About

Thomas Trikalinos is a Professor of Health Services, Policy and Practice at Brown University and serves as the Director of the Center for Evidence Synthesis in Health (CESH). He studied medicine in Greece and has a research focus on developing novel methodologies for comparative effectiveness research, emphasizing evidence synthesis through systematic review and meta-analysis, as well as evidence contextualization via decision and economic analysis. His work aims to modernize and optimize evidence-synthesis processes by integrating methodologies from computer science and applied mathematics. His current research concentrates on decision making under deep uncertainty, contributing to the advancement of evidence-based medicine and health policy.

Research topics

• Medicine
• Political Science
• Computer Science
• Internal medicine
• Psychiatry
• Statistics
• Psychotherapist
• Mathematics
• Clinical psychology
• Nursing
• Geography
• Endocrinology

Selected publications

• Fiber intake and laxation in people with normal bowel function: a systematic review

  American Journal of Clinical Nutrition · 2026-01-27 · 1 citations

  articleOpen access
  PublisherOA PDFDOI

• Differentially Private Modeling of Disease Transmission within Human Contact Networks

  arXiv (Cornell University) · 2026-04-08

  articleOpen access

  Epidemiologic studies of infectious diseases often rely on models of contact networks to capture the complex interactions that govern disease spread, and ongoing projects aim to vastly increase the scale at which such data can be collected. However, contact networks may include sensitive information, such as sexual relationships or drug use behavior. Protecting individual privacy while maintaining the scientific usefulness of the data is crucial. We propose a privacy-preserving pipeline for disease spread simulation studies based on a sensitive network that integrates differential privacy (DP) with statistical network models such as stochastic block models (SBMs) and exponential random graph models (ERGMs). Our pipeline comprises three steps: (1) compute network summary statistics using \emph{node-level} DP (which corresponds to protecting individuals' contributions); (2) fit a statistical model, like an ERGM, using these summaries, which allows generating synthetic networks reflecting the structure of the original network; and (3) simulate disease spread on the synthetic networks using an agent-based model. We evaluate the effectiveness of our approach using a simple Susceptible-Infected-Susceptible (SIS) disease model under multiple configurations. We compare both numerical results, such as simulated disease incidence and prevalence, as well as qualitative conclusions such as intervention effect size, on networks generated with and without differential privacy constraints. Our experiments are based on egocentric sexual network data from the ARTNet study (a survey about HIV-related behaviors). Our results show that the noise added for privacy is small relative to other sources of error (sampling and model misspecification). This suggests that, in principle, curators of such sensitive data can provide valuable epidemiologic insights while protecting privacy.

  PublisherOA PDF

• IP35-01 PREDICTORS OF ADHERENCE TO INTRAVESICAL BACILLUS CALMETTE-GUERIN (BCG) INDUCTION SCHEDULES FOR THE TREATMENT OF EARLY BLADDER CANCER AMONG U. S. MEDICARE BENEFICIARIES FROM 2004 TO 2019

  The Journal of Urology · 2026-04-27

  articleSenior author
  PublisherDOI

• Differences in Bladder Cancer Diagnosis by Demographic Factors: A Simulation Modeling Analysis

  medRxiv · 2026-02-02

  articleOpen access

  Purpose: Bladder cancer is associated with significant morbidity and mortality in the US, with 85,000 new cases and 17,400 deaths expected in 2025. Black patients are more likely than White patients to be diagnosed with bladder cancer at advanced stages, as are female patients compared with male patients. We examine whether differences in cancer diagnosis rates by race and sex can explain the observed variability using a simulation model and project outcomes of potential improvement in diagnosis. Methods: We developed a state transition model for bladder cancer to simulate four cohorts based on sex (males, females) and race (Blacks, Whites) from birth through various health states, including disease-free, preclinical stages (0a/0is - IV), clinical stages (0a/0is - IV), and death (bladder cancer or other cause death). Parameters related to disease onset, progression, and diagnosis were estimated by calibrating the model to race- and sex-specific incidence rates by age, and stage distribution at diagnosis for cases diagnosed between 2015 and 2019 in SEER 17 registry areas. We conducted a scenario analysis to examine the impact of differences in diagnosis rates on stage distribution and life expectancy, assuming that Black males (or females) and White females had diagnosis rates similar to those of White males. Results: The calibrated model attributes the differences in stage distribution to lower diagnosis rates in White females (hazard ratio, [HR] = 0.95, 95% credible interval [CI]: 0.92 - 0.96), Black males (0.80, 95% CI: 0.75 - 0.81) and Black females (0.56, 95% CI: 0.53 - 0.58), relative to White males. If diagnosis rates for all demographic groups were similar to White males, the expected life span of a 65-year-old bladder cancer patient would increase by 0.2 years for White females (from 13.8 to 13.9 years), 0.6 years for Black males (from 10.6 to 11.1 years), and 1.9 years for Black females (from 10.5 to 12.4 years). Conclusions: Differences in diagnosis rates of bladder cancer by race and sex explain the observed differences in stage distribution at diagnosis. Targeted interventions aimed at improving diagnosis rates have the potential to substantially improve survival for patients with bladder cancer.

