CommunityStat: A Public Health Intervention to Reduce Opioid Overdose Deaths in Burlington, Vermont, 2017–2020

Abstract

From 2017 to early 2020, the US city of Burlington, Vermont led a county-wide effort to reduce opioid overdose deaths by concentrating on the widespread, low-barrier distribution of medications for opioid use disorder. As a small city without a public health staff, the initiative was led out of the police department—with an understanding that it would not be enforcement-oriented—and centered on a local adaptation of CompStat, a management and accountability program developed by the New York City Police Department that has been cited as both yielding improvements in public safety and overemphasizing counterproductive police performance metrics if not carefully directed. The initiative was instrumental to the implementation of several novel interventions: low-threshold buprenorphine prescribing at the city’s syringe service program, induction into buprenorphine-based treatment at the local hospital emergency department, elimination of the regional waiting list for medications for opioid use disorder (MOUD), and the de-facto decriminalization of diverted buprenorphine by the chief of police and county prosecutor. An effort by local legislators resulted in a state law requiring all inmates with opioid use disorder be provided with MOUD as well. By the end of 2018, these interventions were collectively associated with a 50% (17 vs. 34) reduction in the county’s fatal overdose deaths, while deaths increased 20% in the remainder of Vermont. The reduction was sustained through the end of 2019. This article describes the effort undertaken by officials in Burlington to implement these interventions. It provides an example that other municipalities can use to take an evidence-based approach to reducing opioid deaths, provided stakeholders assent to sustained collaboration in the furtherance of a commitment to save lives. In doing so, it highlights that police-led public health interventions are the exception, and addressing the overdose crisis will require reform that shifts away from criminalization as a community’s default framework for substance use.

Introduction

In 2018, as the opioid epidemic continued to exert a vise-like grip on communities in the United States, opioid overdose deaths declined by 50% (17 vs. 34) in Chittenden County, Vermont (pop. 163,571), home to Burlington, the state’s largest city. In the remainder of this small state (pop. 624,000), fatal overdoses increased by 20% (Jickling, 2019b). This difference was statistically significant, the percent decrease was one of the steepest observed in any American city contending with the consequences of the opioid crisis that year, and it was maintained through the end of 2019 (Vermont Department of Health, 2020). It occurred during a period when the Burlington city government took several novel and evidence-informed steps to reduce overdose fatalities that were rarely implemented together at full scale in a small city. The city’s progress was reversed in 2020, most likely by the nationwide effects of the COVID-19 pandemic (Schwenk, 2020), which ushered in the worst fatal overdose death toll in the nation’s history (Kallingal & Fox, 2020; NCHS, 2021). Despite the challenges the city and the nation still face, this 2-year trend may be considered an important example of a U.S. municipality’s ability to successfully contend with the lethal consequences of opioid use disorder.¹

The means by which city officials were able to enact innovations and reforms could provide a guide for others to follow. This article describes the process they undertook, demonstrating that evidence-based approaches to reducing overdose mortality associate with positive results in this case, even in the absence of formal evaluation frameworks. It starts with the selection of the ultimate goal of mortality reduction, and explains the surrogate endpoints utilized to pursue this goal. It then examines the leadership, management and accountability mechanisms created to implement and sustain the necessary interventions, the standards of care these interventions sought to embody, and catalogs the range of successful—and less successful—results. It closes by noting that although the simultaneous implementation of several interventions precluded a careful analysis of their relative effectiveness at the population level, the acute nature of the overdose crisis and the evidence in support of the chosen interventions make them worthy of consideration for collective replication. Flexible approaches to evaluation may provide opportunities for research partnerships, especially ones concerning implementation science or a realist approach to program evaluation.

Targeting Fatal Overdoses Through Evidence-Based Surrogate Endpoints

The measures taken by the city and its partners in their response to the opioid crisis consisted of a constellation of interventions, each based in science and evidence. They strove for a single endpoint, the reduction of opioid overdose deaths. It was selected because while deaths were an imperfect measure subject to some inaccuracies (Seth et al., 2018), they were a consistently measured outcome that offered insight into the epidemic’s worst consequences. The others, such as nonfatal overdoses, various drug-related morbidities, and the social and economic effects of the crisis, could not be measured with enough accuracy to assess progress or target interventions. It was believed that while reducing these other outcomes was critical for a community, and fixing the social and economic determinants of substance use disorder (SUD) was required for truly ending the crisis (Dasgupta et al., 2018), success in reducing overdose deaths would necessarily require or yield progress in these other areas.

