The estimated costs and benefits of a hypothetical supervised consumption site in Providence, Rhode Island

Abstract

Background:

Overdose deaths have increased dramatically in the United States, including in Rhode Island. In July 2021, the Rhode Island government passed legislation supporting a two-year pilot program authorizing supervised consumption sites (SCSs) in response to this crisis. We estimated the costs and benefits of a hypothetical SCS in Providence, Rhode Island.

Methods:

We utilized a decision analytic mathematical model to compare costs and outcomes for people who inject drugs under two scenarios: (1) a SCS that includes syringe services provision, and (2) a syringe service program only (i.e., status quo). We assumed 0.95% of injections result in overdose, the SCS would serve 400 clients monthly and have a net cost of $783,899 annually, 46% of overdoses occurring outside of the SCS result in an ambulance run and 43% result in an emergency department (ED) visit, 0.79% of overdoses occurring within the SCS result in an ambulance run and ED visit, and the SCS would lead to a 25.7% reduction in fatal overdoses near the site. The model was developed from a modified societal perspective with a one-year time horizon.

Results:

A hypothetical SCS in Providence would prevent approximately 2 overdose deaths, 261 ambulance runs, 244 ED visits, and 117 inpatient hospitalizations for emergency overdose care annually compared to a scenario that includes a syringe service program only. The SCS would save $1,104,454 annually compared to the syringe service program only, accounting only for facility costs and short-term costs of emergency overdose care and ignoring savings associated with averted deaths. Influential parameters included the percentage of injections resulting in overdose, the total annual injections at the SCS, and the percentage of overdoses outside of the SCS that result in an ED visit.

Conclusion:

A SCS in would result in substantial cost savings due to prevention of costly emergency overdose care.

Introduction

Since 1999, overdose deaths have been increasing in the United States (Hedegaard et al., 2020), including Rhode Island. Additionally, in 2019, the age-adjusted rate of overdose deaths in Rhode Island exceeded the national average (approximately 30 vs. 22 deaths per 100,000 population) (United States Centers for Disease Control and Prevention, 2021). In response, the Rhode Island Department of Health (RIDOH), in collaboration with state and community partners, has implemented numerous interventions to prevent overdose deaths and other drug-related harms. From 2016 to 2019, overdose deaths began to decline in Rhode Island; however, in 2020, Rhode Island experienced its highest number of accidental overdose deaths ever recorded (Rhode Island Department of Health, 2021a). In this context, new overdose prevention approaches are urgently needed.

In July 2021, Rhode Island Governor Daniel McKee signed a bill into law authorizing a two-year pilot program of supervised consumption sites (SCSs) to prevent drug overdoses in Rhode Island (RI Gen. Laws Chapter 23-12.10) (State of Rhode Island, 2021). A SCS is a community-based resource for health screening, disease prevention, and treatment assistance where persons may safely consume pre-obtained controlled substances in a non-judgmental environment without legal repercussions. SCSs have health care and/or harm reduction professionals onsite, as well as supplies and equipment, to treat overdose, reduce drug-related harms, and provide resources and referrals for additional services that may be appropriate for persons utilizing the site (e.g., other harm reduction services, social services, counseling, or other medical treatment). Prior evidence suggests that SCSs are cost-effective and reduce fatal overdoses (Kennedy et al., 2017; Levengood et al., 2021). To date, no SCS client has experienced a fatal overdose within a SCS (Armbrecht et al., 2020). SCSs are sometimes referred to as supervised injection or safer injection sites or facilities, as well as harm reduction centers or drug overdose prevention sites. In Rhode Island, SCSs would complement or expand existing syringe service programs operating in the state (AIDS Care Ocean State, 2021).

SCSs have also been associated with safer drug use practices (e.g., reduced syringe re-using, syringe sharing, and rushed injection), which are critical for preventing human immunodeficiency virus (HIV) and hepatitis C virus (HCV) transmission and skin and soft tissue infections. Additionally, SCS increase uptake of addiction treatment and other health services (DeBeck et al., 2011; Wood et al., 2007). Finally, some studies suggest that SCSs have community and public order benefits (e.g., decrease in public injection), and do not increase drug-related crime or public nuisance in the community (Kennedy et al., 2017; Levengood et al., 2021).

