About

George D. Papandonatos is a Professor of Biostatistics (Research) at Brown University, having joined the institution in 1998. His academic background includes a B.Sc. in Mathematics, Operational Research & Economics from the University of Essex, a Diploma in Statistics from the University of Kent, and both M.S. and Ph.D. degrees in Statistics from the University of Minnesota. His research focuses on behavioral medicine, epidemiology, and biostatistics, with extensive collaborations in areas such as lifestyle interventions and environmental exposure assessment. Dr. Papandonatos has served as Principal Investigator and project biostatistician on numerous NIH-funded projects, and has supervised multiple graduate thesis and dissertation projects. His work encompasses a broad range of public health issues, including smoking cessation, physical activity, cognitive aging, and environmental health, contributing significantly to the understanding of how environmental and behavioral factors influence health outcomes.

Research topics

• Medicine
• Family medicine
• Gerontology
• Computer Science
• Internal medicine
• Audiology
• Physical medicine and rehabilitation
• Simulation
• Endocrinology
• Biology
• Psychiatry
• Nursing
• Physical therapy
• Psychology

Selected publications

• Blood Lead Concentrations and Depressive and Anxiety Symptoms in Childhood

  JAMA Network Open · 2026-01-28

  articleOpen access

  Importance: Low-level lead exposure during early brain development is associated with lower cognitive abilities and externalizing behavioral problems in children. However, the association of lead exposure with depression and anxiety remains understudied, particularly in later childhood, when these symptoms often manifest. Objective: To examine associations between low-level, serial blood lead concentrations in children and symptoms of depression and anxiety and to investigate for periods of susceptibility. Design, Setting, and Participants: This cohort study used data from a prospective population of 218 caregiver-child dyads (218 children and 218 parents in the Health Outcomes and Measures of Environment [HOME] Study) recruited from 2003 to 2006 in Cincinnati, Ohio. Children and their families were followed up from the second trimester to age 12 years (2016-2019). Data were analyzed between June 2024 to November 2025. Exposure: Serial blood lead concentrations in children were measured at ages 1, 2, 3, 4, 5, 8, and 12 years. Main Outcomes and Measures: We measured self- and caregiver-reported child depressive and anxiety symptoms at age 12 years using the Behavioral Assessment System for Children-3 (BASC-3), Children's Depression Inventory-II (CDI-II), and Screen for Child Anxiety Related Disorders (SCARED). Results: Among 218 children (121 female [55.5%]; 78 Black [35.8%] and 140 White and other race or ethnicity [64.2%]; mean [SD] age, 12.4 [0.7] years), the median (IQR; range) mean within-child blood lead concentration was 9.6 (7.8 to 12.6; 4.8 to 32.4) μg/L. Each doubling in mean childhood blood lead concentrations was associated with increased risk of elevated child-reported depressive symptoms on the BASC-3 (relative risk [RR], 1.90; 95% CI, 1.00 to 3.66; P = .05) and increased risk of child- and caregiver-reported child depressive symptoms (RR, 1.76; 95% CI, 1.12 to 2.78; P = .02). By blood lead concentration, there were increasing adjusted mean differences in self-reported BASC-3 depression scores from age 1 year (1.82; 95% CI, -1.10 to 4.74; P = .22), a nonsignificant outcome, through age 8 years (3.22; 95% CI, 0.53 to 5.90; P = .02), a significant outcome. There were no associations between blood lead concentrations and self-reported depression measured with the Children's Depression Inventory-II or anxiety measured with the SCARED. Child sex and race did not modify these associations. Conclusions and Relevance: In this study, low-level childhood blood lead concentrations were associated with self-reported depressive symptoms in later childhood, with particularly large increases in risk for exposures occurring in late childhood and early adolescence. These findings suggest that lead exposure during childhood may be associated with mental health in later childhood, highlighting the need for continued efforts to prevent lead exposure.

