If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Bureau of Alcohol and Drug Use Prevention, Care and Treatment, New York City Department of Health and Mental Hygiene, 42-09 28th St, 19th Floor, CN-14, Queens, NY 11101, USA
Bureau of Alcohol and Drug Use Prevention, Care and Treatment, New York City Department of Health and Mental Hygiene, 42-09 28th St, 19th Floor, CN-14, Queens, NY 11101, USA
Bureau of Alcohol and Drug Use Prevention, Care and Treatment, New York City Department of Health and Mental Hygiene, 42-09 28th St, 19th Floor, CN-14, Queens, NY 11101, USA
Department of Healthcare Policy & Research, Weill Cornell Medical College, 425 E 61st Street, Suite 301, New York, NY 10065, USADepartment of Medicine, Weill Cornell Medical College, 1300 York Avenue Room A421, New York, NY 10065, USA
Centre for Urban Health Solutions, Li Ka Shing Knowledge Institute, Division of General Internal Medicine, St. Michael's Hospital, 209 Victoria Street, 3rd floor, Toronto, Ontario M5B 1T8, CanadaDepartment of Medicine and Institute of Health Policy, Management and Evaluation, University of Toronto, 155 College Street, Suite 425, Toronto, ON M5T 3M6, Canada
Bureau of Alcohol and Drug Use Prevention, Care and Treatment, New York City Department of Health and Mental Hygiene, 42-09 28th St, 19th Floor, CN-14, Queens, NY 11101, USA
Supervised injection facilities save healthcare costs from opioid overdoses.
•
Widespread opioid epidemics as in NYC need multiple supervised injection facilities.
Abstract
Background
The opioid epidemic in the United States has resulted in over 42,000 U.S. opioid overdose fatalities in 2016 alone. In New York City (NYC) opioid overdoses have reached a record high, increasing from 13.6 overdose deaths/100,000 to 19.9/100,000 from 2015 to 2016. Supervised injection facilities (SIFs) provide a hygienic, safe environment in which pre-obtained drugs can be consumed under clinical supervision to quickly reverse opioid overdoses. While SIFs have been implemented worldwide, none have been implemented to date in the United States. This study estimates the potential impact on opioid overdose fatalities and healthcare system costs of implementing SIFs in NYC.
Methods
A deterministic model was used to project the number of fatal opioid overdoses avoided by implementing SIFs in NYC. Model inputs were from 2015 to 2016 NYC provisional overdose data (N = 1852) and the literature. Healthcare utilization and costs were estimated for fatal overdoses that would have been avoided from implementing one or more SIFs.
Results
One optimally placed SIF is estimated to prevent 19–37 opioid overdose fatalities annually, representing a 6–12% decrease in opioid overdose mortality for that neighborhood; four optimally placed SIFs are estimated to prevent 68–131 opioid overdose fatalities. Opioid overdoses cost the NYC healthcare system an estimated $41 million per year for emergency medical services, emergency department visits, and hospitalizations. Implementing one SIF is estimated to save $0.8–$1.6 million, and four SIFs saves $2.9–$5.7 million in annual healthcare costs from opioid overdoses.
Conclusions
Implementing SIFs in NYC would save lives and healthcare system costs, although their overall impact may be limited depending on the geographic characteristic of the local opioid epidemic. In cities with geographically dispersed opioid epidemics such as NYC, multiple SIFs will be required to have a sizeable impact on the total number of opioid overdose fatalities occurring each year.
