Faculty Bibliography

As predictive analytics become more widely integrated into local public health responses to the United States overdose epidemic, community-based substance use service providers have begun to adopt machine learning-based predictive tools to guide the allocation and delivery of overdose prevention services. While these tools hold promise for anticipating community overdose risk and enhancing the efficiency of overdose prevention resource distribution, outreach, and education efforts, their use in community settings raises substantial ethical and practical challenges. In this Viewpoint, we examine the application of predictive analytics to community-based overdose prevention through a public health ethics lens, drawing on principles of distributive justice, transparency, community participation, and implementation readiness. We outline five key ethical considerations for developers (i.e., institutional responsibility, oversimplification of complex social realities, data and algorithmic bias, community displacement in decision making, and equity trade-offs) and corresponding practical challenges for service providers. We offer five recommendations for developers, public health authorities, and frontline organizations to overcome challenges and ensure responsible, equity-driven implementation. As data-driven approaches to overdose prevention proliferate, ethical and participatory frameworks will be essential to ensure predictive tools strengthen, rather than undermine, community trust and health equity.

The overdose epidemic has reshaped law enforcement's relationship with public health, as police increasingly adopt overdose response measures, including naloxone training and use. This study analyzed 15 interviews with New York Police Department (NYPD) officers to examine their experiences administering naloxone in their duties. Naloxone was seen as facilitating a shift from traditional law enforcement to a broader public safety role, and officers noted its potential to improve public perceptions of police. However, tensions emerged as officers navigated dual roles in enforcement and health, with concerns that overdose reversal might enable continued drug use or crime. Additionally, officers expressed frustrations about naloxone's limitations, particularly its inability to address systemic barriers to addiction recovery. These findings underscore the need for clear policies, comprehensive training, and stronger interagency partnerships to enhance the integration of public health strategies within policing and better support officers in responding to the overdose crisis.

INTRODUCTION/BACKGROUND:Police are frequently dispatched to a wide range of 911 calls, including mental and behavioural health crises, despite lacking the training, resources and time to respond effectively. In particular, people with serious mental illness are at elevated risk of experiencing excessive use of force, arrest and continued criminal legal involvement following police contact. Following the murder of George Floyd and other highly publicised police killings, Community Safety Response (CSR) programmes, staffed by unarmed peers, mental health professionals and other trained responders, have proliferated to provide non-police responses to mental and behavioural health and other quality-of-life concerns. CSR programmes have expanded rapidly, yet the evidence base remains fragmented and largely outside the peer-reviewed literature. METHODS AND ANALYSIS/METHODS:This scoping review will synthesise peer-reviewed and grey literature from 2020 to present on CSR programmes operating in North America. Guided by Joanna Briggs Institute methodology and reported according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses Extension for Scoping Reviews (PRISMA-ScR) standards, we will search multiple databases (Medline, PsycINFO, Embase, SocIndex, Web of Science, Policy Commons) and employ complementary grey literature search strategies, including targeted website searches, reference tracking and review of internal and external reports and evaluations. Inclusion criteria require that programmes provide non-police first response to calls traditionally served by law enforcement and include information on programme operations or outcomes. Two reviewers will independently screen and extract data on process metrics including operational characteristics, dispatch, funding, services provided and outcomes such as populations served, diversion from police, service linkage and use of force. ETHICS AND DISSEMINATION/BACKGROUND:No ethical review for this study is required as it will not include human subjects or any identifiable information. Findings will provide the first national synthesis of CSR programme models, operations and outcomes. Results will inform policy-makers, practitioners, researchers and community members. Findings will be disseminated through peer-reviewed publications and public-facing products to support implementation, scale-up and sustainability of CSR programmes.

CONTEXT/BACKGROUND:Predictive modeling can identify neighborhoods at elevated risk of future overdose death and may assist community organizations' decisions about harm reduction resource allocation. In Rhode Island, PROVIDENT is a research initiative and randomized community intervention trial that developed and validated a machine learning model that predicts future overdose at a census block group (CBG) level. The PROVIDENT model prioritizes the top 20th percentile of CBGs at highest risk of future overdose death over the subsequent 6-month period. In CBGs assigned to the trial intervention arm, these predictions are then displayed for partnering community organizations via an interactive mapping dashboard. OBJECTIVE:To evaluate whether CBGs prioritized by the PROVIDENT model were associated with increased user engagement via an online dashboard for fatal overdose forecasting and resource planning. DESIGN/METHODS:We estimated prevalence ratios using modified Poisson regression models, adjusted for CBG-level characteristics that may confound the relationship between model predictions and dashboard engagement. SETTING/METHODS:We used CBG-level data in Rhode Island (N = 809) from November 2021 to July 2024. INTERVENTION/METHODS:Our exposure of interest was whether each CBG was prioritized by the PROVIDENT model and shown as prioritized on the interactive mapping dashboard. MAIN OUTCOME MEASURE/METHODS:Our primary outcome was whether a dashboard user from any partnering community organization engaged (eg, clicked, interacted with dashboard elements, or completed assessment or planning surveys) with each CBG on the interactive mapping dashboard. RESULTS:After adjusting for previous model predictions and dashboard engagement, nonfatal overdose counts, and distribution of race and ethnicity, poverty, unemployment, and rent burden, dashboard users were 1.0 to 2.4 times as likely to engage with CBGs prioritized by the PROVIDENT model that were shown as prioritized on the dashboard as compared to CBGs that were prioritized by the PROVIDENT model that were blinded on the dashboard. CONCLUSIONS:Interactive mapping tools with predictive modeling may be useful to support community-based harm reduction organizations in the allocation of resources to neighborhoods predicted to be at high risk of future overdose death.

