About

Stavroula A. Chrysanthopoulou, PhD, is an Assistant Professor of Biostatistics and the Director of the Master's Program in Biostatistics at Brown University School of Public Health. Her research interests focus on microsimulation modeling (MSM), including the development of the MIcrosimulation Lung Cancer (MILC) model, a streamlined MSM of the natural history of lung cancer, and the publication of a package for its implementation in the R open-source statistical software. Her work spans complex predictive models applied in medical decision making, calibration and predictive accuracy methods, causal inference, missing data, and high-performance computing techniques. Dr. Chrysanthopoulou has extensive teaching experience with graduate-level courses in biostatistics, such as survival and longitudinal data analysis, generalized linear models, and simulation studies. She holds a BSc in Statistics from Athens University of Economics and Business, an MSc in Biostatistics from the University of Athens, and a PhD in Biostatistics from Brown University. Prior to her current role, she was an Instructor of Biostatistics at the University of Massachusetts Medical School.

Research topics

• Medicine
• Internal medicine
• Family medicine
• Nursing
• Environmental health
• Emergency medicine
• Endocrinology
• Psychiatry
• Economic growth

Selected publications

• 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

  article
  PublisherDOI

• Bladder Cancer Burden in the USA: Population Scenarios for 2040

  European Urology Open Science · 2025-12-04 · 2 citations

  articleOpen access

  Background and objective: Bladder cancer is the sixth most common cancer among men and is expensive to manage. We independently developed three microsimulation models that describe its natural history and explain epidemiological trends. We projected bladder cancer burden in the USA through 2040 to inform workforce planning. Methods: We calibrated the models to the Surveillance, Epidemiology and End Results (SEER) program incidence data and standardized key inputs. For White men, the highest-incidence subgroup, the models inferred unobservable epidemiological metrics, including lifetime risks by birth cohort and ages of the key events in the natural history. We simulated individual life histories under calibrated parameter sets and summarized the outcomes as yearly rates and counts. Key findings and limitations: Each model's predictions reproduced SEER age- and stage-specific incidence data. Across models, the lifetime risk of bladder cancer grew from approximately 1.5-2.4% in the 1910 to 3.1-4.4% in the 2010 birth cohorts, consistent with longevity and smoking exposure patterns. Of the cancer cases, 75% instantiate after ages 61-64 yr. The median model durations from when a cancer is screen detectable to its clinical manifestation were 2.1-3.3 yr, with a wide range across individuals. Through 2040, the incidence standardized to the 2000 US population declined by 0.4-0.6%/yr (consistent with the declining smoking rates, the most important environmental risk factor), but the annual incidence and new cases increased by 1.5-1.8%/yr (because the baby boomer population is living longer). Modeling supplements incomplete data with assumptions, but similar findings across independent models suggest some robustness to assumptions. Conclusions and clinical implications: Projected cohort longevity and smoking patterns imply an increased disease burden in the future, which may benefit from commensurate increased research and resources. From the inferred natural history, we speculate a theoretical opportunity for screening, which should be investigated with dedicated modeling and empirical studies. Patient summary: Three computer simulation models predicted the future incidence of bladder cancer burden in White men, in whom this cancer is most common. The models found that although the future incidence of bladder cancer would decrease slightly over time (consistent with the declining smoking rates, the most important environmental risk factor), the overall disease burden increased because the baby boomer population is living longer.