Most prior studies of the cost-effectiveness of SCSs have focused on sites in large urban centers (Kennedy et al., 2017; Levengood et al., 2021). Providence is a smaller urban setting, which may impact the population-level costs and benefits of a SCS due to a lower population density of people who use drugs and different population movement patterns. If fewer clients attend an SCS in a smaller urban setting, then staff would have the opportunity to prevent fewer adverse health outcomes. Additionally, although the total operating costs of a SCS in a smaller urban setting would likely be lower, costs are unlikely to decrease proportionally to the decrease in clients, potentially leading to a higher cost per client served and cost per adverse health outcome prevented. The objective of this analysis was to estimate the costs and benefits of a hypothetical SCS in Providence, the capital city of Rhode Island with a population of about 190,000 people (United States Census Bureau, 2021), to inform implementation efforts.

Methods

Overview

The objective of this analysis was to estimate the incremental costs and benefits of the implementation of a SCS in Rhode Island compared to currently available harm reduction services (i.e., the status quo). Specifically, we compared the addition of a hypothetical SCS to an existing syringe services program in Providence, Rhode Island, versus a Providence-based syringe services program alone (because many SCSs also provide sterile injecting and other drug-using equipment without requiring that persons consume substances on-site). We selected Providence as the location for the hypothetical comparison because it is the city with the largest population and highest number of fatal overdoses (Rhode Island Department of Health, 2021b) in Rhode Island, although the state law allows a SCS in any community with prior municipal-level approval.

Our analysis leveraged a decision analytic mathematical model previously developed by the Institute for Clinical and Economic Review (ICER) in partnership with the Comparative Health Outcomes, Policy, and Economics (CHOICE) Institute at the University of Washington (Armbrecht et al., 2020; University of Washington Comparative Health Outcomes, Policy, and Economics Institute, 2021). ICER and the CHOICE Institute developed this model based on adapted outcome calculations from relevant prior models (Behrends et al., 2019; Hood et al., 2019; Irvine et al., 2019, Irwin et al., 2017a, 2017b; Jozaghi & Vancouver Area Network of Drug Users, 2014; Jozaghi & Vancouver Area Network of Drug Users, 2015), as well as interviews with key staff and researchers of previously implemented SCSs. The model estimates costs and outcomes over a one-year time period, utilizing information predominantly from evaluations of the Insite, a SCS in Vancouver, Canada (P.H.S. Community Services Society, 2021). Although the Rhode Island SCS regulations will permit various modes of consumption (e.g., inhalation), the model described herein focuses on the potential impact of SCSs for people who inject drugs, due to the limited data available to inform overdose risk and other outcomes (e.g., emergency department utilization) averted for other modes of drug use in the context of supervised consumption. The model is described in detail elsewhere (Armbrecht et al., 2020); key aspects and assumptions for the hypothetical Providence-based SCS are summarized below.

Model structure

Our application of the ICER/CHOICE model compares the costs and outcomes for people who inject drugs in Providence under two scenarios: (1) a SCS that includes syringe services provision, and (2) a syringe service program only. In general, the model assumes that a SCS in Providence would provide similar services, have similar cost of living-adjusted operational costs, and have similar protective effects as Insite in Vancouver, but may vary in the number of clients it serves. Insite services include sterile equipment for drug consumption, spectrometer testing of drug contents, immediate response in the event of an overdose, clinical care (e.g., wound management, vaccinations), and connection to substance use disorder treatment, healthcare, and community services (P.H.S. Community Services Society, 2021). Additionally, the model assumes that transmission probabilities of new HIV and HCV infections and rates of initiation and continuation of medications for opioid use disorder (MOUD) would be similar under the two scenarios.

The model was developed from a modified societal perspective (i.e., a perspective that examines some costs and savings outside the health care system) (Neumann et al., 2016) with a one-year time horizon because SCSs are not funded by the health care system or payers of health care. The model was developed in Microsoft Excel for Office 365 (Version 2005) (Armbrecht et al., 2020).

