventure seeking, disinhibition, boredom susceptibility). The four-item scale is examined with a national sample of youth from various racially classified population groups and the eight-item scale is used in a sample of young adult Latinos in two inner cities in the South-west. The key question is whether this construct as measured is adequate for African American youth and whether the multi-dimensional model is valid for young adult Latinos. Because there has been so much attention paid to this construct in health communications and etiological research, a commentary on the papers and construct raises critical questions about research in this area [9].

The third section of this supplemental issue includes three papers, all of which emphasize the need to recognize that broad-band tobacco control strategies often ignore the heterogeneity that exists among population groups that are also influenced by their ecology. The paper by Tauras notes that only a few econometric studies examine the effects on subpopulation groups [10]. The paper by Shu-Hong Zhu and colleagues reports that the overwhelming proportion of Latino smokers in California are low-frequency smokers and raises a question about some of the assumptions underlying current approaches to tobacco control (i.e. withdrawal addiction and harm reduction) for groups in which low frequency of use is the norm [11]. The final paper in this section includes two case studies of attempts at tobacco control among Latinos in California and American Indians in Oklahoma [12].


The paper by Fagan et al.[2] provides a comprehensive and exhaustive review of the existing literature and a detailed examination of tobacco-related health disparities across the tobacco disease continuum for a wide variety of population groups (Latinos, African Americans, American Indians and Alaska Natives, Asian Americans and Pacific Islanders). Within each of these population groups the authors focus on a variety of tobacco disease indicators (initiation and environmental tobacco smoke, current use patterns, cigarettes per day, quitting, treatment and relapse and cancer outcomes). In addition to focusing on tobacco indicators ‘within’ population groups, the authors also review and examine the scientific evidence on these indicators ‘within’ major categories of socio-economic status/position (i.e. poverty, education and occupation). The paper also provides an in-depth review and discussion of a number of constructs (disparities, diversity, inequities and inequalities) that are often used as if they were interchangeable. It is clear that they are not interchangeable.

The authors also reinforce the central and critical importance of understanding that elimination of tobacco-related health disparities requires sustained transdisciplinary collaboration and partnerships at national, state and local levels. In many ways, this paper provides a comprehensive road-map identifying the most important topographical and health/health-care features of the disparities landscape for policy makers, health scientist administrators and community practitioners to note on their journey to eliminating tobacco-related health disparities. The recommendations at the end of the paper emphasize an urgent need for developing and nurturing a pipeline of researchers and practitioners from population groups that have suffered the most from tobacco.

The paper by Moolchan et al.[3] provides a comprehensive transdisciplinary conceptual framework that integrates the developmental life cycle perspective (womb to tomb) with a complex tobacco addiction cycle perspective (from intentions to use to addiction to cessation and relapse). The authors of this paper also provide an analysis of the social, environmental and cultural factors that influence individuals from different population groups to use or not use tobacco, and the ecological contexts within which individuals are embedded throughout the developmental and tobacco addiction cycles. This conceptual model also identifies documented and hypothesized disparities, factors influencing use, clinical stages of addiction and consequences of tobacco use and factors and forces that are related to treatment systems for tobacco addiction.

Noting and discussing the limitations at this point in time of focusing on genetic variations to explain biological differences in tobacco use and its consequences among racially classified social groups, Fernander, Shavers & Hammons [4] describe a biopsychosocial conceptual model. In this paper, the authors review what is known about tobacco use, topography and dependence/addiction with regard to biopsychosocial factors and racially classified social groups. In their discussion of the complex nexus of factors that are biopsychosocial the authors discuss evidence on diet and nutrition, obesity, alcohol consumption, psychosocial stress and occupational factors and how they relate to tobacco-related health disparities within racially classified social groups.

