A systematic review of macro-, meso, and micro-level harm reduction interventions addressing the U.S. opioid overdose epidemic

Abstract

Background:

This systematic review aimed to 1) identify the range of opioid harm reduction interventions implemented at macro-, meso-, and micro-levels in the United States, and 2) summarize the outcomes of these interventions.

Methods:

We conducted a systematic review of academic literature published between 2011-2023 following PRISMA guidelines. Articles were excluded if they reported on research that was not specific to opioids, did not report the effects of an intervention, or focused on a medical treatment for opioid use disorder. Two coders independently extracted data and reconciled discrepancies prior to narrative synthesis.

Results:

Of 6,198 articles initially identified, 36 met inclusion criteria across macro (n=7), meso (n=8), and micro (n=21) domains. Positive evidence for micro- and meso-level interventions is largely consistent, whereas evidence for macro-level interventions is mixed. Among micro- and meso-level interventions, supply distribution interventions were most effective in increasing safe use knowledge and behaviors among people who use drugs.

Discussion:

Most harm reduction interventions demonstrate moderate to strong evidence of effectiveness for addressing the opioid overdose epidemic across domains. Findings revealed a lack of multi-level interventions and a lack of culturally relevant interventions that prioritize Black and Brown communities disproportionately impacted during the opioid overdose epidemic’s latter phases.

Main text introduction

The scope of the U.S. opioid overdose epidemic

Opioid-involved drug overdose deaths have skyrocketed over the past two decades, devastating the United States more than any other country (Yeo, Johnson & Heng, 2022). Between 1999 and 2019, nearly 500,000 individuals died of a drug overdose (Centers for Disease Control and Prevention [CDC], 2020a). More recently, between June 2022 and June 2023, nearly 107,000 individuals died of an opioid-involved overdose, and rates are only expected to worsen (Ahmad et al., 2023).

The opioid overdose epidemic has been particularly challenging to address because there have been three distinct, yet interconnected, phases of mortality, each related to a different class of opioids: prescription opioid pills, heroin, and synthetic opioids (especially fentanyl) (CDC, 2019a; Ciccarone, 2019). Recent data suggest that a fourth wave of the opioid overdose epidemic has emerged characterized by concurrent opioid and psychostimulant use (Ciccarone, 2021; Jenkins, 2021). The COVID-19 pandemic has further complicated this wave due to widespread quarantines and program closures that decreased access to naloxone and treatment services and social (physical) distancing that has forced individuals to use in isolation, thus increasing their overdose risk (CDC, 2020a; CDC, 2020b; Linas et al., 2021). HIV (Alpren et al., 2020) and HCV (Powell, Alpert, & Pacula, 2019) prevalence rates among people who use drugs (PWUD) have also continued to increase, resulting in compounding public health crises (Perlman & Jordan, 2018).

Jalali and colleagues’ (2020) Socio-Ecological Framework of the Opioid Crisis offers a robust contextual framework for understanding the multi-level factors contributing to opioid misuse across the four waves of the opioid overdose epidemic. Individual-level (micro) contributors include sociodemographic, biological, and psychological factors. Interpersonal-level (micro) contributors highlight the role of friends, family, and co-workers in influencing an individual’s attitudes, beliefs, and behaviors related to opioid access, initiation, and misuse. Community-level (meso) contributors include geography and treatment accessibility, among others. Finally, societal-level (macro) contributors include opioid supply and demand, economic conditions, government regulations, law enforcement (e.g., arrest rates), and discrimination, prejudice, and stigma.

Opioid overdose epidemic response efforts

A range of interventions spanning socio-ecological levels have been implemented to address the opioid overdose epidemic. Micro-level interventions, which focus on improving individuals’ physical and behavioral health outcomes, include medications for treating opioid addiction (Kampman & Jarvis, 2015) and psychosocial treatment (Bergman et al., 2019), with the standard of care being a combination of the two (Dugosh et al., 2016). Meso-level interventions, which focus on enhancing group- or community-level outcomes, are diverse and include hospital-based interventions (Weimer, Morford, & Donroe, 2019), academic detailing (physician education to aid in making evidence-based treatment decisions) (Meisenberg et al., 2018), and peer recovery support services (Gormley et al., 2021). Macro-level interventions, which focus on systemic and structural change, have primarily consisted of prescription drug monitoring programs (Fink et al., 2018) but also include laws that expand Medicaid coverage and treatment access for opioid use disorder (OUD) (Sledge et al., 2022).

While these interventions are invaluable in improving health outcomes for individuals with OUD and addressing the opioid overdose epidemic at large, they fail to capture sub-groups that do not engage with the healthcare system or lack access to it. For instance, people experiencing homelessness face significant barriers to obtaining care in traditional healthcare venues, including stigma from healthcare personnel (Omerov et al., 2020) and logistical challenges that disrupt steady access (Brallier, Southworth, & Ryan, 2019). People who inject drugs also consistently experience barriers in accessing and utilizing care, often due to comorbid conditions and stigma related to health status and injection drug use (Lekas, Siegel, & Leider, 2011). People recently released from correctional settings (Csete, 2019) and pregnant and parenting women with OUD (Angelotta et al., 2016) also experience a host of barriers related to accessing and engaging with the healthcare system, such as limited treatment options, interruptions in treatment, and stigma.