Model parameters

The model includes input parameters from three broad categories: (1) city characteristics, (2) primary outcomes, and (3) costs. Table 1 includes our primary estimate, low and high estimates for sensitivity analyses (generally +/−20%), and reference for each parameter.

Table 1.

Input parameter values and sources.

Parameter	Primary estimate (low, high)	References
City characteristic
Cost of living ratio (compared to Vancouver)	0.93* (0.74, 1.12)	Expatistan, 2021
Population density (people per square mile)	9676* (7741, 11,611)	United States Census Bureau, 2021
Commercial property value (cost per square foot)	$302* ($242, $362)	Alixandrescu, 2020
Commercial mortgage loan rates	7.00% (5.00%, 9.00%)	Armbrecht et al., 2020
Primary outcome
Mortality reduction within 0.25 miles of SCS	25.7% (20.6%, 30.8%)	Marshall et al., 2011
Mortality reduction beyond 0.25 miles of SCS	0%	Marshall et al., 2011
Percentage of Providence overdose deaths within 0.25 miles of SCS	8.0%* (6.4%, 9.6%)	Unpublished Rhode Island Department of Health Office of the State Medical Examiners data
Number of overdose deaths per year in Providence	93* (74, 112)	Rhode Island Department of Health, 2021b
Total annual injections in SCS	60,840* (48,672, 73,008)	Tyndall et al., 2006
Unique clients per month at SCS	400* (320, 480)
Percentage of injections resulting in an overdose	0.95% (0.50%, 1.20%)	Colledge et al., 2020; Jacka et al., 2020; Notta et al., 2019; unpublished B.D.L. Marshall RAPIDS clinical trial data
Percentage of SCS overdoses resulting in an ambulance run	0.79% (0.63%, 0.95%)	Irwin et al., 2017b; K.P.M.G., 2010
Percentage of SCS overdoses resulting in an ED visit	0.79% (0.63%, 0.95%)	Irwin et al., 2017b; K.P.M.G., 2010
Percentage of non-SCS overdoses resulting in an ambulance run	46%* (37%, 55%)	Pollini et al., 2006; unpublished Rhode Island Department of Health Harm Reduction Surveillance System data
Percentage of non-SCS overdoses resulting in an ED visit	43%* (34%, 52%)	Pollini et al., 2006; unpublished Rhode Island Department of Health Harm Reduction Surveillance System data
Percentage of ED visits for overdose resulting in inpatient admission	48% (38%, 58%)	Armbrecht et al., 2020
Cost
Insite annual operating costs in Vancouver	$1,687,286 ($1,349,829, $2,024,743)	Armbrecht et al., 2020
Term of commercial loan (years)	15	Armbrecht et al., 2020
SCS square footage	1000	Armbrecht et al., 2020
Annual syringe service program cost†	$863,861* ($691,089, $1,036,633)	Teshale et al., 2019
Ambulance run cost	$466* ($372, $559)	United States Centers for Medicare and Medicaid Services, 2021
ED visit cost	$3451 ($2761, $4141)	Armbrecht et al., 2020
Inpatient hospitalization cost	$7897* ($6318, $9476)	Mallow et al., 2018

Abbreviations: ED, emergency department; SCS, supervised consumption site.

^*Estimate for Providence.

^†Average cost for the United States; weighted by the model to the estimated cost of living in Providence.

City characteristic parameters.

We obtained estimates for Providence for each city characteristic input parameter. Of note, for the estimated commercial property value, we averaged 2019 estimates for the commercial property value in Philadelphia, Pennsylvania ($319 per square foot), and New Haven, Connecticut ($286 per square foot), due to the lack of Providence-specific estimates (Alixandrescu, 2020). For the commercial mortgage loan rate, we retained the estimate utilized by ICER and the CHOICE Institute across cities. We varied these estimates in our sensitivity analysis.

Primary outcome parameters.

