About

Anne N Thorndike is an Associate Professor in the Department of Medicine at Harvard Medical School and a physician at Massachusetts General Hospital. She serves as the Director of the Cardiac Lifestyle Program within the MGH Cardiovascular Disease Prevention Center. Her research focuses on individual and population-level behavioral interventions aimed at preventing cardiometabolic disease. Through implementation research, her team has demonstrated the effectiveness of traffic light labels, choice architecture, social norms, and financial incentives in promoting healthy food choices in real-life settings such as worksite cafeterias and supermarkets. Her work also includes testing new strategies to increase healthy food choices among low-income populations in urban and rural community settings. Recent projects under her leadership examine the impact of food insecurity and social determinants of health on dietary quality and health outcomes. Dr. Thorndike has also served as the chair of the American Heart Association’s Nutrition Committee from 2020 to 2022.

Research topics

• Medicine
• Political Science
• Business
• Biology
• Public economics
• Economic growth
• Environmental health
• Pathology
• Economics
• Food science
• Family medicine

Selected publications

• Lessons from Smart Cart 2.0: Considerations for integrating purchasing data into healthy eating interventions with AI approaches

  Journal of Nutrition · 2026-05-01

  articleOpen accessSenior author

  The increasing prevalence of chronic diseases related to suboptimal diet quality in the US and worldwide contributes to burgeoning healthcare costs and excess death and disability. Changing dietary intake is essential but difficult because dietary behaviors stem from interconnected biological, psychological, environmental, and social factors that are rarely comprehensively addressed in interventions. Emerging digital technologies present a paradox in food environments: some facilitate healthier eating while other practices, like targeted marketing and price manipulation, exploit behavioral vulnerabilities and promote poor dietary habits. Smart Cart 2.0 - a personalized digital healthy eating intervention using artificial intelligence (AI) and machine learning (ML) to detect patterns and influence dietary choices - offers critical lessons for health technology design. This perspective contends that improving digital dietary behavior interventions hinges on responsible implementation: fully leveraging technology's promise while establishing safeguards against premature adoption. With appropriate guardrails, digital technologies provide tremendous opportunity to improve the delivery, personalization, and scalability of healthy eating interventions.

  PublisherDOI

• Changes in health-related social needs and in depression and anxiety symptoms in Massachusetts low-income, community health center patients: an observational cohort study

  Preventive Medicine Reports · 2026-04-25

  articleOpen accessSenior author

  Objective: This observational cohort study examined relationships between changes in health-related social needs (HRSN) and mental health symptoms. Methods: A prospective cohort of adult community health center patients in Boston, Massachusetts completed surveys at baseline and 3-year follow up (LiveWell study, 2020-2024). Changes in HRSNs were categorized as decreased, unchanged, or increased based on the cumulative change in three domains: food insecurity, housing instability, and financial stress. Primary outcomes were changes in depression (Patient Health Questionnaire-8) and anxiety (Generalized Anxiety Disorder-7) scores, modeled as a function of HRSN change group and controlling for baseline scores, age, gender, and household minors using separate linear regressions. Results: = 821) was 42.9 (IQR 32.9-53.1) years and 77.8% were female. At follow-up, 32.6% had decreased, 43.4% had unchanged, and 24.0% had increased HRSN. Compared to the unchanged HRSN group, the decreased HRSN group had decreased depression (adjusted coefficient [aCoef] -1.3 [95%CI: -2.1, -0.5]) and anxiety scores (aCoef -0.9 [95%CI: -1.7, -0.1]); the increased HRSN group had increased depression (aCoef 3.0 [95%CI: 2.1, 3.9]) and anxiety scores (2.8 [95%CI: 2.0, 3.7]). Conclusions: HRSN changes corresponded with changes in depression and anxiety symptoms. Interventions addressing HRSNs may help improve mental health.