Harm reduction’s role in addressing the opioid overdose epidemic

As drug overdose deaths continue to rise, national efforts have increasingly integrated harm reduction strategies into response efforts. Harm reduction is a transformative, three-pronged approach that is simultaneously 1) a movement for social justice that prioritizes PWUD, 2) a range of community-centered services aimed at reducing drug-related harms, and 3) an approach to service provision that centers pragmatism, individual autonomy, and “any positive change” (Drucker et al., 2016; Hawk et al., 2017; National Harm Reduction Coalition, 2020). Harm reduction also emphasizes innovation in drug policy and involving direct recipients of harm reduction services in developing interventions and programs (Allman et al., 2007). Taken together, harm reduction aims to reduce the negative consequences associated with drug use at systemic, community, and individual levels without requiring that those who use drugs reduce or eliminate drug use. As such, this approach includes a broad spectrum of strategies and interventions to reduce the potential harms associated with drug use and other behaviors that may negatively affect health (Drucker et al., 2016; Hawk et al., 2017; National Harm Reduction Coalition, 2020).

Despite long-standing tensions with those who advocate for abstinence-based approaches only, integrating harm reduction into OUD care has facilitated more comprehensive service delivery, particularly for communities with unique considerations. For example, harm reduction interventions intended to reduce opioid use-related harms can include overdose education and naloxone distribution (OEND) programs for individuals as they leave carceral settings (Wakeman et al., 2009), fentanyl test strips for young PWUD (Krieger et al., 2018), and laws that permit the widespread dissemination of naloxone at no cost, often referred to as Naloxone Access Laws (McClellan et al., 2018).

Study aims

Though the opioid overdose epidemic is a major public health concern in several countries, the United States has the highest number of overdose deaths per capita and lags behind in implementing harm reduction services. The various harm reduction interventions that have been implemented across the U.S. differ widely in terms of their implementation settings, goals, and whether they were designed to achieve micro-, meso-, or macro-level effects; thus, a synthesis of these interventions and their effectiveness is needed. Therefore, this systematic review aims to 1) identify the range of harm reduction interventions implemented to address the U.S. opioid overdose epidemic at macro, meso, and micro levels, and 2) summarize the clinical and public health outcomes of these interventions relative to the level at which they were implemented.

Materials and methods

Search strategy

Following consultation with a university librarian, a literature search was conducted that culminated in the Summer of 2023. The following five databases were systematically searched: CINAHL, MEDLINE, PsycINFO, PubMed, and SocINDEX. The search string consisted of the following: ((opioid OR opiate) n1 (use OR misuse OR abuse OR addiction OR overdose OR death OR fatality)) OR SU (opioid OR opiates OR heroin OR fentanyl) OR intervention OR treatment OR therap* OR naloxone OR buprenorphine OR methadone OR needle OR syringe OR program* OR services OR supports OR education OR awareness OR policy OR policies OR legislation OR laws OR legal OR courts OR “good samaritan” OR bystander OR “by stander” and harm n1 reduc.* Since the search focus was on harm reduction interventions as a response to the recent opioid overdose epidemic, a date limiter was set to exclude publications prior to 2011 and after July 2023. Hand searching was not conducted.

Eligibility criteria

For the purposes of this systematic review, intervention was defined as a structured program or other mechanism intended to facilitate change and affect an outcome. Although there is no single definition of harm reduction, in this review it is defined as an approach that aims to reduce the negative consequences and accompanying harms associated with drug (mis)use without necessarily reducing or eliminating drug use (Drucker et al., 2016; Hawk et al., 2017; Owczarzak et al., 2020). The search sought to identify harm reduction interventions at macro, meso, and micro levels. Interventions were categorized according to the unit of analysis (i.e., level of outcome) reported.

The articles included in this systematic review met the following inclusion criteria: (1) were written in English, (2) were published between 2011 and 2023, and (3) reported on opioid-specific harm reduction intervention research conducted in the U.S. Articles were excluded if: (1) written in a language other than English, (2) the intervention(s) occurred outside the U.S., (3) the research was published before 2011 or after July 2023 (4) the research was not specific to opioids, (5) the research was not on an intervention (e.g., reported solely on opioid use prevalence, patterns), and (6) the research focused on medical treatment for OUD. Commentaries, dissertations, systematic reviews/meta-analyses, and non-peer-reviewed articles were also excluded.

Data screening, extraction, and analysis

The first author (KM) implemented the search strategy and removed duplicates using Zotero. KM then independently conducted title and abstract screening according to eligibility criteria using Rayyan and completed a full-text review of articles that met inclusion criteria.

Two authors (KM, JS) independently reviewed and extracted data from included articles on intervention characteristics (e.g., type, length, purpose, intended outcomes), intervention level (macro, meso, micro), study design and methods (e.g., cross-sectional, randomized controlled trial, quasi-experimental), study sample (e.g., individual-level sociodemographic or community characteristics), setting (e.g., hospitals, community-based organizations), and results (e.g., correlations, [non]statistical significance, qualitative themes) using an a priori developed data extraction spreadsheet. Coding discrepancies were discussed and resolved through consensus. Following data extraction, the data were compared and contrasted within and across each of the three domains (macro, meso, and micro). The first and second authors independently assessed the risk of bias for the included studies using Higgins and colleagues’ (2011) guidelines. Risk of bias was assessed across five domains: selection, performance, attrition, reporting, and detection, and assessed as either 1) not applicable, 2) low, 3) high, or 4) unclear risk of bias. Table 1 summarizes the risk of bias of included studies.

