The impact of needle and syringe exchange programs on HIV-related risk behaviors in low- and middle-income countries: A systematic review and meta-analysis examining individual- versus community-level effects

Abstract

We conducted a systematic review and meta-analysis of the impact of needle and syringe exchange programs (NSP) on both individual- and community-level needle-sharing behaviors and other HIV-related outcomes in low- and middle-income countries (LMIC). A search of five databases for peer-reviewed trial or quasi-experimental studies reported through July 2021 identified 42 interventions delivered in 35 studies, with a total of 56,751 participants meeting inclusion criteria. Random-effects meta-analysis showed a significant protective association between NSP exposure and needle-sharing behaviors at the individual-level (odds ratio [OR]=0.25, 95% confidence interval [CI]= 0.16–0.39, 8 trials, n=3,947) and community-level (OR=0.39, CI=0.22–0.69, 12 trials, n=6,850), although with significant heterogeneity. When stratified by needle-sharing directionality, NSP exposure remained associated with reduced receptive sharing, but not distributive sharing. NSP exposure was also associated with reduced HIV incidence and increased HIV testing but there were no consistent associations with prevalence of bloodborne infections. Current evidence suggests positive impacts of NSPs in LMICs.

Introduction

Globally, there are approximately 15.6 million people who inject drugs according to a systematic review published in 2017.¹ People who inject drugs are at increased risk for contracting HIV, hepatitis B, hepatitis C, and other bloodborne diseases.² In many low- and middle-income countries (LMIC), globalization has brought both positive changes, such as improved transportation and international trade, and challenges, such as dissemination of illicit drugs and facilitation of HIV transmission.³ Risk factors associated with HIV infection among people who inject drugs include poor health literacy, low health awareness, limited access to clean needles or other sterile drug injecting equipment, and the practice of sharing needles/syringes in groups.^4,5 In addition, people who inject drugs are often stigmatized and face discrimination, especially in LMICs where the practice of injecting drugs is typically illegal, creating further barriers to care-seeking.⁶

Over the past 30 years, many countries have sought to reduce the risk of disease transmission from sharing contaminated needles through the adoption of harm reduction programs. These programs may include needle and syringe exchange programs (NSP), which strive to reduce high-risk behaviors like needle-sharing and needle reuse,^5,7 among other interventions including overdose prevention, safe injection sites, and opioid substitution therapy. NSPs reduce drug injection-related harm by providing sterile drug injection equipment and a location for safe syringe disposal, and they also often offer other risk reduction interventions, such as condom promotion/provision and other HIV-related services.⁸

Multiple systematic reviews have shown that NSPs are effective in reducing HIV transmission among people who inject drug in high-income settings.^5,9,10 Despite stigma and criminalization of drug use, which impede NSP uptake,¹¹ there is now substantial evidence supporting the effectiveness of NSPs in LMICs as well.¹² However, the most recent reviews of the impact of NSPs in LMICs included studies published through 2011; there is thus a need to document the current state of the published evidence for NSPs. Further, previous reviews have not clearly distinguished between the impact of NSPs on individual-level versus community-level outcomes. At the individual-level, studies may compare individuals who use NSP services to other individuals who do not; this demonstrates the effect of NSPs on those who use them. At the community level, studies may compare communities where NSP services are available to other communities where such services are not available; this demonstrates the broader effect of NSPs, not only on those who use them, but also on the wider community. Separately examining the evidence for an impact at both levels allows us to better understand the full impact of NSPs.

Methods

We sought to evaluate the impact of NSPs on both individual- and community-level needle-sharing behaviors and other HIV-related outcomes in LMICs. This research was conducted as part of a larger series of systematic reviews conducted by the Evidence Project evaluating the effectiveness of behavioral interventions for HIV in LMICs. We conducted this review in accordance with PRISMA guidelines.¹³

Definition

For the purposes of this review, we defined NSPs as programs that (1) provide sterile injecting equipment including needles and syringes to people who inject drugs, whether the needles and syringes are sold, exchanged, or given freely, and (2) help dispose of used needles and syringes safely.

Eligibility Criteria

Studies were included in the review if they met the following criteria:

1. data were presented from a LMIC, defined by combining the World Bank classifications¹⁴ of low-income, lower-middle-income, or upper-middle-income economies at the time of the study;

2. needles and/or syringes were distributed, exchanged, or sold to people who inject drugs as part of the intervention;

3. the study population were people who inject drugs, according to any definition used by study authors (e.g. ever, past year, or other time frames);

4. an evaluation design was employed that compared post-intervention outcomes using either a pre/post or multi-arm study design (this could include randomized trials, non-randomized trials, observational studies, before-after studies, cross-sectional, or serial cross-sectional analyses);

5. specific outcomes of interest including use of new/sterile needles/syringes, return of used needles/syringes, needle borrowing/sharing, and other behavioral, psychological, social, care or biological outcome(s) related to HIV prevention were presented; and

6. the article was published in a peer-reviewed journal.

No language restrictions were used. When an article in a language other than English was found, it was translated into English for full-text review.

Search Strategy

We searched five electronic databases (PubMed, PsycINFO, Sociological Abstracts, the Cumulative Index to Nursing & Allied Health Literature (CINAHL), and EMBASE) for articles published from January 1, 1990 through July 25, 2021. We used the following search terms, adapted for each database (Appendix A): (“needle exchange” OR “needle distribution” OR “needle sales” OR “syringe sales” OR “syringe distribution” OR “syringe exchange” OR “syringe-exchange” OR “needle-exchange” OR “needle syringe programs” OR “needle syringe program” OR “needle syringe programme” OR “needle exchange programmes”) AND (HIV or AIDS).

To identify articles not obtained from electronic database searching, study staff hand-searched the table of contents of the following journals: AIDS, AIDS and Behavior, AIDS Care, AIDS Education and Prevention, and the International Journal of Drug Policy. Finally, we examined the reference lists of articles selected to further identify potential articles for inclusion. This process was iterated until no new articles were found.

Initial screening of studies was done by a member of the study staff, who excluded clearly non-relevant articles based on the titles and abstracts. Two study staff members conducted secondary screening, independently reviewing the abstracts of the remaining citations for inclusion. Differences in screening decisions were discussed to establish consensus. The final inclusion/exclusion of studies identified articles to be included in the qualitative synthesis and quantitative meta-analysis based on a thorough reading of the full-text article.

Data Extraction and Management

Each article meeting the inclusion criteria underwent data abstraction by two study staff members working independently. Data were entered into a standardized, detailed coding form including the following information:

1. study year, study location, and study design;

2. gender and age distribution, sample population, sample size, sampling strategy, loss to follow up, and comparison groups;

3. description of the NSP intervention (name of the program, recruitment strategy, facilitator/role of implementer, ways of exchange/distribution site, and intervention activities);

4. description of the outcome reported (needle/syringe needle sharing, other HIV- or sexually transmitted infection (STI)-related outcomes);

5. outcome measures, statistical tests used, and steps taken to control for covariates;

6. intervention effects and findings.

Completed coding forms from the two study staff members were compared for discrepancies, and differences were resolved through discussion and consensus with a third reviewer from the senior study staff.