  PublisherOA PDFDOI

• Age‐Friendly Health System Implementation in Outpatient Settings: A Systematic Review

  Journal of the American Geriatrics Society · 2026-02-16

  articleOpen access

  INTRODUCTION: The Age-Friendly Health Systems (AFHS) initiative aims to improve care for older adults through the "4Ms" framework: What Matters, Medication, Mentation, and Mobility. Despite national momentum and evidence within individual M domains, limited evidence guides outpatient AFHS implementation of the 4Ms as a set. The objective of this systematic review was to summarize the evidence of the impact of AFHS implementation in outpatient settings. METHODS: We searched Medline, EMBASE, CINAHL, Cochrane, and clinicaltrials.gov from 2015 to November 22, 2024. Comparative studies that implemented all 4Ms in outpatient settings were included. Risk of bias was assessed using questions derived from the Cochrane Risk of Bias tool for RCTs and the Risk of Bias In Non-randomized Studies-of Intervention tool for other study designs, and results were summarized using GRADE methodology. RESULTS: Twelve US-based studies met inclusion criteria. Overall, implementing AFHS interventions was associated with improved process measures across all 4Ms, though the effectiveness of specific implementation strategies could not be determined. Outcome and structural measures were infrequently reported. Study heterogeneity and poor reporting limited generalizability. DISCUSSION: Findings underscore the urgent need for standardized, outcomes-oriented AFHS measurement before policy and payment reforms, such as CMS's Age-Friendly Hospital Measure, are expanded into outpatient settings. To advance meaningful transformation, future research must prioritize implementation fidelity, outcome evaluation, and measures that reflect older adults' values and lived experiences.

  PublisherOA PDFDOI

• Digitally Delivered Cognitive Behavioral Interventions for Alcohol and Other Drug Use: Meta-Analysis Across Consumption and Psychosocial Outcomes

  JMIR Mental Health · 2026-03-17

  articleOpen access

  Background Cognitive behaviorally based interventions have broad appeal and potential for impact when treating adult alcohol and other drug use. Digitally delivered cognitive behaviorally based interventions (dCBIs) may offer this impact with the benefit of increased accessibility. Although prior reviews have indicated the benefits of dCBIs on substance use outcomes, the extension to psychosocial functioning outcomes is unknown. Objective This meta-analysis provides an overview of dCBI effects across a range of functional end points. Methods A literature search was conducted through October 2024. All primary and secondary reports of clinical trials of dCBI were obtained, and all available study end points were eligible for meta-analysis. Descriptive data were extracted and categorized into 1 of 13 different outcome types (eg, abstinence, quantity, cognitive, and quality of life) and into 2 broader outcome classes (ie, consumption and psychosocial). Robust variance estimation was used to conduct hypothesis tests on random effects pooled estimates with outcome class and comparison type as the primary subgroup variables of interest. Results The study sample included 65 randomized trials (K=110 publications; 753 effect sizes) of dCBI for adult alcohol and other drug use. With respect to efficacy, dCBI as a stand-alone treatment in contrast to a minimal treatment control showed positive and statistically significant effects for consumption (g=0.27; P<.001; I2=85.1%; k=31; kes=134) and psychosocial (g=0.16; P=.008; I2=75.2%; k=16; kes=60) outcomes. As an addition to usual care, efficacy was demonstrated for consumption (g=0.23; P<.001; I2=9.8%; k=20; kes=65), but not psychosocial functioning. Efficacy compared to another digital or in-person intervention or cognitive behaviorally based intervention delivered by a therapist was not observed. Within the dCBI condition, large effect sizes were observed for both outcome classes (ie, 60%-80% of participants showed improvement relative to baseline), and effect size magnitude and statistical heterogeneity varied by the type of outcome examined. Conclusions These results show a benefit for dCBI as a stand-alone therapy and an addition to usual care. Importantly, stand-alone effects were observed for both consumption and some psychosocial outcomes. This study is the first to offer a comprehensive look at dCBI intervention effects across a range of functional end points.

