Table 1 Characteristics of systematic reviews on the impact of harm minimization interventions (n = 33)

NSEP approach

Abdul-Quader [22]

SLR

15 interventional studies

Any health setting

NSEP

• HIV prevalence: N = 12/15 studies (80%) show reductions of − 0.6% to − 43%

• HCV prevalence:

N = 5/7 studies (71%) show reductions of − 13% to − 30%

Moderate

Results support NSEP as a structural-level intervention to reduce HIV and HCV infections

Yes

Aspinall [23]

SLRMA

12 observational studies (n = 12,000)

Pharmacies

Outreach services

NSEP

• HIV transmission, all studies: RR 0.66 [0.43–1.01], I2 = 76%

• HIV transmission, high quality studies: RR 0.42 [0.22–0.81], I2 = 79%

Moderate

NSEP is effective in reducing HIV transmission, although other harm reduction interventions can contribute to this

Yes

Davis [24]

SLRMA

6 observational studies (n = 2437)

Any health setting

NSEP

• Association of HCV seroconversion and NSEP participation:

OR 0.51 [0.05–5.15], I2 = 88%

Low

Evidence is mixed and inconclusive; no consistent association on NSEP impact on HCV was found

Inconclusive

Des Jarlais [25]

SLR

11 interventional and observational studies

Low- and middle-income settings

NSEP

• HIV prevalence: N = 5/8 studies (63%) show reductions of − 3% to − 15%

• HCV prevalence:

N = 3/4 studies (75%) show reductions of − 4.2% to − 10.2%

–

While not fully consistent and homogeneous, overall evidence support the effectiveness of NSEP in reducing HIV/HCV in these countries

Yes

Gibson [26]

SLR

42 interventional and observational studies

Any health setting

NSEP

• Sharing needles/syringes: N = 28/42 studies (67%) show reductions

–

Studies show a slightly benefit of NSEP; but methodologic rigor needs to be improved (true impact of the interventions: unknown)

Inconclusive

Gillies [27]

SLR

13 observational studies

Outreach services, SCF/SIF

NSEP

• Sharing needles/syringes: aOR ranging from 0.3 to 0.9

Low

Studies suggest a reduced likelihood of sharing paraphernalia. However, estimates are uncertain given studies’ low quality

Inconclusive

Jones [28]

SLR

16 RCT or observational studies

Any health setting

NSEP

• Injecting behavior:

not statistically significant

• BBV incidence/prevalence:

not statistically significant

• Drug treatment entry:

not statistically significant

–

Most studies (n = 11) showed no evidence of impact of NSEP or syringe dispensation policies on injecting behaviors. Studies are heterogeneous

No

Ksobiech [29]

SLRMA

31 observational studies(n = 52,678)

Any health setting

NSEP

• Injection drug use: Weigh. correlat − 0.189 (SE 0.05)

• Injection frequency:

Weigh. correlat: − 0.024 (SE 0.04)

• Sharing needles/syringes: Weigh. correlat: − 0.059 (SE 0.02)

• Risky behavior: Weigh. correlat: + 0.016 (SE 0.07)

–

NSEP attendance was inversely related to the reduction of most harmful outcomes, however, given the high heterogeneity among studies, data should be carefully interpreted

Partially yes

Sawangjit [30]

SLRMA

14 observational studies (n = 7035)

Pharmacies

NSEP

• Sharing needles/syringes, all studies: OR 0.50 [0.34–0.73], I2 = 60%

• Sharing needles/syringes, high quality studies: OR 0.52 [0.32–0.84], I2 = 41%

• HCV prevalence: OR 0.26 [0.18–0.38], I2 = 0%

• HIV prevalence: OR 0.56 [0.18–1.77], I2 = 92.7%

Low-moderate

Pharmacy-based NSEP programs appear to be effective for reducing risk behaviors, but their effect on HIV/HCV prevalence and economic outcomes are still unclear

Partially yes

OAT approach

Gowing [31]

SLR

38 interventional studies (n = 12,400)

Any health setting

OAT

• Injection drug use:

Reduced by 20–60%

• Illicit opioid use:

Reduced by 32–69%

• Sharing needles/syringes:

Reduced by 25–86%

Low

OAT may reduce drug-related behaviors with a high risk of HIV transmission. The lack of data from RCT limits the strength of the evidence

Yes

Hedrich [32]

SLR

21 RCT or observational studies

Prisons

OAT

• HCV or HIV incidence:

not statistically significant

• Drug treatment entry:

significant positive effect of OAT

• Sharing needles/syringes:

significant positive effect of OAT

• Injecting behavior:

significant positive effect of OAT

• Illicit opioid use:

significant positive effect of OAT

Low-moderate

The evidence is overall consistent and supports the use of OAT/OST to risky behaviors. Yet, for some outcomes, evidence is inconsistent (crime, re-incarceration) or weak (HCV incidence, mortality)

