About

Debbie M. Cheng is a Professor of Biostatistics and Assistant Dean of Data Science at the Boston University School of Public Health. She is the founding Executive Director of the Center for Health Data Science and serves as the Director of Strategy and Partnerships for Boston University’s university-wide AI Developmental Accelerator (AIDA). Her research focuses on applied statistics and clinical trial design and analysis, with particular emphasis on the intersection of substance use and HIV. She leads interdisciplinary initiatives that integrate data science with public health and clinical research to advance evidence-based solutions.

Research topics

• Medicine
• Psychiatry
• Nursing
• Internal medicine
• Environmental health
• Family medicine

Selected publications

• 221. Development of a brief overdose prevention intervention for youth in pediatric and family medicine primary care

  Journal of Adolescent Health · 2026-02-13

  article
  PublisherDOI

• Assessing the impact of nicotinic partial agonists compared to NRT on opioid and cigarette use: A secondary analysis investigating treatment outcomes for co-occurring nicotine and opioid use

  Journal of Substance Use and Addiction Treatment · 2026-01-18

  articleOpen access
  PublisherOA PDFDOI

• Is There an Association Between Cigarette Smoking and Advanced Liver Fibrosis in Smokers with HIV, Heavy Drinking and High Prevalence of HCV?

  Journal of Clinical Medicine · 2025-02-11

  articleOpen access

  Background: Cigarette smoking has been associated with liver fibrosis in the setting of hepatitis C virus (HCV) infection but has not been studied among people with HIV (PWH) who consume alcohol. Methods: This is a cross-sectional study of PWH with heavy drinking and daily smoking in St. Petersburg, Russia. The primary independent variable was past 30-day cigarettes per day (cpd), and the secondary independent variable was pack-years at study entry. Advanced liver fibrosis was defined as FIB-4 > 3.25. Analyses were adjusted for gender, body mass index (BMI), past 30-day number of heavy drinking days, HCV and CD4 count. Results: Participants (n = 400) were two-thirds male (67.3%), young (median age 38 years), lean (median BMI 22), HCV antibody positive (84.5%) and not severely immune suppressed (median CD4 count 351). The median number of past-month cpd was 20 (IQR: 15–25), and the median pack-years was 24 (IQR: 17–31.8). The prevalence of advanced liver fibrosis was 11.3% (45/400). In the adjusted logistic regression analyses, we did not observe a significant association between cpd [middle (10.1–20 cigarettes) vs. lowest (5–10 cigarettes) category (adjusted odds ratio [aOR] (95% confidence interval [CI]): 1.06 (0.40–2.83), highest (>20.0 cigarettes) vs. lowest category aOR (95% CI): 0.65 (0.21–1.99), global p-value = 0.62]. The secondary analysis with pack-years yielded similar results [middle (20.1–30 pack-years) vs. lowest category (≤20 pack-years) aOR (95% CI): 0.81 (0.33–1.99), highest category (>30 pack-years) vs. lowest category aOR (95% CI): 0.91 (0.38–2.19); global p-value = 0.58]. Conclusions: In this Russian cohort of PWH, we did not detect an association between recent cigarette use or mean pack-years and advanced liver fibrosis.

  PublisherDOI

• Gabapentin to achieve HIV viral load suppression in people with risky drinking in Mbarara, Uganda: study protocol for a randomized, double-blinded, placebo-controlled trial (GRAIL)

  Research Square · 2025-10-15

  preprintOpen access
  PublisherOA PDFDOI

• Oral vs Extended-Release Injectable Naltrexone for Hospitalized Patients With Alcohol Use Disorder