). In New York City (NYC), overdose death rates have been increasing since 2010, with the largest increase seen recently from 13.6 overdose deaths per 100,000 in 2015 to 19.9 per 100,000 in 2016 (
). Provisional 2016 data show that of the 1374 drug overdose fatalities in NYC, 82% involved opioids. Among the five boroughs (counties) that comprise NYC, the overdose death rates in Staten Island and the Bronx (31.8 and 28.1 overdose deaths per 100,000, respectively) approach those in the states with the highest drug overdose rates in the country (West Virginia, Ohio, and New Hampshire with 52.0, 39.1, and 39.0 deaths per 100,000, respectively) (
One potential strategy for preventing overdose fatalities is establishing supervised injection facilities (SIFs). SIFs provide a hygienic, safe environment in which pre-obtained drugs can be consumed under clinical supervision to quickly reverse opioid overdoses, provide medical care, deliver harm reduction services, and connect people who inject drugs (PWID) with substance use disorder (SUD) treatments (
). In North America, the first formalized SIF was established in 2003 in Vancouver, Canada. Studies conducted at Vancouver's SIF report reduced opioid overdose mortality (
Reduction in overdose mortality after the opening of North America’s first medically supervised safer injecting facility: A retrospective population-based study.
). Several studies show that implementing SIFs in different Canadian cities are cost-effective or cost saving by preventing new HIV and hepatitis C (HCV) infections (
Because the legal status of SIFs in the United States is uncertain, establishing a SIF requires policy makers to give careful consideration to their potential public health and healthcare system benefits. We estimated the potential impact on opioid overdose fatalities and healthcare system costs of implementing SIFs in NYC, which has had a geographically dispersed and rapidly increasing rate of fatal opioid overdoses.
2. Methods
We estimated the SIF impact on preventing opioid overdose fatalities in NYC using neighborhood-specific overdose mortality data aggregated to the ZIP code level to identify potential fatalities avoided. We used data derived from a survey of PWID in NYC, health care administrative data, and the literature to estimate SIF utilization. We limited our cost analysis to estimated overdose-related costs incurred by the healthcare system, and do not consider SIF operating costs that would be incurred because these costs have not yet been established for NYC. To estimate potential healthcare costs avoided, we first estimated emergency medical services (EMS) transports, emergency department (ED) visits, and hospitalizations associated with fatal and non-fatal opioid overdoses in NYC from healthcare administrative and syndromic data sources, and then applied standardized unit costs to these estimates. We then estimated the costs that would be avoided by establishing SIFs assuming 1) no onsite SIF overdose would require an EMS response or ED visit; or 2) EMS response and ED visit rates for onsite SIF opioid overdoses would be similar to those observed at the Vancouver SIF. In all scenarios we assumed no fatal overdoses would occur onsite at a SIF, consistent with experience observed to date in Vancouver and other SIFs located outside of North America (
). We received guidance on this analysis from a technical advisory group whose members were previously involved in evaluating current and proposed SIFs in Canada.
2.1 Data
The number of fatal opioid overdoses in 2015 and 2016 were from a provisional mortality dataset provided by the NYC Office of the Chief Medical Examiner and the NYC Department of Health and Mental Hygiene (DOHMH) Bureau of Vital Statistics. We used data from two years to provide more stable information at the neighborhood and ZIP code levels. These data include information on the type of opioid involved in overdose death and the ZIP code of overdose location. To estimate whether these fatal overdoses were potentially avoidable had they occurred at a SIF, we used data from the sources described below to estimate the proportion of overdoses that were due to injection drug use (IDU), the proportion of PWID primarily injecting in public spaces (considered most likely to use a SIF), the proportion of PWID willing to use SIFs, and the distance PWID are willing to travel (Table 1).
Table 1Data for estimates of number of opioid overdose fatalities prevented and healthcare costs from implementing supervised injection facilities in New York City.
Applied to public injectors; alternative scenario applies a public injector willingness to use of 0.8 and non-public injector willingness to use of 0.56 (DeBeck et al., 2012).