  PublisherDOI

• Health and Economic Outcomes of Addressing Encampments of Individuals Using Opioids

  JAMA Network Open · 2025-06-27 · 2 citations

  articleOpen access

  Importance: Many US communities face a crisis of people experiencing unsheltered homelessness often intertwined with opioid use. Jurisdictions seek policy options for managing unsanctioned encampments of this population, but their various outcomes are unclear. Objective: To evaluate policy options and their health and economic outcomes for an encampment of people experiencing homelessness and opioid use disorder (OUD). Design, Setting, and Participants: This decision analytical model study conducted a closed-cohort state-transition simulation using the Researching Effective Strategies to Prevent Opioid Death (RESPOND) model from October 2021 to October 2022. The study was based primarily on data from Massachusetts and simulated an urban encampment with a population experiencing homelessness and high-risk opioid use. Data analysis was performed from December 2022 to October 2024. Exposure: The following encampment management strategies were modeled: (1) status quo (no sweep); (2) sweep, a sudden disruption of all residents, followed by no additional resources; (3) housing with medication for opioid use disorder (MOUD) requirement; or (4) housing without MOUD requirement. Main Outcomes and Measures: The primary outcomes were overdose and all-cause mortality per 1000 person-years, weeks spent in housing and taking MOUD, and economic cost from a modified government payer perspective. Sensitivity analyses were conducted by varying uncertain parameters. Results: The simulated cohort included 400 adults (mean [SD] age, 48 [17] years; 232 males [58.0%]). Under the status quo strategy, there were 50.4 (95% uncertainty interval [UI], 48.9-52.2) deaths per 1000 person-years, 15.5 (95% UI, 14.0-17.2) deaths from overdose per 1000 person-years, and 2990 (95% UI, 2897-3081) person-weeks spent taking MOUD for a total cost of $6 583 000 (95% UI, $6 502 000-$6 660 000). A sweep strategy resulted in 53.1 (95% UI, 51.3-55.2) deaths per 1000 person-years, 16.4 (95% UI, 18.2-20.2) deaths from overdose per 1000 person-years, and 1694 (95% UI, 1625-1764) person-weeks spent taking MOUD at a total cost of $6 820 000 (95% UI, $6 736 000-$6 899 000). The housing with medication requirement strategy resulted in 51.2 (95% UI, 49.4-53.0) deaths per 1000 person-years, 16.3 (95% UI, 14.6-18.1) deaths from overdose per 1000 person-years, and 3050 (95% UI, 3025-3075) person-weeks spent taking MOUD and in housing, for a total cost of $7 264 000 (95% UI, $7 188 000-$7 336 000). A housing without MOUD requirement strategy resulted in 49.2 (95% UI, 47.6-51.1) deaths per 1000 person-years, 14.3 (95% UI, 12.7-16.2) deaths from overdose per 1000 person-years, and 5014 (95% UI, 4942-5085) person-weeks spent taking MOUD and 14 511 (95% UI, 14 461-14 562) person-weeks spent in housing, for a total cost of $8 822 000 (95% UI, $8 774 000-$8 868 000). Conclusions and Relevance: In this decision analytical model study of approaches to homeless encampments involving individuals with OUD, sweeps increased mortality and spending. Housing without MOUD requirement was the most costly strategy but saved more lives.

  PublisherOA PDFDOI

• Health and Economic Outcomes of Offering Buprenorphine in Shelters in Massachusetts: Results of a Simulation Model

  Drug and Alcohol Dependence · 2025-02-01

  article
  PublisherDOI

• The Impact of Tobacco Smoking and Smoking Cessation on Lung Cancer and Stroke Incidence Among People With HIV on Antiretroviral Therapy in South Africa: A Simulation Modeling Study

