2024 Funding Cycle Awardees

View other award years: 2025 • 2024 • 2022 • 2021 • 2020 • 2019

Please note: Listing describes appointments and affiliations at the time of award. Please check our Trainees and Alumni pages for current appointments and affiliations.

“Exploring Multiplicative Interactions of Marketing and Disparity Identities on Tobacco Use Behaviors”

PI: Dr. Zongshuan Duan, Assistant Professor at Georgia State University

Abstract and accomplishments +

Dr. Zongshuan DuanGeorgia State University

Abstract

Background: The evolving US tobacco marketplace employs diversified marketing strategies, disproportionately targeting disadvantaged populations. Despite existing knowledge on tobacco-related disparities, literature faces several limitations. First, there is limited research exploring the role of intersectionality in disparities. Second, there is insufficient knowledge of the broad spectrum of tobacco products and multiple tobacco product (MTP) use among minorities and those with intersecting minority identities, particularly in longitudinal or nationally representative studies. Third, limited research has examined the potential multiplicative interaction role of marketing exposure, which may exacerbate disparities in tobacco use among minorities and those with intersecting minority identities. Addressing these gaps is crucial for developing targeted policies and interventions to promote health equity. This proposal aims to examine whether and to what extent tobacco marketing and disparity identities may multiplicatively affect their tobacco use behaviors among US youth and young adults. Specific Aims: In this pilot study, we will focus on analyzing population-based longitudinal data from the Population Assessment of Tobacco and Health (PATH) study, to address 2 aims:
Aim 1: Examine tobacco use behaviors over time (i.e., tobacco product use, MTP use, flavored product use) among youth and young adults representing disparity identities and intersecting disparity identities.
Aim 2: Examine whether and to what extent exposure to tobacco marketing may differentially affect tobacco use behaviors among youth and young adults representing minorities and intersecting disparity identities.

Accomplishments

November 2025: This pilot project aims to examine the multiplicative interactions between tobacco marketing and social identities on tobacco use among U.S. youth and young adults. Using nationally representative Population Assessment of Tobacco and Health (PATH) Study Restricted-Use Files (RUFs), we modeled any-tobacco use and product-specific trajectories via growth mixture models and estimated tobacco use transitions through latent tobacco use transition analyses. Social identities included race/ethnicity, socioeconomic status, rural residence, and internalizing or externalizing mental health conditions. Marketing exposures (e.g., retail, digital/media, coupon) were incorporated as moderators to assess their interaction effects. Preliminary analyses indicated heterogeneous trajectories for any tobacco use. Participants characterized by certain social identity indicators (e.g., lower SES) had greater odds of moving toward any tobacco use. Greater cumulative marketing exposure was consistently associated with faster progression toward higher-risk trajectories. However, estimates for interaction effects were insignificant, likely due to small cell sizes. Major accomplishments included obtaining access to the PATH RUFs and executing pilot GMM and LTA with replicate-weighted balanced repeated replication (BRR; Fay’s k = 0.30). Additionally, current analyses yielded preliminary tables and figures to support subsequent work. As external funding priorities shifted, we redirected the PATH-based modeling to two complementary projects: 1) a mixed methods investigation of nicotine pouch uptake and substitution dynamics; and 2) a policy evaluation quantifying how pandemic-mitigating policies shaped tobacco use and cessation across pre-, during-, and post-pandemic periods. Overall, this pilot project helps advance tobacco regulation research by identifying population groups who are at increased risk of the effects of tobacco marketing. This work was supported in part by pilot funding from the Center for the Assessment of Tobacco Regulations (CAsToR) and a University of Michigan pilot grant. Analyses used the PATH RUFs; we thank the PATH Study team and the National Addiction & HIV Data Archive Program (NAHDAP) at ICPSR for facilitating secure data access. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH, FDA, or any other agency.