To achieve this goal, the city targeted what it believed to be the most critical surrogate endpoint for reducing overdose deaths: the immediate and sustained availability of medications used to treat opioid use disorder (MOUD), specifically the agonist medications buprenorphine and methadone, for any person with a dependence on opioids who desired it. The city’s research led it to believe that these medications were the most effective means to reduce both overdoses and all-cause mortality (Sordo et al., 2017; Wakeman et al., 2020), and functioned as the main component of treatment (Friedmann & Schwartz, 2012). Later studies support officials’ belief at the time that even illicitly-prescribed buprenorphine could reduce incidence of unintentional overdose (Carlson et al., 2020) and its misuse was not likely to contribute to opioid overdose fatalities (Paone et al., 2015; Wightman et al., 2021). City officials concluded that linking people at risk of overdose to effective care through a wide range of touchpoints was critical (Larochelle et al., 2019), and an investment in increasing access to these medications was necessary (Saloner & Barry, 2018). Their strategy was to try to distribute buprenorphine in particular as widely as possible, with concerns about diversion, illicit substance abstinence, compliance with counseling and other treatment requirements secondary to its lifesaving potential (Szalavitz, 2018). In this way, the city pursued a “low-threshold” buprenorphine prescribing strategy, guided by principles that have been subsequently described as “same-day treatment entry; a harm reduction approach; flexibility; and wide availability in places where people with opioid use disorder go” (Jakubowski & Fox, 2020). It should be noted that two interventions were deliberately excluded from the innovation and expansion efforts pursued by the city and its collaborators: overdose prevention sites and drug courts. Local and federal criminal drug courts were operating in the city at the time of the initiative, and they offered defendants an alternative to conviction and incarceration if they successfully completed an treatment program utilizing medications for opioid use disorder. These were excluded from the city’s strategy for overdose reduction because they served relatively few clients (for example, there were 33 total participants in the county program at the time of a 2016 evaluation, with a 45% program completion rate; NPC Research, 2016), the low volume could not be scaled up to achieve a population-level effect, and the court’s graduates were ultimately rearrested at a higher rate than matched cohorts who underwent traditional adjudication (NPC Research, 2016). Efforts to improve outcomes at the county level by altering the processes of the court would be time consuming, while unlikely to yield a noticeable overall effect. It was felt that the added value for the effort would be marginal both in terms of effects at the patient level, and community-level reductions in overdose deaths.

Overdose prevention sites were excluded because while evidence suggested that a well-established site in Vancouver was able to significantly lower fatal overdose in its immediate vicinity (Marshall et al., 2011), and these sites reduce the risks of opioid overdose that result from fentanyl adulteration (Latimer et al., 2016), an attempt to implement a site locally would be highly-politicized, face significant obstruction (Davis, 2017), and would require addressing entrenched moral opprobrium given commonly-held views in opposition (Barry et al., 2019), especially if efforts in other cities served as an example (Hathaway & Tousaw, 2008). For this reason, the establishment of an overdose prevention site was considered a lower priority than other interventions that could be introduced more quickly and with less resistance.

In a related vein, naloxone was also not a focus of the initiative, not because its effectiveness or feasibility was in question, but because the community was already making considerable use of the opioid overdose reversal agent. Local police had been carrying it since early 2016 (True, 2016), the same year that Vermont officials issued a standing order making it available to any person at local pharmacies (Allen, 2016). These measures were complemented by a very robust Good Samaritan law, enacted in 2013, that precluded arrest and prosecution in connection to people calling 911 for help in cases of overdose (Newman & LaSalle, 2013). In the meantime, funding was in place for the free and anonymous distribution of naloxone at the community’s Safe Recovery syringe service program (The Howard Center, 2019). Apart from continuing to secure adequate funding for its free distribution, the belief was that there were few opportunities to substantively expand local access to this lifesaving overdose reversal medication, or to further encourage its use.

Given the scope of the overdose crisis and the need for swift action, the city’s plan was not a centralized, strategic one, but an effort to advance on several axes at once in pursuit of its endpoint. It did not have a timeline, phases, coordinating strategies, or any of the aspects typical of strategic planning or implementation science. Its work was instead based on the premise that the systems in place to contribute to the major surrogate endpoint could so do independently if necessary, but the effects of pursuing these individual measures would culminate at the population level. The challenge was to overcome the bureaucratic, cultural and financial obstacles necessary to bring about innovation and change in several institutions that operated independently.