Quality assessment

Rigor of included articles was assessed using an eight-item risk of bias tool developed for the larger series of systematic reviews by the Evidence Project,¹⁵ conducted independently by two reviewers. Differences were resolved by consensus. The items were: (1) prospective cohort; (2) control or comparison group; (3) pre/post intervention data; (4) random assignment of participants to the intervention; (5) random selection of subjects for assessment; (6) follow-up rate of 80% or more; (7) comparison groups are equivalent on socio-demographic measures; and (8) comparison groups are equivalent at baseline on outcome measures.

Data Analysis

We focused on categorical outcomes of needle/syringe sharing behaviors and NSP exposure. We extracted count and proportion data that evaluated the association between NSP exposure and needle/syringe sharing behaviors based on between-group or pre-post comparisons. In meta-analysis, we only included studies that directly examined the association between needle/syringe sharing behaviors and NSP exposure. Adjusted effect sizes and 95% confidence intervals (CIs) for needle/syringe sharing behaviors associated with exposure to NSP were chosen over crude ones when available. Otherwise, the number and proportion of participants reporting the outcome of interest in both intervention and control/comparison groups or comparing before and after NSP exposure were extracted from each study. We then calculated the study-specific effect size expressed as an unadjusted odds ratio (OR) based on the available count and proportion data. For outcomes measured at multiple time points following NSP exposure, we used the last measurement period unless a study included a longer follow-up than most but also included a time point more similar to the remaining studies. When a study was conducted in two cities or countries and presented data separately, we considered them as separate interventions, so we calculated two location-specific effect sizes.

Meta-analysis was performed using Comprehensive Meta-Analysis, version 3 (CMA; Biostat, Englewood, NJ), using random-effects models to calculate pooled OR estimates and 95% confidence intervals. We assessed statistical heterogeneity using both Q and I² statistics. In the presence of substantial heterogeneity, we conducted sub-group analyses based on whether there was individual- or community-level outcome measurement and the type of needle/syringe-sharing behavior. First, we classified studies according to individual- or community-level outcome measurement. Separate analyses were conducted based on type of needle/syringe-sharing behavior: receptive sharing (i.e., receiving, borrowing, or reusing other persons’ used needles/syringes) and distributive sharing (i.e., distributing, lending, or passing on any used needles/syringes to other persons). Indirect sharing behaviors (e.g., backloading or frontloading needles/syringes²²) were excluded because few studies reported these outcomes. A third analysis combined both outcomes if a study reported both receptive and distributive sharing. In addition, if a study presented needle/syringe-sharing outcomes without specifying sharing directionality, we reported the non-directional estimates of effect sizes as a fourth analysis. Finally, we combined all study-specific estimates (including receptive, distributive, and reported non-directional sharing results) to assess overall NSP impact. We also created funnel plots of standard error by log odds ratio to assess potential reporting bias.

We excluded studies from meta-analysis if they did not report behavioral outcomes of needle/syringe sharing combinable for data synthesis or if we could not calculate the study-specific effect size because of missing or insufficient data, in which case authors were contacted for additional information. For example, articles were excluded if between-group comparisons were based on HIV status instead of needle/syringe sharing behaviors, or sharing behaviors were defined as “ever shared needles/syringes.” Studies were not combinable in meta-analysis if NSP exposure was defined by study authors in a unique way (e.g., exchanging seven or more free needles/syringes per week from the program¹⁶), the study measured syringe sharing as a continuous variable (e.g., mean number of needles/syringes exchanged in the past 30 days¹⁷), or it was not possible to attribute intervention effects to NSP exposure (e.g., the outcome measure was ever shared needles/syringes¹⁸). We reported findings from all studies not included in meta-analysis in the qualitative synthesis, organized by outcome category.

Results

The search generated 4,513 citations from electronic databases and 76 citations by both hand-searching five journals and secondary searching (Figure 1). After removing duplicates, 3,425 references were included for primary screening. We excluded 3,254 citations after examining their titles and abstracts. 171 full-text articles were retrieved to determine eligibility. Ultimately, 35 articles reporting on 42 interventions met the criteria for inclusion in this review and were included in the qualitative synthesis.^16–50

Figure 1.

PRISMA flow diagram of the different phases of the systematic review.

Study characteristics

Table 1 presents summary characteristics of the 42 intervention effects (reported in 35 included articles) and descriptions of reported interventions related to NSP published between 1999 and 2021.^16–50 Two of these articles reported on Avahan, including community interventions for NSP in India, but presented different outcomes;^24,25 three other articles reported different outcomes and timepoints from the Cross-Border Project, a large serial cross-sectional study conducted in multiple sites in Vietnam and China.^22,26,27 The 42 intervention effects included 56,751 total participants; individual study sample sizes ranged from 105 to 20,640. Study designs included 21 cross-sectional studies, six serial cross-sectional studies, two quasi-experimental/non-randomized trials, two time-series studies, one before-after study, one cluster-randomized controlled trial, one prospective cohort study, and one retrospective case-control study. Most of the included articles were conducted in Asia (Bangladesh, China, India, Iran, Nepal, Pakistan, and Vietnam) and Russia. Several trial or cohort designs (one cluster randomized controlled trial, two group non-randomized controlled trials, and one prospective cohort study) were used to evaluate NSP programs in China. Three studies were conducted in Eastern Europe, including one time-series study in Georgia, one cross-sectional study in Croatia, and one cross-sectional study in Bulgaria. All included studies were from lower-middle and upper-middle-income countries; none were from countries classified as low-income by the World Bank. Study participants were predominantly male and young; ages ranged from 18–50 years old. Only one cross-sectional study, from Mexico, focused on female sex workers who injected drugs.

Table 1.