 

“Understanding the Role of Social Network Dynamics on Adolescent Tobacco and Nicotine Use”

PI: Dr. Clinton H. Durney, Research Methodologist at British Columbia Cancer Research Institute

Abstract and accomplishments +

Dr. Clinton H. DurneyBritish Columbia Cancer Research Institute

Abstract

Tobacco use is a significant public health concern, often beginning during adolescence and early adulthood. Existing research frequently examines tobacco and nicotine use at a broad level, without considering the impact of social interactions. This leaves a gap in our understanding of how knowledge, attitudes, perceptions, and behaviors related to tobacco use spread within social networks. Adolescents are heavily influenced by their social circles, both in-person and online, making them particularly vulnerable to peer influence, viral marketing, and targeted products like flavored tobacco. Recognizing the critical role of network characteristics — such as the arrangement and number of connections, as well as other topological features — in shaping tobacco use dynamics is essential. Understanding these influences is crucial for developing effective prevention and intervention strategies tailored to the unique vulnerabilities of youth. Specific Aims:

1. Investigate Peer Influence:
   • We will study how the behaviors and characteristics of friends and social connections affect the initiation and progression of nicotine use among teenagers.
   • Using data from large national surveys, we aim to identify key factors that influence teen tobacco use. This includes examining how peer pressure, perceptions of health risks, and the popularity of different tobacco products affect their choices.
   • We will assess how the interaction between peer influence and peer selection impacts youth decisions regarding tobacco use.
2. Model Social Networks:
   • We will develop a dynamic social network model to simulate how social connections evolve over time and how these changes impact tobacco use among teens.
   • This model will help us understand the role of different network features, such as the number and arrangement of connections and the formation of friend groups, in shaping tobacco use behaviors.
   • By analyzing these network characteristics, we aim to determine how they influence the likelihood of adolescent’s initiation of tobacco products.
3. Explore Key Influences:
   • We will delve deeper into how various factors, such as environmental influences and psychosocial aspects, interact with peer influences to shape adolescent smoking behaviors.
   • Our model will incorporate these factors to refine the probabilities of transitioning to tobacco use within social networks.
   • We will simulate different theoretical scenarios to explore how changes in social networks and smoking habits co-evolve. This includes examining how friends' smoking behaviors influence individual decisions and how network restructuring processes impact overall tobacco use trends.

By examining these social dynamics, we aim to gain comprehensive insights into the factors driving tobacco use among adolescents. This knowledge will be crucial in developing targeted strategies to reduce tobacco use initiation and promote cessation among youth.

Accomplishments

November 2025: This pilot project set out to understand how tobacco and nicotine use spreads through adolescent social networks. Tobacco use often begins during the teenage years, when peer influence is especially strong. Yet most research examines tobacco use at the individual level, without considering how social connections shape behavior. This project addressed that gap by developing a new modeling framework that combines ideas from social network theory and statistical physics to simulate how peer influence, social structure, and individual traits, such as self-esteem, affect who starts using nicotine products, who quits, and how these behaviors evolve over time. In the model, each student is represented as a node in a social network, with relationships capturing the influence of their peers. Three key parameters determine behavior: the strength of peer alignment (how much individuals tend to copy their friends), the bias toward behavioral change (how easily individuals switch behaviors), and personal inertia or self-esteem (how resistant someone is to peer pressure). Using these components, the model produced distinct behavioral regimes that mirrored real-world patterns. When social influence was strong, initiation spread quickly across the network. When individuals had higher self-esteem, abstinence stabilized and the group resisted change. Intermediate levels of social influence and self-esteem generated a mix of users and non-users, suggesting realistic population-level variation. Together, these findings show how a relatively simple set of rules can reproduce complex social dynamics and provide insight into why some youth populations are more vulnerable to tobacco use than others. The project also demonstrated how such models can be extended to study different nicotine products and evolving social environments, such as influencer effects or friendship network changes over time. Beyond scientific results, this project provided valuable mentorship and leadership experience. It offered the opportunity to supervise an undergraduate researcher, manage a project from concept to completion, and build a foundation for future grant proposals. Moving forward, the model will be calibrated with large national datasets to guide prevention strategies and inform tobacco regulatory science.