Table 1.

Risk of bias summary for included studies.

Study	Random Sequence Generation (selection bias)	Allocation Concealment (selection bias)	Blinding of Participants & Personnel (performance bias)	Blinding of Outcome Assessment (detection bias)	Incomplete Outcome Data (attrition bias)	Selective Reporting (reporting bias)
Macro-level Interventions
Atkins et al. (2019)	/	/	/	/	/	−
Drake et al. (2021)	/	/	/	/	/	−
Kelly & Vuolo (2022)						−
Lee et al. (2021)	/	/	/	/	/	−
McClellan et al. (2018)	/	/	/	/	/	−
Nguyen et al. (2018)	/	/	/	/	/	−
Townsend et al. (2022)	/	/	/	/	/	−
Meso-level Interventions
Alexandridis et al (2018)	/	/	−	−	/	−
Allen et al. (2022)	/	/	/	/	/	−
Bornstein et al. (2020)	/	/	/	/	/	+
León et al. (2018)	/	/	+	+	/	+
Naumann et al. (2019)	/	/	−	−	/	−
Rowe et al. (2019)	/	/	/	/	/	−
Walley et al. (2013)	/	/	/	/	/	−
Xuan et al. (2023)	/	/	/	/	/	−
Micro-level Interventions
Arendt (2023)	/	/	−	−	+	−
Behar et al. (2015)	+	?	−	−	−	−
Clarke et al. (2016)	+	+	−	−	?	−
Dahlem et al. (2017)	/	/	+	−	+	+
Dunn et al. (2013)	?	?	−	−	−	−
Fitzpatrick et al., (2022)	−	−	+	−	−	−
Goldman et al. (2019)	?	?	−	−	+	−
Han et al. (2017)	/	/	+	−	?	+
Jones et al. (2014)	+	−	−	−	?	−
Jones et al. (2022)	−	?	?	?	+	−
Katzman et al. (2018)	/	/	−	−	−	+
Katzman et al. (2020)	/	/	+	−	+	+
Krieger et al. (2018)	+	?	−	−	−	−
Lima et al (2022)	/	/	+	+	+	−
Park et al. (2020)	+	−	+	−	+	+
Park et al. (2021)	?	?	−	−	?	−
Phillips et al. (2012)	−	?	?	?	−	−
Pourtaher et al. (2022)	/	/	/	/	/	−
Sisson et al. (2023)	−	−	+	+	−	+
Winhusen et al. (2020)	−	?	?	?	−	−
Yang et al. (2021)	/	/	−	−	?	−

Note. (+) indicates high risk of bias; (−) indicates low risk of bias; (?) indicates unclear risk of bias; (/) indicates not applicable.

Results

The database search identified 6,198 articles. After removing 3,348 duplicates, title and abstract screening was conducted on the remaining 2,850 articles using Rayyan. Of them, 2,754 articles were excluded based on the aforementioned criteria. Full-text review was conducted on the remaining 96 articles. Though several of these articles were topically relevant, 60 were ultimately excluded due to their focus on prescribing/dispensing medications (e.g., Castillo-Carniglia et al., 2019), a hypothetical program (e.g., Hood et al., 2019), or a program description not reporting outcomes (e.g., Avetian et al., 2018). The remaining 36 articles met inclusion criteria across macro (n=7), meso (n=8), and micro (n=21) domains. Figure 1 details the review process.

Figure 1.

PRISMA 2020 flow diagram.

From: Page, M. J., McKenzie, J. E., Bossuyt, P. M., Boutron, I., Hoffmann, T. C., Mulrow, C. D., … & Moher, D. (2021). The PRISMA 2020 statement: An updated guideline for reporting systematic reviews. International Journal of Surgery, 88, 105906.

Synthesis of evidence on macro interventions

Seven articles on macro harm reduction interventions met review inclusion criteria (see Table 2). Studies examined between one and six laws, with the most common being Good Samaritan Laws (GSLs) (n=5) and Naloxone Access Laws (NALs) (n=3). All studies were longitudinal, examining law implementation and impact across two to 17 years.

Table 2.

Summary of Macro Interventions, n=7

Authors	Macro Intervention	Study Design	Study Sample and Time Period	Key Findings
Atkins et al. (2019)	GSL	Secondary analysis of cross-sectional data	n=40 states that have implemented GSLs, 1999-2016	No significant effects of GSLs on overdose mortality.
Drake et al. (2021)	Recreational cannabis laws	Secondary analysis of cross-sectional data	n=29 states that implemented RCLs, 2011-2017	Recreational cannabis laws associated with marginally significant decreases in opioid-related ED visits.
Kelly & Vuolo (2022)	NAL	Secondary analysis of cross-sectional data	n=50 states, 2004-2016	NALs not associated with decreased risk perceptions of heroin use among the general population or vulnerable subgroups.
Lee et al. (2021)	1) PDMP access laws 2) mandatory PDMP 3) prescription limit laws 4) pain clinic laws 5) GSL 6) NAL	Secondary analysis of cross-sectional data	n=50 states, 2007-2018	- PDMP access laws associated with increases in synthetic opioid and cocaine overdose deaths. - Mandatory PDMPs associated with decreases in natural opioid overdose deaths. - Prescription limit laws and pain clinic regulation laws associated with reduced opioid abuse and increase in MAT use. - GSLs and NALs associated with increases in overdose deaths.
McClellan et al. (2018)	1) NAL 2) GSL	Secondary analysis of cross-sectional data	n=50 states, 2000-2014	- NALs and GSLs associated with reduced overdose mortality. - NALs associated with reduced mortality among African Americans. - GSLs associated with reduced mortality among African Americans and Hispanics.
Nguyen et al. (2018)	GSL	Secondary analysis of cross-sectional data	n=270 hospitals in New York and New Jersey, 2010-2012	GSL significantly associated with increased ED visits and hospital admissions for heroin overdoses but not non-heroin overdoses.
Townsend et al. 2022	GSL	Secondary analysis of cross-sectional data	n=7 states 2015-2019	- No significant effects of GSL on overdose overall mortality. - No significant differences of GSL observed on overdose mortality by racial groups.

