The Science of Precision Prevention to Reduce Disparities in Cardiovascular Health

The National Institutes of Health’s (NIH) National Heart, Lung, and Blood Institute (NHLBI) and the Office of Disease Prevention convened a virtual three-day workshop on the “Science of Precision Prevention: Research Opportunities and Clinical Applications to Reduce Disparities in Cardiovascular Health (CVH)” on December 5-7, 2022.

The purpose of the workshop was to a) identify gaps in knowledge and methods used in precision prevention research as foci for future research; b) discuss research opportunities and challenges, and provide information that could potentially expand the number of trials that develop and test the application of precision data science tools; c) discuss ways to use precision prevention approaches to close the health equity gap; and d) discuss how to expand the next generation of the precision prevention workforce to include more diverse investigators.

The attendees of the workshop were a diverse group of research scientists, clinicians, and investigators, representing a wide range of expertise within the field of precision prevention research. These included individuals with expertise in the qualitative and quantitative sciences, such as epidemiology, data science, machine learning (ML), artificial intelligence (AI), bioinformatics and biostatistics, as well as those with expertise in behavioral, public health, population sciences, health equity, clinical and implementation science research.

The three-day workshop was guided by peer-reviewed publications, including a prior NHLBI workshop publication on precision health analytics (Pearson et al. (2020)) that informed the three domains involved in the development of the workshop conceptual model: a) data sciences, b) quantitative sciences, and c) implementation science and interventions. Early stage investigators (ESIs), and individuals with extensive experience in training and educating ESIs, students, policy makers, and clinicians also participated.

Background

Precision prevention has been defined in multiple ways including as a form of tailoring interventions and prevention efforts to improve public health (Gillman and Hammond, 2016). It is comprised of both personalized prevention, and community-wide health conditions. Social determinants of health (SDOH), psychosocial, behavioral, environmental and cultural factors are also considered when making recommendations to improve the quality of life of individuals and communities (Biró et al., 2018). Cardiovascular Disease (CVD) continues to be the leading cause of mortality in the U.S. and it claims a huge proportion of the healthcare cost. Although advances in cardiovascular care has led to a decline in CVD, underrepresented racial/ethnic groups have not reaped the benefits of the decline (Javed et al., 2022). Recent advancements in data science have broadened our understanding of various data tools and their applications in precision prevention intervention research. The primary goal of this workshop was to envision how research in “Precision Prevention” could ultimately translate discoveries into CVH improvements, continue the decline in cardiovascular disease mortality and contribute to closing the health equity gap reported in the literature. The workshop addressed the NHLBI’s strategic vision goals to improve health, reduce human disease and develop the workforce.

Workshop Areas of Focus

The workshop agenda was organized into the following sessions: a) advances in data science; (b) development and validation of tools; c) application of data science in interventions for diverse individuals and communities; and d) needs and opportunities for education and training. Each session was followed by three concurrent breakout sessions during which presenters addressed specific research questions related to studying, optimizing, and implementing precision prevention interventions. The top ranked research opportunities for each session are presented in this report along with cross-cutting themes.

1. Data Science

   Speakers addressed the need to measure individual-level factors as well as social exposome-level factors (i.e., diverse ecosystems, social, physical/chemical, and lifestyle factors) which may partly operate independently, and interact with each other and with biology throughout the lifespan. One example of tools to quantify the exposome includes the Area Deprivation Index (ADI) housed in the Neighborhood Atlas®, a free data democratization tool, which provides current ADI measures for U.S. data harmonization, including linkage and layering of data to be FAIR (findable, accessible, interoperable, and reusable).  There was agreement that such work must be centered in considerations of health equity. Examples of data analytic strategies that could potentially inform health disparities include leveraging spatial information, geocoding, county-level and other data to assess multi-level SDOH, historical and current policies that maintain or exacerbate inequity, as well as using AI to convert unstructured scientific data into structured knowledge across multiple diseases.

