Faculty Bibliography

BACKGROUND:Drug overdose persists as a leading cause of death in the United States, but resources to address it remain limited. As a result, health authorities must consider where to allocate scarce resources within their jurisdictions. Machine learning offers a strategy to identify areas with increased future overdose risk to proactively allocate overdose prevention resources. This modeling study is embedded in a randomized trial to measure the effect of proactive resource allocation on statewide overdose rates in Rhode Island (RI). METHODS:We used statewide data from RI from 2016-2020 to develop an ensemble machine learning model predicting neighborhood-level fatal overdose risk. Our ensemble model integrated gradient boosting machine and Super Learner base models in a moving window framework to make predictions in 6-month intervals. Our performance target, developed a priori with the RI Department of Health, was to identify the 20% of RI neighborhoods containing at least 40% of statewide overdose deaths, including at least one neighborhood per municipality. The model was validated after trial launch. RESULTS:Our model selected priority neighborhoods capturing 40.2% of statewide overdose deaths during the test periods and 44.1% of statewide overdose deaths during validation periods. Our ensemble outperformed the base models during the test periods and performed comparably to the best-performing base model during the validation periods. CONCLUSIONS:We demonstrated the capacity for machine learning models to predict neighborhood-level fatal overdose risk to a degree of accuracy suitable for practitioners. Jurisdictions may consider predictive modeling as a tool to guide allocation of scarce resources.

Prior applications of machine learning to population health have relied on conventional model assessment criteria, limiting the utility of models as decision supports for public health practitioners. To facilitate practitioner use of machine learning as decision support for area-level intervention, this study developed and applied four practice-based predictive model evaluation criteria (implementation capacity, preventive potential, health equity, and jurisdictional practicalities). We used a case study of overdose prevention in Rhode Island to illustrate how these criteria could inform public health practice and health equity promotion. We used Rhode Island overdose mortality records from January 2016 to June 2020 (N=1,408) and neighborhood-level Census data. We learned two disparate machine learning models, Gaussian process and random forest, to illustrate the comparative utility of our criteria to guide interventions. Our models predicted 7.5-36.4% of overdose deaths during the test period, illustrating the preventive potential of overdose interventions assuming 5-20% statewide implementation capacities for neighborhood-level resource deployment. We described the health equity implications of predictive modeling to guide interventions along urbanicity, racial/ethnic composition, and poverty. In sum, our study discussed considerations to complement predictive model evaluation criteria and inform the prevention and mitigation of spatially dynamic public health problems across the breadth of practice.

Substance use disorders (SUD) are associated with increased risk of worse COVID-19 outcomes. Likewise, racial/ethnic minority patients experience greater risk of severe COVID-19 disease compared to white patients. Providers should understand the role of race and ethnicity as an effect modifier on COVID-19 severity among individuals with SUD. This retrospective cohort study assessed patient race/ethnicity as an effect modifier of the risk of severe COVID-19 disease among patients with histories of SUD and overdose. We used merged electronic health record data from 116,471 adult patients with a COVID-19 encounter between March 2020 and February 2021 across five healthcare systems in New York City. Exposures were patient histories of SUD and overdose. Outcomes were risk of COVID-19 hospitalization and subsequent COVID-19-related ventilation, acute kidney failure, sepsis, and mortality. Risk factors included patient age, sex, and race/ethnicity, as well as medical comorbidities associated with COVID-19 severity. We tested for interaction between SUD and patient race/ethnicity on COVID-19 outcomes. Findings showed that Non-Hispanic Black, Hispanic/Latino, and Asian/Pacific Islander patients experienced a higher prevalence of all adverse COVID-19 outcomes compared to non-Hispanic white patients. Past-year alcohol (OR 1.24 [1.01-1.53]) and opioid use disorders (OR 1.91 [1.46-2.49]), as well as overdose history (OR 4.45 [3.62-5.46]), were predictive of COVID-19 mortality, as well as other adverse COVID-19 outcomes. Among patients with SUD, significant differences in outcome risk were detected between patients of different race/ethnicity groups. Findings indicate that providers should consider multiple dimensions of vulnerability to adequately manage COVID-19 disease among populations with SUDs.