Abbreviations: ED = Emergency Department; GSL = Good Samaritan Law; NAL = Naloxone Access Law; PDMP = Prescription Drug Monitoring Program.

Only one study examined NALs alone. Kelly & Vuolo (2022) found that NALs were not associated with decreased risk perceptions of heroin use among the general population, vulnerable subgroups (e.g., young people), or diverse subpopulations (e.g., race/ethnicity). Of the five studies that examined GSLs, three examined these laws alone. Atkins and colleagues (2019) estimated the effect of GSLs on overdose mortality in 40 states from 1999-2016 controlling for other harm reduction and substance use prevention laws (e.g., prescription drug monitoring programs) and found that GSLs were not significantly associated with overdose mortality. Similarly, Townsend and colleagues (2022) estimated the effect of GSLs in seven states from 2015-2019 and found that GSLs did not have an effect on overdose mortality overall, nor when stratified by Black/White race/ethnicity. Nguyen and Parker (2018) examined the effects of New York’s GSL on emergency department visits and hospital admissions, finding that this law was significantly associated with increased emergency department visits and hospital admissions for heroin overdoses (but not non-heroin opioid overdoses).

Two of the four GSL studies examined this law in combination with other laws. Both examined mortality rates, but their findings differed. McClellan and colleagues (2018) examined GSLs and NALs in all 50 states between 2000-2014 and concluded that both types of laws were significantly associated with reduced overdose mortality (15% and 14%, respectively). NALs were associated with a 23% lower incidence of overdose deaths among African Americans, and GSLs were associated with reduced mortality among both African Americans and Hispanics (26% and 16%, respectively). Lee and colleagues (2021) examined outcomes related to six drug laws based on data from a commercially insured population in all 50 states from 2007-2018 (given the focus of this review, only findings related to harm reduction laws are discussed here, though findings for all six laws are detailed in Table 4). In contrast to McLellan and colleagues (2019), they found that GSLs and NALs were associated with increases in overdose deaths, a possible unintended consequence of supply-side laws limiting the availability of prescription opioids, and as a result, driving individuals to the illicit drug market.

Table 4.