   Research Opportunities

   • Include data on the social exposome in research and link it to actionable real-world interventions and policy changes
   • Improve ways to streamline acquisition of raw data from research- or consumer-grade wearable devices and harmonize metrics across different fields (e.g., physical activity and sleep)
   • Create tools that can analyze and integrate different data types across the lifespan, including quantifying the dynamics of the social exposome over time and their effects on lifestyle and health
   • Assess how historical and current local and national program and policies have influenced lifestyle behaviors, social determinants of health, and health conditions, and how future changes might address these issues
   • Promote the development and availability of rigorous and harmonized measures that can be integrated across research projects
   • Address risks in analyzing heterogeneous/unharmonized data, including the use of ML to mitigate bias across layers
   • Maximize available data by developing implementation methods that permit utilization of both centralized federated data aggregation approaches
   • Encourage collaborations between AI developers and end users throughout the modeling and framework adaption process
2. Development and Validation of Tools and their Applications

   Efforts to develop and validate data tools and applications for precision prevention require an understanding of human behavior, social and cultural structures, and extant and potential programs and policies, as well as consideration of resilience factors to leverage emerging opportunities in data science to open new frontiers for CVH research. For individual assessment, speakers discussed new technologies such as using home blood tests, glucose monitors, smart bandages, health apps, AI coaches, and social robots as tools for implementing precision prevention. Speakers recognized the fewer established tools to assess sociocultural, programming, and policy influences in a systematic, quantitative fashion. They acknowledged the need to engage stakeholders in all aspects of AI development and to build trustworthiness, enhance resilience, and reduce biases in diverse datasets. Tools that are developed and validated using ML models to estimate individual’s overall CVH with Life’s Essential 8 Metrics were presented, but questions remain on factors that impact the life course of CVH trajectories and health disparities.

   Research Opportunities

   • Create and/or refine tools to integrate, link, and append exposome metrics
   • Customize tools that incorporate environmental, policy, sociocultural, as well as behavioral change while linking individual, community, and policy changes to health outcomes
   • Assess the best methods to balance individual versus population approaches
   • Use qualitative and existing data to receive input from the community about research priorities, variables to include in data tools, and feedback on interventions
   • Systematically identify the community domains that need to be assessed, points of convergence, and high-impact measures that can be consistently applied 
   • Address reverse causation by using temporal and intervention data
3. Application of Data Science in Interventions for Diverse Individuals and Communities

   Two sessions covered this topic, where participants discussed several behavioral interventions that apply data science approaches for diverse populations. It was noted that traditional behavioral intervention development often follows multiple interrelated stages (i.e., basic science, intervention generation and refinement, efficacy and effectiveness, implementation, and dissemination).  A variety of precision prevention intervention study designs are available to streamline the intervention development process to identify the most appropriate intervention strategies for individuals. These include individually assigned and delivered study designs ((e.g., randomized controlled trials (RCTs), Multiphase Optimization Strategy (MOST) trials, N-of-1 trials, and Sequential Multiple Assignment Randomized Trials (SMART)). Also discussed were group assigned or non-independently delivered study designs (e.g., Group or Cluster-Randomized Trials (GRTs), Stepped-Wedge Group- or Cluster-Randomized Trials (SWGRTs),  and Individually Randomized Group Treatment (IRGT)Trials). Assignment of participants and delivery of interventions to participants were two factors that helped identify the most appropriate study design(s). An example of an N-of-1 study (Davidson et al., 2021) identified which behavior change techniques were most effective in increasing physical activity. Another example that focused on precision lifestyle medicine with physical activity as its aim demonstrated the heterogeneity in responses to physical activity and in CVH outcomes. A discussion on how data science can be leveraged to reduce disparities in cardiovascular care suggested improving healthy behaviors using electronic health records to nudge patients and providers, and automated messaging and linkage via telemedicine and data portals. An intervention that focused on patient and organizational readiness to address trauma and CVD risk among Black and Latino persons living with HIV is currently ongoing and could provide information on outcomes such as physical and mental health, and implementation (e.g., adoption, guideline adherence, satisfaction, feasibility, and maintenance/sustainability). Discussions about a community precision health research project in a developing country, noted that community precision health approaches could provide a means to address rural health disparities, both globally and domestically. The following were ranked as the top research opportunities from the two sessions:

   Research Opportunities

   • Encourage population-based intervention research that addresses community- and society-informed study aims and intervention strategies.
   • Understand how different communities in various racial groups respond to culturally tailored interventions and assess study outcomes keeping the heterogeneity of various communities in mind.
   • Use pragmatic trial study designs and analytic plans to deliver interventions to participants in real-world settings.
   • Leverage existing tools, techniques, frameworks, models, and lessons learned to analyze diverse datasets and to inform level of interventions and the development of effective implementation strategies.
   • Encourage community and population-based research that emphasizes diversity, equity, inclusivity, and accessibility.
   • Generate and validate intervention models to determine when it is appropriate to use an individual versus systems-wide, environmental intervention.
   • Understand how and why different communities respond differently to the same interventions and how these interventions work in different health systems.
   • Integrate electronic interventions that encourage the most at-risk individuals to seek timely preventative healthcare services, with the goal of strengthening population health, and reducing health disparities.
   • Encourage precision prevention approaches that are multimodal, multi-level and enhanced with real-time and dynamic provider-patient decision support.
   • Use multiple lifestyle behavior and disease screening tools (e.g., dietary screeners, BP screenings, etc.)  to reach at-risk and underserved communities to provide population-level benefits (i.e., reach the right people with the right interventions at the right time).
   • Establish a consensus regarding common metric and equity measures (e.g., all-cause mortality, quality of life, measures of equity) that researchers should generally incorporate.
   • Leverage partnerships among communities, academic partners, private sector, and government in the application of data science.
4. Education and Training

   Participants discussed the education and training of students and early-stage investigators in precision prevention. One example from project EAT (Eating and Activity over Time) was presented as a training opportunity with a focus on weight-related problems in marginalized populations. Another project REACH (Research on Eating and Activity for Community Health) is an applied epidemiology training program to promote weight-related health in youth and families from diverse communities. Both projects enhanced the education and training of students and early-stage investigators. Trainees work with young people and families with the goal of reducing disparities on weight related outcomes. Key challenges for both projects were related to the translation of findings into community interventions to address health disparities. Optimal strategies to design training programs for clinicians included clinical trial design, biostatistics, grant writing, incorporation of SDOH and implementation science into curricula. Discussions on the education needs of community stakeholders and policy makers focused on democratizing precision prevention, clarifying the meaning of precision prevention versus precision medicine, and gaining community trust. In addition, communication with organizational leaders to understand how implementation science can align with stakeholder priorities was emphasized. The following were ranked as the top research opportunities:

   Research Opportunities

   • Diversify the research workforce  to ensure representation of backgrounds and experiences relevant to the priority of communities for interventions.
   • Train investigators on the design and analytic methods appropriate for designs in which groups or clusters are assigned to study arms or in which interventions are delivered to groups or clusters of participants.
   • Develop the workforce and train investigators in community-level engagement from the early stages of their training to support an inclusive environment that follow the principles of trustworthiness ( https://www.aamchealthjustice.org/resources/trustworthiness-toolkit)
   • In a bi-directional process of listening and learning, clarify for community stakeholders the definitions of precision prevention, data science, and the difference between implementation science versus implementation in clinical and community settings.
   • Involve community representatives in the development and implementation of guidelines in a way that is aligned with cultural values.
   • Build the scientific workforce and educate investigators about the social and policy exposome and use of geographic analytic techniques
   • Bridge the gap between implementation science and practice by generating opportunities for implementation scientists, clinicians, community leaders, and policy makers to partner and co-design interventions
5. Cross Cutting Themes

   Cross cutting themes included processes to integrate data science, data tools, implementation science and interventions to address health disparities. Below are examples of potential research opportunities that were identified through the workshop:

   • Studies of precision prevention interventions that would influence major behavioral health outcomes (e.g., dietary intake, physical activity, and weight status) related to CVH and that are implemented throughout the life course to promote health equity and meet the needs of subgroups of the population at greatest risk
   • Studies of precision prevention interventions that would leverage data and use appropriate analytic methods to determine the best and most appropriate interventions for low-resourced communities
   • Multidisciplinary studies that focus on health and resilience at various levels from environmental or community to an individual’s psychological, physiological, and molecular capacity.
   • Interventions that would address social, structural, and policy determinants of CVH and include diverse participants and communities to change behaviors and promote CVH in disparate populations.
   • Use of design thinking to devise and carryout preliminary evaluation of potential interventions targeted to specific population groups at high risk for CVD as a precision prevention strategy.
   • Consider adding standardized measurement of the exposome into most research projects to inform interventions and modeling effects of intervening on various levels (e.g., structural vs. environmental vs. policy vs. individual).
   • Community-engaged precision prevention research that includes input from community members, addresses structural inequities, and incentivizes community partners to engage and collaborate.
6. Select References