Creation of a Small-City Public Health Apparatus

On the day of his hire, the mayor of Burlington assigned the city’s new chief of police responsibility for responding to the opioid crisis (Vane, 2016). He was specifically charged with taking a public health approach, not a law-enforcement centered one. Without a department of health, locating the effort in the police department was a matter of available leadership capacity rather than a focus on police-related or punitive interventions. Due to its size, the city had no full-time public health officials, Vermont had no county-level government to speak of, and the state government’s public health apparatus was too centralized to focus intense efforts on one county, even the most populous one. In the midst of a crisis, the alternative was for a city’s police leader to use the principles of public health to help lead the community towards a population-level response to the fatal overdoses (del Pozo, 2018).

To build out capacity, the city created the position of Opioid Policy Coordinator in November 2016, and staffed it with a person with over 30 years of clinical and policy experience in substance use disorder treatment (Vane, 2016). Her goal was to vet both police department operations and the city’s general approach to the opioid crisis for compatibility with sound principles of public health and SUD treatment. The chief of police seated this coordinator in the office closest to his own to signal that she would be an integral part of the leadership team, and supplemented her with an analyst with a Masters of Public Health trained in epidemiology, and a data manager (Aikman, 2016a), both of whom were funded by the United Way of Northwest Vermont, a branch of an international philanthropic organization that funds health and welfare initiatives tailored to local conditions and needs (United Way, 2021). In addition to these positions, the Burlington Police Department appointed Dr. Richard Rawson its scientific advisor on treatment. Dr. Rawson was the retired Co-Director of UCLA’s Integrated Substance Abuse Programs, lived in Vermont, and assumed the position on a volunteer basis. With a team in place, the city began its CommunityStat and SubStat programs.

The CommunityStat Model

Commenced in November 2016 (Vane, 2016) under the funding umbrella of a county opioid alliance (Aikman, 2016b), CommunityStat was a monthly coordination meeting held on neutral ground attended by executive-level stakeholders of organizations and agencies with a role in addressing the opioid crisis and the ability to rapidly make changes to policy or allocate resources. Participants included the Mayor of Burlington and his Chief of Police, the senior executives of the hospital system, the local state-designated SUD treatment provider, the Federally Qualified Health Center, halfway houses, the state Department of Health, as well as defense attorneys, prosecutors and representatives for state and federal elected officials. The meetings were run based on adherence to four guiding principles:

• Timely and accurate intelligence and information about the epidemic

• Effective tactics and strategies to reduce fatal overdose

• Rapid deployment of resources to implement interventions

• Relentless assessment and follow-up of results (Aikman, 2016a; Vane, 2016)

CommunityStat was based on the New York City Police Department’s CompStat program, a widely-adopted model of police management accountability and problem solving (Weisburd et al., 2003). It was implemented in the tradition of “problem-oriented policing,” especially the framework’s tenets that once the community identifies a significant problem, results rather than methods are paramount, an iterative, data-based evaluation should guide the process, and the most effective means may indicate the need for an interdisciplinary or non-police approach (Goldstein, 1979). Adherence to the framework has been shown to produce statistically significant improvement in a range of community problems from substance use to violent crime (Weisburd et al., 2010). Implementing the framework was not without its risks, however: its intense focus on data and performance metrics does not ensure that what is being measured is equitable or effective, or achieves its goals without negative collateral consequences neglected in the process of measurement (Eterno & Silverman, 2006). The CompStat model had been shown in some cases to incentivize data manipulation, overly-aggressive policing, and to encourage the pursuit of performance measures—especially enforcement-related ones—regardless of their association with negative outcomes for vulnerable and minoritized populations (Eterno & Silverman, 2010).

To guard against these concerns, the conveners emphasized that a focus on MOUD was evidence-based and meant to be implemented in ways that benefitted all members of the community rather than just subsets. In contrast to traditional police-focused CompStat programs, CommunityStat guarded against negative and inequitable outcomes by including actors with a role in protecting the vulnerable, such as supervising defense attorneys, criminal justice reform advocates, and city councilors representing diverse constituencies. The presumption was that all participants shared the goal of reducing overdose mortality, which distinguished Burlington’s use of the CompStat model from applications in police-dominated settings by its exclusive focus on a health-related outcome rather than crime or enforcement statistics.