Summary of Micro Interventions, n=21

Authors	Micro Intervention	Study Design	Study Setting and Sample	Key Findings
Arendt (2023)	Harm reduction dispensing machine	Prospective observational	Ohio n=637 PWUD 41% aged 35-44 50% male and female 90% White	- Dispensing machine distributed more harm reduction supplies than any other SSP in the county in the previous year. - Implementation of the dispensing machine associated with lower incidence of unintentional overdose death and HIV in the county.
Behar et al. (2015)	OEND	Quasi-experimental	California n=60 SSP clients Average age=45 73% male 51% Caucasian	- Intervention significantly associated with increased comfort recognizing, responding to, and administering naloxone for opioid overdose among SSP clients.
Clarke et al. (2016)	NEP	Descriptive exploratory	California n=106 PWID Average age=46 68% male Race not reported	- NEP users reported less risky injection drug use behaviors: 48% stopped sharing needles, 18% stopped reusing needles, 55-91% used clean paraphernalia.
Dahlem et al. (2017)	OEND	Descriptive program evaluation	Michigan n=109 sherriff’s deputies Average age=40 87% male Race not reported	- OEND significantly associated with increased overdose knowledge. - OEND equipped deputies to reverse 31 overdoses and enroll 19% of those individuals in treatment within one year of intervention.
Dunn et al. (2013)	HIV/HCV educational intervention	Randomized controlled trial	Setting not reported n=54 PWOUD Average age=27 70% male 98% Caucasian	- Intervention significantly associated with increases in HIV and HCV knowledge, with the large gains in HCV knowledge. - Intervention significantly associated with increased self-reported likelihood of using a condom use.
Fitzpatrick et al., (2022)	Heroin pipe distribution program	Quasi-experimental	Washington n=694 SSP clients Average age=39 61% male 63% White	After program implementation: - Proportion of participants that exclusively injected heroin was lower; - Proportion that injected and smoked heroin was higher; - Proportion that reused syringes was lower.
Goldman et al. (2019)	FTS	Sequential explanatory mixed methods	Rhode Island n=81 young adults who use drugs Average age=26 56% male 56% White	- FTS is feasible and acceptable to young adults who use drugs, with residue FTS being more convenient. - Positive FTS results facilitated changes in young adults' drug use behaviors. - Young adults distributed to peers within their social networks.
Han et al. (2017)	OEND	Quasi-experimental	Pennsylvania n=22 providers n=16 PWUO Demographics not reported	- Providers reported improved knowledge about naloxone prescribing and increased satisfaction caring for patients who use opioids. - PWUO reported high levels of comfort discussing opioid use with their provider. - OEND equipped individuals to successfully reverse overdoses.
Jones et al. (2014)	Peer-based OEND	Quasi-experimental	New York n=44 heroin users Average age=41 84% male 62% African American or Latino/Hispanic	- Intervention increased heroin users’ ability to identify opioid overdoses and when to use naloxone. - Intervention did not increase heroin users’ ability to identify non-opioid overdose situations.
Jones et al. (2022)	OEND	Prospective randomized-controlled trial	New York n=321 PWUO 45% aged 50+ 78% male 43% Non-Hispanic Black	- All three types of OEND training (minimal vs. extended vs. extended with a significant other) significantly associated with sustained increased overdose knowledge and attitudes. - All three types of OEND training had equivalent rates of successful naloxone use.
Katzman et al. (2018)	OEND	Prospective cohort study	New Mexico n=244 PWOUD 38% aged 20-29 71% female 63% Hispanic White	- 13% of PWOUD administered naloxone, indicating that PWOUD are willing to administer naloxone during an opioid overdose if naloxone is available to them. - 87% of reversals were with PWOUDs’ friends/family, indicating that OEND participants are usually not the naloxone recipient.
Katzman et al. (2020)	OEND	Prospective cohort study	New Mexico n=395 PWOUD Average age=35 68% female 66% Hispanic White	-18% of PWOUD administered naloxone, indicating that PWOUD are willing to administer naloxone during an opioid overdose if naloxone is available to them. -87% of reversals were performed on someone the participant knew, indicating that OEND participants are usually not the naloxone recipient.
Krieger et al. (2018)	FTS + OEND	Quasi-experimental	Rhode Island n=81 young adults who use drugs Average age=27 56% male 52% White	- FTS is acceptable to young adults who use drugs. - Positive FTS results significantly associated with changes in young adults’ drug use behaviors.
Lima et al. (2022)	FTS	Feasibility Study	Illinois n=23 opioid-related emergency department patients Average age=40 74% male 57% Black	- Of the 23 patients offered FTS, 18 accepted them. - Of the 3 patients reached at follow-up, all reported using the FTS, receiving a positive test result, and modifing their substance use behaviors as a result.
Park et al. (2020)	FTS	Quasi-experimental	Maryland n=103 female sex workers 32% aged 30-39 100% cisgender women 59% Non-Hispanic White	- FTS are acceptable to and highly utilized by female sex workers. - Positive FTS results facilitated changes in female sex workers’ drug use behaviors. - FTS significantly associated with decreased daily illicit opioid use, injection frequency, benzodiazepine use, and solitary drug use.
Park et al. (2021)	FTS	Cross-sectional survey	Maryland n=123 SSP clients Average age=42 59% male 52% White Delaware n=102 SSP clients Average age=37 55% female 76% White	- FTS are acceptable to and highly utilized by SSP clients. - Positive FTS results significantly associated with changes in drug use behaviors.
Phillips et al. (2012)	Skin and needle hygiene intervention	Pilot randomized controlled trial	Colorado n=41 heroin users Average age=43 75% male 54% Caucasian	- Intervention increased heroin users’ skin and needle cleaning skills but change not sustained over time. - Reductions in overall frequency of drug use and injection drug use days were not significant.
Pourtaher et al. 2022	OEND	Observational analysis	New York n=9133 naloxone administration forms submitted by 5835 unique law enforcement officers Average age of naloxone recipients=34 Naloxone recipients 70% male Naloxone recipients’ race not reported	- 86% of the time, law enforcement officers arrived at the scene of an overdose before EMS. - Law enforcement officers administered an average of 2 doses of naloxone. - Successful reversals were indicated in 87% of cases.
Sisson et al. (2023)	OEND delivered via online video format	Pilot randomized controlled trial	16 Southeastern cities n=98 PWUO Average age= 37 61% female 85% White/Caucasian	- Intervention significantly associated with increases in knowledge of opioid overdose and naloxone. - Naloxone possession higher in group compensated to purchase personal naloxone. - Opioid use frequency and naloxone possession positively associated in group that only received overdose education only.
Winhusen et al. (2020)	Personally-tailored OEND	Secondary analysis of a randomized controlled trial	Ohio n=80 PWUD Average ag=39 55% male 13% African American	- Intervention significantly associated with increases in overdose and treatment knowledge and treatment readiness. - Intervention significantly associated with decreases in opioid use and overdose risk behaviors.
Yang et al. (2021)	Mail-based OEND and SSP	Descriptive program evaluation	50 states n=3926 PWUD 60% age 26-45 61% female 88% White	- Intervention feasible and acceptable to PWUD. - 9% of PWUD administered naloxone, indicating that PWOUD are willing to administer naloxone during an opioid overdose if naloxone is available to them.

Abbreviations: EMS = Emergency Medical Services; OEND = Overdose Education and Naloxone Distribution; NEP = Needle Exchange Program; FTS = Fentanyl Test Strip; PWUD = People Who Use Drugs; PWOUD = People with Opioid Use Disorder; PWID = People Who Inject Drugs; PWUO = People Who Use Opioids.

Drake and colleagues (2021) examined recreational cannabis laws using data from 29 states from 2011-2017 and found they were initially associated with decreases in opioid-related emergency department visits, but effects dissipated six months after implementation. They speculated that upon the passage of a recreational cannabis law, people with OUD may have attempted to use cannabis as a substitute for opioids but found cannabis an insufficient substitute and returned to opioid use.