  American Journal of Respiratory and Critical Care Medicine · 2025-05-01

  article

  Abstract Rationale: With declining HIV-related mortality, over 20% of people with HIV (PWH) in South Africa are now over age 50y, and tobacco-related non-communicable disease burden is increasing. Lung cancer and stroke incidence are increasing among PWH in South Africa and may occur at earlier ages compared to people without HIV. Tobacco smoking is a strong risk factor for lung cancer and stroke, but little is known about the potential benefit of smoking cessation interventions on non-communicable disease incidence among PWH in low- and middle-income countries. We quantified the impact of smoking and smoking cessation on lung cancer and stroke incidence among PWH in South Africa. Quantifying potential reductions in disease burden can help inform policymakers considering integration of smoking cessation interventions with HIV care in South Africa and similar settings. Methods: Using a microsimulation model, we simulated 18 cohorts of initially virologically suppressed PWH over their lifetime, categorized by sex, initial age (35y/45y/55y), and smoking status (current/former/never). Smoking status remains constant throughout the simulation; individuals with former smoking status quit at model start. PWH can disengage from HIV care and experience virologic rebound. We modeled the relative risk of lung cancer for females (males) with current versus never smoking status as 16.69 (15.83), and for females (males) with former versus never smoking status as 1.99-8.80 (1.90-6.18), depending on age at cessation. Corresponding modeled relative risks of stroke were 1.79 (1.54) for current versus never smoking, and 1.00-1.29 (1.00-1.12) for former versus never smoking. We varied HIV-related and smoking-related parameters in sensitivity analyses. Results: Modeled female (male) PWH who stop smoking at age 45y experience 61.3% (70.9%) and 35.6% (18.6%) lower cumulative lung cancer and stroke incidence over 25y compared to people who continue smoking. Through smoking cessation, the proportion alive and lung cancer-free or alive and stroke-free over 25y would increase by 10.4 (9.5) or 10.5 (8.5) percentage points (Figure 1). In sensitivity analysis, smoking and smoking cessation have a greater impact on lung cancer and stroke cumulative incidence if competing HIV-related mortality risks are lower or if PWH experience higher lung cancer and stroke risk compared to people without HIV. Conclusion: Smoking cessation could substantially reduce lung cancer and stroke risk among PWH in South Africa. To reduce the rising non-communicable disease burden among PWH, smoking cessation should become part of routine care of PWH.

  PublisherDOI

• Projected Impact of Higher E-cigarette or Combustible Cigarette Taxes on Use of Both Products in the U.S.

  AJPM Focus · 2025-08-06

  articleOpen accessSenior author

  <h2>ABSTRACT</h2><h3>Background</h3> <b>:</b> Higher taxes on e-cigarettes (e-cigs) could reduce their use but might encourage combustible tobacco cigarette use. Simulation models enable evaluation of dynamic transitions between cigarette and e-cig use or non-use and population-level projections in response to a policy change, such as higher taxes. <h3>Methods</h3> <b>:</b> We used a microsimulation model to project the relative impact of higher taxes on cigarettes, e-cigs, or both on the prevalence of cigarette use and e-cig use among U.S. youth (ages 12-17y), young adults (ages 18-24y), and adults (ages ≥25y). We used multi-state models to derive probabilities of transition across states of current/former/never cigarette smoking and e-cig use from Waves 2-5 (2015-2019) of the Population Assessment of Tobacco and Health (PATH) Study. We applied these transition probabilities to a simulated cohort in STOP, reflective of PATH Wave 5, to project cigarette use and e-cig use over five years in a <i>Status Quo</i> scenario. For tax scenarios, we estimated the impact of a $1 increase in cigarette tax (<i>Cigarette Tax</i>), e-cig tax (<i>E-cig Tax</i>), or both (<i>Combined Tax</i>). We performed sensitivity analysis around transition parameters and tax effects to evaluate the influence of uncertainty in these estimates. <h3>Results</h3> Compared with <i>Status Quo</i>, the relative changes among youth/young adults/adults in prevalence of use at five years would be: <i>Cigarette Tax</i>, -4/-19/-3% (cigarette use) and 0/+3/+3% (e-cig use); <i>E-cig Tax</i>, 0/-10/-2% (cigarette use) and -27/-63/-25% (e-cig use); and <i>Combined Tax</i>, -5/-28/-3% (cigarette use) and -27/-62/-25% (e-cig use). Model results are most variable in young adults, reflecting greater uncertainty around tax effects in this population. Cigarette smoking and e-cig use prevalence results, regardless of tax impact, are sensitive to probabilities of initiation among youth and young adults and probabilities of cessation among adults. <h3>Conclusions</h3> <b>:</b> A higher tax on e-cigs alone would likely reduce e-cig use prevalence at 5 years and might not increase cigarette smoking prevalence. A combined tax on both products would likely produce larger reductions in cigarette and e-cig use, maximizing public health benefits.

