Faculty Bibliography

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.

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.

We estimated the extent to which different pain management practices, considered together as well as individually, mediated the relationship between chronic pain or physical disability and new-onset opioid use disorder (OUD) in a large cohort of adult Medicaid patients. Considering the plausibility of the assumptions required to identify different mediational estimands, we estimated natural indirect effects when considering mediation through the group of mediators together and estimated interventional indirect effects when considering mediation through each pain management practice individually. We estimated each effect using a nonparametric one-step estimator. The pain management variables we examined mediated all of the total effect of chronic pain on OUD risk and nearly half of the total effect of physical disability on OUD risk. High-dose, long-duration opioid prescribing and co-prescription of opioids with benzodiazepines, gabapentinoids, and muscle relaxants each contributed substantially to the increased risk of OUD due to chronic pain (contributing to 10-37% of the overall effect) and more moderately to the increased risk of OUD due to physical disability (contributing to 3-19% of the overall effect). Antidepressant or anti-inflammatory prescribing and physical therapy generally did not contribute to increased OUD risk, and, in some cases, even contributed to small reductions in risk.

BACKGROUND:Buprenorphine and methadone are US Food and Drug Administration-approved medications for opioid use disorder (MOUD). Although utilization of MOUD was increasing pre-COVID-19, it is not well understood how this trend shifted during and "after" the COVID-19 pandemic in Rhode Island. This analysis will consider the differential utilization of MOUD over time and by key demographic factors. METHODS:We utilized two of Rhode Island's statewide databases to examine aggregate counts of dispensed buprenorphine and methadone from January 1, 2017, to December 31, 2023. Data were stratified by age group, sex assigned at birth, and race/ethnicity (where available). Counts were stratified into pre-COVID-19 (Q1 2017-Q1 2020), COVID-19 (Q2 2020-Q4 2022), and endemic COVID-19 (2023) eras. Averages and annualized percent change for each period were calculated to understand how utilization changed over time. RESULTS:Before COVID-19, buprenorphine and methadone utilization were increasing annually. During COVID-19, utilization declined annually by 0.40% and 0.43%, respectively. In the endemic COVID-19 time period, buprenorphine and methadone utilization declined more rapidly at 2.59% and 1.77%, respectively. Declines were more dramatic for adults aged 18-34. CONCLUSIONS:We observed a decline in MOUD utilization during and after COVID-19 in Rhode Island, primarily driven by substantial decreases in MOUD use among the youngest group of adult residents. Interventions specifically tailored to youth, such as school-based or primary healthcare-based programs, may be particularly effective in engaging with youth in substance use disorder treatment.

BACKGROUND AND AIM/OBJECTIVE:Sedative, hypnotic or anxiolytic use disorders (SHA-UD) are defined by significant impairment or distress caused by recurrent sedative, hypnotic or anxiolytic use. This study aimed to measure trends in the prevalence of SHA-UD diagnoses in adolescent and young adult US Medicaid enrollees from 2001 to 2019. DESIGN/METHODS:Annual, cross-sectional study, 2001-2019. SETTING/METHODS:Medicaid Analytic eXtracts (MAX) and Transformed Medicaid Analytic Files (TAF) from 42 US states with complete data. PARTICIPANTS/CASES/METHODS:Adolescents (13-17 years) and young adults (18-29 years) with ≥10 months Medicaid enrollment in the calendar year; analytic sample contained 5.7 (2001) to 13.2 (2019) million persons per year. MEASUREMENTS/METHODS:Annual prevalence of SHA-UD in adolescent and young adult Medicaid enrollees [defined as an inpatient or outpatient ICD code (304.1x, 305.4x, F13.1x, F13.2x) in the calendar year] was stratified by sex, race/ethnicity, receipt of a benzodiazepine, z-hypnotic or barbiturate prescription, and selected mental health diagnoses. Absolute and relative percent-changes from 2001 vs. 2019 were summarized. Secondary analyses were restricted to states with more consistent data capture. FINDINGS/RESULTS:The prevalence of SHA-UD diagnoses statistically significantly increased for adolescents (0.01% to 0.04%) and young adults (0.05% to 0.24%) from 2001 to 2019. Increasing trends were observed in sex and race/ethnicity subgroups, with greatest relative increases among Non-Hispanic Black (624%) and Hispanic (529%) young adults. The trend increased among those with and without a benzodiazepine, z-hypnotic or barbiturate prescription; i.e. young adults with (2001 = 0.39% to 2019 = 1.77%) and without (2001 = 0.03% to 2019 = 0.18%) a prescription. Most adolescents (76%) and young adults (91%) with a SHA-UD diagnosis in 2019 had a comorbid substance use disorder. CONCLUSIONS:Sedative, hypnotic or anxiolytic use disorders (SHA-UD) diagnoses increased 3- to 5-fold between 2001 and 2019 for adolescent and young adult US Medicaid enrollees, with prevalence remaining low in adolescents. The increase over two decades may be attributed to changes in the availability, use and misuse of sedative, hypnotic and anxiolytic medications and to increased detection, awareness and diagnosing of SHA-UD.