Description of included studies and reported NSP interventions

Study Citation	Study Design (Study method: study location; study years; length of follow-up/exposure)	Sample size (baseline number; gender percentage (%); age distribution: mean / median age (years) ± standard deviation;	Intervention Description (level of outcome measurement; recruitment method; facilitator or role of implementer)	Intervention Activity (Ways of exchange or distribution site for NSP and other intervention activity)
Badrieva et al., 2007	cross-sectional: Kazan, Russia; 1999–2001; > 4 months, ≤ 4 months	554 (78.3% male; 21.7% female); mean age: 22.7 (range=13–47)	harm reduction program: individual level; social network recruitment; facilitated by gatekeepers	NSP delivered by secondary exchange and on-site exchange alongside gatekeepers as harm reduction educator
Broadhead, et al., 2006 (Bragino)	before-after study: Bragino, Russia; 2003–2004; baseline, 6-month	490 (75.3% male; 24.7% female); 20–29 years: 350 (71.7%)	standard peer-driven intervention: individual level; peer-driven recruitment; facilitated by peer educators	NSP delivered by on-site exchange alongside HIV education, peer recruitment, nominal monetary rewards for recruitment and education, HIV-test counseling, and harm reduction kits including condoms, alcohol-wipes and bleach tablets
Broadhead et al., 2006 (Rybinsk)	before-after study: Rybinsk, Russia; 2003–2004; baseline, 6-month	364 (70.6% male; 29.4% female); 20–29 years: 165 (45.7%)	simplified peer-driven intervention: individual level; peer-driven recruitment; facilitated by peer educators	NSP delivered by on-site exchange alongside HIV education, peer recruitment, nominal monetary rewards for recruitment and education, HIV-test counseling, and harm reduction kits including condoms, alcohol-wipes and bleach tablets
Des Jarlais et al., 2002	cross-sectional: Prague, Czech Republic; Budapest, Hungary; Skopje, Former Yugoslavian Republic of Macedonia; Krakow, Poland; Poltava, Ukraine; Russian: Nizhniy Novgorod, Pskov, Rostov-Na-Donu, St. Petersburg, and Volgograd; 1999; <30 days	1671: male percentages varied by city (range 64%−92%); female percentages varied by city (range 8%−36%); 20–29 years: age percentages varied by city (range 41%−79%)	NSP: individual level; recruitment and facilitator not reported	NSP delivery method not reported, but delivered alongside information about HIV/AIDS
Des Jarlais et al., 2007	serial cross-sectional: Ning Ming County, Guangxi Province, China; Lang Son Province, Vietnam; 2002–2005; baseline, 6-, 12-, 18-, 24-, 36-month	Ning Ming: 235 (89% male); Lang Son: 190 (99% male); mean age in Ning Ming: 27.2; mean age in Lang Son: 27.7	HIV prevention intervention: individual level; peer-driven recruitment; facilitated by peer educators	NSP delivered by secondary exchange and pharmacy vouchers alongside information on reducing drug use and sexual risk behaviors, distribution of ampoules of sterile water for injection and condoms, and free vouchers that can be redeemed for sterile injection equipment and condoms in pharmacies.
Eicher et al., 2000	cross-sectional: Manipur, India; 1996; 12-month	191 (85.3% male); mean age: 22.6±4.9 (range=14–40)	NSP: individual level; privileged access and social network recruitment; facilitated by health workers	NSP delivered by on-site and outreach exchange alongside free condoms distribution, HIV and HCV testing
Ganju et al., 2016	cross-sectional: Manipur and Nagaland, India; 2009; 6-month	1650 (100% male); 18–34 years: 83.8%	Avahan comprehensive harm reduction program: individual level; peer-driven recruitment; facilitator not reported	not reported
Goswami et al., 2014 (Manipur)	serial cross-sectional: Manipur, India; 2006, 2009; baseline, 36-month	1660 (100% male); 18–30 years: 77.8% (in survey 2006)	Avahan intervention: community level; peer-driven recruitment; facilitator not reported	NSP delivery method not reported, but delivered alongside abscess management, clinic services for treating of STIs, condom promotion and distribution, and community empowerment
Goswami et al., 2014 (Nagaland)	serial cross-sectional: Nagaland, India; 2006, 2009; baseline, 36-month	1689 (100% male); 18–30 years: 93.2% (in survey conducted in 2006)	Avahan intervention: community level; peer-driven recruitment; facilitator not reported	NSP delivery method not reported, but delivered alongside abscess management, clinic services for treating of STIs, condom promotion and distribution, and community empowerment
Hammett et al., 2006 (China)	serial cross-sectional: Ning Ming County, Guangxi Province, China; 2002–2006; baseline, 6-, 12-, 18-, 24-month	1433 (>90% male); 21–30 years: 68% (baseline pre-NSP)	HIV prevention intervention: community level; peer-driven recruitment; facilitated by peer educators	NSP delivered by secondary exchange and pharmacy vouchers alongside distribution of ampoules of sterile water for injection and condoms, and information on reducing drug use-related and sexual HIV risks orally and through distribution of brochures
Hammett et al., 2006 (Vietnam)	serial cross-sectional: Lang Son Province, Vietnam; 2002–2006; baseline, 6-, 12-, 18-, 24-month	1677 (>90% male); 21–30 years: 72% (baseline pre-NSP)	HIV prevention intervention: community level; peer-driven recruitment; facilitated by peer educators	NSP delivered by secondary exchange and pharmacy vouchers alongside distribution of ampoules of sterile water for injection and condoms, and information on reducing drug use-related and sexual HIV risks orally and through distribution of brochures
Hammett et al., 2012 (Nang Min County)	serial cross-sectional: Ning Ming County, Guangxi Province, China; 2002–2008; baseline, 6-, 12-, 18-, 24-, 36-, 48-, 72-month	2125 (mostly male); mostly 21–30 years	HIV prevention intervention: community level; peer-driven recruitment; facilitated by peer educators	NSP delivered by secondary exchange and pharmacy vouchers alongside distribution of ampoules of sterile water for injection and condoms, information on reducing drug use-related and sexual HIV risks orally and through distribution of brochures, and collecting and disposing of used needles/syringes
Hammett et al., 2012 (Lang Son)	serial cross-sectional: Lang Son Province, Vietnam; 2002–2009; baseline, 6-, 12-, 18-, 24-, 48-, 72-, 84-month	2677 (mostly male); mostly 21–30 years	HIV prevention intervention: community level; peer-driven recruitment; facilitated by peer educators	NSP delivered by secondary exchange and pharmacy vouchers alongside distribution of ampoules of sterile water for injection and condoms, information on reducing drug use-related and sexual HIV risks orally and through distribution of brochures, and collecting and disposing of used needles/syringes
Hammett et al., 2012 (Ha Giang Town)	serial cross-sectional: Ha Giang Town, Vietnam; 2003–2011; baseline, 6-, 12-, 18-, 24-, 36-, 48-, 60-, 72-, 96-month	793 (mostly male); mostly 21–30 years	HIV prevention intervention: community level; peer-driven recruitment; facilitated by peer educators	NSP delivered by secondary exchange and pharmacy vouchers alongside distribution of ampoules of sterile water for injection and condoms, information on reducing drug use-related and sexual HIV risks orally and through distribution of brochures, and collecting and disposing of used needles/syringes
Handanagic et al., 2017	cross-sectional: Rijeka and Split, Croatia; 2014–2015	654 (Rijeka: 74.2% male; Split: 78% male); median age: Rijeka: 34 (IQR=30–41); Split: 37 (IQR=30–41)	NSP: individual level; recruitment and facilitator not reported	not reported
Jenkins et al., 2001 (Dhaka)	cross-sectional: Dhaka, Bangladesh; 2000	682 (100% male); mean age: 35.6	NSP: community level; randomization recruitment; facilitated by peer educators	NSP delivered by on-site and outreach exchange
Jenkins et al., 2001 (Rajshahi)	cross-sectional: Rajshahi, Bangladesh; 2000	512 (100% male); mean age: 34.5	NSP: community level; randomization recruitment; facilitated by peer educators	NSP delivered by secondary exchange and outreach exchange