  PublisherOA PDFDOI

• Reliability and validity of the Rhode Island Mobile Cognitive Assessment Tool

  Alzheimer s & Dementia Diagnosis Assessment & Disease Monitoring · 2026-01-01

  articleOpen access

  Abstract INTRODUCTION This study evaluated the reliability and validity of the Rhode Island Mobile Cognitive Assessment Tool (RIMCAT), a proctored digital cognitive screening test. METHODS One hundred older adult participants (healthy controls [HC] n = 50; cognitively impaired [CI], n = 50) completed the RIMCAT. RESULTS Test–retest reliability across 4 weeks was excellent. Internal consistency was high. RIMCAT total scores were significantly correlated with the Mini‐Mental State Examination and the Mattis Dementia Rating Scale. RIMCAT component scores were significantly correlated with corresponding in‐office neuropsychological measures. Logistic regression analysis of RIMCAT total scores correctly classified 77.9% of individuals as cognitively impaired or cognitively healthy. Receiver operating characteristic analysis revealed an optimal sensitivity of 84.8% and specificity of 79.6%. A second model, including total reaction time, correctly classified 83.2% of cases and resulted in sensitivity and specificity of 84.8% and 83.7%, respectively. DISCUSSION Results supported RIMCAT as an effective digital tool for cognitive screening of older adults in supervised settings.

  PublisherDOI

• Cumulative childhood lead exposure estimation and school-age IQ in a prospective birth cohort

  Environmental Health · 2025-11-27

  articleOpen access

  BACKGROUND: Lead is a well-known neurotoxicant with no identified safe level. Prior studies found that childhood lead exposure is associated with decreased intelligence quotient (IQ) scores. However, most studies rely on a limited number of blood lead measurements. In this prospective pregnancy and birth cohort, we estimated cumulative childhood lead exposure using repeated blood lead concentrations and regression calibration, allowing for more accurate assessment of lead burden over time and its association with IQ. METHODS: This prospective study included 262 mother-child dyads from Greater Cincinnati enrolled in the Health Outcomes and Measures of the Environment (HOME) Study from 2003 to 2006. We obtained serial blood lead measurements and estimated cumulative childhood lead exposure using a regression calibration method. Outcome was assessed via Wechsler-based IQ testing at ages 5-12 years. We examined the association between estimated cumulative childhood lead exposure and child IQ using linear regression models. RESULTS: Our cohort had low levels of estimated lifetime average lead exposure (geometric mean: 1.21 μg/dL). Overall, estimated lead exposure decreased from age 12 months to time of IQ test. Cumulative childhood lead exposure estimate was associated with decreased IQ at ages 5-12 years in unadjusted analyses, but not after adjusting for maternal IQ, household income, reported prenatal vitamin use, Home Observation for Measurement of the Environment score, and maternal serum cotinine. Sensitivity analyses additionally adjusting for prenatal total folate did not markedly change our results. We assessed early-life, school-age, or concurrent blood lead exposure estimate in place of cumulative childhood lead exposure estimate and observed a similar pattern of results. CONCLUSIONS: We used a regression calibration method to leverage robust, repeated lead exposure data in our prospective pregnancy and birth cohort. In this cohort with low levels of lead exposure, cumulative childhood lead exposure estimate was negatively associated with school-age IQ in unadjusted analyses but not adjusted analyses. We considered sociodemographic and maternal factors previously associated with cognitive development. Our results suggest these factors may confound the association between low-level child lead exposure and child IQ.

  PublisherDOI

• Baseline predictors of smoking cessation among individuals with current or past major depressive disorder following treatment with varenicline and intensive behavioral treatment

  Drug and Alcohol Dependence · 2025-06-21

  articleSenior author
  PublisherDOI

• Dual abstinence from nicotine vaping and cannabis use among young people: secondary analyses from two U.S.-based randomized controlled trials of vaping cessation

  Substance Abuse Treatment Prevention and Policy · 2025-11-02

  articleOpen accessSenior author

  OBJECTIVE: To examine patterns of abstinence from nicotine vaping and cannabis use among adolescent and young adult (YA) e-cigarette users in two text message vaping cessation trials. METHODS: Among adolescents with complete 7-month data (n = 1,016) at baseline, 25.4% were Exclusive E-cigarette Users (no past 30-day cannabis use) and 74.6% were Dual Users (past 30-day cannabis use). Among YAs with complete 7-month data (n = 1,829), 40.8% were Exclusive E-cigarette Users and 59.2% were Dual Users at baseline. Primary analyses examined the proportion of participants who were Dual Abstinent at 7-months by treatment arm differences. We also examined for interaction effects between baseline product use and vaping status at 7 months on cannabis use outcomes. RESULTS: At 7-months, adolescent categories of use were: Dual Abstinent, 31.7% (95% CI: 28.8, 34.6); Exclusive E-cigarette Users, 18.2% (95% CI: 15.9, 20.7); Exclusive Cannabis Users, 15.1% (95% CI: 12.9, 17.4); Dual Users, 35.0% (95% CI: 32.1, 38.1). Among YAs: Dual Abstinent, 15.6% (95% CI: 13.9, 17.3); Exclusive E-cigarette Users, 29.4% (95% CI: 27.3, 31.6); Exclusive Cannabis Users, 12.8% (95% CI: 11.3, 14.5); Dual Users, 42.2% (95% CI: 39.9, 44.5). Intervention outperformed Control in promoting rates of Dual Abstinence among adolescents (38.5% vs. 25.0%, p < 0.0001) and YAs (17.9% vs. 13.3%, p = 0.007). A higher proportion of Exclusive E-cigarette Users compared to Dual Users were Dual Abstinent at follow-up (adolescents: 37.6% vs. 29.7%, p = 0.019; YAs: 25.8% vs. 8.5%, p < 0.001). CONCLUSION: A text message nicotine vaping cessation intervention promoted dual abstinence from e-cigarettes and cannabis among adolescents and YAs. Dual abstinence rates were higher among exclusive vapers than dual users, signaling the need to optimize cessation programs for dual users. TRIAL REGISTRATION: Studies included were registered on ClinicalTrials.gov (NCT04251273, registered on January 31, 2020; NCT04919590, registered on June 9, 2021).