  JAMA Internal Medicine · 2025-04-21 · 13 citations

  letterOpen access

  Importance: Alcohol use disorder (AUD) is common in hospital patients. AUD medications are not typically initiated in that setting. The comparative effectiveness between initiation of oral naltrexone and extended-release injectable naltrexone in the hospital is not known. Objective: To compare the effectiveness of initiating oral naltrexone vs extended-release injectable naltrexone on reduction in alcohol use and health care utilization among medical inpatients with AUD. Design, Setting, and Participants: The Alcohol Disorder Hospital Treatment (ADOPT) study is a randomized clinical trial conducted at an urban teaching hospital in the US, with enrollment between June 2016 and March 2020. Inpatients were screened for eligibility, and those with AUD and recent heavy drinking (defined as 5 or more drinks for men and 4 or more drinks for women) were enrolled. Outcomes were assessed at 3-month follow-up; assessors were not blinded to treatment assignment. Data were analyzed from May 2021 to September 2023. Interventions: Participants received either daily oral naltrexone or monthly extended-release injectable naltrexone. All received medical management with a research nurse who specialized in addiction. Main Measures and Outcomes: The primary outcome was change in percentage of heavy drinking days (HDDs) over the past 30 days from baseline to 3-month follow-up, assessed by validated instrument. The secondary outcome was any acute health care utilization (emergency department or hospitalization) at 3-month follow-up over the past 90 days. Results: Of 248 participants, 199 (80.2%) were male, and the mean (SD) age was 49.4 (10.4) years. The baseline median (IQR) percentage of HDDs in the past 30 days was 80.0% (43.3-100). At 3-month follow-up, the mean percentage of HDDs in the past 30 days was reduced in both groups (oral naltrexone: baseline, 66.7% HDDs; 3-month follow-up, 27.4% HDDs; difference, -38.4 percentage points; 95% CI, -125.0 to 48.2; extended-release injectable naltrexone: baseline, 70.7% HDDs; 3-month follow-up, 23.8% HDDs; difference, -46.4 percentage points; 95% CI, -123.4 to 30.6; P = .14). At follow-up, 59 of 109 in the oral naltrexone arm (54.1%) and 66 of 108 in the extended-release injectable naltrexone arm (61.1%) reported acute health care utilization in the prior 3 months; the odds of this utilization were not significantly different between groups (adjusted odds ratio, 1.34; 95% CI, 0.77-2.33). Conclusions and Relevance: In this randomized clinical trial, when initiated at hospital discharge, oral and extended-release injectable naltrexone did not differ in effectiveness. Participants had substantial reductions in HDDs in both treatment groups; however, there was not a significant difference in the reduction of percentage of HDDs in the past 30 days or acute health care utilization between groups. Hospitalization represents an opportunity to start AUD pharmacotherapy; choice of oral naltrexone vs extended-release injectable naltrexone should be directed by factors such as patient preference and insurance. Trial Registration: ClinicalTrials.gov Identifier: NCT02478489.

  PublisherOA PDFDOI

• Communities that Heal Intervention, Data 2000 Waivered Providers and Buprenorphine Prescribing: A Randomized Clinical Trial

  Drug and Alcohol Dependence · 2025-02-01

  article
  PublisherDOI

• Sex differences in beneficial and pathogenic bacteria in People With HIV (PWH) with a history of heavy alcohol drinking

  Frontiers in Microbiology · 2025-11-21 · 2 citations

  articleOpen access

  Background HIV-1 infection and hazardous levels of alcohol consumption have been independently linked to gut dysbiosis affecting beneficial butyrate-producing bacteria. However, sex-based differences in the composition and function of gut microbiome of People With HIV (PWH) with a history of heavy alcohol drinking remain undetermined, which is the focus of this study. Methods Cross-sectional study examining structural and functional features of the gut microbiome in PWH between men and women with a history of hazardous alcohol drinking recruited at St. Petersburg, Russia. 16S rDNA sequencing information was used for metataxonomic, Phylogenetic Investigation of Communities by Reconstruction of Unobserved States (PICRUSt2) and Linear Discriminant Analysis Effect Size (LEfSe) analyses. Group-wise comparisons were done using Mann–Whitney U -test. Further, linear and logistic regression models were used to evaluate the association between sex and measures of gut microbial dysbiosis and Firmicutes/Bacteroidota (F/B) ratio, respectively. Data were adjusted for confounding covariates particularly, HIV-viral load, Anti-retroviral Therapy (ART) and alcohol usage. Results Metataxonomic analysis demonstrated that women depicted significantly higher microbial diversity (Operational Taxonomic Units, OTUs and Shannon Index), higher percent relative abundance (%RA) of Firmicutes , lower %RA of Bacteroidota and higher F/B ratio. Importantly, logistic regression revealed that women had twice the odds of having F/B ratio > 1. Notably, women demonstrated significantly higher %RA of butyrate-producing bacterial families, i.e., Lachnospiraceae, Oscillospiraceae, Rikenellaceae and Marinifilaceae and genera. Correspondingly, significantly greater expression of bacterial genes involved in butyrate synthesis in women was demonstrated by PICRUSt2 analysis. Additionally, women depicted lower %RA of pathobiont, Prevotellaceae particularly, Prevotella_9 genus. Conclusion Overall, we observed significant sex-based differences in the relative abundances of beneficial bacterial communities such as butyrate producers and potential pathogenic Prevotella community in the gut microbiome of PWH with a history of heavy alcohol consumption. The observed sex-based differences are clinically relevant and could inform therapeutic strategies with evidence-based probiotics.