Khan et al., 2011	cross-sectional: Karachi, Lahore, Sargodha, Sialkot, Faisalabad, Quetta, and Peshawar, Pakistan; 2006–2008	20640	national enhanced AIDS control program: community level; recruitment not reported; facilitated by NGO	NSP delivered by on-site and outreach exchange alongside condoms distribution, voluntary counseling and testing, detoxification, wound care, primary healthcare and social support, and non-pharmacologic rehabilitation
Lin et al., 2004	group non-randomized controlled trial: Yangjiang City and Yunfu City, Guangdong Province, China; 2002–2003; baseline, 10-month	857 (96% male); mean age: 29.9 ± 6.2	NSP: community level; snowball recruitment; facilitated by peer educators and health workers	NSP delivered by on-site and outreach exchange alongside distribution of brochures, posters, exhibitions, video shows, and lectures on drug and HIV/AIDS knowledge in the rehabilitation centers; health education and peer education in the communities
Luo et al., 2015	cross-sectional: Yunnan, Guangdong, Sichuan, Guizhou and Hunan provinces and Xinjiang Uygur and Guangxi Zhuang Autonomous Regions, China; 2010–2011	3494 (85.9% male); median age: 35	NSP: level not reported; peer-driven recruitment; facilitator not reported	not reported
Mehrabi et al., 2020	cross-sectional: Kermanshah, Iran; 2017	606 (100% male); mean age: 36.7±8.5	NSP: individual level; convenience sampling recruitment; facilitator not reported	NSP delivered by on-site exchange
Naserirad et al., 2020	cross-sectional: Golestan, Ardabil, Sistan and Baluchestan Province, Iran; 2018–2019	634 (94% male); age: 18–33 years (49%);	NSP: community level; respondent-driven sampling recruitment; facilitated by peer outreach teams	NSP delivered by on-site and outreach exchange
Nazari et al., 2016	cross-sectional: Kermanshah, Iran; 2014	455 (100% male); majority under 30 years old (54.6%)	NSP: community level; convenience sampling recruitment; facilitator not reported	NSP delivered by on-site and outreach exchange
Noroozi et al., 2018	cross-sectional: Kermanshah, Iran; 2014	500 (100% male); mean age: 33.5±7.6	NSP: individual level; snowball sampling recruitment; facilitator not reported	NSP delivered by on-site and outreach exchange
Noroozi et al., 2019	cross-sectional: Tehran and Kermanshah, Iran; 2016	960; age: mostly 30–40	NSP: individual level; convenience and snowball sampling recruitment; facilitator not reported	NSP delivered by on-site exchange
Otiashvili et. al., 2006	time-series: Tbilisi, Georgia; baseline, 3-, 6-month	300 enrolled: 140 completed intake assessment; 132 at 3 months; 100 at 6 months	NSP: individual level; recruitment not reported; facilitated by peer educators	NSP delivered by on-site exchange alongside booklets and fliers about safe injection practice and overdose prevention
Power et al., 2004	serial cross-sectional: Sverdlovsk, Russia; 1999–2002; baseline, 12-, 19-, 31-month	1342 (79% male); mean age: 23.4 (intervention, N=810); 22.0 (control, N=532)	NSP: Individual level; convenience sampling, snowball recruitment; facilitated by NGO	NSP delivered by on-site and outreach exchange
Reynolds et al., 2000	cross-sectional: Kathmandu, Nepal; 1997–1998	200	The Life Saving and Life Giving NSP: level and recruitment not reported; facilitated by NGO	NSP delivery method not reported, but delivered alongside HIV education, counselling, referral, outreach work, and distribution of bleach, sterile water and condoms
Ruan et al., 2013	prospective cohort study: Xichan, Sichuan, China; pre-intervention cohort (2002–2004), post-intervention cohort (2006–2008); every 6-month	pre-intervention: 333 (81.7% male); mean age 28±5; post-intervention: 376 (77.4% male); mean age 32±5	harm reduction program: community level; peer-driven recruitment; facilitated by peer educators, health workers	NSP delivered by on-site exchange alongside methadone maintenance therapy, condom distribution, and addiction counseling
Samo et al., 2013	case-control study: Karachi, Pakistan; 2013	140 (100% male); mean age 34.17±10.74	NSP: individual level; convenience sampling recruitment; facilitator not reported	NSP delivered by on-site exchange
Saukhat et al., 2000	serial cross-sectional: Rostov-on-Don, Russia; 1999; first survey in 1997–1998; second survey in 1999	465: 180 in the first survey; 285 in the second survey 20–29 years: 43.3%, 30–39 years: 35.9%	NSP: Individual level; convenience sampling recruitment; facilitated by NGO and peer educators	NSP delivered by on-site exchange alongside psychosocial consulting on HIV infection; distribution of alcohol tissues, wraps of sterile cotton, condoms, and education materials that are specifically developed for people who inject drugs.
Sergeyev et al., 1999	time-series: Yaroslavl, Russia; 1996–1998; three follow-up time-points not specified	484 (71.7% male); 20–29 years: 60.2%	peer-driven outreach intervention: individual level; peer-driven recruitment; facilitated by peer educators	NSP delivered by outreach distribution alongside distribution of educational pamphlet, condoms, disinfectant, and band aids
Sharma et al., 2003	cross-sectional: Manipur, India; 2000	201 (about 90% male); mean age: 26 (intervention); 29 (control)	The Rapid Intervention and Care: individual level; convenience sampling recruitment; facilitator not reported	NSP delivered by on-site and outreach exchange alongside condom promotion and provision, referral for HIV testing, STI treatment, home-based care, and counseling
Strathdee et al., 2011	cross-sectional: Tijuana, Ciudad Juarez, Mexico; 2008–2010	620 (100% female); median age: 30 (IQR= 27–40)	NSP: individual level; convenience sampling recruitment; facilitator not reported	not reported
Vassilev et al., 2006	cross-sectional: Sofia, Bulgaria; 2003	773 (79.4% male); mean age: 25.9±5.7	NSP: individual level; convenience sampling recruitment; facilitator not reported	NSP delivered by on-site exchange
Vazirian et al., 2005	cross-sectional: Tehran, Iran; 2004	105 (100% male); mean age: 32	NSP: individual level; convenience sampling recruitment; facilitated by NGO	NSP delivered by on-site and outreach exchange alongside water vials, filters, alcohol pads, and 2 condoms provided at each visit
Wu et al., 2007 (Guangdong)	cluster randomized controlled trial: Guangdong, China; 2002–2003; baseline, 9-month	428 (>90% male); mostly 20–30 years	needle social marketing strategy: community level; peer-driven recruitment; facilitated by peer educators and health workers	NSP delivered by on-site and outreach exchange alongside health education, educational pamphlet, displaying posters, videos etc. related to drug abuse and HIV
Wu et al., 2007 (Guangxi)	cluster randomized controlled trial: Guangxi, China; 2002–2003; baseline, 9-month	395 (>90% male); mostly 20–30 years	needle social marketing strategy: community level; peer-driven recruitment; facilitated by peer educators and health workers	NSP delivered by on-site and outreach exchange alongside health education, educational pamphlet, displaying posters, videos etc. related to drug abuse and HIV
Wu et al., 2017	cross-sectional: Yunnan, China; 2006	685 (78.7% male); mean age: 31.5	NSP: individual level; recruitment not reported; facilitated by NGO	NSP delivered by on-site exchange
Zamani et al., 2010	cross-sectional: Tehran, Iran; 2005	419 (100% male); mean age: 30	NSP: community level; peer-driven recruitment; facilitated by NGO	NSP delivered by on-site and outreach exchange
Zhang et al., 2013	cross-sectional: Hunan, China; 2010	402 (82.8% male); mean age: 34.6±7.1	NSP: community level; recruitment not reported; facilitated by peer educators, health workers	Not reported
Zhou et al., 2009	serial cross-sectional Sichuan, China; 2002–2005; 12-, 24-, 36-month	2066 (77.3% male); mean age: intervention: 30.7±6.5, control: 28.2±5.6	Expanded and Comprehensive Response: community level; recruitment not reported; facilitated by peer educators and health workers	NSP delivered by on-site exchange alongside health education, condom distribution, material distribution, and methadone maintenance therapy