Synthesis of evidence on meso interventions

Eight articles on meso-level harm reduction interventions met review inclusion criteria (see Table 3). Multiple studies (n=3) examined OEND interventions, while the others examined a syringe service program and an overdose monitoring facility. Study samples ranged from a single program (n=2) to hospital admissions (n=1), communities (n=1), and counties (n=2).

Table 3.

Summary of Meso Interventions, n=8

Authors	Meso Intervention	Study Design	Study Setting and Sample	Key Findings
Alexandridis et al. (2018)	1) community education 2) provider education 3) hospital ED policies 4) diversion control 5) support programs for pain patients 6) naloxone policies 7) addiction treatment	Interrupted time series analysis	North Carolina, 2009-2014 n=74 counties implemented n=26 counties did not implement	- Intervention counties consistently had lower overdose rates than non-intervention counties amid increases in absolute numbers of overdoses statewide. - Provider education associated with 9% lower mortality rate but little change in ED visits. - Hospital policies, community education and supports for pain patients associated with small increases in overdose mortality. - Diversion control and addiction treatment associated with higher mortality rates. - Diversion control associated with gradual increases in ED visits after 3 months. - Naloxone policies associated with gradual decreases in ED visits at 5 months. - Addiction treatment associated with decrease in ED visits within 2-5 months but returned to baseline at 6 months.
Allen et al. (2022)	Public health vending machine	Interrupted time series analysis	Nevada, 2015-2020 n=7 vending machines	Naloxone dispens via public health vending machines associated with immediate reductions in opioid-involved overdose fatalities.
Bornstein et al. (2020)	SSP	Retrospective analysis	Florida, 2015-2018 n=302 PWIO admissions	- ED visits significantly decreased post-SSP implementation. - Opioid-related deaths decreased in county that implemented the SSP, despite increases statewide.
León et al. (2018)	Overdose monitoring facility	Pre-post comparison of observational data	Massachusetts, 2016 n=1 program	- Overdose monitoring facility significantly associated with decreases in over-sedated individuals observed in public. - Overdose monitoring facility not significantly associated with reductions in public drug use, public drug exchange, publicly discarded syringes, or injection-related litter. - Correlation between overdose monitoring facility use and an increase in publicly discarded syringes approached significance. - Decrease in public drug use and exchange after the overdose monitoring facility opened approached significance.
Naumann et al. (2019)	OEND	One group pre-post design	North Carolina, 2000-2016 n=38 counties	- OEND associated with lower overdose death rates. - For every dollar spent on OEND, over $2,700 in benefits were generated by averting an overdose death.
Rowe et al. (2019)	OEND	Retrospective Analysis	California, 2014-2015 n=1 program	Amid a fentanyl outbreak, naloxone distribution and use increased significantly but the proportion of overdose deaths did not during study period.
Walley et al. (2013)	OEND	Interrupted time series analysis	Massachusetts, 2004-2006 n=19 communities	- Low and high OEND implementation associated with lower overdose death rates. - ED visit and hospital admission rates not significantly different between low and high OEND implementation.
Xuan et al. 2023	Post-overdose Outreach Programs	Interrupted time series analysis	Massachusetts, 2015 n=93 municipalities	Post-overdose outreach programs significantly associated with lower opioid fatality rates and lower opioid-related EMS rates over time.

Abbreviations: ED = Emergency Department; EMS = Emergency Medical Services; OEND = Overdose Education and Naloxone Distribution; PWIO = People Who Inject Opioids; PWUD = People Who Use Drugs; SSP = Syringe Service Program.

Though OEND interventions were tailored to the setting and population, each consisted of a relatively brief training on how to recognize, respond to, and reverse an overdose along with naloxone distribution (i.e., nasal spray or syringes and vials). All three studies found that OEND interventions were significantly associated with lower overdose death rates. Rowe and colleagues (2019) found that despite increases in opioid-involved overdoses, the proportion of overdose deaths due to opioid use did not increase in the one-year period, likely a result of increased naloxone distribution and use facilitated by local OEND programs. Walley and colleagues (2013) examined whether the extent of city/town-level OEND implementation mattered, concluding that high versus low rates of community implementation did not differentially affect emergency department visit and hospital admission rates, but rather, whether or not OEND was implemented at all. Naumann and colleagues (2019) conducted cost-benefit and cost-effectiveness analyses of an OEND intervention and determined that for every dollar spent on naloxone kits and their distribution, over $2,700 in benefits (quantified as deaths avoided and monetized using the value of a “statistical” life) were generated by preventing an overdose death.

The remaining studies examined five distinct meso-level harm reduction interventions. Allen and colleagues (2022) found that naloxone dispensation via public health vending machines was associated with immediate reductions in opioid-involved overdose deaths. Bornstein and colleagues’ (2020) found that emergency department visits significantly decreased after a community-based syringe service program was implemented. They also found that while overdose deaths increased across the state, counties that implemented syringe service programs saw significant decreases in overdose deaths. Xuan and colleagues (2023) examined the impact of post-overdose outreach program implementation and found that it was significantly associated with lower opioid fatality rates and lower opioid-related EMS rates over time. León and colleagues (2018) conducted an observational field study using environmental scanning methods to understand changes in public order (e.g., public drug use) before and after the implementation of an overdose monitoring facility. Although the intervention was associated with decreases in publicly over-sedated individuals, it was not associated with decreases in public drug use, public drug exchange, publicly discarded syringes, or injection-related litter. This was the only study of supervised consumption sites.