Partially yes

Karki [33]

SLR

12 RCT or observational studies (n = 16,195)

Any health setting

OAT

• Risky behavior reduction:

significant positive effect of MOUD/OAT

–

MOUD is associated with significant decrease in injecting drug use and sharing of injecting equipment. Evidence for other outcomes limited

Partially yes

Larney [34]

SLR

5 interventional studies

Prisons

OAT

• Injection drug use: Reduced by 55–75%

• Sharing needles/syringes: Reduced by 47–73%

• Illicit opioid use:

Reduced by 62–91%

Low-moderate

There may be a role for OAT in preventing HIV in prisons, but rigorous research addressing this question is required

Yes

MacArthur [35]

SLRMA

12 observational studies

Any health setting

OAT

• HIV transmission:

RR 0.60 [0.42–0.85], I2 = 23%

• HIV incidence:

RR 0.46 [0.32–0.67], I2 = 60%

Low-moderate

OAT by means of MOUD is associated with reduction in the risk of HIV infection among PWID, although some heterogeneity/bias among studies exist

Yes

Moore [36]

SLRMA

24 interventional studies (n = 807)

Prisons

OAT

• Injection drug use:

OR 0.26 [0.12–0.56], I2 = 62%

• Illicit opioid use:

OR 0.22 [0.15–0.32], I2 = 0%

• Recidivism:

OR 0.93 [0.51–1.68], I2 = 46%

Moderate

Data support the use of MOUD/OAT in prisons for reducing some outcomes; yet, additional work is needed to understand the its impact on other health risk behaviors

Partially yes

Behavioral or educational interventions

Copenhaver [37]

SLRMA

37 RCT (n = 10,190)

Any health setting

Behavioral interventions

• Injection drug use: WM 0.08 [0.03–0.13], p < 0.001

• Drug treatment entry: WM 0.11 [0.02–0.21], p = 0.013

• Sharing needles/syringes: WM 0.03 [− 0.04, 0.10],p = 0.062

• Frequency of trading sex for drugs: WM 0.33 [0.10–0.57], p = 0.052

–

Behavioral interventions were effective in reducing some HIV-risk behaviors; yet no benefits were observed for reducing needle or syringe borrowing

Partially yes

Deuba [38]

SLRMA

43 RCT (n = 15,642)

Any health setting in low-income countries

Behavioral interventions (peer-based)

• HIV prevalence, PWID studies: 11.9% [8.3–16.7]

• Unsafe injections, PWID studies:

OR 0.942 [0.726–1.222], I2 = 0%

Moderate

None of the included interventions were found to be effective for reducing unsafe injection practices among PWID in low-income countries

No

Gilchrist [39]

SLRMA

24 RCT (n = 12,840)

Any health setting

Psychosocial interventions

• Any injecting behavior:

SMD − 0.29 [− 0.42, − 0.15], I2 = 61%

• Sharing needles/syringes:

SMD − 0.43 [− 0.69, − 0.18], I2 = 68%

• Injecting drug use:

SMD − 0.17 [− 0.35, 0.00], I2 = 61%

• Sexual risk behavior:

SMD − 0.19 [− 0.30, 0.01], I2 = 58%

• HIV testing/counseling:

SMD − 0.24 [− 0.44, − 0.03], I2 = 0%

Low-moderate

Psychosocial interventions appear to reduce some risky behavior outcomes, but moderate heterogeneity was reported. Such interventions can be used to prevent BBV

Partially yes

Meader [40]

SLRMA

35 interventional studies (n = 11,867)

Any health setting

Psychosocial interventions

Educational interventions

• Injection risk behaviour

SMD − 0.04 [− 0.31, 0.23], I2 = 69%

• Sexual risk behavior:

SMA − 0.12 [− 0.33, 0.08], I2 = 49%

Low-moderate

Both multi-session psychosocial interventions and standard education can reduce injection and sexual risk behaviour; minimal differences between interventions were found

Yes

Prendergast [41]

SLRMA

18 controlled studies

Any health setting

Behavioral interventions

• HIV risk-reduction, overall:

WM 0.31 [0.20–0.42], p < 0.01

• Injection drug use: WM 0.04 [− 0.14, 0.22], p > 0.05

• Sexual risk behavior:

WM 0.26 [0.15–0.38], p < 0.01

–

The overall effect sizes suggests that HIV interventions within drug treatment have a reliable effect. Yet, heterogeneity among studies is high

Partially yes

Sacks-Davis [42]

SLR

6 RCT (n = 5472)

Any health setting

Behavioral interventions

• HCV incidence:

not statistically significant

• Sexual risk behavior:

not statistically significant

• Injecting behavior:

significant positive effect of intervention

Low-moderate

Behavioral approach can have some effects on HCV transmission; yet considerable variations in study design, outcomes, magnitude/direction of effect exist