Implementation of CommunityStat in Burlington can therefore be distinguished from a more general recurring community stakeholder meeting about the opioid crisis in key ways. Instead of fostering a general discussion guided by an agenda, it followed a detailed “script” that was prepared in advance by the city’s opioid policy coordinator and only distributed to the leaders of each meeting. It consisted of a chronological list of the topics to be presented, the corresponding slides and other visuals, key conclusions and data, and then a series of prompts for the leaders to emphasize certain points, ask specific accountability questions of the various participants in the room, and call for discrete deliverables or clarifications at a future date.

The script ensured that the meeting would remain tightly focused, draw out key facts and findings, and yield deliverables that would later be the subject of scripted inquiries in later meetings. People who would be expected to comment on difficult or sensitive matters would be notified in advance, but the format of the meeting meant that some of the exchanges would be difficult, perhaps adversarial, or might involve conceding a marked need for improvement in processes. One of the goals was to create a social pressure to meet the expectations set by peers and colleagues in the meetings (True, 2017a). For example, one meeting dissected the failure of coordination that resulted in a woman’s advocacy group paying for a defendant’s bail when both the prosecution and defense agreed she was in imminent danger of fatal overdose and were working on a plan to voluntarily send her out of state for MOUD-based treatment. For these reasons, the meetings were not open to the press or the public.

The SubStat Model

Commenced in mid-2017, SubStat (“Substance Stat”) was a biweekly meeting led by the city’s opioid policy coordinator that convened practitioners from different agencies to discuss specific clients who were deemed at elevated risk of fatal overdose based on recent behavioral trends such as criminal activity, nonfatal overdose, domestic discord, or child welfare issues (Baker, 2017; True, 2017c). These meetings were tactical in nature and focused on responses to individual problems. They facilitated collaboration between the actors likely to encounter the person at risk, with a plan to offer them treatment at the various touchpoints provided by the roles of the meeting participants. The premise was that criminal justice actors and other government institutions were critical touchpoints for people at elevated risk of overdose (Larochelle et al., 2019), and coordination among these actors was a way to maximize the probability of effective intervention.

Articulating Standards of Care

In an effort to learn the science necessary to implement the “effective tactics and strategies” that would comprise the city’s response to the opioid crisis, city officials developed an informal partnership with the Bloomberg American Health Initiative (AHI), which was engaged in a national campaign to reverse the opioid epidemic, and with researchers at the Johns Hopkins Bloomberg School of Public Health. The Burlington side of the relationship was led by the chief of police, who had a broad mandate to develop evidence-based plans to address the overdose epidemic. In June of 2018, an AHI-funded meeting in Washington, D.C. at the Police Executive Research Forum, along with Johns Hopkins researchers, police leaders, and two former United States drug czars yielded “Ten standards of care: Policing and the opioid crisis” (Bloomberg American Health Initiative, 2018). The standards were meant to provide a policy guide for the nation’s police departments. They stressed the efficacy of MOUD and the effectiveness of treatments that centered on it, and urged police to support SSPs, MOUD treatment in jails and prisons, and proactive efforts to reduce stigma. The standards were unveiled at an annual meeting of police executives in Nashville, Tennessee in May of 2018, but the science and principles they were based on corresponded to the CommunityStat initiatives being undertaken in Burlington.

Interventions to Reduce Overdose Deaths

Over the course of 2017 and 2018, CommunityStat participants and the city’s policing and public health team were instrumental to innovations and reforms that were collectively associated with a significant decline in overdose deaths. Some were implemented in the latter half of 2018, limiting their possible role in the reduction of deaths for that calendar year. It is worth noting, however, that as opioid overdose deaths resumed their national upward march in 2019, they stayed level in Burlington, which was able to sustain its reductions. The interventions were:

• Elimination of the regional wait list for medication-assisted treatment. Chittenden County had only one treatment facility that could prescribe both methadone and buprenorphine to patients, located in South Burlington. For the duration of the opioid epidemic, it had waiting lists for induction into treatment that could last from weeks to months. State government authorized the creation of an additional facility in St. Albans, a small city to the north of Burlington. Slated for a January 1, 2017 opening (Shumlin, 2016), construction was subject to over 9 months of delays when the CommunityStat group began tracking its progress and began lobbying the governor’s office for expedited issuance of the construction permits at the heart of the delay. When the facility opened, the increased regional capacity all but eliminated the county’s wait list for MOUD treatment (Ready-Campbell, 2017).