HCV: hepatitis C virus

IQR: interquartile range.

NGO: non-governmental organization.

NSP: needle and syringe program, or needle and syringe exchange program.

STI: sexually transmitted infection

Intervention characteristics varied across studies. Peer-driven and social networks were the most common methods to recruit people who inject drugs. Peer educators, non-governmental organizations, and outreach health workers usually partnered with the NSP intervention to facilitate on-site and outreach needle/syringe exchange. Secondary exchange by peers and pharmacy vouchers also served as sources of clean needles and syringes. Most participants were also exposed to other intervention activities, such as HIV information sessions, health education, distribution of free condoms and injection equipment, and voluntary HIV counseling and testing.

Table 2 presents a summary of comparative outcomes reported, including needle-sharing behaviors and HIV and other STI-related outcomes, by study. Overall, 21 interventions presented outcomes measured at the individual-level and 19 at the community-level. We could not determine level of outcome measurements in two interventions. Among 30 interventions that reported needle-sharing behavior outcomes, 15 did not specify the direction of sharing. One reported the outcome of lifetime/ever shared needles/syringes. The other 14 included descriptions of how needles/syringes were shared, which we used to assign receptive and distributive needle-sharing outcomes. The most frequent additional HIV-related outcomes reported were HIV incidence and HIV prevalence, while a few studies measured hepatitis B, hepatitis C, syphilis, and chlamydia prevalence.

Table 2.

Comparative outcomes reported by included studies^a

Study citation	Needle-sharing behaviors					Shared injecting paraphernaliac	Other injection outcomesd	HIV- or STI-related outcomes
	Overall	Distributive	Receptive	Combinedb	Non-directional			HIV incidence	HIV prevalence		HIV testing		Other STIs incidence or prevalencee
Badrieva et al., 2007
Broadhead, et al., 2006 (Bragino)
Broadhead et al., 2006 (Rybinsk)
Des Jarlais et al., 2002
Des Jarlais et al., 2007
Eicher et al., 2000
Ganju et al., 2016f
Goswami et al., 2014 (Manipur)
Goswami et al., 2014 (Nagaland)
Hammett et al., 2006 (China)
Hammett et al., 2006 (Vietnam)
Hammett et al., 2012 (Nang Min County)
Hammett et al., 2012 (Lang Son)
Hammett et al., 2012 (Ha Giang Town)
Handanagic et al., 2017
Jenkins et al., 2001 (Dhaka)
Jenkins et al., 2001 (Rajshahi)
Khan et al., 2011
Lin et al., 2004
Luo et al., 2015g
Mehrabi et al, 2020
Naserirad et al., 2020h
Nazari et al., 2016
Noroozi et al., 2018i
Noroozi et al., 2019
Otiashvili et. al., 2006
Power et al., 2004
Reynolds et al., 2000
Ruan et al., 2013
Samo et al., 2013
Saukhat et al., 2000
Sergeyev et al., 1999
Sharma et al., 2003
Strathdee et al., 2011
Vassilev et al., 2006
Vazirian et al., 2005j
Wu et al., 2007 (Guangdong)
Wu et al., 2007 (Guangxi)
Wu et al., 2017
Zamani et al., 2010
Zhang et al., 2013
Zhou et al., 2009

^a.Cell shade in light green: outcome reported; in dark green: outcome reported and included in meta-analysis.

^b.Combined receptive and distributive outcomes per study, when both reported.

^c.Shared injecting paraphernalia include containers, cooker, cotton, filter, rinse water, solutions, etc.

^d.Other injection outcomes include return of needles, frontloading/backloading, needle sharing frequency, injection frequency, number of sharing partner, etc.

^e.Other STIs include chlamydia, gonorrhea, HBV, HCV, and syphilis.

^f.Outcome reported is HIV testing prior to the survey. Duration or time frame of the measurement is unknown.

^g.Needle-sharing outcome reported is ever shared needles/syringes.

^h.Unique categorical measure of exposure: accessibility of NSP services (low, middle, high).

^i.Unique categorical measure of exposure: low or high NSP user.

^j.Unique categorical measure of exposure: whether received 7 or more free needles/syringes.

Table 3 presents the risk of bias assessment for each intervention. Most interventions did not employ a cohort design (N=34) or perform randomization when enrolling participants and assigning interventions (N=40). Most interventions (N=32) were evaluated through cross-sectional or serial cross-sectional methods. Therefore, those study results are subject to limitations due to cross-sectional design, weak participant representativeness, and non-equivalence of comparison groups (if applicable). One prospective cohort study, two before-after studies and two time-series studies were rated low rigor because of non-randomized designs and low follow-up rates, which were all below 80%. Only one of two non-randomized trials had a follow-up rate higher than 80%. Only one randomized trial was included, which showed the greatest rigor. This cluster randomized controlled trial in China investigated a needle social marketing strategy at the community level among 823 young people who inject drugs from 2002 to 2003.⁴³ Although its findings suggested significant NSP intervention benefits, it did not randomly select participants for assessment and the comparison groups were not equivalent at baseline.⁴³ In fact, none of the included studies reported equivalent group characteristics at baseline.

Table 3.

Quality assessment of included articles.

Domain	Study design			Participant representativeness			Equivalence of comparison groups
Study	Cohort	Control or comparison group	Pre/post intervention data	Random assignment of participants to the intervention	Random selection of participants for assessment	Follow-up rate of 80% or more	Comparison groups equivalent on socio-demographics	Comparison groups equivalent at baseline on outcome measure
Badrieva et al., 2007*	No	Yes	No	No	No	NA	No	NA
Broadhead, et al., 2006 (Bragino)*	Yes	No	Yes	No	No	No	NA	NA
Broadhead et al., 2006 (Rybinsk)*	Yes	No	Yes	No	No	No	NA	NA
Des Jarlais et al., 2002	No	No	Yes	No	Yes	NA	NA	NA
Des Jarlais et al., 2007	No	No	Yes	No	Yes	NA	NA	NA
Eicher et al., 2000	No	Yes	No	No	No	NA	No	NA
Ganju et al., 2016	No	Yes	No	No	Yes	NA	NR	NA
Goswami et al., 2014 (Manipur)*	No	Yes	No	No	Yes	NA	No	NA
Goswami et al., 2014 (Nagaland)	No	Yes	No	No	Yes	NA	No	NA
Hammett et al., 2006 (China)*	No	No	Yes	No	No	NA	NA	NA
Hammett et al., 2006 (Vietnam)*	No	No	Yes	No	No	NA	NA	NA
Hammett et al., 2012 (Nang Min County)	No	No	Yes	No	No	NA	NA	NA
Hammett et al., 2012 (Lang Son)	No	No	Yes	No	No	NA	NA	NA
Hammett et al., 2012 (Ha Giang Town)	No	No	Yes	No	No	NA	NA	NA
Handanagic et al., 2017	No	Yes	No	No	No	NA	NR	NA
Jenkins et al., 2001 (Dhaka)*	No	Yes	No	No	Yes	NA	NR	NA
Jenkins et al., 2001 (Rajshahi)*	No	Yes	No	No	Yes	NA	NR	NA
Khan et al., 2011	No	Yes	No	No	Yes	NA	NR	NA
Lin et al., 2004*	Yes	Yes	Yes	No	No	Yes	NR	No
Luo et al., 2015	No	Yes	No	No	No	NA	No	NA
Mehrabi et al, 2020*	No	Yes	No	No	No	NA	No	NA
Naserirad et al., 2020	No	Yes	No	No	No	NA	NR	NA
Nazari et al., 2016*	No	Yes	No	No	No	NA	NR	NA
Noroozi et al., 2018	No	Yes	No	No	No	NA	NR	NA
Noroozi et al., 2019	No	Yes	No	No	No	NA	NR	NA
Otiashvili et.al., 2006*	Yes	No	Yes	No	No	No	NA	NA
Power et al., 2004*	No	Yes	Yes	No	No	NA	NR	NR
Reynolds et al., 2000	No	Yes	No	No	No	NA	NR	NA
Ruan et al., 2013	Yes	No	Yes	No	No	No	NA	NA
Samo et al., 2013	No	Yes	No	No	No	NA	No	NA
Saukhat et al., 2000*	No	No	Yes	No	Yes	NA	NA	NA
Sergeyev et al., 1999	Yes	No	Yes	No	No	No	NA	NA
Sharma et al., 2003*	No	Yes	No	No	No	NA	NR	NA
Strathdee et.al., 2011	No	Yes	No	No	No	NA	NR	NA
Vassilev et al., 2006	No	Yes	No	No	No	NA	NR	NA
Vazirian et al., 2005	No	Yes	No	No	No	NA	NR	NA
Wu et al., 2007 (Guangdong)*	Yes	Yes	Yes	Yes	No	Yes	No	No
Wu et al., 2007 (Guangxi)*	Yes	Yes	Yes	Yes	No	Yes	No	No
Wu et al., 2017	No	Yes	No	No	No	NA	NR	NA
Zamani et al., 2010*	No	Yes	No	No	Yes	NA	No	NA
Zhang et al., 2013*	No	Yes	No	No	No	NA	NR	NA
Zhou et al., 2009*	No	Yes	Yes	No	No	NA	NR	NR