Note: OASAS=Office of Alcoholism and Substance Abuse Services; IDUCS=Injection Drug Users Health Alliance Citywide Study; SPARCS = New York Statewide Planning and Research Cooperative System; MEPS = Medical Expenditure Panel Survey.
a Public injectors = primary injection location is outside own home or the home of a friend or family member.
b Applied to public injectors; alternative scenario applies a public injector willingness to use of 0.8 and non-public injector willingness to use of 0.56 (
To estimate the proportion of opioid overdoses due to IDU, we used the average percentage of heroin and non-heroin opioid admissions among individuals who entered detoxification (crisis admissions) in NYC from the New York State Office of Alcoholism and Substance Abuse Services (OASAS). Estimates of the proportion of PWID who primarily inject in public (any location other than their own home or a friend's home) and willingness to travel to a SIF were derived from the Injection Drug Users Health Alliance Citywide Study (IDUCS) (
). The IDUCS survey collected information from clients of all 14 syringe exchange programs in NYC with a sample size of 814 for two combined years of data (June 2013–June 2015). To determine estimates of willingness to use a SIF, we conducted literature reviews of studies conducted in North America. We received feedback on the face validity of assumptions and assistance in identifying additional relevant sources from the technical advisory group.
For the cost analysis, a syndromic emergency room surveillance data set from 2015 was used to estimate the ratio of non-fatal opioid overdoses to fatal opioid overdoses and the proportion of non-fatal opioid overdoses treated at a hospital that resulted from an EMS call. ED visit and inpatient healthcare utilization were identified from the January–September 2015 New York Statewide Planning and Research Cooperative System (SPARCS) data set for NYC. SPARCS is a comprehensive all payer data reporting system that covers outpatient and inpatient discharges, including ED visits. We defined opioid overdoses in SPARCS using ICD9 code 965.0x, (poisoning by opiates and related narcotics) or 967.xx, 969.xx, 970.xx (any drug poisonings) combined with codes 304.0x, 304.7x, 305.5x (opioid use disorder). We selected codes that would result in few cases of false positives for an opioid overdose (high specificity), recognizing we would not capture all overdose hospitalizations and ED visits (low sensitivity) as described in the literature (
Assessing the accuracy of opioid overdose and poisoning codes in diagnostic information from electronic health records, claims data, and death records.
Use of diagnosis codes for detection of clinically significant opioid poisoning in the emergency department: A retrospective analysis of a surveillance case definition.
Costs were assigned to each outcome from the perspective of the NYC healthcare system. We used NYC area Medicare fee-for-service payment estimates as proxies for costs providers actually incur, consistent with other recent studies (
). Costs are understated in that they do not include physician costs, but this approach is consistent with recent poison control studies where physician costs were not included or reported separately (
), because no relevant local data were available. The mean Medicare reimbursement rate for an ED visit was estimated using data from the Medical Expenditure Panel Survey (
). For inpatient Medicare costs, data on opioid overdose diagnosis-related group (DRG) inpatient discharges in SPARCS were used to determine costs for both fatal and non-fatal opioid overdoses (
). Four ICD-9 codes (918, 917, 871, and 004) accounted for approximately 86% of all opioid overdose inpatient discharges in SPARCS and were used to estimate average inpatient Medicare fee-for-service costs in NYC (Appendix A). This study is not human subjects research because all input values were obtained from the literature or from the health department as summary values.
2.2 Analysis
2.2.1 Geographical distribution of opioid overdose fatalities
NYC is made up of five boroughs with 42 United Hospital Fund (UHF) neighborhoods (
). We selected UHF neighborhoods as the main unit of geographical analysis for identifying opioid overdose hotspots for SIF placement, with the advantages of being a measure frequently used by DOHMH for evaluation and having ZIP codes uniquely assigned to only one UHF neighborhood. The combined numbers of fatal opioid overdoses for 2015–2016 were mapped at the ZIP code and UHF neighborhood levels. UHF neighborhoods with a greater than average number of opioid overdose fatalities were chosen for further analysis as potential sites for SIF placement, resulting in 16 UHF neighborhoods being selected for further evaluation. These 16 UHF neighborhoods accounted for approximately 60% of the 1852 fatal opioid overdoses recorded in the provisional 2015–2016 overdose data.