• Prescribing low-barrier buprenorphine at the city hospital’s emergency department. Emergency department physicians were asked to present monthly data on overdose outcomes at CommunityStat. The most prevalent outcome was a reversal of the overdose using naloxone and discharge with a brochure about local treatment programs (Mackey, 2019). In many cases, the patients were in danger of entering a state of withdrawal that would send them in search of illicit opioids to relieve it, sometimes taking risks in doing so. The second most prevalent outcome was death. Patients were almost never referred to MOUD as a result of presenting with an overdose at the emergency department. After months of disclosing this data to colleagues and peers at monthly meetings, the University of Vermont Medical Center secured funding to begin a buprenorphine prescribing program at its emergency department (Freese, 2018), and went on to provide the medication to anyone who screened positive for OUD, regardless of what brought them in for treatment. The plan was to offer patients an immediate supply of sublingual buprenorphine of up to 3 days, and a corresponding appointment with a psychiatrist within that timeframe. Patients were free to return to the ED to re-enroll as desired, and were prescribed additional buprenorphine if they did. Of the first 18,731 emergency department visitors screened, 4.4% had indicators of OUD, and 61 were enrolled in buprenorphine-based MOUD treatment (Mackey, 2019).

• Prescribing low-barrier buprenorphine at the city’s syringe service program. A series of CommunityStat meetings yielded a consensus around the need for expanded access to buprenorphine at Safe Recovery, Burlington’s SSP. It was premised on evidence that clients seeking treatment would find the location an appealing alternative to more formal treatment settings they were less familiar with and less comfortable in (Fox et al., 2015), especially if they were equivocal about their commitment to treatment. With the allocation of federal funding for a pilot, the program began prescribing the medicine to any client who desired it, with minimal additional requirements (Lopez, 2018). The program characterized the people it treated as patients who either did not consider themselves ready for the demands of more formal treatment, or who were ejected from formal treatment programs for criminal activity, misbehavior or relapse. To promote retention among this population, patients were offered take-home doses of sublingual buprenorphine with minimal counseling requirements, no obligation to demonstrate abstinence from illicit substances, and no expectation of a treatment completion date. Basic retention was considered success and a means to expand the scope and intensity of treatment when patients were ready. The physician working at the site was an obstetrician/gynecologist who had lost her son to an opioid overdose, which motivated her to obtain a federal waiver to prescribe buprenorphine (Lopez, 2018).

• Statewide MOUD-based treatment in Vermont’s jails and prisons. With the support of city government, a Burlington resident who served in the state legislature authored a bill mandating all inmates in Vermont jails or prisons who screened positive for OUD be offered one of the three medications proven effective at treating the condition, and all prisoners who were in treatment at the time of incarceration be able to continue their treatment. CommunityStat participants aggressively lobbied for passage of the legislation (True, 2017a, 2017b), citing success with a similar program in Rhode Island (Green et al., 2018) and calling upon the researchers who implemented it there to share lessons learned. The bill became law on July 1, 2018. When the requirements of the law were met, Vermont’s prison system was regularly treating up to 31% percent of its inmate population with MOUD “for as long as medically necessary” (Faher, 2019).

• Expanded primary care physician treatment capacity of the community’s Federally-Qualified Health Center (FQHC). Attendance at monthly CommunityStat meetings by executives from the Community Health Center of Burlington, a federally-funded health care provider for the community’s low income residents, encouraged the center to expand its treatment of OUD by its staff of primary care physicians. Over the course of approximately 18 months, the center went from treating several dozen patients with MOUD to consistently treating over 150. The people it treated were from socioeconomically disadvantaged backgrounds and would likely be less resilient at recovering from the disruptions to employment that can accompany more acute cases of OUD (Alexandre & French, 2004). The treatment therefore provided this population of patients with the safety net of an intervention that could not only reduce the likelihood of overdose, but also reduce their economic vulnerability in many cases.

• De-facto decriminalization of the possession of unprescribed buprenorphine. Burlington’s chief of police and the county’s prosecutor came to an agreement that their agencies would not arrest or prosecute defendants for the possession of personal quantities of unprescribed buprenorphine, under the logic that even illicit buprenorphine served as an overdose prophylaxis and could contribute to a population-level overdose reduction (Szalavitz, 2018). They were the first public officials in the nation to take this position, and it was later adopted by officials in Philadelphia (del Pozo et al., 2020). This was based on research that most diverted buprenorphine is consumed as an attempt at self-treatment and results from scarcity of treatment options (Carroll et al., 2018), and is associated with a decreased risk of overdose (Carlson et al., 2020). Few to no adverse consequences were observed, and in 2021 Vermont legislators went on to remove unprescribed possession of buprenorphine in personal use amounts from its criminal code, formally legalizing the practice. It was the first state in the nation to do so (Landen, 2021), soon followed by Rhode Island (Mulvaney, 2021).