  PublisherOA PDFDOI

• Empirical calibration of a simulation model of opioid use disorder

  PLoS ONE · 2025-03-27

  articleOpen accessSenior authorCorresponding

  BACKGROUND: Simulation models of opioid use disorder (OUD) aim at evaluating the impact of different treatment strategies on population-level outcomes. Researching Effective Strategies to Prevent Opioid Death (RESPOND) is a dynamic population, state-transition model that simulates the Massachusetts OUD population synthesizing data from multiple sources. Structural complexity and scarcity of available data for opioid modeling pose a special challenge to model calibration. We propose an empirical calibration approach applicable to complex simulation models in general. METHODS: We implement an empirical approach to calibrate RESPOND to multiple targets: annual fatal opioid-related overdoses, detox admissions, and OUD population sizes. The empirical calibration involves Latin hypercube sampling for searching a multidimensional parameter space comprising arrivals, overdose rates, treatment transition rates, and substance use state transition probabilities. The algorithm accepts proposed parameters when the respective model outputs lie within pre-determined target uncertainty ranges. This is an iterative process resulting in a set of parameter values for which the model closely fits all the calibration targets. We validated the model assessing its accuracy to projections important for shared decision-making of OUD outside the training data. RESULTS: The empirical calibration resulted in a model that fits well both calibration and validation targets. The flexibility of the algorithm allowed us to explore structural and parameter uncertainty, reveal underlying relationships between model parameters and identify areas of model improvement for a more accurate representation of the OUD dynamics. DISCUSSION: The proposed empirical calibration approach is an efficient tool for approximating parameter distributions of complex models, especially under complete uncertainty. Empirically calibrated parameters can be used as a starting point for a more comprehensive calibration exercise, e.g., to inform priors of a Bayesian calibration. The calibrated RESPOND model can be used to improve shared decision-making for OUD.

  PublisherOA PDFDOI

• Modeling Recurrent Events: A Tutorial Based on Relapse and Remitting Episodes during Medication-Assisted Treatment for Opioid Use Disorder

  Medical Decision Making · 2025-12-14

  articleOpen access1st authorCorresponding

  In health or medical studies, participants can often experience the outcome(s) of interest multiple times during the observation period, creating recurrent event data. Depending on the primary research objective, advanced statistical methods are required to correctly analyze this special type of data. This tutorial discusses 4 general frameworks, appropriate for analyzing recurrent events data: 1) extended Cox, 2) parametric survival, 3) longitudinal, and 4) multistate models. We present in detail the implementation of these methods, including a description of the required dataset structure, R code, and interpretation of results, using data from the CTN-0051 study, a randomized clinical trial comparing the effectiveness of opioid use disorder treatments. The objectives of 3 use case scenarios exemplify the usage and relevance of the methods for the analysis of recurrent events: 1) estimate adjusted effects, 2) make individual-level predictions, and 3) model a complicated process involving multidirectional transitions between disease states. We compare the methods, comment on their strengths and limitations, and make recommendations on the preferred method depending on the primary research objective.HighlightsRecurrent events are a common phenomenon in experimental research settings, and their analysis requires advanced survival modeling approaches. This tutorial aims to explain and make these approaches more accessible with code and detailed instructions.We compare a detailed list of statistical methods for analyzing recurrent events and make suggestions on which one should be used depending on the study objective.This tutorial will enable researchers to make better use of recurrent events data.