Alexandridis and colleagues (2018) examined an intervention that consisted of seven strategies implemented by 74 (of 100) counties in a single state (given the focus of this review, only findings related to harm reduction strategies are discussed here, though findings for all seven strategies are reported in Table 3). They found that policies promoting the naloxone dissemination were associated with small increases in overdose mortality, but the effect was time-delayed.

Synthesis of evidence on micro interventions

Twenty-one articles on micro-level harm reduction interventions met review inclusion criteria (see Table 4). The most common types of interventions examined were OEND programs (n=11) and fentanyl test strips (FTS) (n=5). Study sample sizes ranged from 16 to 9,133 individuals (removing two outliers, the range caps at 395). Among studies that reported sample demographics, most participants were male (n=14) and Caucasian/White (n=12).

Micro-level OEND intervention studies were similar to meso-level OEND intervention studies in that they examined effects of a relatively brief training on how to recognize, respond to, and reverse an overdose, and naloxone distribution. Findings were similar among these micro-level OEND studies; nearly all reported the intervention as feasible and acceptable to PWUD. Six studies found OEND interventions were significantly associated with increasing individuals’ knowledge, skills, and comfort in identifying and responding to an overdose. Winhusen and colleagues (2020) examined the effects of a personally-tailored OEND intervention and found that in addition to significant increases in knowledge regarding overdoses, the intervention was also associated with decreases in opioid use and overdose risk behaviors, as well as increases in treatment readiness. Jones and colleagues (2014) reported similar findings, but found that the intervention did not increase individuals’ ability to differentiate between opioid and non-opioid overdoses. Han and colleagues (2017) studied the effects of an OEND intervention among both healthcare providers and people who use opioids (PWUO) and found that the intervention improved provider comfort prescribing naloxone and providing healthcare to PWUO, and also increased PWUOs’ comfort discussing opioid use with their providers. Eight studies reported that the OEND intervention resulted in trainees successfully reversing overdoses. Katzman and colleagues (2018, 2020) reported that for the majority of overdose reversals (up to 87%), intervention participants were not the naloxone recipient, but rather PWUDs’ friends, family, or acquaintances. Similarly, Pourtaher and colleagues (2022) examined data from a statewide law enforcement naloxone administration program and found that 86% of the time, officers were the first to aid a person with a suspected overdose (87% of naloxone recipients survived the suspected overdose), demonstrating that law enforcement officers can successfully identify and reverse overdoses.

FTS interventions were the second most common (n=5) micro-level harm reduction intervention. All FTSstudies reported that the intervention was associated with changes in individuals’ drug use behaviors. Park and colleagues (2020) found the FTS intervention was also associated with decreased daily illicit opioid use, injection frequency, benzodiazepine use, and solitary drug use among female sex workers. Two studies also found that FTS intervention participants redistributed their FTS to others in their network whom they perceived to be at-risk for experiencing an overdose.

The remaining studies examined five distinct micro-level harm reduction interventions. Clarke and colleagues’ (2016) examination of a syringe service program found that injection drug users who participated in the program reported less risky drug use behaviors. Arendt and colleagues (2023) found that harm reduction dispensing machines distributed more harm reduction supplies than any other syringe service program in the county had in the previous year and that implementation was associated with lower incidence of unintentional overdose deaths and HIV in the county. Fitzpatrick and colleagues (2022) examined outcomes of a heroin pipe distribution program implemented at a syringe service program and found that the proportion of participants that exclusively injected heroin and that reused syringes was lower after program implementation, suggesting the intervention may reduce drug use-related harms by changing drug consumption patterns. Phillips and colleagues (2012) conducted a randomized controlled trial of a two-session skin and needle hygiene intervention among heroin users and found decreases in the number of days any drugs were used and the number of days drugs were injected. The intervention was also found to increase individuals’ skin and needle cleaning skills, but practices were not sustained over time. Similarly, Dunn and colleagues (2013) found that a HIV/hepatitis educational intervention for people with OUD was associated with increases in HIV and hepatitis knowledge and a greater likelihood of using condoms.

Several limitations were evident across macro-, meso-, and micro-level intervention studies. The majority of studies were observational, thus precluding researchers from making causal inferences about intervention effects. The underreporting of overdose reversal reports and unverified reversal reports were another important limitation across relevant studies. Authors of macro- and meso-level intervention studies noted the potential for random and systematic measurement errors creating inaccurate and incomplete datasets that may have produced under/overestimates of intervention effects. Among macro-level intervention studies specifically, authors commonly noted 1) the challenge of estimating the effects of a single law due to an inability to examine all potential confounders, and 2) difficulty comparing effects of law interventions in differing implementation contexts (e.g., counties vs. states). Among micro-level intervention studies, authors noted limitations relevant to participant self-report (e.g., social desirability) and generalizability (e.g., underpowered analyses due to participant attrition).

Discussion

This systematic review identified and examined the outcomes and evidence-base of harm reduction interventions implemented at macro-, meso-, and micro-levels to mitigate effects of the opioid overdose epidemic. Evidence for micro- and meso-level interventions is largely consistent (i.e., interventions were found to be significantly associated with outcomes that reduce opioid-related harm), whereas evidence for macro-level interventions is mixed. Study findings also revealed two additional themes across levels of interventions: 1) a lack of interventions that span multiple socio-ecological levels, and 2) a lack of attention to race, ethnicity, and gender in study contextualization and intervention reach.