All ZIP codes within each of the 16 UHF neighborhoods were ranked according to the number of opioid overdose fatalities. One SIF was optimally placed within each of the UHF neighborhoods, as determined by the number of opioid overdose fatalities within each ZIP code. Therefore, SIFs were placed close to the geographic centers of ZIP codes containing the most opioid overdose fatalities. SIF placements were implemented in ArcGIS, version 10.2.1 and were not constrained by actual geographical physical limitations such as highways or parks.
2.2.2 Projection of opioid overdose fatalities prevented
We calculated the number of fatal opioid overdoses that could have been avoided by establishing each SIF. This was based on willingness to travel estimates derived from IDUCS, including distance from the syringe exchange address to the client's ZIP code. We established concentric rings around each hypothetical SIF at different distances that represented different probabilities of traveling to a SIF, based on willingness to travel to a syringe exchange at the same distances, up to a maximum of 3 miles. Opioid overdose fatalities within each ring were then assigned a probability of being avoided (Appendix B). We assumed that if the individual who experienced the fatal overdose had been attending a SIF at that time, no overdose would have occurred or the individual would have experienced a non-fatal overdose reversed successfully at the SIF. We also assumed, as a simplification, that SIF access would be unconstrained by hours of operation or physical capacity.
We made a conservative base case assumption that the 61% of PWID in NYC who primarily inject in their home or the home of a friend or family member would not attend a SIF regularly and therefore would not have any overdoses avoided by attending a SIF. We apply a willingness to use a SIF of 86% from the literature to the remaining 39% of PWID (Table 1) (
We also evaluated an alternative scenario where we assumed 56% of PWID who primarily inject in their home or the home of a friend or family would be willing to use a SIF regularly and 80% of PWID who primarily inject elsewhere would be willing to use a SIF (
), substantially increasing the overall potential number of opioid overdoses avoided (Table 1). We consider this an upper range estimate of the potential impact of SIFs on the number of fatal opioid overdoses averted.
We applied these estimates using the following equation:
Potential fatal overdoses avoided
= Number of fatal overdoses
× Proportion of fatal overdoses due to IDU
× Proportion of IDU overdoses that occur outside the home
× Proportion of PWID willing to use a SIF
× Proportion of PWID willing to travel to a SIF at a given distance from the theoretical SIF
We compared the results of the base and alternative cases to a scenario based on the outcomes reported for the Vancouver SIF, which was associated with a 35% reduction in overdose fatality rates within 500 m of the SIF and no impact beyond that distance (
Reduction in overdose mortality after the opening of North America’s first medically supervised safer injecting facility: A retrospective population-based study.
). We assumed a 35% reduction in all opioid overdose fatalities within a half mile radius (approximately 800 m) around each hypothetical SIF given the estimated high willingness of PWID to travel to an SEP within half a mile (Appendix B).
2.2.3 Projection of healthcare costs avoided
A decision analytic framework was used to assign utilization of EMS, ED, and inpatient services and their respective costs avoided, taking into consideration both fatal and non-fatal opioid overdoses (Fig. 1). For fatal overdoses, we assumed that 25% of overdose fatalities die in the ED or inpatient setting as assessed from the provisional mortality data. We assumed that among all fatal opioid overdoses an EMS call also occurred 90% of the time, with the response for the other 10% coming directly from the Office of the Medical Examiner with no additional costs to the healthcare system. For fatal overdoses that occur in the ED or inpatient setting, we applied the utilization of health services by fatal overdoses in SPARCS.
Fig. 1Decision tree of healthcare services utilization from opioid overdoses in New York City.
For non-fatal overdoses, we used an estimate of 10 non-fatal opioid overdoses that result in receipt of EMS, ED, or inpatient services occurring for each fatal opioid overdose in NYC, based on the NYC DOHMH syndromic data. The proportion of non-fatal overdoses using EMS was also estimated from the 2015 syndromic data. For non-fatal ED utilization and proportion of ED to inpatient admissions, we used data from a chart review of 165 suspected overdoses in 2013 performed by DOHMH staff (unpublished data). The proportion of direct inpatient admissions were derived from SPARCS.
