About

Scarlett L. Bellamy, Sc.D., is an Adjunct Professor of Biostatistics in the Department of Biostatistics and Epidemiology at the University of Pennsylvania's Perelman School of Medicine. She holds a Doctor of Science degree in Biostatistics from Harvard University and a Master’s degree in Biostatistics from Harvard University, as well as a Bachelor of Arts in Mathematics from Hampton University. Her research expertise focuses on biostatistics and epidemiology, with a particular emphasis on public health issues. She is affiliated with the Department of Epidemiology and Biostatistics at Drexel University’s Dornsife School of Public Health and has contributed to various research projects and publications in her field.

Research topics

• Medicine
• Environmental health
• Demography
• Internal medicine
• Psychology
• Demographic economics
• Gerontology
• Economics
• Geography
• Social psychology
• Psychiatry
• Family medicine
• Obstetrics

Selected publications

• Cocaine Use, Unhealthy Alcohol Use, and PainInterference Among People with HIV

  AIDS and Behavior · 2026-01-05

  articleOpen access

  Pain is prevalent among people with HIV (PWH), and many PWH who experience pain also use substances (illicit drug and/or unhealthy alcohol use). While cocaine use and cocaine and alcohol co-use are prevalent in this population, their effects on pain in PWH are unknown. This study aims to investigate the association of cocaine use and co-use of cocaine and alcohol with pain interference among PWH. We completed a secondary analysis of the Boston Alcohol Research Collaboration on HIV/AIDS (ARCH) study, a longitudinal cohort of PWH with a history of substance use. The outcome was pain interference (Brief Pain Inventory). Exposures were recent cocaine use (Addiction Severity Index) and recent unhealthy alcohol use (Timeline Follow Back). Generalized Estimating Equation (GEE) ordinal logistic regression models were employed, adjusted for demographic factors, illicit/non-medical opioid use and cannabis use. Among 251 participants, 22.3% reported unhealthy alcohol use only, 11.1% reported cocaine use only, and 13.2% reported use of both. Cocaine use was associated with greater pain interference (adjusted odds ratio [aOR]: 1.73, 95% confidence interval [CI]: 1.15-2.60), whether or not participants had unhealthy alcohol use (interaction term, p = 0.695). Participants reporting both cocaine and unhealthy alcohol use had greater pain interference than participants reporting neither (aOR: 2.27, 95%CI: 1.35-3.79). Cocaine use was associated with greater pain interference as was co-use of cocaine and unhealthy alcohol among PWH. Considering patterns of substance use can inform clinicians conversations with PWH who may be using substances for pain management.

  PublisherDOI

• Implementation of contingency management with women engaging in polysubstance use

  Addiction Science & Clinical Practice · 2025-08-08

  articleOpen access

  BACKGROUND: Contingency management (CM) is an effective intervention that provides financial incentives as positive reinforcement for reducing opioid or stimulant use. However, it has not been tested in populations of women who inject drugs (WWID) engaging in polysubstance use. METHODS: We aimed to compare the feasibility of two CM protocols designed to encourage illicit stimulant and opioid abstinence among WWID participating in an ongoing HIV prevention trial. Participants completed a 3-month CM period during which they submitted thrice weekly urine toxicology screenings (UTOX). In the 'abstinence from stimulants and opioids' protocol, participants received a $5 USD incentive when metabolites of stimulants and opioids were not detected in urine. In the 'partial-abstinence protocol', they received a $5 USD incentive when metabolites of stimulants or opioids were not detected, thus doubling the potential incentive obtained each visit. Women also received scaling bonuses after three consecutive negative UTOX ($5-$15 USD). We used descriptive statistics to summarize the total number of (1) UTOXs completed and (2) bonuses distributed. Rates of engagement per person per month were calculated (i.e., total number of completed UTOX/3 months*24 participants). Rates of engagement were compared by CM protocol period. RESULTS: Participants were primarily White women (67%) with an average age of 47 years. Self-reported polysubstance use was common (96%) with women reporting injecting an average of 5 times daily (Interquartile Range: 2-7). Participants (N = 24) collectively submitted 177 UTOX during their 3-month CM periods. Rates of non-reactive UTOX results were slightly higher in the partial-abstinence protocol compared to the abstinence from stimulants and opioids protocol (2.9 per month versus 1.0 per month). More bonuses were earned in the partial-abstinence protocol (0.50 bonuses per participant per month) compared to the abstinence from stimulants and opioids protocol (none). There were no study related adverse events in either protocol group during the CM period. CONCLUSIONS: Findings demonstrate the feasibility of a CM protocol that provided financial incentives for partial abstinence, periods with documented stimulant or opioid abstinence, as well as abstinence to both, without the occurrence of iatrogenic effects. Future research focusing on CM protocols with more flexible incentive structures remains critical. TRIAL REGISTRATION: NCT05192434.