December 2025: CAsToR Pilot Projects Final Report: Dr. Clinton H. Durney (Research Methodologist, British Columbia Cancer Research Institute) (VIDEO 🎥)

 

“Estimating Unbiased and Precise Effects of Oral Nicotine Product Use on Inhaled Tobacco Product Use Persistence and Progression in Adolescence”

PI: Dr. Dae Hee Han, Postdoctoral Scholar at University of Southern California

Abstract and accomplishments +

Dr. Dae Hee HanUniversity of Southern California

Abstract

Background: Non-tobacco oral nicotine products (ONP) are a relatively new type of commercial tobacco product and one of the fastest-growing commercial tobacco product categories in the U.S. market. During 2019–2022, U.S. sales of nicotine pouches, one of the most prevalent ONPs among U.S. adolescents, have increased by 540% (from 126 million units to 808 million units). ONPs may be of particular interest to young people who use inhaled tobacco products (ITPs; e.g., combustible cigarettes and e-cigarettes) because ONPs can be used discreetly where ITP use is not allowed. Due to these concerns and adverse health effects of ONP use (e.g., oral toxicity), it is important to understand whether ONP-ITP co-use promotes ITP use persistence and progression among adolescents who use ITPs. Inverse probability weighting (IPW) has been one of the most widely used approaches to reduce confounding effects in epidemiologic and public health research. Augmented inverse probability weighting (AIPW) combines both the properties of the regression-based estimator (balancing differences in outcomes) and IPW estimator (balancing differences in exposure) and is therefore a “doubly robust” method in that it requires only either the propensity or outcome model to be correctly specified but not both. Machine learning can be used to enhance AIPW (AIPW-ML) to produce more accurate standard error estimates, narrowing the confidence interval (CI) for exposure effect estimates, improving statistical accuracy and efficiency. Yet, AIPW-ML has never been applied in tobacco regulatory science research, including analysis of ONP use effects on subsequent ITP use patterns. Specific Aims:
Aim 1: Examine the association of ONP and ITP co-use with subsequent ITP use persistence and progression using a conventional IPW approach. The average treatment effects of ONP-ITP co-use on ITP persistence and progression among adolescents will be estimated using a conventional inverse probability weighting approach in which participants are weighted by the inverse probability of their treatment assignment (i.e., ONP co-use). Hypothesis: Among adolescents who used ITPs at baseline, those with (vs. without) baseline ONP co-use would be more likely to continue to use ITPs and increase ITP use frequency at follow-up.
Aim 2: Examine the association of ONP-ITP co-use with subsequent ITP use persistence and progression using machine learning enhanced augmented inverse probability weighting (AIPW-ML) and compare estimates to Aim 1 results. We will use AIPW-ML that estimates less biased average treatment effects of adolescent co-use of ONP and ITP on ITP use persistence and progression. We will use the ensemble (i.e., super learner) algorithm for machine leaning analysis. Hypothesis: 95% CIs of ONP co-use effect estimates using AIPW-ML will be narrower than those estimated using the conventional IPW.

Accomplishments

November 2025: This project explored how new oral nicotine products (ONPs), such as nicotine pouches, gums, and lozenges, may influence tobacco use patterns among teens who already use e-cigarettes. Because ONPs are easy to conceal, come in appealing flavors, and are often marketed as “tobacco-free,” they have become increasingly popular among young people. Yet it remains unclear whether these products help youth move away vaping—or instead keep them using nicotine in another form. Using data from two Southern California cohorts of adolescents and young adults, the study followed participants over six months (2022–2023) to see whether those who used both ONPs and e-cigarettes were more likely to continue or quit their e-cigarette use compared to those who used only e-cigarettes. To ensure robust results possible, I used advanced statistical methods called inverse probability weighting and augmented inverse probability weighting with machine learning (AIPW-ML). These approaches help account for differences in background factors such as age, gender, mental health, and other substance use that might otherwise bias results. The findings showed that adolescents who had used ONPs in the past six months were less likely to still be using e-cigarettes at the next survey compared to those who had not used ONPs. However, this pattern did not appear among young adults. These findings suggest that ONPs may act as a partial replacement for ecigarettes among adolescents, but not among young adults. However, this pattern likely reflects behavioral substitution, switching between nicotine products, rather than a true attempt to quit or evidence that ONPs help with cessation, since these products are not approved as quit aids. Regulatory policies should carefully weigh potential harm-reduction benefits against age-specific risks and address the growing issue of dual ONP and ecigarette use among young adults.