  PublisherOA PDFDOI

• Simulation Models for Bladder Cancer: A Scoping Review

  medRxiv · 2025-03-18

  reviewOpen access1st authorCorresponding

  Objectives: The study identifies and summarizes information from manuscripts using simulation models for Bladder Cancer (BCA) research. Methods: We conducted and presented results of a systematic literature search of Medline, Web of Science, and Google scholar, following the PRISMA guidelines for scoping reviews. We summarized extracted key components of the methodology, data sources, and software used for the development of simulation models and classify eligible articles in terms of the study objectives and conclusions. Results: The 97 identified modeling studies simulating aspects of BCA included models that (1) describe the biological process of carcinogenesis and tumor progression (mostly compartmental models); (2) examine the impact of screening protocols and interventions on disease progression and prognosis (mostly microsimulation models); and (3) assess the cost-effectiveness of BCA treatment and control strategies (cohort-based simulation models or simpler decision tree structures). The scope, objectives, and conclusions of these studies varied substantially. Most focused on evaluating treatments, mostly for non-muscle invasive bladder cancer, with some examining BCA screening and surveillance. Their objectives, methods, and analyses were inconsistently and often incompletely reported. Conclusions: Simulation models in bladder cancer examine questions that span the range from tumor kinetics to cost effectiveness of tumor management, but shortcomings in their reporting hinder assessments of their applicability and methodological rigor, severely limiting their practical usefulness. Highlight statements: We assessed the available landscape of simulation modeling for health decision making in BCA research.Shortcomings in the reporting of this research severely limit their practical usefulness.Future population modeling should assess BCA screening and surveillance. Strengths: This is the first, to our knowledge, systematic appraisal of simulation models in bladder cancer. Simulation modeling will be a key technology to assess the utility of highly promising novel diagnostics and treatments, while evidence accumulates.The described variation in the objectives, methodological rigor, and reporting of models' development, validation, and analysis likely generalize to other disease areas. Limitations: This descriptive compendium does not explicitly compare the results of different models between them or with observed data.

  PublisherOA PDFDOI

• Likes and Hashtags: Exploring the Potential Relationship Between Social Media Use and the Emotional Wellbeing of Oncology Professionals

  Pediatric Blood & Cancer · 2025-01-25 · 1 citations

  articleOpen access

  PURPOSE: To explore the potential relationship between social media (SoMe) and burnout or overall wellbeing within the field of oncology. DESIGN: A cross-sectional study of adult and pediatric oncology professionals conducted using an anonymous electronic survey. The survey was disseminated through the Children's Oncology Group (COG) and the SWOG Cancer Research Network (SWOG) member listservs. RESULTS: The majority of pediatric and adult oncology professionals are not engaging on, with only 873/3000 (29%) using SoMe professionally. Use of SoMe was associated with statistically significant higher incidence of self-reported burnout and poorer self-reported work‒life integration (WLI). However, both groups reported the same degree of career satisfaction and choosing the same career/job again. SoMe users and non-users reported similar overall psychological distress, although the use of SoMe was associated with less severe psychological distress. CONCLUSION: While SoMe users reported higher rates of burnout and poorer WLI compared to non-users, it was not accompanied by higher levels of psychological distress. Furthermore, there were no differences in career satisfaction. These misalignments require further study.

  PublisherOA PDFDOI

Frequent coauthors

• Rebecca L. Kinney

  University of Massachusetts Chan Medical School

  25 shared

• Catarina I. Kiefe

  University of Massachusetts Chan Medical School

  25 shared

• Randolph S. Devereaux

  HCA Healthcare

  25 shared

• Darleen Lessard

  University of Massachusetts Chan Medical School

  25 shared

• Molly E. Waring

  University of Connecticut

  25 shared

• Barbara Gandek

  University of Massachusetts Chan Medical School

  25 shared

• Nathaniel Erskine

  University of North Carolina at Chapel Hill

  25 shared

• Robert J. Goldberg

  University of Massachusetts Chan Medical School

  25 shared

Education

• B.S., Statistics

  Athens University of Economics and Business

• M.S., Biostatistics

  University of Athens

• Ph.D., Biostatistics

  Brown University