BACKGROUND AND AIMS/OBJECTIVE:Individuals with opioid use disorder (OUD) suffer disproportionately from COVID-19. To inform clinical management of OUD patients, research is needed to identify characteristics associated with COVID-19 progression and death among this population. We aimed to investigate the role of OUD and specific comorbidities on COVID-19 progression among hospitalized OUD patients. DESIGN/METHODS:Retrospective cohort study of merged electronic health records (EHR) from five large private health systems SETTING: New York City, New York, USA, 2011-2021 PARTICIPANTS: Adults with a COVID-19 encounter and OUD or opioid overdose diagnosis between March 2020-February 2021 MEASUREMENTS: Primary exposure included diagnosis of OUD/opioid overdose. Risk factors included age, sex, race/ethnicity and common medical, substance use and psychiatric co-morbidities known to be associated with COVID-19 severity. Outcomes included COVID-19 hospitalization and subsequent intubation, acute kidney failure, severe sepsis and death. FINDINGS/RESULTS:Of 110,917 COVID-19+ adults, 1.17% were ever diagnosed with OUD/opioid overdose. OUD patients had higher risk of COVID-19 hospitalization (adjusted risk ratio [aRR]: 1.40 [95% confidence interval (CI) 1.33, 1.47]), intubation (adjusted aood ratio [aOR]: 2.05 [95% CI 1.74, 2.42]), kidney failure (aRR: 1.51 [95% CI 1.34, 1.70)), sepsis (aRR: 2.30 [95% CI 1.88, 2.81]), and death (aRR: 2.10 [95% CI 1.84, 2.40)]. Among hospitalized OUD patients, risks for COVID-19 outcomes included being male; older; of a race/ethnicity other than white, Black or Hispanic; and having co-morbid chronic kidney disease, diabetes, obesity or cancer. Protective factors included having asthma, hepatitis-C, and chronic pain. CONCLUSIONS:Opioid use disorder patients appears to have a substantial risk for COVID-19-associated morbidity and mortality, with particular comorbidities and treatments potentially moderating this risk.

Pre-existing mental disorders are linked to COVID-19-related outcomes. However, the findings are inconsistent and a thorough analysis of a broader spectrum of outcomes such as COVID-19 infection severity, morbidity, and mortality is required. We investigated whether the presence of psychiatric diagnoses and/or the use of antidepressants influenced the severity of the outcome of COVID-19. This retrospective cohort study evaluated electronic health records from the INSIGHT Clinical Research Network in 116,498 individuals who were diagnosed with COVID-19 between March 1, 2020, and February 23, 2021. We examined hospitalization, intubation/mechanical ventilation, acute kidney failure, severe sepsis, and death as COVID-19-related outcomes. After using propensity score matching to control for demographics and medical comorbidities, we used contingency tables to assess whether patients with (1) a history of psychiatric disorders were at higher risk of more severe COVID-19-related outcomes and (2) if use of antidepressants decreased the risk of more severe COVID-19 infection. Pre-existing psychiatric disorders were associated with an increased risk for hospitalization, and subsequent outcomes such as acute kidney failure and severe sepsis, including an increased risk of death in patients with schizophrenia spectrum disorders or bipolar disorders. The use of antidepressants was associated with significantly reduced risk of sepsis (p = 0.033), death (p = 0.026). Psychiatric disorder diagnosis prior to a COVID-19-related healthcare encounter increased the risk of more severe COVID-19-related outcomes as well as subsequent health complications. However, there are indications that the use of antidepressants might decrease this risk. This may have significant implications for the treatment and prognosis of patients with COVID-19.

The association between recent release from incarceration and dramatically increased risk of fatal overdose is well-established at the individual level. Fatal overdose and. arrest/release are spatially clustered, suggesting that this association may persist at the neighborhood level. We analyzed multicomponent data from Rhode Island, 2016-2020, and observed a modest association at the census tract level between rates of release per 1000 population and fatal overdose per 100,000 person-years, adjusting for spatial autocorrelation in both the exposure and outcome. Our results suggest that for each additional person released to a given census tract per 1000 population, there is a corresponding increase in the rate of fatal overdose by 2 per 100,000 person years. This association is more pronounced in suburban tracts, where each additional release awaiting trial is associated with an increase in the rate of fatal overdose of 4 per 100,000 person-years and 6 per 100,000 person-years for each additional release following sentence expiration. This association is not modified by the presence or absence of a licensed medication for opioid use disorder (MOUD) treatment provider in the same or surrounding tracts. Our results suggest that neighborhood-level release rates are moderately informative as to tract-level rates of fatal overdose and underscore the importance of expanding pre-release MOUD access in correctional settings. Future research should explore risk and resource environments particularly in suburban and rural areas and their impacts on overdose risk among individuals returning to the community.