Most of our primary outcome input parameter estimates were consistent with those based on Vancouver Insite studies (Armbrecht et al., 2020), except for six that are specific to Providence. First, we utilized 2020 RIDOH Office of the State Medical Examiners data to estimate the number of overdose deaths occurring in Providence per year (Rhode Island Department of Health, 2021b). Second, we utilized 2020 Office of the State Medical Examiners data to identify locations where the highest percentage of fatal overdoses occur within 0.25 miles (8%) (unpublished Rhode Island Department of Health data). Additionally, compared to the parameter established by ICER and the CHOICE Institute, we decreased the number of unique clients per month expected at the SCS (from 2100 to 400). We derived this expected number of SCS clients per month for Providence from discussions with leadership of the state’s largest syringe service program and consideration of their monthly and annual number of clients in Providence, review of the monthly number of clients attending a non-profit organization that serves people who use drugs in Providence, and discussions with personnel highly involved in SCS planning discussions with people who use drugs in Providence. We then used previously published estimates for the monthly number of visits per client from Insite to estimate the total annual injections in the SCS among 400 unique clients (Tyndall et al., 2006). Additionally, we used 2021 RIDOH Harm Reduction Surveillance System data (unpublished Rhode Island Department of Health data), in combination with published estimates of the percentage of overdoses that are witnessed (Pollini et al., 2006), to estimate the percentage of non-SCS overdoses resulting in an ambulance run and an ED visit, respectively. In sum, key assumptions that are built into our primary parameter estimates include: (1) there would be a 25.7% reduction in fatal overdoses within a 0.25 mile radius of the SCS and a 0% reduction elsewhere in Providence, based on a prior study in Vancouver by Marshall and colleagues (Marshall et al., 2011); (2) prior to the implementation of a SCS, 8% of fatal overdoses in Providence occur within 0.25 miles of the facility, which assumes that the facility is located in a neighborhood where the highest percentage of overdoses occur (unpublished Rhode Island Department of Health data); (3) 0.95% of injections result in an overdose, based on studies in Providence and Vancouver (Colledge et al., 2020; Jacka et al., 2020; Notta et al., 2019; unpublished B.D.L. Marshall RAPIDS clinical trial data); (4) 46% of overdoses occurring outside of a SCS in Providence result in an ambulance run and 43% result in an ED visit, based on RIDOH’s Harm Reduction Surveillance System data (unpublished Rhode Island Department of Health data); and (5) 0.79% of overdoses that occur within a SCS would result in an ambulance run and ED visit, based on studies in Baltimore (Irwin et al., 2017b) and Sydney (K.P.M.G., 2010).

Cost parameters.

Most of our cost input parameter estimates were consistent with those used by ICER and the CHOICE Institute, except for three that were specific to Rhode Island or were otherwise adjusted. First, for the cost per ambulance run, we utilized an estimate from the 2021 Centers for Medicare and Medicaid Services Ambulance Fee Schedule for Rhode Island, Healthcare Common Procedure Coding System code A0427, urban base rate (United States Centers for Medicare and Medicaid Services, 2021). Second, for the cost per inpatient hospitalization, we utilized an adjusted New England average payment to hospitals for opioid-related visits (Mallow et al., 2018). Finally, we utilized previously published cost estimates for a medium-sized, urban syringe services program in the United States (Teshale et al., 2019); the estimated cost of naloxone was subtracted, and the resulting cost weighted by the model to the estimated cost of living in Providence. Of note, we estimated the SCS annual operating cost based on the full operating budget of Insite; this is a highly conservative estimate for the smaller facility considered in our analysis.

Model outcomes

The outcomes of the model for each scenario included total short-term costs, number of overdose deaths within 0.25 miles of the SCS location, and number of EMS runs, ED visits, and inpatient hospitalizations resulting from the injections that would occur at the SCS, if available. More detailed costs (e.g., annual facility costs and ambulance run, ED visit, and inpatient hospitalization costs for emergency overdose care) were also produced. All results were undiscounted values because of the one-year time horizon.

Analyses

Using our primary parameter estimates, we conducted base-case analyses to estimate the costs and health outcomes in an incremental fashion comparing the two scenarios (i.e., a SCS that includes syringe services vs. a syringe service program only). We also conducted one-way sensitivity analyses, using the low and high estimates for each parameter, to identify critical parameters that have a large impact on costs and health outcomes. Finally, we conducted scenario analyses utilizing our base-case parameters and further varying only the number of unique SCS clients per month to understand the impact of client volume on the difference in total short-term costs. Of note, as described above, we calculated the total annual injections within the SCS using the assumed number of unique SCS clients per month and published estimates for the monthly number of visits per client from Insite (Tyndall et al., 2006).