The Burlington Police Department also launched two programs that were met with little success. In 2018, it began an effort to visit the homes of people who recently overdosed in order to offer them treatment and harm reduction resources (The Associated Press, 2018). The team conducting the visits was composed of the Opioid Policy Coordinator and a senior police officer. After several attempts, the program was terminated due to lack of interest on the part of the target population, who did not respond positively to visits from the police. A second program screened every person taken into custody for OUD, with the goal of offering people who screened positive immediate transportation to either the SSP or the hospital emergency department for induction into MOUD-based treatment (Jickling, 2019a). The initiative lasted approximately 6 months, and after little success, led to the conclusion that the time of arrest was not an effective point for induction into treatment unless it was accompanied by an incentive such as diversion from prosecution, which the program was not designed to do. The very modest results of this effort yielded a decision to continue a focus on referring people to treatment using the ongoing SubStat, process rather than at the time of arrest. People with repeated contact with police and elevated risk of overdose would be identified and discussed at SubStat meetings, and interventions could be formulated that did not rely on an interaction under the circumstances of arrest.

Another point to acknowledge is the reliance on OUD screeners for identifying candidates for treatment, and the strategy’s focus on people whose substance use had reached acute stages as targets for intervention. Given the potency of fentanyl and its growing omnipresence in the drug supply (Ciccarone, 2019; Ivsins et al., 2020; Mars et al., 2019), vulnerability to overdose is not limited to people with substance use disorders, but includes substance users who moderate their use, and who show few or none of the markers of a disorder. The distribution of naloxone along with advice not to use alone was one way of protecting this population, as was the fact that they could present and request treatment themselves at low-barrier facilities if they desired it. Widespread distribution of fentanyl test strips, which were not a substantial component of the city’s strategy, would also have helped this population better manage the risks it faced (Peiper et al., 2019; Weicker et al., 2020), and the use of buprenorphine as an intermittent overdose prophylaxis during times of perceived danger or uncertainty of supply could have provided a novel form of safe supply for such a group (del Pozo & Rawson, 2020). Nonetheless, reliance on screeners meant to detect opioid use disorder as a means for intervention assumed that such a population was the one to prioritize, at the expense of measures that would better protect substance users who were indeed in danger but would not have indicated a disorder when administered a screening instrument.

Success Without a Ready Means for Evaluation

As noted, the initiatives put in place during the CommunityStat process associated with a 50% reduction in fatal overdoses in the county as compared to a 20% increase in the remainder of Vermont (Jickling, 2019b), a gain which was sustained from 2018 into 2019 (Vermont Department of Health, 2020). Researchers affiliated with the University of Vermont and the Johns Hopkins Bloomberg School of Public Health observed that despite being successful, the community’s interventions were not designed or implemented in a way that permitted rigorous evaluation. They were enacted without regard to randomization or control, and there was no attempt to prevent the interventions from confounding each other.

The city’s premise was that the community was responding to an acute public health emergency, and its actions were based on scientific principles that were already well-supported by evidence. The same confounding effects observed by implementing them together could alternatively be viewed as an attempt to maximize the effectiveness of the response. Moreover, attempts to measure success may have impeded progress. For example, the low-barrier program at the emergency department initially called for randomized assignment to a buprenorphine arm or a non-MOUD control arm. People going through withdrawal in the aftermath of a naloxone-induced reversal were asked to enroll in a trial where an outcome might be that they leave the hospital with no medicine. Recruitment for the first few months was poor. The study was revised to screen all people who came to the emergency department for OUD and prescribe buprenorphine to any person who screened positive and requested it.

Ultimately, the low-threshold MOUD programs at the Safe Recovery SSP and the emergency department had an evaluation component tied to their funding, but they were principally concerned with the programs in isolation rather than their effect on community health. The former received funding from the US Substance Abuse and Mental Health Services Administration, which required evaluating the adherence and retention of program participants, while the latter received National Institutes of Health funding as a feasibility study that required assessing if the program can “produce positive outcomes in terms of treatment retention and negative screens for illicit opioid use at 1 week, 3 months and 6 months after treatment initiation” (Mackey et al., 2019). In each case, however, the evaluations did not assess relative patient preferences for either modality, or community-level effects on substance use-related morbidity and mortality.