Given the complexity of the opioid overdose epidemic, which involves the interplay of micro-, meso-, and macro-level factors, stakeholders have increasingly called for a multifaceted approach, inclusive of interventions that span at least two of these levels (HHS, 2021a; Jalali et al., 2020). While some do exist (see Windsor et al., 2018), they are few and far between. This is especially so for multilevel harm reduction interventions, of which this systematic review found only one (see Han et al., 2017 in which the intervention aimed to improve both client and provider outcomes). In fact, of the 36 included articles, the majority (n=21) examined micro-level interventions. This focus largely reflects the United States’ approach to drug (mis)use in which individuals and their behaviors are the focal point of change, while the context in which they are situated is ignored (Rhodes, 2009). Although the exact causes of substance use disorders are not known, the individual is generally regarded as the locus of the problem, rather than the social and structural forces that shape their environments, access to resources, choices/behaviors, and health outcomes (Brothers et al., 2023; Kolla et al., 2019; Rhodes et al., 2006). Researchers risk subverting the philosophy of harm reduction if their studies of micro-level interventions do not situate interventions or findings within the broader context of structural and systemic factors that make individuals vulnerable to drug (mis)use-related harms. Researchers can avoid this pitfall by conducting community-engaged research in partnership with PWUD and PWUD-serving organizations throughout the entire research process (Boucher et al., 2017; Claborn et al., 2023; Morgan et al., 2023). Study findings underscore the need for developing and examining community-informed harm reduction interventions designed to impact at least two socio-ecological levels, which may produce synergistic effects and therefore greater benefits.

Authors of three studies did not report participants’ demographic characteristics. Of the 18 studies that did, the majority of participants were males (n=14 studies) and Caucasian/White (n=12 studies). Only two studies in this review (McClellan et al., 2018; Townsend et al., 2022) explicitly addressed race/ethnicity. Given that this systematic review examined scientific literature from 2011-2023, we expected to identify more articles detailing interventions that reached Black and Hispanic individuals and communities since they have been disproportionately impacted by the third wave of the opioid overdose epidemic that emerged in 2013 (Cano, 2021; El-Bassel et al., 2021; Shiels et al., 2018). It is possible that the effects of racism are an undergirding factor, as Black communities have long resisted harm reduction interventions and research due in part to the distrust engendered by historical racism (Woods, 1998). Recent research has both echoed this and added that the prevalence of structural and institutional racism in harm reduction services are deterrents to utilization among Black and Brown communities (Godkhindi, Nussey & O’Shea, 2022; Lopez et al., 2022; Owczarzak et al., 2020). It is also possible that there are alternative interventions that have been understudied in the scientific literature on harm reduction that target the substantive harms experienced by racialized communities, such as cannabis legalization or drug possession decriminalization (Earp et al., 2021; Rouhani et al., 2023). Research related to understanding inequities in harm reduction interventions is needed.

Throughout the title and abstract screening stage of this review, technology-enabled harm reduction interventions emerged as a budding line of research. These interventions included Bluetooth-enabled naloxone kits (Lai et al., 2020), wearable naloxone injector systems (Chan et al., 2021), mobile apps (Aggarwal & Borycki, 2019), info boxes prompted by keyword Google searches (Arendt, 2021), and virtual reality overdose response training (Herbert et al., 2020). Though these articles did not meet this study’s inclusion criteria, this secondary finding suggests an opportunity for further study. Researchers should examine gray literature that captures community-based innovations that are being implemented in the field but that may not be reported in the academic research literature such as FIX, an app-based syringe service program (https://eficcs.org/first-integrated-xchangefix/). Additionally, research is needed regarding whether these interventions are acceptable to PWUD and service providers, if they are feasible in real-world settings, and what the possible unintended consequences and implications are of widespread dissemination (e.g., inequitable access, privacy violations).

Limitations

Findings from this systematic review should be considered in light of several limitations. First, this study focused solely on harm reduction interventions implemented in the U.S. and thus excludes studies detailing interventions developed and implemented in contexts that have more experience delivering harm reduction services (e.g., supervised consumption sites in Canada). Despite this, findings may be informative to other settings and countries that have less experience with harm reduction interventions. Future studies should examine micro-, meso-, and macro-level interventions in other contexts. Second, due to capacity constraints, only one team member conducted the screening phase of this review. To avoid oversights and potential Type I and/or Type II errors, the first author flagged any articles for which inclusion criteria were unclear upon initial screening, and regularly met with co-authors to discuss and categorize accordingly. Additionally, the screening process and decision-making methods were rigorously and transparently detailed to allow for replication. Third, the broad scope of this systematic review resulted in identifying a range of interventions with varying study designs and degrees of reliability, thus preventing the authors from comparing effectiveness across included interventions. Lastly, the differing overall levels of bias paired with specific sources of bias (e.g., non-response bias, confounding factors) across intervention levels further restricts the interpretation of results.

Conclusion

The United States is in the midst of a public health crisis having experienced historic rates of overdose deaths. With mortality rates continuing to rise, the need to leverage, invest in, and expand evidence-based approaches that reduce harms associated with drug use and the opioid overdose epidemic is pressing. Review findings provide insight into the current state of scientific literature on harm reduction interventions that have been implemented to address the opioid overdose epidemic and pose notable implications for research, policy, and practice.