  PublisherOA PDFDOI

• Care processes and racial/ethnic differences in family reports of end‐of‐life care among Veterans: A mediation analysis

  UNC Libraries · 2025-05-01

  articleOpen access

  BACKGROUND: Bereaved family members of racial/ethnic minority Veterans are less likely than families of White Veterans to provide favorable overall ratings of end-of-life (EOL) care quality; however, the underlying mechanisms for these differences have not been explored. The objective of this study was to examine whether a set of EOL care process measures mediated the association between Veteran race/ethnicity and bereaved families' overall rating of the quality of EOL care in VA medical centers (VAMCs). METHODS: A retrospective, cross-sectional analysis of linked Bereaved Family Survey (BFS), administrative and clinical data was conducted. The sample included 17,911 Veterans (mean age: 73.7; SD: 11.6) who died on an acute or intensive care unit across 121 VAMCs between October 2010 and September 2015. Mediation analyses were used to assess whether five care processes (potentially burdensome transitions, high-intensity EOL treatment, and the BFS factors of Care and Communication, Emotional and Spiritual Support, and Death Benefits) significantly affected the association between Veteran race/ethnicity and a poor/fair BFS overall rating. RESULTS: Potentially burdensome transitions, high-intensity EOL treatment, and the three BFS factors of Care and Communication, Emotional and Spiritual Support, and Death Benefits did not substantially mediate the relationship between Veteran race/ethnicity and poor/fair overall ratings of quality of EOL care by bereaved family members. CONCLUSIONS: The reasons underlying poorer ratings of quality of EOL care among bereaved family members of racial/ethnic minority Veterans remain largely unexplained. More research on identifying potential mechanisms, including experiences of racism, and the unique EOL care needs of racial and ethnic minority Veterans and their families is warranted.

  PublisherDOI

• Supportive noninvasive ventilation for acute chest syndrome prevention in hospitalized children with sickle cell disease: Determinants of implementation

  Journal of Hospital Medicine · 2025-07-17 · 1 citations

  articleOpen access

  BACKGROUND: In 2017 Boston Medical Center's (BMC) general pediatric inpatient unit implemented bi-level positive airway pressure (BiPAP) as supportive noninvasive ventilation for acute chest syndrome prevention (SNAP) to prevent adverse respiratory outcomes among medically stable, hospitalized patients with sickle cell disease. Barriers and facilitators to SNAP implementation at BMC differ from those in other settings. OBJECTIVE: To examine contextual determinants of SNAP implementation across different settings. METHODS: We conducted semi-structured interviews with inpatient clinicians at three sites (Site 1: extensive implementation, Site 2: limited implementation, Site 3: not implemented) about factors that would affect SNAP implementation. Interviews and analysis were guided by the Promoting Action on Research Implementation in Health Services (PARiHS) framework. RESULTS: We interviewed 29 healthcare workers (physicians, nurses, physician assistants, child life specialists, respiratory therapists and psychologists). Five major themes emerged: (1) Communication between staff, particularly between the night and day shift, is critical for BiPAP initiation and success. (2) Nurses are key to successful SNAP implementation. (3) SNAP implementation requires multidisciplinary support, including integration of respiratory therapists and child life. (4) Individual unit level size, culture and workflow influence implementation (5) Hospital resources and leadership support are important for SNAP implementation. CONCLUSIONS: Successful SNAP implementation across different contexts will require optimized communication between healthcare team members and the integration of respiratory therapy and child life early in the process of BiPAP initiation. Adapting to the unique features of each institution will be critical.