Healthcare system cost savings were assigned to each potential fatal and non-fatal overdose avoided by each hypothetical SIF according to the decision analytic framework. We present costs avoided by the healthcare system under the two different assumptions: 1) all opioid overdoses that occur at the SIF are avoided or require no additional healthcare services outside of the SIF, and 2) all opioid overdoses occur at the SIF but are non-fatal, 39% result in an EMS call, and 28% of the EMS calls result in a transfer to the hospital, as was observed at the Vancouver SIF (
Using 2015–2016 provisional mortality data, implementing one SIF in the neighborhood with the most opioid overdose fatalities could prevent an estimated 19 opioid overdose fatalities per year in the base case scenario and 37 opioid overdose fatalities in the alternative scenario, assuming no operational constraints on SIF hours or capacity (Table 2). This is within range of the number of opioid overdose fatalities estimated based on the Vancouver's SIF experience (28 per year). For an area defined by a 3-mile radius around the hypothetical SIF in this highest priority neighborhood, the estimates of 19 and 37 fatal opioid overdoses avoided represent a 2% and 4% decrease in total annual NYC fatal opioid overdoses (6% and 12% neighborhood decrease), respectively. Between 68 and 131 opioid overdose fatalities per year could be prevented if 4 SIFs were placed in the four neighborhoods with the most overdose fatalities (Table 2), representing 7% and 14% decreases of total average annual fatal opioid overdoses in NYC.
Table 2Projected Number of Fatal Opioid Overdoses Avoided Annually by Implementing a Supervised Injection Facility in Each of the Most Affected NYC Neighborhoods (Based on 2015–2016 Overdose Fatality Data).
Represents the opioid overdoses prevented based on Vancouver's percentage decrease in opioid overdoses within 500 m applied to a half mile radius of each hypothetical SIF in NYC. These numbers do not represent the actual number of overdoses prevented in Vancouver, which were reported to be 23 per year (Marshall et al., 2011).
1
19
19
37
37
24
2
18
37
35
73
31
3
15
53
29
102
37
4
15
68
29
131
46
5
15
82
28
160
16
6
11
93
21
181
18
7
10
104
20
201
26
8
10
114
20
221
14
9
10
123
19
240
20
10
8
131
15
255
10
11
7
138
13
267
21
12
6
144
12
280
21
13
5
149
10
290
7
14
4
153
9
298
9
15
4
157
8
306
10
16
4
161
7
313
3
a Represents the opioid overdoses prevented based on Vancouver's percentage decrease in opioid overdoses within 500 m applied to a half mile radius of each hypothetical SIF in NYC. These numbers do not represent the actual number of overdoses prevented in Vancouver, which were reported to be 23 per year (
Reduction in overdose mortality after the opening of North America’s first medically supervised safer injecting facility: A retrospective population-based study.
Table 3 presents estimates of current opioid overdose costs to the healthcare system and costs avoided from SIF implementation in the base case scenario. Opioid overdoses cost the NYC healthcare system an estimated $41 million per year for EMS calls, ED visits, and hospitalizations. Approximately $4 million of these costs are associated with fatal opioid overdoses (Appendix C). The average cost per opioid overdose is approximately $3980 ($3995 per non-fatal overdose; $3827 per fatal overdose).
Table 3Annual Base Case Costs to the Healthcare System: Current Costs and Projected Costs of implementing One and Four SIFs.