^*Study outcomes included in meta-analysis.

Below, we present findings for needle-sharing, HIV incidence and prevalence, and other HIV-related outcomes.

Needle/syringe sharing

Overall, 25 articles measured the effects of NSP exposure on needle/syringe sharing behaviors.^{16–21,25–27,29–34,38,39,43–50} Of these, 16 articles provided comparable outcome data and were included in meta-analysis; 13 presented additional outcomes that were analyzed qualitatively.

Meta-analysis

Sixteen articles presenting 20 interventions^{19,20,25,27,29,31–34,38,39,43–47} were meta-analyzed, resulting in a total sample size of 11,075, with 6,024 people who inject drugs exposed to the intervention and 5,051 in the reference group. Meta-analysis results are summarized in Table 4.

Table 4.

Summary of meta-analyses of needle-sharing outcomes.

	Individual-level			Community-level
Outcome	n	OR (95% CI)	p-value	n	OR (95% CI)	p-value
Overall sharinga	8	0.25 (0.16–0.39)	<0.001	12	0.39 (0.22–0.69)	0.001
Receptive sharing	3	0.39 (0.23–0.65)	<0.001	6	0.30 (0.11–0.78)	0.014
Distributive sharing	2	0.61 (0.31–1.21)	0.160	5	0.40 (0.14–1.15)	0.090
Non-directional sharingb	5	0.18 (0.11–0.31)	<0.001	8	0.42 (0.19–0.91)	0.027
Combined sharingc	2	0.54 (0.34–0.87)	0.011	4	0.31 (0.09–1.10)	0.069

^aReceptive, Distributive, and Any (non-directional) needle sharing combined per study, when any of the three were available for a study.

^bNon-directional needle sharing estimate reported.

^cReceptive and Distributive needle sharing combined per study, when both were reported.

For overall needle-sharing behaviors (combining receptive, distributive, and non-directional needle sharing), NSP exposure was associated with a statistically significant protective effect when assessed at both the individual- and community-level (Figure 2; individual-level: odds ratio [OR]=0.25, 95% confidence interval [CI]=0.16–0.39, p<0.001, I²=86%, 8 trials, n=3,947; community-level: OR=0.39, CI=0.22–0.69, 12 trials, n=6,850, p=0.001, I²=95%).^{19,20,25,27,29,31–34,38,39,43–47} However, there was high statistical heterogeneity, likely due to significant differences in outcome measurements, intervention modality, intervention duration, and exposure types.

Figure 2.

Meta-analysis of overall needle sharing (receptive, distributive, and non-directional needle sharing combined), exposure to NSP versus control/comparison group.

Separate analyses examined pooled effects of NSP exposure on receptive sharing and distributive sharing. For receptive sharing, we combined the effects of nine interventions in meta-analysis. We found that NSP exposure was associated with a lower odds of receptive sharing behaviors (Figure 3) at both the individual-level (OR=0.39, CI=0.23–0.65, p<0.001, 3 trials, I²=70%) and the community-level (OR=0.30, CI=0.11–0.78, p=0.001, 6 trials, I²=95%).^{19,27,29,32,34,39,44} Seven interventions examining the associations between NSP exposure and distributive sharing did not show a statistically significant association when combined in meta-analysis at either the individual- or community-level (Figure 4; individual-level: OR=0.61, CI=0.31–1.21, p=0.16, 2 trials, I²=72%; community-level: OR=0.40, CI=0.14–1.15, p=0.09, 5 trials, I²=96%).^{19,27,29,32,39}

Figure 3.

Meta-analysis of receptive needle sharing, exposure to NSP versus control/comparison group.

Figure 4.

Meta-analysis of distributive needle sharing, exposure to NSP versus control/comparison group.

When we combined the effect sizes from studies that did not specify directionality of sharing (Figure 5),^{20,25,29,31,33,38,43,45–47} a protective association between exposure to NSP and needle/syringe sharing was observed among the five interventions at the individual-level (OR=0.18, CI=0.11–0.31, p<0.001, 5 trials, I²=83%). We also observed 58% lower odds of needle/syringe sharing at the community-level (OR=0.42, CI=0.19–0.91, p=0.027, 8 trials, I²=96%), although heterogeneity was significant.

Figure 5.

Meta-analysis of non-directional needle sharing, exposure to NSP versus control/comparison group.

In addition, six interventions reported both receptive and distributive sharing outcomes.^19,27,29,39 The pooled within-study effect of needle sharing was estimated per study, and we found statistically significant lower odds of this combined receptive and distributive needle/syringe sharing outcome among people who inject drugs who participated in NSP (OR=0.54, CI=0.34–0.87, p=0.01, 2 trials, I²=44%) with moderate heterogeneity at the individual-level. However, the protective association did not retain statistical significance when analyzed at the community-level (OR=0.31, CI=0.09–1.10, p=0.07, 4 trials, I²=97%).

Figure 6 shows a scatter plot of 20 intervention effect estimates (log odds ratio) measured in the meta-analysis against each study’s standard error. We observed a slightly asymmetric shape with more effect estimates from large studies scattered widely in a horizontal line at the middle top of the graph. The funnel plot suggests most included studies with similar standard errors reported protective effects of NSPs, potentially reflecting some reporting bias.

Figure 6.