December 2025: CAsToR Pilot Projects Final Report: Dr. Dae Hee Han (Assistant Professor, Emory University) (VIDEO 🎥)

 

“A Machine Learning Approach to Cluster Similar Flavors of Electronic Nicotine Delivery Systems”

PI: Dr. Mona Issabakhsh, Research Instructor at Georgetown University

Abstract and accomplishments +

Dr. Mona IssabakhshGeorgetown University

Abstract

Background: Despite a significant decline in smoking rates, cigarette use remains a leading cause of preventable mortality in the US, responsible for 480,000 deaths annually. Smoking uptake prevention and smoking cessation are among the most cost-effective and powerful solutions for tobacco-related disease prevention. Of US adult tobacco users, 24% use electronic nicotine delivery systems (ENDS). ENDS are available in hundreds of flavors beyond tobacco, which is one of the most commonly reported reasons for their use. The wide variety of ENDS flavors is a persistent concern among policymakers since studies indicate that flavored ENDS attract youth to ENDS use with the risk of subsequent cigarette initiation. Flavored ENDS use may also encourage cigarette smoking cessation; flavors, such as fruit, dessert, and menthol, have been reported to attract adult smokers trying to quit or reduce cigarette use. Studies, however, reported outcomes ranging from no effect to a significant positive effect for the association between flavored ENDS use and smoking cessation. The range of outcomes reported in the literature may stem from the wide variety of ENDS flavors and the heterogeneity of their users. Therefore, a systematic classification of ENDS flavors (considering flavor types, user characteristics, and ENDS devices) can improve empirical studies and simulation models' accuracy and increase the comparability of research results. The proposed research will apply machine learning (ML) clustering models to distinguish related clusters of ENDS flavors using the Nielson consumer panel data. Specific Aims:
Aim 1: Develop a flavored ENDS clustering model based on flavor types, devices, and user characteristics. To develop a clustering model, first, a variable selection step will be performed, which determines essential features to assign the most similar flavors to the same cluster while assigning the least similar flavors to separate clusters. Variables will include flavor types, ENDS devices, and consumers’ characteristics and will be inputted into the variable selection analysis to find the best subset of variables to develop an optimized clustering model. The most significant variables will be selected using an efficient tree-based algorithm that applies a “decision tree” structure to determine variable importance.
Aim 2: Train the flavored ENDS clustering model. Using the selected important variables, the dissimilarities between ENDS flavors will be calculated using the Gower distance method, a measure used to calculate the distance between two data points whose attribute has a mix of categorical and numerical values. The sample will be divided into a training and a testing dataset. The training dataset will be used to “train” the model, while the testing dataset (which shares the same distribution and properties as the training dataset) will be used to test the model's accuracy.
Aim 3: Test and validate the flavored ENDS clustering model. The trained model will cluster the testing set instances (i.e., unclustered data). The model's performance will then be evaluated by assigning the most similar flavors of ENDS to the same cluster or, in other words, minimizing dissimilarity within each cluster. While the model will be internally validated using a testing set, external validation can be done by developing ENDS clusters using the latent class analysis (LCA) statistical approach and comparing the results to the literature that has developed ENDS flavor categories.

Accomplishments

November 2025: This pilot project applies machine learning to advance understanding of how flavored electronic nicotine delivery systems (ENDS) relate to cigarette use and cessation behaviors. Recognizing the conflicting evidence surrounding flavored ENDS (some studies suggesting that ENDS use can help adult smokers quit cigarette use, while others linking them to youth initiation), this study aims to systematically cluster similar flavored ENDS users by incorporating flavor characteristics and user demographics. Applying Wave 7 of the Population Assessment of Tobacco and Health (PATH) survey, the study follows three aims: (1) identify key variables contributing to flavored ENDS use similarity; (2) train an optimized clustering model using Gower distance and k-means algorithm to group analogous flavored ENDS users; and (3) validate model performance with testing data. Analyzing flavored ENDS use among 1600 adults in PATH Wave 7 revealed two behavioral clusters of e-cigarette users. The first cluster comprised younger, mostly female users (aged 18–24) who reported lower cigarette use (15% dual users of cigarettes and ENDS), stronger tobacco cravings, and a preference for fruit flavors. The second cluster included older, mostly male dual users who were more likely to smoke cigarettes, express intentions to quit using e-cigarettes, and reside in the western part of the US. These findings suggest that cigarette use, craving, quit intention, sex, and region are the top behavioral differentiators among flavored ENDS users. By uncovering latent behavioral and flavor-based subgroups, this project developed a data-driven framework for classifying flavored ENDS users that can inform regulatory science, support evaluations of flavor restriction policies, and enhance predictive models of smoking behavior. Ultimately, this machine learning based clustering model lays the groundwork for future research exploring how flavorspecific ENDS use influences cigarette smoking initiation, cessation, and relapse, aligning directly with the FDA Center for Tobacco Products’ priority area of Behavior and Impact Analysis.