  PublisherDOI

• Network threats to causal inference: Variations in network position by participation in randomized controlled trials

  Social Networks · 2026-03-28

  articleOpen access

  Researchers and practitioners rely on randomized controlled trials (RCTs) to make causal inferences. However, most people who participate in RCTs are part of multiple, overlapping social networks that shape their behaviors and attitudes. As a result, variations in trial participants’ and non-participants’ network positions may impede the generalizability of a RCT’s conclusions. The current project evaluates the extent and impact of these variations by comparing the network positions of RCT participants to RCT non-participants. As an informative case study, we considered a workplace-based RCT at a large hospital where a subset of employees was randomized to a healthy eating intervention or control group from 2016 to 2019. We constructed longitudinal social networks from data about employees’ cafeteria purchases and applied stochastic actor-oriented models (SAOMs) to determine whether RCT participants and non-participants occupied significantly different structural positions. Then, we performed a series of computational knockout experiments to assess whether the elimination of specific network-related phenomena impacted estimates of the intervention’s effect. Results suggest that RCT participants made cafeteria co-purchases with more of their colleagues than non-participants did. These differences downwardly biased estimates of the intervention’s impact, both with respect to the trial’s efficacy among participants and its expected effectiveness in the larger population of employees. • Limited research considers how variations in network position impact RCT findings • Focus on a healthy eating workplace-based intervention as a case study • Construct co-purchasing networks from employees’ cafeteria transactions • RCT participants are highly connected and there is homophily by RCT enrollment • Differences in network position impacted estimates of the intervention’s effect

  PublisherDOI

• A metabolite index of ultra-processed food intake is associated with stroke, cancer mortality, and all-cause mortality

  medRxiv · 2026-01-12

  articleOpen access

  Ultra-processed foods (UPFs) are associated with adverse health outcomes, but measurement of UPF intake in epidemiological studies remains challenging. Dietary assessments typically employ recall questionnaires, which may lead to measurement errors and misclassifications. Here, we develop a plasma metabolite index of UPF consumption to objectively measure UPF intake. Using two years of authentic food purchasing records from the ChooseWell 365 cohort, we identify metabolites associated with long-term UPF intake. We then apply the UPF-metabolite index to the REasons for Geographic and Racial Differences in Stroke (REGARDS) cohort, a population-based cohort of U.S. adults followed for incident health outcomes. The UPF-metabolite index was associated with leading causes of morbidity and mortality, including incident stroke, all-cause mortality, and cancer mortality. Mediation analysis revealed that the UPF-metabolite index accounts for 62% of the association between UPF intake and stroke risk. These findings suggest that the UPF-metabolite index offers an objective method for assessing UPF intake and its contribution to diet-related disease risk.

  PublisherOA PDFDOI

• SNAP Decisions: Qualitative Research on the Perceived Influence of the COVID-19 Pandemic and SNAP Benefit Changes on Food Purchases

  Journal of the Academy of Nutrition and Dietetics · 2026-05-01

  article
  PublisherDOI

• Community Health Worker Perspectives on Home-Based Care and Remote Monitoring in Heart Failure

  Journal of Primary Care & Community Health · 2026-04-01

  articleOpen accessSenior author

  BACKGROUND: Heart failure (HF) affects over 6 million people in the United States and is a frequent cause of hospital readmissions. While interest in individual HF interventions like digital platforms and home-based care delivered by community health workers (CHWs) has risen, few studies have explored barriers and facilitators to engaging CHWs with digital or remote monitoring devices in HF populations. METHODS: Three virtual focus groups were conducted from December 2020 to January 2021 with 14 CHWs. Key domains included: barriers and facilitators to managing HF at home; CHW care in HF; remote monitoring and technology in HF; and CHW ability to assist with remote technology in HF. OBJECTIVE: To explore CHW perceptions about managing HF at home relevant to the use of home-based remote monitoring. FINDINGS: Focus groups produced 4 themes: (1) patients with HF struggle to follow clinical care plans at home (i.e., medication adherence, diet, fluid intake); (2) CHWs clarify clinical care plans and provide psychosocial support; (3) CHWs connect patients to resources addressing unmet health-related social needs (e.g., transportation, rental assistance, and other needs); (4) patients face challenges understanding how to use clinician-prescribed technologies for remote monitoring that can be resolved with CHW coaching and instruction. CONCLUSIONS: CHWs reported specific ways they were able to assist patients with clinical, social, and clinician-prescribed health-related technology in HF. Partnerships with CHWs are well positioned to add important value to HF home management for clinical care plan adherence and engagement, remote monitoring, and technology use.