   The following references were provided to workshop participants in preparation for the meeting or provided after the workshop:

   1. Bíró K, Dombrádi V, Jani A, Boruzs K, Gray M. Creating a common language: defining individualized, personalized and precision prevention in public health. J Public Health (Oxf). 2018;40(4):e552-e559. doi:10.1093/pubmed/fdy066
   2. Bradwell KR, Wooldridge JT, Amor B, et al. Harmonizing units and values of quantitative data elements in a very large nationally pooled electronic health record (EHR) dataset. J Am Med Inform Assoc. 2022;29(7):1172-1182. doi:10.1093/jamia/ocac054
   3. Collins KA, Huffman KM, Wolever RQ, et al. Determinants of dropout from and variation in adherence to an exercise intervention: the STRRIDE randomized trials. Transl J Am Coll Sports Med. 2022;7(1):e000190. doi:10.1249/tjx.0000000000000190
   4. Davidson KW, Silverstein M, Cheung K, Paluch RA, Epstein LH. Experimental designs to optimize treatments for individuals: personalized N-of-1 trials. JAMA Pediatr. 2021;175(4):404-409. doi:10.1001/jamapediatrics.2020.5801
   5. Diez Roux AV. Social epidemiology: past, present, and future. Annu Rev Public Health. 2022;43:79-98. doi:10.1146/annurev-publhealth-060220-042648
   6. Gillman MW and Hammond RA. Precision treatment and precision prevention. JAMA Pediatrics, 2016, 170 (1), 9-10.
   7. Haendel MA, Chute CG, Bennett TD, et al. The National COVID cohort collaborative (N3C): rationale, design, infrastructure, and deployment. J Am Med Inform Assoc. 2021;28(3):427-443. doi:10.1093/jamia/ocaa196
   8. Haendel MA, Chute CG, Robinson PN. Classification, Ontology, and Precision Medicine. N Engl J Med. 2018 Oct 11;379(15):1452-1462. doi: 10.1056/NEJMra1615014. PMID: 30304648; PMCID: PMC6503847.
   9. Havlir DV, Balzer LB, Charlebois ED, et al. HIV testing and treatment with the use of a community health approach in rural Africa. N Engl J Med. 2019;381(3):219-229. doi:10.1056/NEJMoa1809866
   10. Hilton Boon M, Craig P, Thomson H, Campbell M, Moore L. Regression discontinuity designs in health: a systematic review [published correction appears in Epidemiology. 2021 Jul 1;32(4):e15]. Epidemiology. 2021;32(1):87-93. doi:10.1097/EDE.0000000000001274
   11. Javed Z, Maqsood MH, Yahya T, Amin Z, Acquah I, et al. Race, racism, and cardiovascular health: Applying a social determinants of health framework to racial/ethnic disparities in cardiovascular disease. Circulation: Cardiovascular Quality and Outcomes, 2022,15 (1), e007917. doi.org/10.1161/CIRCOUTCOMES.121.007917
   12. Kind AJH, Buckingham WR. Making neighborhood-disadvantage metrics accessible - The Neighborhood Atlas. N Engl J Med. 2018;378(26):2456-2458. doi:10.1056/NEJMp1802313
   13. Lloyd-Jones DM, Allen NB, Anderson CAM, et al. Life's essential 8: updating and enhancing the American Heart Association's construct of cardiovascular health: a presidential advisory from the American Heart Association. Circulation. 2022;146(5):e18-e43. doi:10.1161/CIR.0000000000001078
   14. Murray DM, Ganoza LF, Vargas AJ, et al. New NIH primary and secondary prevention research during 2012-2019. Am J Prev Med. 2021;60(6):e261-e268. doi:10.1016/j.amepre.2021.01.006
   15. Murray DM, Taljaard M, Turner EL, George SM. Essential Ingredients and Innovations in the Design and Analysis of Group-Randomized Trials. Annu Rev Public Health. 2020;41:1-19. Epub 2019/12/24. PMID: 31869281.
   16. Pearson TA, Califf RM, Roper R, et al. Precision health analytics with predictive analytics and implementation research: JACC state-of-the-art review. J Am Coll Cardiol. 2020;76(3):306-320. doi:10.1016/j.jacc.2020.05.043
   17. Yoon S, Schwartz JE, Burg MM, et al. Using behavioral analytics to increase exercise: a randomized N-of-1 study. Am J Prev Med. 2018;54(4):559-567. doi:10.1016/j.amepre.2017.12.011