Going forward, the experience in Burlington highlights important opportunities for research even when a community acts quickly and broadly in a time of crisis. While quasi-experimental methods such as interrupted time series analyses would be confounded by several “interruptions” as interventions are implemented in rapid succession, more flexible methods such as a realist evaluation could prove valuable (Pawson & Tilley, 1997). Premised on the theory of scientific realism, it would evaluate CommunityStat as a complex mechanism that operates in a specific social and political context, offering insights that may not be precisely generalizable in other times and places. In doing so, an in-situ realist evaluation would provide other practitioners with practical insights into whether Burlington’s interventions would be feasible for them.

Burlington’s efforts may also illustrate that there is a point where continued evaluations of interventions with a firm basis in evidence are counterproductive during a crisis, as may be the case with MOUD. In such cases, the most valuable research agenda may call for a focus on the high-fidelity implementation of promising programs, as well as researching ways to make successful implementation feasible across different jurisdictions. The Consolidated Framework for Implementation Research (Damschroder et al., 2009) has been shown effective at providing the means to evaluate and improve interventions in a wide range of settings (Kirk et al., 2016), and has the flexibility to produce insights at a rapid pace (Keith et al., 2017). In sum, at the present stage of the opioid crisis, public officials disinclined to carefully structure their work for clinical evaluation may still produce valuable research in the field of implementation science.

Reconsidering the Relationship Between Research and Practice

The experience of CommunityStat in Burlington allows us to step back and ask larger questions about the relationship between research and practice, or evidence and methods, as we grapple with the overdose epidemic and our generally lackluster response. It is problematic to spend taxpayer money on programs based on longstanding but unfounded beliefs, or that rely on an untested novelty as a selling point. At the same time, we should ask why some of the innovations that make the most extensive use of MOUD in low threshold settings are still being piloted as clinical trials when nothing prevents much more widespread implementation. It is commonly asserted that the lag between building an evidence base for the basic science of an intervention and translating it to practice it is about seventeen years in health services (Bauer et al., 2015; Morris et al., 2011). It is unclear where that puts low-threshold MOUD on such a timeline, but it has yet to be widely implemented. SSPs across the United States would likely embrace the opportunity to directly induct clients into low threshold buprenorphine treatment when they desire it, and the chances of this drawing patients away from more established treatment programs are slim, considering the shortage of MOUD treatment generally.

Traditional approaches to research remain an obstacle. Consider the initial need to randomly assign patients to a non-MOUD arm in the low-threshold buprenorphine pilot at the UVM Medical Center’s emergency department as a condition of evaluation by the “gold standard” of a randomized controlled trial (Hariton & Locascio, 2018). It was not only unsuccessful in terms of recruitment when people recovering from an overdose were told they would have to go through an enrollment process and might leave with nothing but a referral, but arguably put the control arm at greater risk of ensuing overdoses based on what we have learned about the effectiveness of buprenorphine treatment (Wakeman et al., 2020). The initial IRB process also took months, and then took considerable time to amend. Research ethics are not the only relevant ethics for implementing interventions; the need for timely, far-reaching measures in a time of crisis should carry ethical weight as well. When people are at an acute risk of fatal overdose, part of the calculation about preserving their rights and dignity should center on rapidly implementing and scaling up the systems that stand the best chance of protecting them. These imperatives carry considerable normative weight in communities of practice, where experimental design and the IRB can seem unnecessarily arcane or burdensome.

Perhaps we should more carefully consider that the people scientifically situated to recommend effective interventions are often a product of, and funded by, institutional processes that are rigorous as a tradeoff with speed and agility. Their habits and incentives may therefore be aligned accordingly. We might see faster and greater improvements to public health if the funding for innovative programs to address emergent needs was decoupled from processes that emphasize or insist on rigorous experimental design, and shifted to people with facility at taking basic proven concepts, such as the effectiveness of MOUD, and implementing them in complex systems. Doing so might involve placing practitioners at key points in an accelerated grant funding cycle as reviewers of proposals, for example. The resulting studies might be observational and quasi-experimental rather than randomized and controlled, but the public might benefit more quickly. When the city of Burlington held monthly meetings imploring stakeholders to act fast and worry about measurement later, it was based on the premise that practice should guide research during a genuine crisis, rather than the other way around.