Results

In base-case analyses, the annual operating cost for a fixed site, standalone SCS that includes syringe services provision was $1,602,334, while the cost of operating a syringe service program only was $818,435 (Table 2, Fig. 1A). The hypothetical SCS that includes syringe services is estimated to prevent 1.9 deaths per year (Fig. 1B). This outcome was sensitive to the percentage of overdose deaths within 0.25 miles of the SCS, the number of overdose deaths per year in Providence, and the mortality reduction within 0.25 miles of the SCS (1.5 to 2.3 deaths prevented annually). Additionally, each year, the SCS would prevent 261 ambulance runs, 244 ED visits, and 117 inpatient hospitalizations for emergency overdose care. In total, accounting for the annual operating costs of the SCS as well as short-term medical costs of emergency overdose care (i.e., ambulance runs, ED visits, and inpatient hospitalizations), the SCS would be cost saving; specifically, the total short-term savings would be approximately $1,104,454 annually.

Table 2.

Base-case analysis results.

Outcome	Scenario with SCS including syringe services	Scenario with syringe service program only	Difference
Overdose-related health outcomes
Overdoses within 0.25 miles of SCS	578.0	578.0	0.0
Overdose deaths within 0.25 miles of SCS	5.5	7.4	−1.9
Ambulance runs*	4.6	265.9	−261.3
ED visits*	4.6	248.5	−244.0
Inpatient hospitalizations*	2.2	119.3	−117.1
Total short-term costs	$1,637,525	$2,741,979	−$1,104,454†
Annual facility costs	$1,602,334	$818,435	$783,899
Upfront loan	$302,000	–	–
Loan annual payment	$33,158	–	–
Operating cost	$1,569,176	–	–
Ambulance run costs*	$2126	$123,789	−$121,663†
ED visit costs*	$15,757	$857,682	−$841,924†
Inpatient hospitalization costs*	$17,308	$942,073	−$924,765†
Cost-benefit	The addition of a SCS prevents adverse overdose-related health outcomes and is cost-saving.

Abbreviations: ED, emergency department; SCS, supervised consumption site.

^*Resulting from injections that would occur at the SCS, if available.

^†A negative cost difference represents savings.

Fig. 1.

Base-case analysis overdose-related health outcome results (A) and short-term cost results (B).

Abbreviations: ED, emergency department; SCS, supervised consumption site.

* Within 0.25 miles of SCS.

^† Resulting from injections that would occur at the SCS, if available.

In sensitivity analyses varying input parameters across plausible ranges to evaluate the impact on short-term cost savings per overdose death prevented, we found that the most influential parameters were the percentage of injections resulting in overdose, the total annual injections at the SCS, the percentage of overdoses outside of the SCS that result in an ED visit, the reduction in overdose mortality within 0.25 miles of the SCS, the SCS annual operating costs, and the cost-of-living ratio comparing Providence to Vancouver (Fig. 2). Across one-way sensitivity analyses, for each overdose death prevented, society would save a maximum of $837,512 and a minimum of $109,813 in the short term. The most influential parameters when evaluating the impact on short-term cost savings per ambulance run, ED visit, and inpatient hospitalization prevented are included in the list above (Supplemental Material, Figs. S1-S3).

Fig. 2.

Sensitivity analysis results, short-term cost savings per overdose death prevented.

Abbreviations: ED, emergency department; OD, overdose; SCS, supervised consumption site.

In scenario analyses utilizing our base-case parameters and varying only the number of unique SCS clients per month, the difference in total short-term cost savings ranged from $176,462 (with 200 unique clients per month) to $8,992,390 (with 2100 unique clients per month similar to Insite), comparing the scenario with a SCS including syringe services provision to the scenario with a syringe service program only (Table 3).

Table 3.

Scenario analyses varying only the number of unique SCS clients per month.