Health care services
Ambulance
ED visit
Inpatient discharge
Total Cost per year
Healthcare costs averted per year
Current Costs
Non-Fatal
$3,271,600
$4,594,100
$29,127,700
$36,993,400
–
Fatal
$322,000
$4600
$3,217,100
$3,543,700
–
Total
$3,593,600
$4,598,700
$32,344,800
$40,537,100
–
Costs with 1 SIF, avoiding all healthcare costs from overdoses prevented
One optimally placed SIF could save $831,700 in healthcare system costs annually if all onsite opioid overdoses were avoided or required no additional healthcare services offsite, according to our base case scenario for overdoses avoided (Table 3). If overdoses continued to occur onsite but EMS response and ED admission rates were similar to those observed in Vancouver, $703,200 in healthcare system costs would be avoided. If four SIFs were optimally placed, $2.2–$2.9 million in healthcare system costs would be saved ($553,700–$736,200 per SIF) under these two conditions. Healthcare costs saved are primarily from avoided hospitalizations (80% of savings), followed by avoided ED visits (11%), and avoided EMS transports (9%). Approximately 91% of savings are from non-fatal opioid overdoses avoided and 9% are from fatal opioid overdoses avoided. In the alternative scenario that includes SIF use by people who primarily inject in their home or friend's home, the higher number of fatal overdoses avoided results in healthcare costs saved that are almost double the base case scenario for one SIF ($1.4–$1.6 million) and four SIFs ($4.0–$5.7 million) (Appendix C).
4. Discussion
The implementation of one SIF is estimated to prevent 19 to 37 opioid overdose fatalities per year in NYC, assuming optimal placement and no operational constraints on SIF hours or capacity. These estimates are consistent with the estimated 28 opioid overdose fatalities that were prevented based at Vancouver's SIF. We estimate a greater impact of one SIF on preventing overdose fatalities than studies that estimate 0.24 and 5.9 lives saved per year in San Francisco and Baltimore, respectively (
). Differences in these estimates are likely related to NYC having a larger population and number of overdoses than San Francisco and Baltimore. Moreover, given the geographically dispersed opioid epidemic in NYC, our study demonstrates the potential value of implementing multiple SIFs in different high-impact neighborhoods. We estimate that between 68 and 131 opioid overdose fatalities per year could be prevented if 4 SIFs were placed in the four neighborhoods with the most overdose fatalities.
Opioid overdoses cost the NYC healthcare system an estimated $41 million per year in EMS calls, ED visits, and hospitalizations. If one to four SIFs were optimally placed, we estimate between $831,700 and $2.9 million in costs would be saved by the healthcare system if all opioid overdose-related healthcare costs were avoided. If healthcare system costs accrued similarly to those associated with the SIF in Vancouver, $703,200 to $2.2 million in health care costs would be avoided. Our estimates fall within the wide range of savings from averted overdose fatalities projected in other US studies, which estimate $284,000 and $3.0 million of societal savings for San Francisco and Baltimore respectively (
). Our estimates of healthcare costs averted, however, do not include savings from avoiding lost wages and productivity that result from preventing overdose fatalities as other studies. We also do not include savings from other non-overdose health outcomes (HIV and HCV prevention, increased entry into SUD treatment, and improved wound care) and non-healthcare cost savings such as from reduced crime, which are considered in these and other US-based studies (
). We are unable to estimate the impact on cost savings of including SIF operating costs because these cost estimates are not available for NYC; the two published estimated annual operating cost of a stand-alone SIF in the US are $1.8 and $2.6 million (
), but operating costs may be lower if implemented in an SEP facility. We recommend future in-depth analysis of many of these additional economic benefits to estimate the net monetary benefit of establishing SIFs in NYC from multiple perspectives.
We used an innovative methodology to estimate the impact of SIFs in NYC that takes account of local geography and uses local data sources whenever possible. We also included different scenarios to account for uncertainty regarding SIF attendance by people who inject at their home or the home of a friend, as well as uncertainty about the ability of SIF personnel to avoid non-fatal overdoses or manage them onsite without additional assistance. Nevertheless, there are some limitations related to our approach. We relied on data from SUD treatment program admissions to determine the proportion of people who inject heroin and other opioids. Because of unreliable local cost estimates, a national ED cost was estimated instead of a NYC-specific cost for opioid overdoses, which may underestimate actual ED costs for NYC. NYC-specific estimates derived from SPARCS were based on claims recorded as being associated with an ED visit or hospitalization for an opioid overdose. These claims do not represent all opioid-related ED and hospital admissions due to well-known challenges with overdose identification in health care claims data (
Assessing the accuracy of opioid overdose and poisoning codes in diagnostic information from electronic health records, claims data, and death records.