  PublisherOA PDFDOI

• A cluster‐randomized controlled trial of a combination HIV risk reduction and economic empowerment intervention for women engaged in sex work in Uganda

  Journal of the International AIDS Society · 2025-11-01

  articleOpen access

  INTRODUCTION: Women engaged in sex work (WESW) in Uganda face a high risk of HIV and other sexually transmitted infections (STIs), driven by the intersection of gender inequality, poverty and structural barriers. This paper reports on the Kyaterekera Project, a cluster-randomized controlled trial (c-RCT) testing the efficacy of a combined HIV risk reduction (HIVRR) and economic empowerment intervention to reduce biologically confirmed STIs and HIV risk behaviours. METHODS: The study recruited 542 WESW from 19 HIV hotspots across four districts in Uganda between June 2019 and March 2020. Participants were randomized into three groups: (1) HIVRR intervention alone; (2) HIVRR combined with financial literacy training and an unconditional matched savings account; or (3) HIVRR combined with financial literacy training and an unconditional matched savings account and vocational training. Although initially implemented as a three-arm c-RCT, the COVID-19 lockdown prevented the implementation of the vocational training component. Therefore, the two treatment groups were combined, and the trial was re-approved as a two-arm c-RCT. Biological assessments were conducted at baseline, 18 and 24 months. Behavioural assessments were conducted at baseline, 6, 12, 18 and 24 months from April 2019 to December 2023. Primary outcomes included incident HIV acquisitions (seroconversions among baseline HIV-negative participants), point prevalence of STIs at each visit, and the number/proportion of unprotected sexual acts with paying and regular partners. This study utilized community-based participatory research methods, engaging a community advisory board to ensure the study's alignment with local needs. RESULTS: Across follow-up, condomless sex with paying partners decreased and income shifted towards non-sex work in both arms; no between-group differences were detected. Eighteen incident HIV acquisitions occurred (14 by 18 months; 4 additional by 24 months) with no between-group differences. STI prevalence was lower at 18 months compared to baseline, but not sustained at 24 months. CONCLUSIONS: In an environment of high baseline HIV prevalence, substantial pre-exposure prophylaxis uptake and COVID-19 disruptions, the added financial literacy/savings components did not yield measurable incremental benefits over HIVRR alone. Integrating an unconditional matched-savings model within an HIVRR platform was feasible. CLINICAL TRIAL NUMBER: NCT03583541.

  PublisherOA PDFDOI

• Residential Evictions by Life Course, Type, and Timing, and Associations with Self-rated Health: Social Epidemiology to Combat Unjust Residential Evictions (SECURE) Study

  Journal of Urban Health · 2025-05-14 · 1 citations

  articleOpen accessSenior author

  Few existing data sources quantify the magnitude of court-ordered and illegal residential evictions, among historically marginalized groups. We describe the Social Epidemiology to Combat Unjust Residential Evictions (SECURE) Study (2021-2024; n = 1,428; 91.1% response rate) methodology and participant characteristics. Univariable and multivariable statistics including Spearman correlations were used to describe data. Unadjusted and adjusted modified Poisson regression with robust error variance estimated relative risk (RR) and associated 95% confidence intervals (95% CI) for associations between five eviction measures and self-rated health (SRH), and self-rated health relative to most similarly aged peers (RSRH). A quarter of the participants reported experiencing an eviction as a child (n = 354). Over half of the study sample reported ever experiencing a court-ordered (n = 432) and/or an illegal eviction (n = 360). In the past 2 years, 15.2% of the sample reported experiencing a court-ordered (n = 122) and/or illegal eviction (n = 95). Eviction during childhood, and ever experiencing both court-ordered and/or illegal eviction was associated with between 12 and 17% higher risk of poor SRH, and childhood eviction and ever experiencing illegal eviction was associated with between 34 and 37% higher risk of worse RSRH among reproductive age Black women. More community-partnered research using participatory action research methods are needed to understand and intervene upon the health impacts of residential evictions among disproportionately impacted groups.