The fourth paper in this first section of the supplemental issue examines the role of genetics in health disparities among racially classified social groups. The paper by Fernander [5] is short, relatively speaking, but is packed scientifically and substantively with evidence about race as biological versus socially constructed. The author argues for the search for relevant genes by environment interactions (g–e interactions), understanding that tobacco-related health disparities exist ‘within’ as well as ‘between’ these groups and perhaps because of socio-political factors. The author ends with a series of questions designed to stimulate rigorous research on the role of genetics in understanding the etiology of tobacco-related health disparities.

The final paper in this section describes an empirical study that is intended to address one of the more complex and vexing issues in health disparities research: the confounding influences of socio-economic status and race on smoking and the health status of disparate groups. Traditionally, national data sets are designed to examine prevalence rates in a given population. A common limitation is simultaneously examining variables that may confound the interpretation of exposure to risk factors. LaVeist and colleagues [6] utilize a unique sampling technique (yoking neighborhood studies to national samples) to account for smoking among African Americans and whites taking into account the racial make-up (i.e. amount of integration) within their neighborhoods. Findings from this study highlight the limitations of traditional methodologies to investigate contextual and socio-political influences on smoking to understand race and socio-economic disparities.


The second section of the supplemental issue begins with two empirical papers that focus on a specific individual-level risk factor, sensation seeking. Within psychology, sensation seeking has received increasing recent attention because it is part of broader discussions and debates about the five-factor structure of personality. Some personality theorists have asserted that there is a genetic and biological basis for sensation seeking. Experts on adolescent health and risk-related behavior patterns have drawn attention to sensation seeking as an important risk factor for predicting involvement in a wide range of unhealthy behaviors. Health communications experts have used sensation seeking as a psychographic variable for audience segmentation for mass media campaigns. The most widely used empirical measure of sensation seeking is quite long. Therefore, particularly for its use in health communications research, there has been an effort to create a reliable and valid ‘short’ measure of sensation seeking to evaluate the efficacy of this construct as an audience segmentation variable.

The paper by Vallone et al.[7] uses large samples from a national media tracking survey of youth as part of the American Legacy Foundation's overall communications strategy. Using a four-item brief sensation-seeking scale, the authors examined the reliability and validity of this scale among white, African American, Hispanic and Asian youth. As expected, the higher the score on the brief sensation-seeking scale, the higher the risk for established smoking. However, the scale was a less reliable measure of sensation seeking for African American youth compared to white and Hispanic youth. The authors raise the question about the utility of this and other sensation-seeking scales across race/ethnicity population groups and whether it would be prudent to develop a scale that would measure more reliably the sensation-seeking construct for African American youth.

The paper by Stephenson et al.[8] uses data from a study on the prevention of tobacco use by 789 high-risk, young (18–30 years old) Latino workers in two inner cities in Texas. An exclusionary variable was enrollment in college. These researchers used a brief, eight-item sensation-seeking scale. They found that the eight-item scale was a reliable and valid index of sensation-seeking tendencies among English-speaking participants in this study, but the results were less clear and more ambiguous among Spanish-speaking Latino participants.

In a commentary on these two papers, Clayton, Segress & Caudill [9] raise several questions about the conceptualization and measurement of sensation seeking. They note that the original conceptualization of sensation seeking has four putative factors or dimensions (experience seeking, thrill and adventure seeking, disinhibition, boredom susceptibility). The eight-item and the four-item scales used in the two papers by Vallone et al. [7] and Stephenson et al. [8] attempt to reflect the putative factors by having either one or two items for each factor represented in the scale. The commentators note that one or two items in a scale to reflect multiple dimensions are not adequate. They also note that throughout the research literature it has been common practice to simply sum the scores and split the sample into high- and low-sensation seekers at the median, even when an adequate number of items were available to measure each dimension, thus ignoring the putative four dimensions of sensation seeking. In this commentary, it is posited that there are a number of potentially important conceptual and measurement development issues that should be addressed, including differences across racial and ethnic groups (and perhaps other classificatory schemes as well) and the utility of this construct for audience segmentation.