Use of diagnosis codes for detection of clinically significant opioid poisoning in the emergency department: A retrospective analysis of a surveillance case definition.
), which is why we also used local syndromic data whenever possible.
The estimated combined impact of implementing multiple SIFs may be overstated for some combinations of neighborhoods, because the impact of each SIF was estimated independently without accounting for overlapping impact of SIFs implemented in adjacent neighborhoods. We also assumed optimal placement and no operational constraints on SIF hours or capacity, suggesting that our estimates represent maximum impact. On the other hand, this methodology does not directly account for additional community effects in preventing opioid overdose fatalities outside of the SIF due to naloxone distribution and overdose prevention education provided to PWID who attend the SIF. We anticipate that SIFs will closely collaborate with local SUD treatment programs, SEPs, and other opioid overdose prevention programs by creating referral networks that link people to additional services that prevent overdose fatalities. Our Vancouver-based estimates implicitly account for these additional effects by using Vancouver's actual measure of impact, but the estimates are limited by only applying to a small area around the SIF. While we adjusted the Vancouver impact range to apply to a slightly larger catchment area, these estimates are still limited by potential differences in willingness to travel to SIFs versus SEPs. We did not estimate the impact of substituting non-fatal for fatal overdoses, which could lead to an overstatement of benefits because people who survive a non-fatal overdose are at high risk for repeat overdose. Our estimate of a SIF's impact on reducing opioid overdose fatalities may be conservative, however, given current trends related to fentanyl use in NYC. The proportion of overdose fatalities related to fentanyl in NYC has increased from less than 10% in 2014 to 44% in 2016 (
). SIF staff are more likely to respond more quickly and effectively than community members, since fentanyl-related overdose may require timely administration and higher naloxone doses to avoid a fatality (
). SIFs may also play a role in preventing non-opioid overdoses by promoting safer injection practices or non-injection use if they become supervised consumption facilities where other forms of use are allowed. While our analysis does not account for this additional impact on preventing non-opioid overdoses, future research should consider these scenarios, particularly in regions with high stimulant use.
5. Conclusions
This study provides evidence that establishing SIFs in NYC can save lives and avert healthcare system costs. SIFs can be an innovative and effective strategy to prevent opioid overdose fatalities, although their overall impact may be limited depending on the geographic characteristic of the local opioid epidemic. Geographic hotspots of opioid overdoses may shift over time and require reassessment of locations or expansion through additional permanent or temporary sites (e.g., mobile units, pop ups within SEPs), although there are challenges to implementing temporary sites (
). In cities with geographically dispersed opioid epidemics such as NYC, multiple SIFs will be required to have a sizeable impact on the total number of opioid overdose fatalities occurring each year.
Funding
The project described was supported by 2017 New York City Council Executive Budget Funding. The contents of this publication are solely the responsibility of the authors and do not necessarily represent the official views of the New York City Department of Health and Mental Hygiene. The authors have no conflicts of interest to disclose.
Acknowledgements
We would like to thank Eva Enns, Bohdan Nosyk, Carole Strike, and Greg Zaric, who provided guidance as members of the technical advisory group for this economic analysis.
Appendix A. Healthcare service costs for inpatient services
The following tables show the weighted average cost for inpatient care when considering the leading or most expensive DRGs identified in SPARCS based on utilization by people with an opioid overdose.
Table A.1Non-fatal inpatient utilization and average Medicare payment rates.
Appendix B. Model inputs for SIF impact on opioid overdoses
Potential fatal overdoses avoided (assuming all fatal overdoses are avoided in a SIF).
= Number of fatal overdoses
× Proportion of IDU willing to travel to a SIF at a given distance from the theoretical SIF
× Proportion of IDU willing to use a SIF
× Proportion of fatal overdoses due to IDU
× Proportion of IDU overdoses that occur outside the home
I.