  PublisherOA PDFDOI

• Does the timing of sample collection confound the association between prenatal serum PFAS concentrations and birthweight: results from two prospective cohort studies

  Environment International · 2025-12-06 · 1 citations

  articleOpen access

  BACKGROUND: Maternal serum concentrations of per- and polyfluoroalkyl substances (PFAS) during pregnancy have been consistently associated with reduced birthweight. However, these associations may be confounded by the timing of sample collection due to pregnancy-related changes in pharmacokinetics that impact both serum PFAS concentrations and birthweight. METHODS: We measured nine serum PFAS concentrations in each trimester among 728 women in the MABC Study (Ma'anshan, China, enrolled 2013-2014). We estimated covariate-adjusted differences in birthweight for individual PFAS and their mixture, as well as percent change in PFAS by infant birthweight. We repeated our analyses in 70 women-infant dyads from the HOME Study (Cincinnati, Ohio, enrolled 2003-2006) with repeated PFAS measures in 2nd trimester and at delivery. RESULTS: -transformed 1st and 3rd trimester PFOA was associated with a 6-gram (95 % CI:-35, 22) and 26-gram (95 % CI:-54, 3) decrease in birthweight, respectively. Concentrations of six PFAS declined more steeply among women who gave birth to larger vs. smaller infants. PFOA declined by 11.3 % (95 % CI:-12.7, -9.9) vs. 8.9 % (95 % CI:-10.3, -7.5) per 13 weeks among women who gave birth to infants born at the 90th vs. 10th birthweight percentiles, respectively. We observed similar findings in the HOME Study. CONCLUSIONS: These findings suggest that physiological changes in pregnancy may confound associations between serum PFAS and birthweight. Studies with PFAS measures later in pregnancy may overestimate the true effect of PFAS on fetal growth.

  PublisherDOI

• Prenatal PFAS exposures and cardiometabolic health in middle childhood in the MIREC cohort

  Environmental Research · 2025-03-06 · 1 citations

  article
  PublisherDOI

• Successful Digital Recruitment of Adolescents for Population-Based Research: A Case Study of Methods and Strategies

  Academic Pediatrics · 2025-11-09

  review
  PublisherDOI

• Evaluating neonatal cord serum metabolome in association with adolescent cardiometabolic risk factors

  Pediatric Research · 2025-08-16 · 2 citations

  article
  PublisherDOI

• The impact of early childhood adversity on adult cognitive performance: The mediating roles of childhood cognitive ability and educational attainment

  Social Science & Medicine · 2025-07-21

  article
  PublisherDOI

Frequent coauthors

• Amanda L. Graham

  Georgetown University Medical Center

  172 shared

• Joseph M. Braun

  Providence College

  171 shared

• Brian R. Ott

  Brown University

  120 shared

• Kimberly Yolton

  Cincinnati Children's Hospital Medical Center

  116 shared

• Bruce P. Lanphear

  Child and Family Research Institute

  116 shared

• Raymond Niaura

  New York University

  92 shared

• Laura R. Stroud

  Brown University

  73 shared

• Jennifer Davis

  Brown University

  72 shared

Education

• B.S., Mathematics, Operational Research & Economics

  University of Essex, UK

• Other, Statistics

  University of Kent, UK

• M.S., Statistics

  University of Minnesota (Twin Cities), USA

• Ph.D., Statistics

  University of Minnesota (Twin Cities), USA