  PublisherOA PDFDOI

• Neighborhood Threat of Eviction over Time and Risk of Preterm Birth in Black American Women

  Journal of Racial and Ethnic Health Disparities · 2025-05-13 · 1 citations

  articleOpen accessSenior author

  BACKGROUND: Black communities are disproportionately impacted by dual crises: residential evictions and adverse birth outcomes. A growing literature has documented the spill-over effects of neighborhood evictions on adverse birth outcomes, but none have examined associations between these exposures over time and risk of preterm birth (PTB) among Black women. METHODS: We linked survey data from the Life-course Influences on Fetal Environments Study (n = 807) and publicly available block group-level eviction filing rate data. Addresses from the preconception (from 2007 to 2009) and during pregnancy neighborhoods (from 2009 to 2011) were linked to data from the Eviction Lab. Eviction filing rate trajectories included (1) steady low (referent), (2) steady high, (3) decreasing, and (4) increasing categories. PTB was defined as birth before 37 completed weeks of gestation and was abstracted from participant medical records. Modified Poisson regression with robust error variance estimated relative risk and 95% confidence intervals. Models were adjusted for predictors of residential selection (income, education, marital/cohabiting status, and age), as well as duration of residence in current neighborhood, current neighborhood sociodemographic disadvantage, and residential move from the before pregnancy to during pregnancy neighborhood. RESULTS: Preterm birth was experienced in 16.2% of the sample (n = 131), and the mean age of participants was 27 years. In adjusted models, PTB risk was strongly associated with increasing eviction filing rates (compared to steady low) (relative risk: 1.68, 95% CI: 1.05, 2.68). CONCLUSION: Our results provide new evidence about the spillover effects of increased neighborhood threat of eviction over time, on risk of PTB among Black women. Future interventions, including policy solutions aimed at addressing the eviction and PTB crises, are warranted.

  PublisherOA PDFDOI

• Breaking tradition: should biostatistics doctoral qualifying exams evolve to better serve our students’ ability to demonstrate readiness to conduct independent research?

  Frontiers in Public Health · 2025-06-11 · 1 citations

  articleOpen access1st authorCorresponding

  Doctoral programs in science, technology, engineering, and mathematics (STEM) education often include qualifying exams as a central component of the curriculum. While these exams are designed to assess a student's knowledge and potential to conduct independent research as part of the culminating dissertation phase of their studies, they can also inadvertently perpetuate structural biases and barriers for underrepresented groups. Biostatistics programs have increasingly focused on efforts to address diversity. While some programs had long-standing initiatives, others began following the summer of 2020. The momentum following some of these efforts has been disrupted following the recent Supreme Court ruling around the college admissions process. In response to the Association of Schools and Programs of Public Health's Framing the Future 2030 (ASPPH FTF2030) call to action, most specifically to "create and support inclusive and anti-racist teaching, learning, and working environments," we propose examining the structure of the written qualifying examination to mitigate potential disparities in student success in doctoral training programs including the format of the exams, the evaluation criteria, and the support available to students as they prepare for the exam. In this paper, we briefly review the history and founding of our discipline, present data on the continuing under-representation of historically marginalized groups in our field, review the basic structure and purported purpose of the qualifying exam, and finally we propose several recommendations to address this potential structural barrier and encourage others to engage in critical reflection of their curricular requirements to assess whether they promote inclusive excellence.

  PublisherOA PDFDOI

• Sex Work Stigma as a Partial Mediator Between Social Support and Depressive Symptoms Among Vulnerable Women Engaged in Sex Work in Southern Uganda

  AIDS and Behavior · 2025-10-27

  article
  PublisherDOI

• The relationship between milk fat globule size, structure of its membrane and the dairy cow diet

  Animal - science proceedings · 2025-06-01

  articleSenior author
  PublisherDOI

Recent grants

• Fostering Diversity in Biostatistics Workshop

  NIH · $455k · 2005–2021

Frequent coauthors

• Jason D. Christie

  University of Pennsylvania

  128 shared

• Lorraine B. Ware

  Vanderbilt University Medical Center

  100 shared

• Jonathan B. Orens

  Johns Hopkins University

  93 shared

• Vibha N. Lama

  Emory University

  90 shared

• Ann Weinacker

  Stanford Medicine

  90 shared

• Steven M. Kawut

  University of Pennsylvania

  90 shared

• Ejigayehu Demissie

  National Heart Lung and Blood Institute

  81 shared

• João A. de Andrade

  Vanderbilt University Medical Center

  65 shared