Funnel plot of standard error by log odds ratio, N=20.

Qualitative synthesis

In addition, 13 articles presented associations between NSP exposure and needle/syringe sharing related outcomes, but could not be included in meta-analysis as described in the methods section above.^{16–21,31,32,34,39,48–50} Overall, six articles reported an association between NSP exposure and reduced needle/syringe sharing behaviors, although their definitions of NSP exposure varied widely.^{16,17,21,48–50} Four cross-sectional studies in Iran published between 2004 and 2020 suggested that the odds of reporting needle/syringe sharing behaviors were lower among people who inject drugs who had high accessibility to NSP, or who received more sterile needles/syringes from NSP, or who attended NSP regularly, compared with their comparator groups.^16,48–50 Another cross-sectional study in 10 central and eastern European cities found statistically significant reductions in receptive needle/syringe sharing.²¹ Less needle/syringe sharing was reported among NSP users in China¹⁸ and Russia¹⁷ as well. Other articles reported on sharing injection paraphernalia (cookers, filters, rinse water, solutions, containers, etc.) but effects of NSP varied widely – positive, neutral and negative – depending on specific study sites, comparators, and length of time from intervention.^{16,17,19–21,31,32,34,39,48,49}

HIV incidence and prevalence

HIV incidence was measured or estimated in 5 articles.^{22,26,34,36,43} One cohort study in China found a statistically significant and substantial decrease in HIV incidence comparing participants in a cohort before NSP was available to a post-NSP cohort.³⁶ A second study from Russia presented government statistics on new HIV diagnoses (per 100,000 population) in the cities where NSPs were implemented, finding decreases in new diagnosis rate over the study period in one city (Verknaya Salda) and increases then decreases in new diagnoses in two other cities (Pervouralsk and Ekaterinburg).³⁴ The remaining studies calculated estimated measures of HIV incidence, based on the assumption that prevalent cases of HIV among new injectors were comparable to incidence. The strongest study design, a cluster-randomized trial in China, found that the NSP intervention was associated with reduced HIV incidence in one study site (Guangdong), but not the other (Guangxi); combined, the effect was not statistically significant.⁴³ One serial cross-sectional study, the Cross-Border Project, conducted in Vietnam and China reported estimated HIV incidence data in two articles.^22,26 Both articles found that estimated HIV incidence decreased substantially to low levels among new injectors by 24 months, although one article²² found that estimated incidence continued to decline by 36 months while the other²⁶ saw some rebound after 36 months, particularly in one study site (Ning Ming, China). Four of these five articles also reported other behavioral or disease outcomes; generally, where HIV incidence decreased, sharing behaviors and other infections decreased as well.

Fifteen articles (20 interventions), mostly cross-sectional studies, reported associations between NSP and HIV prevalence.^{18,22,23,25–27,30,31,35,37,38,40–43} Across these, we found no consistent association between NSP exposure and HIV prevalence.

Other HIV-related outcomes

Several studies reported positive changes in other HIV-related outcomes, such as significantly higher HIV testing uptake.^32,49 Additional studies showed inconsistent associations between NSP interventions and incidence or prevalence of other infectious diseases (chlamydia, gonorrhea, hepatitis B, hepatitis C, and syphilis).^{25,28,31,36,41,43}

Discussion

We systematically reviewed the current evidence base for the effectiveness of NSP at both the individual- and community-level, with a focus on needle-sharing behaviors and other HIV-related outcomes. Although few rigorous trials were identified, we did find a relatively strong evidence base of mostly cross-sectional studies examining the association between NSPs and needle sharing, and a more limited evidence base for HIV incidence, HIV testing, and other infectious diseases.

In meta-analysis, NSPs were associated with fewer people who inject drugs receiving used needles/syringes at both the individual- and community-level. The effect of NSPs on distributive sharing was not statistically significant at either level. When we combined distributive and receptive needle/syringe sharing outcomes, we observed a more significant protective association for NSPs whose outcomes were measured at the individual-level than at the community-level. These findings are consistent with a direct benefit of NSPs for those who use them. However, this benefit seems to be primarily for receptive sharing – the impact of which is substantial enough that it can be seen even when measured at the community-level. Compared with other reviews exploring the effectiveness of NSP in reducing HIV incidence,^5,10,12,51 our approach to stratifying effects by level and directionality provides additional insight on specific risks. While the provision of clean, sterile needles/syringes is fundamental and necessary to reduce receptive sharing, it is not sufficient to reduce distributive sharing.

Beyond needle/syringe sharing, we also found that NSP exposure was associated with lower HIV incidence and higher HIV testing uptake. Findings on HIV prevalence and other infectious diseases were mixed; however, we attribute this to the cross-sectional nature of the study designs which could not assess causal relationships. Overall, findings from our review support the conclusion that NSPs are associated with better HIV outcomes.

We defined NSPs as programs both providing clean, sterile injecting equipment and also disposing of used injection equipment according to safety regulations. The included studies involved a variety of service delivery mechanisms, including pharmacy-based services and peer outreach. While several studies mentioned the cost of clean needles through pharmacies, not enough studies provided sufficient detail to analyze whether distribution modality made a difference in needle-sharing behaviors and other outcomes. However, programs which sell or otherwise provide needles/syringes to people who inject drugs through pharmacies have a number of benefits: they are relatively inexpensive to operate, are often open for long hours, are typically accessible in low-resourced settings including in rural areas, and may provide discretion to their clients. Though some research has already explored the benefits of pharmacy-based compared to facility-based NSP,⁵² further work could examine the benefits of such programs.

In addition to understanding the individual- or community-level measurement and directionality of needle-sharing behaviors, it is also important to understand how NSPs influence needle disposal behaviors. Studies in high-income settings have posited that unsafe needle disposal outside of hospitals or other healthcare settings by people who inject drugs could increase risk of bloodborne infectious disease in the broader community;^53,54 NSPs could ameliorate this risk by reducing unsafe disposal behaviors.⁵⁵ However, none of the included articles reported the exchange rate of used needles/syringes or other unsafe disposal behaviors. No included articles provided details on whether all used needles/syringes were retrieved when study participants interacted with the NSP, or whether study participants were able to obtain new injection kits without having to return used needles/syringes. Future research could explore unsafe disposal behaviors in LMICs.

Because we included studies published from 1990 to 2021, we were able to qualitatively explore temporal changes in NSP implementation in countries like Bangladesh, China, India, Iran, Pakistan, Russia, and Vietnam which provided data from both the 1990s/2000s and more recently. Earlier NSP activities primarily focused on needle distribution and needle exchange, along with counselling and health education/promotion about drug use reduction and HIV prevention. Earlier programs tended to use non-governmental organizations or primary care workers, but more recent studies tended to involve peer educators as facilitators.