BACKGROUND:Neighborhood-level overdose risk may vary over time. In Rhode Island, we developed and validated a machine learning model to identify the 20 percent of census block groups (CBGs) at the highest predicted risk of future overdose death. We updated this model periodically between November 2021 and August 2024 to generate six sets of predictions. This study aims to characterize the trajectory of each CBG's predicted overdose risk over time across these six periods. METHODS:In each prediction period, CBGs were designated as "high risk" or not designated as "high risk" based on our model's 20 percent predicted overdose risk threshold. We implemented sequence analysis to describe unique trajectories in each CBG's risk designation over each prediction period. We then calculated optimal matching distances to estimate dissimilarity between each pair of trajectories and applied agglomerative hierarchical clustering to group similar trajectories. RESULTS:The 809 CBGs included in this study followed 60 unique trajectories in predicted overdose risk designation over the six prediction periods. Clustering of trajectories favored a solution with five trajectory groups. Most CBGs (73.4 %) were rarely or never designated as "high risk", 7.9 % of CBGs were always designated as "high risk", and the remaining 18.7 % were designated as "high risk" in multiple prediction periods, represented by trajectory groups with different patterning over time. CONCLUSIONS:Given the substantial variability in which CBGs were at highest overdose risk over time, dynamic machine learning predictions may inform harm reduction resource allocation by identifying neighborhoods with emerging needs.

Following federal regulatory changes during the COVID-19 pandemic, Rhode Island expanded methadone access for opioid treatment programs (OTPs) in March 2020. The policy, which permitted take-home dosing for patients, contrasted with longstanding restrictions on methadone. This study used patient-level OTP admission and discharge records to compare six-month retention before and after the policy change. We conducted a retrospective cohort study of 1,248 patients newly admitted to OTPs between March 18 and June 30 of 2019 (pre-policy) and 2020 (post-policy). We used logistic regression to estimate associations with retention before and after the policy and used a machine learning approach, the Heterogeneous Treatment Effect (HTE)-Scan, to explore heterogeneity in retention across subgroups. Overall, we found no change in retention following the policy, with an adjusted OR of 1.08 (95% CI: 0.80-1.45) and adjusted RR of 1.03 (0.90-1.18). Using HTE-Scan, we identified two subgroups with significantly increased retention above the overall cohort: (1) patients with below high school education and past-month arrest and (2) male, non-Hispanic white or Hispanic/Latino patients reporting heroin or fentanyl use with past-month arrest. We identified no subgroups with significantly decreased retention. Collectively, findings suggest that expanded methadone access may benefit vulnerable populations without harming overall retention.

More than one million people have died from drug overdose in the United States in the past 20 years. The overdose crisis started in the late 1990s with the proliferation of overdoses involving prescription opioids, transitioned to heroin-involved overdoses in 2010, and is currently driven by illegally manufactured synthetic opioids such as fentanyl. In response to this crisis, New York City implemented two publicly recognized overdose prevention centers (OPCs) in the nation in November 2021. Rhode Island became the first US state to authorize OPCs through state legislation and will open a site in Fall 2024. We are conducting a rigorous, multi-site, multi-component evaluation of OPCs in New York City and Rhode Island. At the individual level, we assess whether a cohort of 500 persons utilizing OPCs experience lower rates of overdose, other health problems (e.g., hepatitis C, skin infections), and emergency department use, and a higher rate of substance use treatment initiation, compared to a cohort of 500 persons who use drugs but do not utilize OPCs. At the community level, we examine whether neighborhoods surrounding the OPCs experience a greater change in overdose, measures of drug-related public disorder, and acute economic conditions following the opening of OPCs, compared to neighborhoods with no OPCs. Third, we delve into the role that the operational context, including neighborhood location, program models, and operating procedures, plays in shaping the effectiveness of OPCs using qualitative and ethnographic approaches. Fourth, we estimate the costs and cost savings associated with starting up and operating OPCs. In this paper, we: (1) present the study design and harm reduction framework which is used to evaluate the impact of OPCs in New York City and Rhode Island; (2) share the types of assessment instruments and data sources used to measure changes at the individual and community level; and (3) discuss the strengths and limitations associated with the planned approach to evaluate the health and community effects of OPCs.