Number of unique SCS clients per month	Total annual injections in SCS*	Difference in total short-term costs comparing the SCS including syringe services to the syringe service program only
200	30,420	−$176,462
300	45,630	−$640,458
400†	60,840	−$1,104,454
500	76,050	−$1,568,450
750	114,075	−$2,728,441
1000	152,100	−$3,888,432
1500	228,150	−$6,208,413
2000	304,200	−$8,528,394
2100‡	319,410	−$8,992,390

Abbreviations: SCS, supervised consumption site.

^*Calculated using the number of unique SCS clients per month and published estimates for the monthly number of visits monthly per client from Insite (Tyndall et al., 2006).

^†Base-case analysis.

^‡Unique number of SCS clients per month at Insite (Milloy et al., 2008).

Discussion

In our primary analysis, we found that a hypothetical SCS in Providence that includes syringe services provision and serves 400 clients per month would prevent approximately 2 deaths, 261 ambulance runs, 244 ED visits, and 117 inpatient hospitalizations for emergency overdose care per year compared to a scenario that includes a syringe service program only (i.e., the status quo). Additionally, when only accounting for annual facility costs and short-term costs of emergency overdose care (i.e., ambulance runs, ED visits, and inpatient hospitalizations), we found that the SCS would be cost saving overall compared to the syringe service program only, with an estimated $1,104,454 saved per year from a modified societal perspective. In contrast, the Rhode Island Office of Regulatory Reform generally considers it reasonable to spend up to $9,100,000 to prevent one death (Rhode Island Office of Management and Budget, 2015), which further highlights the value of SCS in Rhode Island.

Our results for this small urban context are generally consistent with prior studies of the costs and benefits of hypothetical SCSs in other large urban centers of the United States. In the ICER/CHOICE study of hypothetical SCSs in six cities in the United States, compared to a syringe service program only, adding a standalone, fixed-site SCS (like Insite) was estimated to prevent 3 overdose deaths per year in each of Boston and Seattle, 4 overdose deaths per year in San Francisco, 6 overdose deaths per year in Atlanta, 9 overdose deaths per year in Baltimore, and 15 overdose deaths per year in Philadelphia. SCSs were consistently found to be cost-saving for all six cities, with short-term annual savings ranging from $3,623,000 for Atlanta to $4,199,000 for Seattle (Armbrecht et al., 2020). Separate studies using different models found similar results for hypothetical SCSs like Insite in Denver (2.8 overdose deaths prevented and roughly $6,900,000 saved per year) (Irwin et al., 2019), San Francisco (0.24 overdose deaths prevented and $3,500,000 saved per year) (Irwin et al., 2017a), and Baltimore (5.9 overdose deaths prevented and $7,800,000 saved per year) (Irwin et al., 2017b). One study of a hypothetical SCS in Seattle considered an estimated budget and smaller volume of clients tailored specifically to their hypothetical pilot site and found that the SCS would prevent 6 overdose deaths and save $3,933,687 per year (Hood et al., 2019). Our estimates for the short-term annual cost savings were lower than those for other cities, likely because we conservatively assumed similar SCS operating costs to those of Insite despite the SCS serving substantially fewer clients (400 vs. 2100 per month). In our scenario analyses varying the number of unique SCS clients per month but retaining the assumption that SCS operating costs would be similar to Insite, the difference in total short-term cost savings ranged from $176,462 (with 200 unique clients per month) to $8,992,390 (with 2100 unique clients per month similar to Insite). These results demonstrate that a SCS serving a larger number of clients will avert more emergency overdose care services and, thus, lead to greater savings. Nonetheless, our finding that a hypothetical SCS in a smaller urban setting with the full operating cost of Insite but serving more than 80% fewer clients (400 vs. 2100 per month) was still cost saving is novel. Our results may be useful for professionals in several other countries that are considering opening sites in smaller urban settings (e.g., in Ireland, Mexico, and Portugal (Harm Reduction International, 2019)). Our methods may also be modified and used to inform program planning in these settings.