  PublisherDOI

• Understanding Bladder Cancer Screening Limits Through Comparative Modeling: The Maximum Clinical Incidence Reduction (MCLIR) Methodology

  medRxiv · 2026-01-30

  articleOpen accessCorresponding

  Background: Bladder cancer imposes substantial clinical and economic burden, yet key natural-history quantities that determine the potential effectiveness of screening-such as the size of the screen-detectable preclinical reservoir and the preclinical sojourn time-are largely unobservable. The Cancer Intervention and Surveillance Modeling Network (CISNET) uses standardized stress tests to compare independently developed microsimulation models and to clarify how differences in model structure translate into differences in intervention impact. The Maximum Clinical Incidence Reduction (MCLIR) framework estimates the maximum achievable reduction in clinically detected incidence following a one-time, perfect screening intervention, while Realistic Clinical Incidence Reduction (RCLIR) relaxes the perfect-test and/or perfect-treatment assumptions. Methods: We applied the CISNET MCLIR/RCLIR protocol to three independently developed bladder cancer microsimulation models (COBRAS, Kystis, and SCOUT), each calibrated to common U.S. epidemiologic targets. We simulated a U.S. birth cohort born in 1950 and compared: (1) no-screening baseline; (2) one-time perfect screening with universally curative treatment (MCLIR) at ages 60, 65, 70, and 75; and (3) one-time realistic screening with cystoscopy sensitivity of 80% and perfect treatment (RCLIR-1) or usual-care treatment effectiveness (RCLIR-2). Outcomes included age-specific clinical incidence and incidence-reduction curves relative to baseline, as well as cumulative reductions over follow-up. Results: Across models, median ages at first lesion emergence, clinical diagnosis, and onset of muscle-invasive disease were similar, but preclinical sojourn time differed meaningfully: COBRAS produced the shortest median sojourn time (2.1 years) compared with Kystis (3.3 years) and SCOUT (3.1 years). Under MCLIR, all models predicted an immediate drop in clinical incidence followed by attenuation toward zero as new lesions emerged after the screening age. Peak MCLIR at age 65 among White men ranged from 20% (COBRAS) to 21% (Kystis) and 32% (SCOUT), with reductions dissipating within ∼8 years in COBRAS and persisting ∼10 years in Kystis and SCOUT. In COBRAS and Kystis, incidence reductions later became negative, consistent with rebound emergence of new lesions among individuals whose first lesions were removed by screening. Under RCLIR-1, peak reductions were smaller and varied substantially across models (13%, 4%, and 37% for COBRAS, Kystis, and SCOUT, respectively). RCLIR-2 further reduced gains and produced more gradual decay, reflecting incomplete prevention under usual-care treatment. Across models, most residual post-screening incidence was attributable to new lesion emergence rather than missed detection or incomplete treatment. Conclusions: In a standardized CISNET stress test, three bladder cancer microsimulation models imply a relatively short detectable preclinical phase, placing a modest upper bound on the effectiveness of one-time screening in the general population. Differences in MCLIR/RCLIR magnitude and persistence are explained by differences in implied sojourn time and detectable reservoir size. These findings motivate evaluation of risk-targeted and repeated early-detection strategies and highlight key empirical priorities for improving inference on bladder cancer natural history.

  PublisherOA PDFDOI

• Treatment Sequence in Advanced Small Bowel Neuroendocrine Neoplasms: A Simulation Study

  Endocrine Abstracts · 2026-03-04

  articleSenior author
  PublisherDOI

• On Representations and Quantifications of Uncertainty

  Medical Decision Making · 2026-01-13 · 1 citations

  articleOpen accessSenior authorCorresponding
  PublisherOA PDFDOI

• Machine Learning Tools To (Semi-)Automate Evidence Synthesis: A Rapid Review and Evidence Map Surveillance Report

  2026-02-13

  reportSenior author
  PublisherDOI

Recent grants

• NIH Grant R01HS018494

  NIH · $1.2M · 2013

• Population Modeling of Bladder Cancer Detection and Control

  NIH · $3.4M · 2021–2026

• Develop Patient Centered Outcomes Scholars for Comparative Effectiveness Research

  NIH · $2.6M · 2014–2020

• Semi-Automating Data Extraction for Systematic Reviews

  NIH · $2.4M · 2015–2024

Frequent coauthors

• Ethan M. Balk

  Brown University

  302 shared

• Mei Chung

  Tufts University

  260 shared

• Gowri Raman

  Vel Tech Rangarajan Dr. Sagunthala R&D Institute of Science and Technology

  240 shared

• Issa J Dahabreh

  221 shared

• John P. A. Ioannidis

  Stanford University

  207 shared

• Stanley Ip

  184 shared

• Alice H. Lichtenstein

  United States Department of Agriculture

  180 shared

• Joseph Lau

  176 shared

Education

• M.D.

  Greece