Workshop Organizing Committee and Leadership

Workshop Chairs

• Thomas A. Pearson, MD, MPH, PhD, University of Florida
• Charlotte Pratt, MSc, PhD, RD, NHLBI

Workshop Planning Committee:

• Gabriel Anaya, MD, MSc, NHLBI
• Olga Brazhnik, PhD, NHLBI
• Alison Brown, PhD, MS, RDN, NHLBI
• Rebecca Campo, PhD, MS, NHLBI
• Dave Clark, DrPH, NHLBI
• Laurie Friedman Donze, PhD, NHLBI
• Lawrence J Fine, MD, DrPH, NHLBI
• Bramaramba Kowtha, MS, RDN, LDN, ODP/NIH
• Thomas Pearson, MD, MPH, PhD, University of Florida
• Charlotte A. Pratt, MS, PhD, RD, NHLBI
• Nicole Redmond, MD, PhD, MPH, NHLBI
• Eric Shiroma, ScD, NHLBI
• Debbie Vitalis, PhD, MPH, NHLBI
• Deborah Young-Hyman, PhD, OBSSR/NIH

Workshop Speakers and Moderators:

• David Au, MD, MS, Center of Innovation for Veteran-Centered and Value-Driven Care
• Bettina Beech, DrPH, MPH, University of Houston
• Olga Brazhnik, PhD, NHLBI
• Christopher Chute, MD, DrPH, MPH, Johns Hopkins Medicine, Institute for Clinical and Translational Research
• Karina Davidson, PhD, MASc, Institute of Health System Science, Feinstein Institutes for Medical Research, Northwell Health
• Ana Diez Roux, MD, PhD, MPH, Dornsife School of Public Health, Drexel University
• Davera Gabriel, RN, FHL7, FAMIA, Johns Hopkins University, Biomedical Informatics & Data Science
• David Goff, MD, PhD, NHLBI
• Peter Groeneveld, MD, MS, Perelman School of Medicine, University of Pennsylvania, VA Center for Health Equity Research and Promotion.
• Jaclyn Hall, PhD, University of Florida
• Alison B Hamilton, PhD, MPH, VA Center for the Study of Healthcare Innovation, Implementation & Policy at the Greater Los Angeles Healthcare System
• Hui Hu, PhD, Harvard Medical School, Brigham and Women’s Hospital, and Channing Division of Network Medicine
• Heng Ji, MA, MS, PhD, University of Illinois at Urbana-Champaign
• Amy Kind, MD, PhD, University of Wisconsin School of Medicine, and Public Health
• William Kraus, MD, Duke University
• Harlan Krumholz, MD, SM, Yale School of Medicine
• George Mensah, MD, FACC, NHLBI/CTRIS
• Raina Merchant, MD, MSHP, FAHA, University of Pennsylvania
• Dariush Mozaffarian, MD, MPH, DrPH, Tufts Friedman School of Nutrition Science and Policy
• David Murray, PhD, ODP/NIH
• Dianne Neumark-Sztainer, PhD, MPH, University of Minnesota
• Maya Petersen, MD, PhD, UC Berkeley
• Charlotte A. Pratt, MS, PhD, RD, NHLBI

NIH Participants:

• Gabriel Anaya, MD, MSc
• Vanessa Barnes, BS
• Olga Brazhnik, PhD
• Alison Brown, PhD, MS, RDN
• Rebecca Campo, PhD
• Dave Clark, PhD, NHLBI, NIH
• Lawrence J Fine, MD, DrPH
• Laurie Friedman Donze, PhD
• Davera Gabriel, RN, FHL7, FAMIA 
• Bramaramba Kowtha, MS, RDN, LDN
• George Mensah, MD, FACC
• David Murray, PhD,
• Charlotte A. Pratt, MSc, PhD, RD
• Nicole Redmond, MD, PhD, MPH
• Eric Shiroma, ScD
• Deborah Young-Hyman, PhD
• Debbie Vitalis, PhD, MPH, NHLBI