  PublisherDOI

• Design of Heart Well: A randomized pilot study of a community health worker-delivered nutrition intervention for low-income adults with hypertension

  Contemporary Clinical Trials · 2026-01-28

  articleSenior author
  PublisherDOI

• Network threats to causal inference: Variations in network position by participation in randomized controlled trials

  CrimRxiv · 2026-04-07

  articleOpen access

  Researchers and practitioners rely on randomized controlled trials (RCTs) to make causal inferences. However, most people who participate in RCTs are part of multiple, overlapping social networks that shape their behaviors and attitudes. As a result, variations in trial participants’ and non-participants’ network positions may impede the generalizability of a RCT’s conclusions. The current project evaluates the extent and impact of these variations by comparing the network positions of RCT participants to RCT non-participants. As an informative case study, we considered a workplace-based RCT at a large hospital where a subset of employees was randomized to a healthy eating intervention or control group from 2016 to 2019. We constructed longitudinal social networks from data about employees’ cafeteria purchases and applied stochastic actor-oriented models (SAOMs) to determine whether RCT participants and non-participants occupied significantly different structural positions. Then, we performed a series of computational knockout experiments to assess whether the elimination of specific network-related phenomena impacted estimates of the intervention’s effect. Results suggest that RCT participants made cafeteria co-purchases with more of their colleagues than non-participants did. These differences downwardly biased estimates of the intervention’s impact, both with respect to the trial’s efficacy among participants and its expected effectiveness in the larger population of employees.

  PublisherOA PDFDOI

• Interventions to Optimize Dietary Patterns in Black Pregnant Individuals: An Integrative Review

  Birth · 2026-03-02

  article

  INTRODUCTION: Black pregnant individuals bear an inequitable burden of maternal morbidity and mortality in many high income countries (HICs). Adverse social determinants of health and health-related factors, including dietary patterns, influence these disproportional rates. Dietary patterns are highly individualized and are associated with health and pregnancy outcomes in distinctive ways across to races and ethnic groups. While nutrition interventions are effective in improving dietary patterns in pregnant individuals, trial participants are predominantly non-Hispanic White. The objective of this integrative review is to identify and evaluate research exploring interventions to optimize dietary patterns in Black pregnant individuals in HICs. METHODS: Electronic searches were conducted in October, 2023 and June, 2025 and reported according to the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) 2020 statement. Eligibility was determined based on select criteria specific to the study participants, setting, methodological approach, intervention, and publication type. Final articles were critically appraised using the Joanna Briggs Institute and Consolidated Standards of Reporting Trials tools. Data extraction and synthesis of eligible articles was conducted as outlined by Whittemore and Knafl. RESULTS: Fourteen articles from ten studies were included in the final review and were overall favorably appraised. Half of these studies enrolled Black or African American pregnant women exclusively. Interventions included nutrition education, behavioral counseling, and nutritious food delivery and were delivered in a variety of settings and modalities. Measurement and analysis of dietary assessments were inconsistently reported, and change in diet quality was the primary outcome for only three articles. Only one intervention was effective in improving participants' dietary intake. CONCLUSION: Results from this integrative review illustrate the paucity of research on interventions to optimize dietary patterns in Black pregnant individuals in (HICs). Future nutrition intervention studies are needed among Black pregnant individuals, that include comprehensive reporting of participant demographics, social determinants of health, and dietary patterns. This will allow better understanding of the relationship between these factors and pregnancy and health outcomes for Black pregnant individuals.

  PublisherDOI

• Sociodemographic and behavioral factors associated with diet quality in low-income adults with prediabetes and type 2 diabetes

  Primary care diabetes · 2025-11-26

  articleOpen accessSenior author
  PublisherOA PDFDOI

Recent grants

• LiveWell 2: Assessment of a Medicaid ACO program to address food and housing insecurity

  NIH · $736k · 2019–2024

• Psychological, cognitive, and genetic factors in a behavioral intervention to prevent weight gain

  NIH · $1.7M · 2017–2021

• NIH Grant K23HL093221

  NIH · $785k · 2015

• Mentorship in patient-oriented research on behavioral and social determinants of cardiometabolic health

  NIH · $504k · 2023–2028

• LiveWell 2: Assessment of a Medicaid ACO program to address food and housing insecurity

  NIH · $3.9M · 2019–2029

Frequent coauthors

• Douglas E. Levy

  Harvard University

  177 shared

• Eric B. Rimm

  Brigham and Women's Hospital

  114 shared

• Jessica L. McCurley

  San Diego State University

  86 shared

• Vicki Fung

  Harvard University

  70 shared

• Rebecca L. Franckle

  Mathematica Policy Research

  55 shared

• Michele Polacsek

  University of New England

  55 shared

• Alyssa J. Moran

  Johns Hopkins University

  53 shared

• Emily D. Gelsomin

  Massachusetts General Hospital

  44 shared

Awards & honors

• Chair of the American Heart Association’s Nutrition Committe…