The third and final section of this issue includes three papers that reflect a reality honored in our rhetoric but sometimes ignored in the implementation of policy and programming—‘one size does not fit all’. The first paper in this section by Tauras [10] notes that tobacco use and the consequences of that use are distributed disproportionately among certain racial and ethnic population groups, yet few econometric studies examining of the effects of state-level tobacco control policies have focused on the effects partitioned by race and ethnicity. This paper reflects the theme that permeates a supplemental issue of the Journal of Epidemiology & Community Health (Vol. 60, Supplement II, 2006) produced by TReND on tobacco control policy and low socio-economic status women and girls. The second paper in this section is by Shu-Hong Zhu et al.[11]. Using data from California, a state that has been at the forefront of tobacco control policy innovations in the United States, these authors show that 70% of Latino smokers are low-frequency smokers. This paper then raises a provocative question: are current approaches to tobacco control, driven by ‘withdrawal addiction’ or ‘harm reduction’ models of smoking, appropriate for population groups in which the modal pattern is low average numbers of cigarettes per day and some-day smoking? The third paper in this section then arrives at where the proverbial ‘rubber meets the road’, capacity building at the community level for specific population groups. This paper, by Báezconde-Garbanati, Beebe & Pérez-Stable [12], involves case studies of attempts to address tobacco-related health disparities among American Indians in Oklahoma and examines capacity building in Hispanic/Latino Tobacco Education Partnership organizations in California. This paper reinforces that effectiveness is contingent upon leadership, collaboration, programs, funding and resources, development of policies and a pervasive awareness and sensitivity to the communities in which tobacco control is implemented. The need for tobacco control strategies to be culturally and linguistically appropriate is also reinforced.

TReND is committed to eliminating tobacco-related health disparities through transdisciplinary research that advances the science, translates that scientific knowledge into practice and informs public policy. There has been significant progress documenting the epidemiology and etiology of tobacco-related health disparities. As the papers in this supplemental issue show, however, there is substantial heterogeneity ‘within’ and ‘between’ population groups on the nature and extent of tobacco use, and the consequences of tobacco use reflect significant disparities. The burdens of ill health and disease caused by tobacco are concentrated disproportionately in racial and ethnic population groups and those groups have disproportionately the fewest resources.

Because tobacco use is the largest preventable cause of ill health and death in the United States, it is especially important to draw attention to the distribution of tobacco use and its consequences within and across population groups defined largely by race and ethnicity and socio-economic status. The papers in this issue underscore the catalytic role that a broad-based emphasis on tobacco-related health disparities can play, not just for the tobacco control field but for academic disciplines ranging from cells to society. A dedicated and sustained scientific, public policy and practice emphasis on tobacco-related health disparities could provide a vehicle for the implementing initiatives reflecting the value of social justice in this society, and lead to the elimination of the opposite of social justice, a continuation of tobacco-related health disparities.

TReND network members

Linda Alexander, University of Kentucky; Lourdes Báezconde-Garbanati, University of Southern California; Gillian Barclay, Pan American Health Organization; Laura Beebe, University of Oklahoma; Robert Bendel, Washington State University; Richard Clayton, University of Kentucky; Jennifer Doucet, University of Rhode Island; Sherry Emery, University of Illinois at Chicago; Pebbles Fagan, National Cancer Institute; Anita Fernander, University of Kentucky; Brian Flaherty, University of Washington; George Hammons, Philander Smith College; Mark Hayward, University of Texas at Austin; Deirdre Lawrence, National Cancer Institute; Deborah McLellan, Harvard University; Eric Moolchan, National Institute on Drug Abuse; Eliseo Pérez-Stable, University of California, San Francisco; Melissa Segress, University of Kentucky; Vickie Shavers, National Cancer Institute; Dennis Trinidad, University of California, San Diego; Donna Vallone, American Legacy Foundation; Wayne Velicer, University of Rhode Island; K. Vish Viswanath, Harvard School of Public Health; D. Gant Ward, private practice.


  • *

    TReND members are listed at the end of the paper.