Conclusion

The overdose death reductions achieved in Chittenden county in 2018 were sustained through the end of 2019, but the COVID-19 pandemic erased them for the county and the state. Vermont witnessed a 57.6% increase in fatal overdoses in 2020, the largest increase in the United States (Bernstein & Achenbach, 2021). Research is needed to determine why this was the case, but preliminary theories seem plausible. One is the disruption of MOUD treatment availability and continuity as health care resources were shifted to contend with the pandemic and emphasized telemedicine and other remote services (Vermont Department of Health, 2021). At the same time, CommunityStat meetings moved to an online format, while the majority of participants were from sectors and institutions charged with the pandemic’s emergent crisis response. The Burlington city government meanwhile repurposed analysts, staff and leaders to focus on preventing the spread of the virus. This shift in resources and priorities, combined with pronounced social isolation and the stressors and instabilities of the pandemic, may have increased the risk of overdose, especially as people were more likely to use drugs alone, while reducing the effectiveness of the CommunityStat response.

Even before the CommunityStat initiative, the state of Vermont was not new to innovative responses to SUD and overdose. Its physicians developed the hub-and-spoke model of treatment (Brooklyn & Sigmon, 2017), which has been shown to be an effective way to meet the different needs of people seeking treatment for opioid use disorder (Rawson et al., 2019). It has since been copied as a model in other states such as California and Washington (Darfler et al., 2020; Reif et al., 2020). By itself, however, it did not contain or reduce overdose fatalities in the years prior to the implementation of the CommunityStat model in Burlington. This can be attributed to the fact that people who required hub treatment, which was administered out of a highly structured opioid treatment program, had to contend with waiting lists for years (Ready-Campbell, 2017). Many died while waiting. Meanwhile, most spokes were primary care physicians, which were an awkward and unwelcoming environment for many people who use drugs, especially those who had a difficult time adhering to formal treatment protocols in a medical environment.

To move past these limitations, Burlington’s approach went beyond a rigid conception of hubs and spokes to include delivery mechanisms for MOUD that appealed to people who were either more likely to trust places like an SSP or who personally conceived of these low-threshold options as something other than formal treatment per se, which they believed they did not want. City staff began referring to these other mechanisms as “nodes” to conceptually distinguish them from their counterparts. It will require more research to determine why people find certain modalities appealing while they eschew others, even when the end result is often being prescribed the same type of medicine. The experience of Burlington, Vermont also makes the argument that such research has its limits, however.

As the opioid overdose crisis resurges to record levels in the wake of the pandemic (Katz et al., 2020), Burlington’s work from 2017 through early 2020 can serve as a model for other municipalities. Other cities’ leaders should consider bringing different institutions together into a common setting that fosters mutual accountability in addressing the challenges of SUD and overdose. At the same time, it is worth noting that centering the city’s work on a police accountability model and utilizing police leadership to effectively implement a public health response is an exception to the rule. Such a focus on health metrics would be considered a paradigm shift for most police departments, and police response to people who use drugs is often characterized by harmful and counterproductive measures that erode mutual trust (Friedman et al., 2021; van der Meulen et al., 2021). Indeed, police attempts at overdose outreach in Burlington were not met with success. This suggests that small cities without departments of health should give careful thought to investments in public health infrastructure, even on a modest scale, that can provide effective leadership in lieu of police and other actors.

In the meantime, municipalities without the type of police departments capable of leading a collaborative, multidisciplinary approach to the overdose crisis can pursue an agenda of police innovation and reform based on a public health ethic (del Pozo et al., 2021). They can set the expectation that their police adhere to public health measures and metrics in responding to risk behaviors and the sequalae of substance use (Goulka et al., 2021), as well as mental illness and other behavioral health crises. This will likely involve the extensive use of discretion in response to drug possession and offenses related to drug use, and an emphasis on effective treatment in lieu of prosecution. Insofar an entire community’s response to substance use and overdose is shaped and constrained by their criminalization (Earp et al., 2021), abandoning the failed “War on Drugs” approach to substance use is critical to addressing the present crisis as we slowly emerge from that of the pandemic (del Pozo & Beletsky, 2020).

Most importantly, all the agencies and institutions responsible for a community’s multi-pronged response to the opioid crisis should see themselves as contributing in their own way to a collective goal. Their leadership should encourage each of them to do the things within their own sphere that place a very effective means for preventing opioid overdose within reach of as many people with OUD as possible: agonist medications for the treatment of opioid use disorder (NASEM, 2019).

Author Biography

Brandon del Pozo is a postdoctoral research fellow in addiction, substance use, policing and public health at The Miriam Hospital/Warren Alpert Medical School of Brown University. Prior to his work in research, he served as a police officer for 23 years: 19 in the New York City Police Department, and 4 as the Chief of Police of Burlington, Vermont.