BACKGROUND AND AIMS/OBJECTIVE:Evidence from high-income countries has linked duration and compliance with treatment for substance use disorders (SUDs) with reductions in substance use and improvements in mental health. Generalizing these findings to other regions like South America, where opioid and injection drug use is uncommon, is not straightforward. We examined if length of time in treatment and compliance with treatment reduced subsequent substance use and presence of psychiatric comorbidities. DESIGN/METHODS:Prospective cohort analysis (3 assessments, 18 months) using inverse probability weighting to account for confounding and loss to follow-up. SETTINGS/METHODS:Outpatient/inpatient programs in Región Metropolitana, Chile. PARTICIPANTS/METHODS:Individuals initiating publicly funded treatment (n = 399). MEASUREMENTS/METHODS:Exposures included length of time in (0-3, 4-7, 8 + months, currently in) and compliance with treatment (not completed, completed, currently in) measured in the intermediate assessment (12 months). Primary outcomes were past-month use of primary substance (most problematic) and current psychiatric comorbidities (major depressive episode, panic, anxiety or post-traumatic stress disorders) measured 6 months later (18 months). Secondary outcomes included past month use of alcohol, cannabis, cocaine powder and cocaine paste. FINDINGS/RESULTS:18.3% [95% confidence interval (CI) = 14.7%-22.6%] of individuals participated for 3 or fewer months in treatment and 50.1% (95% CI = 45.2%-55.1%) did not complete their treatment plan at 12 months. Participating for 8 + months in treatment was associated with lower risk of past month use of primary substance at 18 months [vs. 0-3 months, risk ratio (RR) = 0.62, 95% CI = 0.38-1.00] and completion of treatment (vs. not completed, RR = 0.49, 95% CI = 0.30-0.80). Neither participating 8 + months (vs. 0-3 months, RR = 0.83, 95% CI = 0.57-1.22) nor treatment completion (vs. not completed, RR = 1.02, 95% CI = 0.72-1.46) were associated with lower risk of psychiatric comorbidity at 18 months. CONCLUSIONS:Longer time in treatment and compliance with treatment for substance use disorders in Chile appears to be associated with lower risk of substance use but not current comorbid psychiatric conditions 18 months after treatment initiation.

OBJECTIVES/OBJECTIVE:To evaluate the predictive performance of different data sources to forecast fatal overdose in Rhode Island neighborhoods, with the goal of providing a template for other jurisdictions interested in predictive analytics to direct overdose prevention resources. METHODS:We evaluated seven combinations of data from six administrative data sources (American Community Survey (ACS) five-year estimates, built environment, emergency medical services non-fatal overdose response, prescription drug monitoring program, carceral release, and historical fatal overdose data). Fatal overdoses in Rhode Island census block groups (CBGs) were predicted using two machine learning approaches: linear regressions and random forests embedded in a nested cross-validation design. We evaluated performance using mean squared error and the percentage of statewide overdoses captured by CBGs forecast to be in top percentiles from 2019 to 2021. RESULTS:Linear models trained on ACS data combined with one other data source performed well, and comparably to models trained on all available data. Those including emergency medical service, prescription drug monitoring program, or carceral release data with ACS data achieved a priori goals for percentage of statewide overdoses captured by CBGs prioritized by models on average. CONCLUSIONS:Prioritizing neighborhoods for overdose prevention with forecasting is feasible using a simple-to-implement model trained on publicly available ACS data combined with only one other administrative data source in Rhode Island, offering a starting point for other jurisdictions.

OBJECTIVES/OBJECTIVE:The escalating overdose crisis in the United States points to the urgent need for new and novel data tools. Overdose data tools are growing in popularity but still face timely delays in surveillance data availability, lack of completeness, and wide variability in quality by region. As such, we need innovative tools to identify and prioritize emerging and high-need areas. Forecasting offers one such solution. Machine learning methods leverage numerous datasets that could be used to predict future vulnerability to overdose at the regional, town, and even neighborhood levels. This study aimed to understand the multi-level factors affecting the early stages of implementation for an overdose forecasting dashboard. This dashboard was developed with and for statewide harm reduction providers to increase data-driven response and resource distribution at the neighborhood level. METHODS:As part of PROVIDENT (Preventing OVerdose using Information and Data from the EnvironmeNT), a randomized, statewide community trial, we conducted an implementation study where we facilitated three focus groups with harm reduction organizations enrolled in the larger trial. Focus group participants held titles such as peer outreach workers, case managers, and program coordinators/managers. We employed the Exploration, Preparation, Implementation, Sustainment (EPIS) Framework to guide our analysis. This framework offers a multi-level, four-phase analysis unique to implementation within a human services environment to assess the exploration and preparation phases that influenced the early launch of the intervention. RESULTS:Multiple themes centering on organizational culture and resources emerged, including limited staff capacity for new interventions and repeated exposure to stress and trauma, which could limit intervention uptake. Community-level themes included the burden of data collection for program funding and statewide efforts to build stronger networks for data collection and dashboarding and data-driven resource allocation. DISCUSSION/CONCLUSIONS:Using an implementation framework within the larger study allowed us to identify multi-level and contextual factors affecting the early implementation of a forecasting dashboard within the PROVIDENT community trial. Additional investments to build organizational and community capacity may be required to create the optimal implementation setting and integration of forecasting tools.