Importantly, our analyses incorporated only the annual facility costs and short-term costs of emergency overdose care over a one-year period. There would likely be additional lifetime medical costs saved by society due to the HIV and HCV infections prevented by the SCS. However, our analysis did not include these benefits due to the short (i.e., one-year) time horizon. Other studies have found that a reduction in HIV and HCV treatment costs due to prevention of HIV and HCV infections is among the largest contributors to total savings over a longer time horizon (Irwin et al., 2017a,b), suggesting the estimated savings in our study is quite conservative. For example, if we assumed that people who inject drugs in Providence and acquire HIV have similar care-seeking behavior and life expectancy to other people living with HIV in the United States, the lifetime medical costs saved by preventing one HIV infection would be approximately $261,675 (in 2020 dollars) (Schackman et al., 2015). Similarly, for each course of HCV treatment potentially avoided by preventing one HCV infection, society would save approximately $38,552 (in 2020 dollars) (Jozaghi & Vancouver Area Network of Drug Users, 2014).

Given that the SCS and syringe service program only scenarios considered in our analysis are hypothetical, there may be substantial uncertainty surrounding estimated outcomes due to parameter uncertainty. In sensitivity analyses varying each input parameter across plausible ranges to evaluate the impact on costs per adverse outcome prevented, we found that the most influential parameters were the percentage of injections resulting in overdose, the total annual injections at the SCS, the percentage of overdoses outside of the SCS that result in an ED visit, the reduction in overdose mortality within 0.25 miles of the SCS, the SCS annual operating costs, and the cost-of-living ratio comparing Providence to Vancouver, but none of them were able to alter the conclusion that the SCS would be cost-saving. Across these one-way sensitivity analyses, for each overdose death prevented, Providence would save a maximum of $837,512 and a minimum of $109,813 in the short term.

Importantly, the assumed reduction in overdose mortality within 0.25 miles of the SCS in our study (25.7%) was based on a single, high-quality study of Insite. Causal evidence of a population-level effect of SCS on overdose mortality is relatively limited, given the small number of SCS with robust evaluations and limitations of observational and simulation studies (Caulkins et al., 2019). Across two systematic reviews, of five total observational and simulation studies that evaluated the impact of SCS on overdose mortality, four suggested that SCS reduce overdose mortality while one found no impact (Kennedy et al., 2017; Levengood et al., 2021). When we varied our parameter for the reduction in overdose mortality within 0.25 miles of the SCS from 20.6% to 30.8%, the estimated number of overdose deaths prevented annually by the SCS ranged from 1.5 to 2.3.

Given the large impact of the estimated annual operating costs of the SCS on our findings, it is important to consider the plausible range of options for SCS implementation in Rhode Island. We intended to be conservative in our analysis by basing our modeled SCS costs on the full annual operating costs of Insite in Vancouver, despite expecting that a SCS in Rhode Island would likely serve fewer clients than Insite (400 vs. 2100 per month) due to the smaller population and lower population density of Providence and other municipalities. With similar services but fewer clients, a SCS in Rhode Island would likely have a lower annual operating budget by requiring fewer staff and supplies, as well as a smaller space. Thus, we expect that an actual SCS in Providence would be even more cost-saving than the scenario considered in our analysis. Similarly, depending on the service delivery model of potential SCSs in Rhode Island, additional costs may be saved. For example, leveraging existing infrastructure by co-locating a SCS in an existing organization or clinic may decrease annual operating costs for a SCS compared to what was considered in our analysis. Cost and health outcomes for other service delivery models, such as mobile or temporary (e.g., tent) SCSs, are less well understood. These models may be subject to different unit costs per client served, depending on the number of clients served and operational costs. The 2-year pilot program of SCSs in Rhode Island may provide an opportunity to evaluate the actual costs and benefits of a variety of service delivery models.

The location of the SCS is also likely to have large impact on the cost-effectiveness of the SCS. Our primary analysis considered a hypothetical location in Providence where 8% of the overdose deaths occur within 0.25 miles of the SCS (prior to the implementation of a SCS). Based on 2020 data, there are two locations in Providence where the greatest percentage (8.6%) of fatal overdoses occur within a 0.25 miles radius (Supplemental Material, Fig. S4). Other things being equal, locating the SCS in one of these areas is likely to be more cost-effective than locating the SCS in other parts of Providence. Additionally, a single organization opening more than one SCSs within Providence may incur lower unit costs per client served; this would be an interesting area for future research. Similar maps of overdose fatalities and EMS runs for suspected opioid overdoses may be useful for informing potential SCS implementation in Rhode Island and other settings.