Number of Opioid overdoses/Willingness to travel:
For each SIF, the number of fatal opioid overdoses that may be prevented is partly influenced by the number of PWID who would have traveled to the SIF location. We used the Injection Drug Users Health Alliance Citywide Study (IDUCS) data set to estimate the distance between the address of the SEP used and the centroid of the ZIP code where the participant last slept. While there were PWID who reported traveling further than 3 miles to the SEP, we assumed that PWID would not travel more than 3 miles to a SIF because of more frequent expected use of a SIF versus an SEP. The median distance traveled to an SEP calculated from IDUCS data is similar to that reported in the literature.
We established concentric rings around each hypothetical SIF at different distances (0.25, 0.5, 0.75, 1.0, 1.5, 2, 2.5, 3 miles) that represented different probabilities of traveling to a SIF that were reduced as distances from the SIF increased. The proportions of PWID who would travel to a SIF at each distance were estimated directly from the IDUCS data, and applied as cumulative proportions (see below).
Fig. B1Cumulative Proportion Estimated Willing to Travel to a SIF or SEP (Source: IDUCS Survey).
Based on the distribution of willingness to travel at each distance from the SIF and the number of opioid overdose fatalities around each theoretical SIF, the proportion of PWID who experienced a fatal opioid overdose and would have traveled to a SIF is then estimated for each individual SIF. In order to estimate the maximum potential impact of each SIF, we include opioid overdoses in contiguous UHFs and do not consider whether or not SIFs have been established in adjacent UHFs when calculating the potential number of fatal opioid overdoses that may be reached by each SIF within a 1.5-mile radius.
II.
Willingness to use a SIF:
We conducted a literature review regarding the willingness of PWID to use a SIF because we did not have any SIF-specific data from representative samples of PWID in NYC. Since SIF knowledge and education has improved over time, we focused on the most recent studies to estimate the percentage of PWID who would be willing to use a SIF (
Proportion of overdoses that result from injection drug use:
Because SIFs currently under consideration for NYC are spaces for PWID only, we limited the potential number of fatal opioid overdoses avoided to those associated with injection drug use. We estimate the proportion of heroin and non-heroin opioid users who inject from the OASAS crisis admissions data set for NYC, and also conducted a literature review in which we found results consistent with the estimates from the OASAS data.
IV.
Proportion of overdoses that result from injection in public spaces (outside the home)
In our base case scenario, we focused on individuals who primarily do not inject at home or the home of friends or family (i.e., “public injectors”), assuming this population would be most likely to use a SIF regularly for their injections. Given this assumption, we used IDUCS data to determine the proportion of PWID who fit this definition. We also conducted a literature review to assess the range of estimates for proportion of people who primarily inject in public spaces, but found the definitions of “public injection” varied across studies.
Appendix C. Alternative scenario cost estimates
Table C.1Annual alternative case costs to the healthcare system: current costs and projected costs of Implementing one and four SIFs.
Health care services
Ambulance
ED visit
Inpatient discharge
Total cost per year
Healthcare costs averted per year
Current Costs
Non-Fatal
$3,271,600
$4,594,100
$29,127,700
$36,993,400
–
Fatal
$322,000
$4600
$3,217,100
$3,543,700
–
Total
$3,593,600
$4,598,700
$32,344,800
$40,537,100
–
Costs with 1 SIF, avoiding all healthcare costs from overdoses prevented
Assessing the accuracy of opioid overdose and poisoning codes in diagnostic information from electronic health records, claims data, and death records.
Reduction in overdose mortality after the opening of North America’s first medically supervised safer injecting facility: A retrospective population-based study.
Use of diagnosis codes for detection of clinically significant opioid poisoning in the emergency department: A retrospective analysis of a surveillance case definition.
30080-7&id=gr1.jpg){kind=link}
30080-7&id=gr2.jpg){kind=link}