  PublisherOA PDFDOI

• Response to the Editor: Re: registration and multiple outcome testing in the HEALing Communities Study by Dennis M. Gorman and Ben G. Fitzpatrick

  International Journal of Drug Policy · 2025-10-11

  article
  PublisherDOI

• Effects of the Communities That HEAL intervention on initiation, retention, and linkage to medications for opioid use disorder (MOUD): A cluster randomized wait-list controlled trial

  Drug and Alcohol Dependence · 2025-07-15 · 1 citations

  articleOpen access

  Medications for opioid use disorder (MOUD) can reduce opioid use and overdose deaths. This study examined whether the Communities That HEAL (CTH) intervention increased MOUD initiation, retention, and linkage. The HEALing Communities Study was a multi-site, 2-arm, parallel, community-level, cluster-randomized, unblinded, wait-list controlled trial conducted in 67 communities (n = 34 intervention, n = 33 control). Using Prescription Drug Monitoring Programs and Medicaid claims data, we compared mean community-level rates of MOUD outcomes during the 1-year comparison period (July 2021-June 2022) for: (a) MOUD receipt at least once; (b) continuous MOUD receipt for 180 days; and (c) MOUD linkage within 31 days following an opioid-related emergency department or hospital encounter. For intervention and control communities, adjusted rates of receiving MOUD at least once were 578 (95 % CI: 562, 594) and 596 (95 % CI: 572, 621) per 1000 Medicaid enrollees, respectively [adjusted Relative Rate (aRR)= 0.97 (95 % CI: 0.93, 1.01)]. Adjusted rates of receiving MOUD for 180 consecutive days (retention) were 614 (95 % CI: 595, 634) and 620 (95 % CI: 603, 638) per 1000 Medicaid enrollees receiving MOUD at least once for intervention and control communities, respectively [aRR= 0.99 (95 % CI: 0.95, 1.04)]. The adjusted rate of linkage was 280 (95 % CI: 254, 310) and 252 (95 % CI: 226, 281) per 1000 encounters for intervention and control communities, respectively [aRR= 1.11 (95 % CI: 0.96, 1.28). Compared to control communities, communities that received the CTH intervention did not demonstrate higher rates of MOUD use, retention, or linkage. Additional efforts are needed to improve uptake and sustained use of MOUD. TRIAL REGISTRATION: ClinicalTrials.gov Identifier: NCT04111939.

  PublisherDOI

• The Sense and Sensibility of Sensitivity Analyses

  New England Journal of Medicine · 2024-09-14 · 10 citations

  article1st authorCorresponding
  PublisherDOI

Recent grants

• Core L - Clinical & Behavioral Sciences Core

  NIH · $21.7M · 1998–2028

• NIH Grant U24AA020779

  NIH · $1.3M · 2022

• NIH Grant P01AG004390

  NIH · $17.9M · 2010

Frequent coauthors

• Jeffrey H. Samet

  Boston Medical Center

  608 shared

• Richard Saitz

  319 shared

• Carly Bridden

  Boston Medical Center

  115 shared

• Evgeny Krupitsky

  112 shared

• Alexander Y. Walley

  Boston University

  110 shared

• Tibor P. Palfai

  Boston University

  95 shared

• Daniel P. Alford

  Boston University

  89 shared

• Elena Blokhina

  First Pavlov State Medical University of St. Petersburg

  65 shared

Labs

• Admissions TeamPI