Our analysis adapted a previously published model developed through a process that incorporated feedback from external reviewers and other stakeholders along with public review by a board of experts. We tailored a variety of city characteristics, outcome parameters, and cost parameters to Providence using real-world data, likely improving the utility of our results. Nonetheless, our analysis has important limitations. While many model parameters were estimated from local data or obtained from local stakeholders, others were extrapolated from national sources or experiences of SCS in large urban centers in settings outside of the United States. Given that Insite in Vancouver may present the only SCS in North America that has been extensively studied, many of our cost, outcome, and other parameter estimates were based on data from Insite. The findings from Insite/Vancouver may not always be gen-eralizable to the Rhode Island context, particularly due to the different health care and social systems. In particular, our estimate for the total annual injections at the SCS, which was one of the most influential parameters in one-way sensitivity analyses, was based in part on the percentage of Insite clients who attended the center at specific frequencies (e.g., 31.5% of clients attended 2-5 visits per month) (Tyndall et al., 2006). Service utilization at a SCS in Rhode Island may be meaningfully different and will depend on the trust built and effectiveness of engagement with people who use drugs. Additionally, as described above, our estimate of the SCS annual operating cost was based on the operating budget of Insite, which is likely an overestimate of the costs for the smaller facility in Rhode Island that was considered in our analysis. We utilized multiple data sources to inform our estimate of the probability of overdose per injection, which was another highly influential parameter in our sensitivity analyses. However, risk of overdose is likely to change over time and by location depending on the drug supply and use patterns. Importantly, none of these uncertainties were found to alter our study conclusions based on the sensitivity analyses.

There were also important factors that were not included in our analysis. As noted above, we did not account for potential costs and benefits associated with prevention of HIV and HCV infections due to the short (one-year) time horizon of our analysis. Additionally, the model did not incorporate the potential for drug checking and other SCS services to reduce the risk of overdose within and outside the SCS, the potential for SCSs to reduce risk of subcutaneous injection-related infections and endocarditis (which can lead to costly hospital stays if left untreated (Fleischauer et al., 2017)), nor the potential impact of the site on the surrounding community. Although some studies suggest that SCSs have community and public order benefits, such as a decrease in public injection without an increase in drug-related crime or public nuisance in the community (Kennedy et al., 2017; Levengood et al., 2021), these were not captured in our model. We also did not account for potential increased initiation of MOUD and other treatment and recovery modalities among SCS clients compared to the syringe services program only scenario. Previous studies have shown that frequent SCS utilization is associated with increased uptake of various treatment services (DeBeck et al., 2011; Wood et al., 2006). Because engagement in MOUD and other services have numerous health benefits, including a decreased risk of mortality (Larochelle et al., 2018), our cost benefit analysis is likely conservative. Finally, our study did not account for cost savings that may accrue as a result of reduced criminal justice involvement among clients of the SCS; this is an important area for future research. Importantly, the 2-year pilot program of SCSs in Rhode Island will provide an opportunity to collect additional cost, outcome, and other data that can improve future analyses and decision-making.

In conclusion, we found that a hypothetical SCS in Providence that includes syringe service provision would prevent nearly two deaths annually, as well as prevent a substantial number of costly emergency overdose care services (i.e., ambulance runs, ED visits, and inpatient hospitalizations), compared to the status quo which includes a syringe service program only. The SCS would save more than $1,100,000 annually, accounting only for annual facility costs and short-term costs of emergency overdose care. Especially important factors that will influence the costs and benefits of potential SCSs in Rhode Island include the service delivery model, location of the site, actual service utilization, and actual operational costs. Rigorous collection of detailed operational, cost, and health outcome data at SCSs in Rhode Island can inform quality improvement activities and improve our understanding of SCS impacts.