December 2025: CAsToR Pilot Projects Final Report: Dr. Mona Issabakhsh (Research Faculty, Georgetown University) (VIDEO 🎥)

 

“Impact of Flavored E-Cigarette Use and Nicotine Metabolism on Systemic Inflammatory Biomarkers of Cardiovascular Disease Risk”

PI: Dr. Nancy Jao, Assistant Professor at Rosalind Franklin University of Medicine and Science

Abstract and accomplishments +

Dr. Nancy JaoRosalind Franklin University of Medicine and Science

Abstract

Background: In 2020, the United States (US) Food and Drug Administration (FDA) issued a new enforcement policy against the sale of certain flavored e-cigarette (EC) products in an effort to limit the popularity of flavored ECs among youth, especially fruit and mint flavors. However, the overall use of flavored ECs has continued to increase since the implementation of the FDA policy. Additionally, despite the efforts to curb flavored EC use among youth, the 2022 National Youth Tobacco Survey found that 84.9% of high and middle school students who use ECs prefer flavored ECs, including a rise in menthol-flavored EC use. With the continued rise of flavored ECs, it remains vital for the FDA to continue to prioritize understanding how flavored EC use can impact overall public health. Although the use of combustible cigarettes has been long connected to increased risk for cardiovascular disease (CVD), the investigation of the short- and long-term health effects of EC use is still emerging. One important gap in the literature is the understanding of e-cigarette use with subclinical CVD outcomes in well-defined longitudinal cohorts. Although basic science studies have shown that flavoring in EC products can lead to increases in inflammatory response and dysfunction beyond the effects of nicotine, no studies to date have examined whether flavored EC use may increase CVD risk compared to unflavored (i.e., tobacco) EC use. Individual differences in nicotine processing not only influence smoking behavior but also have direct implications for adverse health outcomes. As individuals adjust their smoking behavior based on their rate of nicotine processing, individual differences in nicotine processing have been shown to play a role in influencing clinical outcomes. The nicotine metabolite ratio (NMR) is a validated phenotypic biomarker measurement of nicotine processing and clearance. However, while flavored EC use has been shown to suppress NMR to prolong the duration of nicotine bioavailability, no studies to date have examined how flavored EC use may impact NMR to impact levels of systemic inflammation. Specific Aims: The proposed study will examine the potential role of flavored EC use and NMR on biomarkers of CVD risk in the Population Assessment of Tobacco and Health (PATH) Study conducted by NIH/FDA. Specifically, the proposed study will examine population-level differences in systemic biomarkers of inflammation and nicotine processing between individuals who use flavored vs. unflavored ECs. The specific aims of this project are to:
Aim 1: Assess the direct and interactive effects of EC type (flavored vs. unflavored) and NMR on levels of systemic inflammation at Wave 1. Hypothesis: Individuals who use flavored ECs with slower NMR will have higher levels of systemic inflammation.
Aim 2: Compare the level of risk for CVD diagnosis based on flavored EC use and NMR at baseline and follow up waves. Hypothesis: Individuals who use flavored ECs and have slower NMR will have a greater likelihood of CVD diagnosis at Wave 1 and follow up Waves.