NYC RxStat, the United States' first public health and public safety partnership aiming to reduce overdose deaths, began in 2012 and established a national model for cross-sector partnerships. The partnership aimed to integrate data-driven policing with actionable public health interventions and surveillance to develop and implement cross-sector overdose responses. With federal support, jurisdictions nationally have implemented public health and public safety partnerships modeled on RxStat. To inform partnership replication efforts, we conducted a stakeholder evaluation of RxStat. We conducted in-depth, semi-structured interviews with 25 current and former RxStat stakeholders. Interviews probed stakeholder perceptions of RxStat's successes, challenges, and opportunities for growth. Interview data were iteratively coded and thematically analyzed. Stakeholders reported certainty about the need for cross-sector collaboration and described cross-disciplinary tensions, challenges to collaboration and implementation, and opportunities for partnership optimization and growth. Findings informed 12 strategies to improve RxStat and partnerships in its model, organized into three opportunity areas: (1) ensure stakeholder and agency accountability; (2) build secure and mutually beneficial data systems; and (3) structure partnerships to facilitate equitable collaboration. Cross-sector partnerships offer a promising strategy to integrate the public health and safety sectors, but disciplinary tensions in approach may hamper implementation. Findings can inform efforts to implement and scale cross-sector partnerships.

BACKGROUND:Mandates for prescriber use of prescription drug monitoring programs (PDMPs), databases tracking controlled substance prescriptions, are associated with reduced opioid analgesic (OA) prescribing but may contribute to care discontinuity and chronic opioid therapy (COT) cycling, or multiple initiations and terminations. OBJECTIVE:To estimate risks of COT cycling in New York City (NYC) due to the New York State (NYS) PDMP mandate, compared to risks in neighboring New Jersey (NJ) counties. DESIGN/METHODS:We estimated cycling risk using Prentice, Williams, and Peterson gap-time models adjusted for age, sex, OA dose, payment type, and county population density, using a life-table difference-in-differences design. Failure time was duration between cycles. In a subgroup analysis, we estimated risk among patients receiving high-dose prescriptions. Sensitivity analyses tested robustness to cycle volume considering only first cycles using Cox proportional hazard models. PARTICIPANTS/METHODS:The cohort included 7604 patients dispensed 12,695 prescriptions. INTERVENTIONS/METHODS:The exposure was the August 2013 enactment of the NYS PDMP prescriber use mandate. MAIN MEASURES/METHODS:We used monthly, patient-level data on OA prescriptions dispensed in NYC and NJ between August 2011 and July 2015. We defined COT as three sequential months of prescriptions, permitting 1-month gaps. We defined recurrence as re-initiation of COT after at least 2 months without prescriptions. The exposure was enactment of the PDMP mandate in NYC; NJ was unexposed. KEY RESULTS/RESULTS:Enactment of the NYS PDMP mandate was associated with an adjusted hazard ratio (HR) for cycling of 1.01 (95% CI, 0.94-1.08) in NYC. For high-dose prescriptions, the risk was 1.16 (95% CI, 1.01-1.34). Sensitivity analyses estimated an overall risk of 1.01 (95% CI, 0.94-1.11) and high-dose risk of 1.09 (95% CI, 0.91-1.31). CONCLUSIONS:The PDMP mandate had no overall effect on COT cycling in NYC but increased cycling risk among patients receiving high-dose opioid prescriptions by 16%, highlighting care discontinuity.

OBJECTIVES:To compare the characteristics of individual overdose decedents in Rhode Island, 2016-2020 to the neighborhoods where fatal overdoses occurred over the same time period. METHODS:We conducted a retrospective analysis of fatal overdoses occurring between January 1, 2016 and June 30, 2020. Using individual- and neighborhood-level data, we conducted descriptive analyses to explore the characteristics of individuals and neighborhoods most affected by overdose. RESULTS:Most overdose decedents during the study period were non-Hispanic White. Across increasingly more White and non-Hispanic neighborhoods, rates of fatal overdose per 100,000 person-years decreased. An opposite pattern was observed across quintiles of average neighborhood poverty. CONCLUSIONS:Rates of fatal overdose were higher in less White, more Hispanic, and poorer neighborhoods, suggesting modest divergence between the characteristics of individuals and the neighborhoods most severely affected. These impacts may not be uniform across space and may accrue differentially to more disadvantaged and racially/ethnically diverse neighborhoods.

Opioid analgesics and benzodiazepines remain substantial contributors to unintentional drug overdose deaths in the United States. To promote judicious prescribing and improve care for patients with substance use disorders, the New York City Department of Health and Mental Hygiene piloted the Prescriber Notification Program, an educational initiative to deliver targeted public health messaging to providers who had prescribed opioid analgesics and/or benzodiazepines to patients who died from overdose in New York City. This article reports on provider responses to receipt of patient death notifications and program feasibility. Findings demonstrate that a majority of prescribers were not aware of patient deaths prior to receiving notification letters. Public health authorities considering prescriber notification systems should address barriers to implementation and sustainability-in particular, consistent and routine access to and linkage of overdose mortality and prescription monitoring data-as part of planning such programs.