Participants in these included studies were predominantly male. While this may reflect local patterns of drug use in these particular LMIC settings, women who inject drugs are often underrepresented in the scientific and medical literature.⁵⁶ Previous research has found gender differences among people who inject drugs, in terms of HIV risk and needle-sharing behaviors.^2,57 Women who inject drugs have a higher likelihood of engaging in unsafe sexual practices and a compounded risk of contracting HIV.⁵⁸ The complex intersection of HIV and injection drug use could benefit from a socio-ecological approach and gender-specific interventions tailored to specific populations’ needs. For example, recruiting female outreach workers or female peer educators, coupled with women-focused behavioral interventions, has been found effective.⁵⁹

Despite heterogeneity in intervention modalities and outcome measures, we found evidence that NSP is associated with reduced receptive and overall needle-sharing behaviors in both individual-level and community-level NSP in LMICs. Interestingly, these behaviors which typically are only measured at the individual-level were strong enough to demonstrate community-level effects. NSPs appear to be an effective intervention component to prevent HIV and other blood-borne infections by improving access to sterile needles/syringes and discouraging people who inject drugs from sharing used injection equipment. There is also a literature gap in African, Latin American, and Caribbean countries where NSP implementation is less common, though HIV prevalence may be high.

NSPs have been promoted as an effective intervention to reduce HIV, hepatitis C, and hepatitis B transmission among people who inject drugs in high-income settings.⁵¹ We identified several LMICs in Asia and Eastern Europe that implemented large-scale NSPs as part of harm reduction programs. According to the latest Global State of Harm Reduction report, although NSPs are officially allowed and scaling up in many Asian and Eastern European countries where our included studies were based, their service capacity is relatively limited because of weak political support and shrinking funds.⁶⁰ For example, illicit drug use is an indictable act in China. If arrested and found to have relapsed, people who inject drugs could be directed to receive compulsory methadone maintenance therapy or referred to mandatory detoxification centers.⁴⁵ Stringent drug-control regulations and police confinement appear to limit access to NSPs by people who inject drugs. Some Asian countries, including Laos, the Philippines, and Mongolia, still prohibit the implementation of NSPs even though NSPs have been shown as an effective harm-reduction approach in nearby countries based on the 2020 Harm Reduction International report.⁶⁰ Punitive drug policies, limited funding, and a high level of stigma also substantially limit the coverage and quality of NSPs in Eastern European countries.^60,61 For example, Bulgaria suspended a NSP site in July 2020 due to unstable funding after one year of reopening operation.⁶² Punitive and stringent regulations around injection drug use also make people who inject drugs seeking NSPs in Russia and Ukraine at high risk of discrimination, police hostility, and arrest,⁶⁰ which could be further exacerbated by the current conflict. Therefore, adapting NSPs to local policies and maintaining NSPs’ accessibility in LMICs deserve further investigation.

This review has several strengths. We explored both individual- and community-level outcome measurements and included different types of needle/syringe-sharing behaviors. Use of an eight-item risk of bias tool allowed us to assess the rigor of included studies. By not restricting inclusion by language and searching multiple databases, we minimized the possibility of missing relevant studies. We also synthesized the most recent empirical evidence of NSP’s impact on HIV-related outcomes through July 2021. Finally, we tried to achieve quality assurance by having two independent study reviewers/assistants screen articles, abstract data, and resolve any discrepancies.

However, our study also has several limitations. First, our search strategy was focused on HIV-related outcomes; we therefore may have missed other studies examining the impact of NSPs on needle/syringe sharing behaviors without mentioning HIV, or that focused exclusively on hepatitis C or other outcomes. Second, most of the included studies were cross-sectional in design and did not provide data adjusted for confounders, which limits the inferences we can draw. Further, there was significant heterogeneity across studies in intervention length/frequency, recruitment strategy, choice of facilitators/implementers, delivery methods, and other concurrent intervention activities. Due to the limited number of studies for each outcome, we were also unable to conduct meta-regression to examine study-level factors associated with positive outcomes. Future studies could consider evaluating NSP designs or procedures in difference settings and regions. How to customize program designs and intervention activities considering local patterns of drug use, population characteristics, and regulations to enhance NSP’s positive impacts is worth further academic attention. Third, we were unable to include several effect measures in meta-analysis due to insufficient information, and we failed to obtain responses from correspondent authors after email inquiries. This reduced the number of included interventions for the meta-analysis. Fourth, there were no standardized measurements for NSP exposure, and most studies relied on participants’ self-reported behavioral outcomes. Due to the nature of the intervention, convenience and purposive sampling methods were also the most used recruitment methods. Therefore, the pooled estimates of NSP impacts are likely affected by measurement bias, social-desirability bias, and selection bias. Finally, implementation science research in both high- and low-resource settings has generally shifted from focusing on NSP to providing “comprehensive HIV prevention and care” including NSP, opioid agonist treatment, methadone maintenance treatment, and antiretroviral therapy for people who inject drugs;⁶³ two recent studies of such comprehensive programs in LMICs (HPTN 074 in Eastern Europe and Southeast Asia,⁶⁴ the DRIVE study in Vietnam⁶⁵) have found very low HIV incidence among people who inject drugs. Better understanding how NSPs contribute to the overall success of such comprehensive packages is challenging but worth future investigation.

In conclusion, we examined the association of NSPs with HIV-related outcomes in LMICs over the past three decades. We found that exposure to NSP across studies was overall associated with reduced needle/syringe sharing and reduced HIV incidence among people who inject drugs. The meta-analysis also found that NSPs showed a statistically significant beneficial association with reducing receptive sharing, non-directional sharing, and overall needle-sharing behaviors among people who inject drugs when measured at the individual-level. However, we did not find a statistically significant protective association with distributive needle-sharing behaviors at either the individual- or community-level.

Appendix A. Search strategy for all databases, last searched on July 25, 2021

PubMed

(“needle exchange” OR “needle distribution” OR “needle sales” OR “syringe sales” OR “syringe distribution” OR “syringe exchange” OR “syringe-exchange” OR “needle-exchange” OR “needle syringe programs” OR “needle syringe program” OR “needle syringe programme” OR “needle exchange programmes”) AND (HIV or AIDS)

PsycINFO

(“needle exchange” OR “needle distribution” OR “needle sales” OR “syringe sales” OR “syringe distribution” OR “syringe exchange” OR “syringe-exchange” OR “needle-exchange” OR “needle syringe programs” OR “needle syringe program” OR “needle syringe programme” OR “needle exchange programmes”) AND (HIV or AIDS)

CINAHL

(“needle exchange” OR “needle distribution” OR “needle sales” OR “syringe sales” OR “syringe distribution” OR “syringe exchange” OR “syringe-exchange” OR “needle-exchange” OR “needle syringe programs” OR “needle syringe program” OR “needle syringe programme” OR “needle exchange programmes”) AND (HIV or AIDS)

Sociological Abstracts

(“needle exchange” OR “needle distribution” OR “needle sales” OR “syringe sales” OR “syringe distribution” OR “syringe exchange” OR “syringe-exchange” OR “needle-exchange” OR “needle syringe programs” OR “needle syringe program” OR “needle syringe programme” OR “needle exchange programmes”) AND (HIV or AIDS)

Embase

(‘needle exchange’ OR ‘needle distribution’ OR ‘needle sales’ OR ‘syringe sales’ OR ‘syringe distribution’ OR ‘syringe exchange’ OR ‘syringe-exchange’ OR ‘needle-exchange’ OR ‘needle syringe programs’ OR ‘needle syringe program’ OR ‘needle syringe programme’ OR ‘needle exchange programmes’) AND (HIV or AIDS)

Hand search of the following journals:

AIDS

AIDS and Behavior

AIDS Care

AIDS Education and Prevention

the International Journal of Drug Policy

Availability of data and material

Available upon request to the corresponding author.