As the United States grapples with an ongoing overdose crisis, states and jurisdictions are adopting novel approaches to reduce overdose mortality. In one novel approach, public health researchers and leaders in Rhode Island leveraged the state's robust surveillance data and collaborations between government, academic, and community-based organizations (CBOs) to launch the PROVIDENT (PReventing OVerdose using Information and Data from the EnvironmeNT) project, a population-based randomized controlled research trial (NCT05096429) in December 2019. The PROVIDENT trial utilizes machine learning (ML) methods to identify neighborhoods at risk of future overdose deaths at the census-block-group level to inform community-level overdose prevention resource distribution. To disseminate the ML model predictions, our research team developed an interactive, online mapping dashboard in close collaboration with three statewide CBOs. We measured whether these organizations utilized the PROVIDENT dashboard to allocate harm-reduction services based on ML model predictions and collected information about their data-driven decision-making processes. This case study describes how we assembled and piloted this overdose forecasting dashboard for use by CBOs between November 2021 and August 2024. By measuring dashboard logins, completed surveys, and engagement with ongoing training, we illustrate how organizations utilized ML and surveillance data to inform their outreach efforts and generate valuable insights at a neighborhood level.

To meet the needs of diverse communities, public health authorities are increasingly reliant on hyperlocal interventions targeting specific health issues and distinct populations. To facilitate epidemiological evaluation of hyperlocal interventions on community-level outcomes, we developed a framework of six practice-based considerations for researchers: spatial zone of impact, temporal resolution of impact, outcome of interest, definition of a plausible comparison group, micro vs. macro impacts, and practitioner engagement. We applied this framework to a case study of an impact evaluation of the New York City (NYC) overdose prevention centers (OPCs) on neighborhood-level drug-related arrests. We used drug arrest data from NYC from January 1, 2014, to September 30, 2023 and US Census data to conduct synthetic control modeling, comparing pre- and post-OPC arrests in the neighborhoods surrounding the two NYC OPCs (East Harlem and Washington Heights). We conducted sensitivity analyses to validate our results and compare our findings with those from a prior published study. Our findings indicate no significant change in drug-related arrests following the OPC openings. The mean absolute differences in daily drug-related arrests between the OPCs and their synthetic controls were 0.63 (p = 0.19) in East Harlem and 0.14 (p = 0.22) in Washington Heights. Sensitivity analyses corroborated our main results. Overall, findings demonstrate how our framework can be used to guide future epidemiological evaluations of diverse, hyperlocal public health interventions.

BACKGROUND:Concerned significant others (CSOs) play a significant role in supporting individuals with substance use disorders. There is a lack of tailored support services for these CSOs, despite their substantial contributions to the well-being of their loved ones (LOs). The emergence of helplines as a potential avenue for CSO support is outlined, culminating in the focus on the Partnership to End Addiction's helpline service, an innovative public health intervention aimed at aiding CSOs concerned about an LO's substance use. OBJECTIVE:The article analyzes the demographics and use patterns of the Partnership to End Addiction's helpline service, highlighting the critical role of such services, and advocating for expanded, tailored support models. METHODS:This observational study draws data from 8 data platforms spanning April 2011 to December 2021, encompassing 24,096 client records. Surveys were completed by helpline specialists during synchronous telephone calls or self-reported by CSOs before helpline engagement. Collected information encompasses demographics, interaction language, substance of concern, CSO-LO relationship, and the LO's "use state," that is, their location on the continuum of substance use. RESULTS:CSOs primarily comprised women (13,980/18,373, 76.1%) seeking support for their children (1062/1542, 68.9%). LOs were mostly male (1090/1738, 62.7%), aged 18-25 years (2380/7208, 33%), with primary substance concerns being cannabis (5266/12,817, 40.9%), opioids (2445/12,817, 19%), and stimulants (1563/12,817, 12.1%). CSOs primarily sought aid for LOs struggling with substances who were not in treatment (1102/1753, 62.9%). The majority of CSOs were looking for support in English (14,738/17,920, 82.2%), while the rest (3182/17,920, 17.8%) preferred to communicate in Spanish. Spanish-speaking CSOs were significantly more likely to call about cannabis (n=963, 53.7% vs n=4026, 38.6%) and stimulants (n=304, 16.9% vs n=1185, 11.3%) than English-speaking CSOs (P<.001). On the other hand, English-speaking CSOs were more likely to be concerned about opioids than Spanish-speaking CSOs (n=2215, 21.3% vs n=94, 5.2%; P<.001). CONCLUSIONS:The study illuminates the helpline's pioneering role in aiding CSOs grappling with an LO's substance use. It highlights helplines as crucial resources for CSOs, revealing key demographic, substance-related, and use-state trends. The dominant presence of women among users aligns with other helpline patterns and reflects traditional caregiving roles. While parents form a significant percentage of those reaching out, support is also sought by siblings, friends, and other family members, emphasizing the need for assistance for other members of an LO's social network. Spanish-speaking individuals' significant outreach underscores the necessity for bilingual support services. Substance concerns revolve around cannabis, opioids, and stimulants, influenced by age and language preferences. The helpline serves as an essential intermediary for CSOs, filling a gap between acute crisis intervention services and formalized health care and treatment services. Overall, the study highlights this helpline's crucial role in aiding CSOs with tailored, accessible support services.