Accomplishments

November 2025: Although combustible cigarette use has been long connected to increased risk for cardiovascular disease (CVD), investigation of the health effects of e-cigarette use is still emerging. One important gap in the literature is understanding the relationship between e-cigarette use and subclinical CVD outcomes, such as chronic inflammation, in national longitudinal cohorts. With the rising popularity of flavored nicotine products, understanding how product characteristics (e.g., flavorings) and individual characteristics (e.g., nicotine metabolism) may impact clinical outcomes can help regulatory agencies better identify and quantify the potential health hazards of flavored e-cigarettes. This project explored the potential role of flavored e-cigarette use and nicotine metabolism ratio (NMR) on biomarkers of systemic inflammation and CVD risk using data from the Population Assessment of Tobacco and Health (PATH) Study. The study aimed to: 1) assess the direct and interactive effects of e-cigarette type (flavored vs. unflavored) and NMR on levels of systemic inflammation at baseline, and 2) examine the risk for CVD diagnoses based on flavored e-cigarette use and NMR over multiple follow-up waves. Preliminary findings suggest that flavored e-cigarette use was not consistently associated with biomarkers of systemic inflammation after accounting for demographic and behavioral factors. However, findings did indicate that flavored e-cigarette use may be associated with CVD-related diagnoses during follow-up waves, and that these associations may vary by NMR and over time. These findings underscore the need for ongoing research to examine potential biological mechanisms and temporal dynamics underlying associations between nicotine product use and CVD risk. Overall, understanding how nicotine product characteristics and individual biological differences may influence cardiovascular health outcomes remains critical for evidence-based tobacco regulation.

December 2025: CAsToR Pilot Projects Final Report: Dr. Nancy Jao (Assistant Professor, Rosalind Franklin University) (VIDEO 🎥)

 

“Forecasting the Impact of a U.S. Nicotine Reduction Standard on People with Disabilities”

PI: Dana Rubenstein, MHS, Medical Student at Duke University School of Medicine

Abstract and accomplishments +

Dana Rubenstein, MHSDuke University School of Medicine

Abstract

Background: People with disabilities use tobacco at disproportionately high prevalences and experience the greatest burden of related morbidity and mortality. National survey data show that adults with disabilities—including those with physical, cognitive, sensory, and other functional impairments—are significantly more likely to smoke cigarettes than adults without disabilities. This elevated prevalence contributes to persistent health disparities, including increased risk of cardiovascular disease, cancer, and premature death. One public policy approach that may reduce smoking and improve outcomes in people with disabilities—and in the U.S. population overall—is a nicotine reduction standard (NRS). This approach is based on nearly a decade of clinical trials demonstrating improved smoking outcomes for individuals randomized to very low nicotine content (VLNC) versus normal nicotine content cigarettes. The NRS was announced as a potential proposed rule by the U.S. Food and Drug Administration (FDA) in 2022 and remains under consideration. While the net public health impact of an NRS is anticipated to be groundbreaking (e.g., estimated to drop U.S. smoking prevalence from 12% to <2%), it is unclear whether these benefits will be equitably distributed. In particular, people with disabilities face unique structural, social, and economic barriers to cessation and may experience different outcomes under such a policy. Although clinical trials have shown positive effects of VLNC cigarettes in several tobacco disparity populations, the long-term effects of an NRS in people with disabilities—a large, heterogeneous, and underserved group—remain unknown. Specific Aims: The objective of this study is to extend the Mendez-Warner simulation model—a well-established model of smoking prevalence and public health effects—to adults with disabilities, a priority population with disproportionately high tobacco burden. In particular, we will look at adults with disabilities overall and specific types of disabilities, such as mobility, hearing, vision, and cognitive disabilities, as well as number of disabilities. Aims 1-3 will be replicated within each of these groups. Specifically:

• Aim 1: Acquire relevant input parameters (e.g., smoking prevalence, net cessation rate, and initiation rate) for adults with disabilities using nationally representative datasets such as the National Health Interview Survey (NHIS).
• Aim 2: Use the Mendez-Warner simulation model to project smoking prevalence and public health harms from 2023–2100 in people with disabilities, with and without implementation of an NRS.
• Aim 3: Evaluate the potential public health benefits of an NRS for people with disabilities.
  • Aim 3.1: Compare projected outcomes under the NRS versus no-NRS scenario to quantify the potential impact in this population.
  • Aim 3.2: Compare projected outcomes in people with disabilities to those in the overall U.S. population to assess equity in policy effects.