Inconclusive

Semaan [43]

SLRMA

33 interventional studies

Any health setting in the United States

Behavioral interventions

• Sexual risk behavior:

OR 0.86 [0.76–0.98], p < 0.05, I2 = 47%

–

Behavioral-based interventions may lead to reduction on sexual risk behavior among drug users, but data is heterogeneous

Yes

SCF/SIF studies

Kennedy [44]

SLR

47 observational studies

SCF/SIF

SCF/SIF

• Overdose mortality/morbidity: N = 6/8 studies (75%) show reductions

• Risky behavior:

N = 4/9 studies (44%) show reductions

–

These facilities may mitigate overdose-related harms and unsafe drug use. Yet, no meta-analyses were performed; outcomes are not standardized

Inconclusive

Levengood [45]

SLR

22 observational studies

SCF/SIF

SCF/SIF

• Overdose mortality/morbidity:

N = 4/5 studies (80%) show reductions

• Injecting behavior:

N = 6/7 studies (86%) show reductions

• Drug treatment entry:

N = 6/7 studies (86%) show increases

• Crime and public nuisance:

N = 5/7 studies (70%) show stability

Low-moderate

These facilities may reduce overdose-related risks and improve access to care, while not increasing crime or publicnuisance to the surrounding community. Inconsistent outcomes across studies prevent further conclusions

Inconclusive

THN approach

McAuley [46]

SLRMA

9 observational studies

SCF/SIF

Drug treatment centers

Prisons

THN

• Proportion of naloxone use (every 3 months / 100 trained users): WM: 0.092 [0.052–0.131]

–

Around 9% of naloxone kits distributed are likely to be used for peer administration within three months’ supply. The evidence for THN is limited

Inconclusive

McDonald [47]

SLR

22 observational studies (n = 2912)

Any health setting

THN

• Overdose reversals:

96.3% [95.5–97.1] (n = 2249/2336 THN administrations)

• Deaths:

0.9% [0.5–1.2] (n = 24/2336 THN administrations)

Low-moderate

THN programmes can reduce overdose mortality with a low rate of adverse events. However, there is a large variability in the size and quality studies

Inconclusive

Other combined interventions and interventions’ comparisons

Bouzanis [48]

SLR

97 interventional and observational studies

Any health setting in Canada

NSEP

SCF/SIF

OAT

PoC

Behavioral interventions

Evidence indicates advantages of multifaceted care programmes for PWID, which include harm reduction, medical/pharmaceutical treatments, social support and education

–

The included studies call for exploratory work in facilitators and barriers to treatment and care, more robust study designs, and attention to contextual factors and more complex interventions

Inconclusive

Cross [49]

SLRMA

18 interventional studies (n = 7926)

Any health setting

Educational interventions

NSEP

• Risky behavior, educational studies: WM: 0.749 [0.708–0.790]

• Risky behavior, NSEP studies:

WM: 0.279 [0.207–0.352]

–

Both interventions had a positive impact on reducing HIV risk behaviors, however, results are dependent upon research design, outcomes, follow-up

Yes

Hagan [50]

SLRMA

26 interventional and observational studies

Any health setting

NSEP

OAT

Behavioral interventions

• HCV incidence, behavioral studies: RR: 1.18 [0.77–1.81], I2 = 0%

• HCV incidence, OAT studies:

RR: 0.60 [0.35–1.03], I2 = 45%

• HCV incidence, NSEP studies:

RR: 1.62 [1.04–2.52], I2 = 81%

• HCV incidence, multi-component:

RR: 0.25 [0.07–0.83], I2 = 55%

Moderate

Multi-component interventions using strategies that combined substance-use treatment and support for safe injection were most effective at reducing HCV seroconversion

Yes

McNeil [51]

SLR

21 qualitative studies (n = 800)

Urban or semi-urban settings

NSEP

SCF/SIF

Behavioral interventions (peer-based)

Interventions are potentially associated with:

(1) providing refuge from streets

(2) enabling safer injecting environments

(3) mediating access to resources and health care services

(4) constrained by drug prohibition and law enforcement activities

–

Safer environment interventions may mitigate drug-related harms. Further qualitative and quantitative evidence syntheses in this field are needed

Inconclusive

Platt 2017, [52,53,54]

SLRMA

28 observational studies (n = 6279)

Any health setting

NSEP

OAT

• HCV incidence, all studies: RR 0.26 [0.07–0.59], I2 = 80%

• HCV incidence, OAT studies: RR 0.50 [0.40–0.63], I2 = 0%

• HCV incidence, NSEP studies: RR 0.79 [0.39–1.61], I2 = 77%

Low-moderate

Although the evidence is still of low quality and should be strengthened, it seems that the combination of OAT and NSEP significantly reduce the risk of HCV acquisition

Partially yes

Turner [55]

SLRMA

6 individual-level data studies (n = 2986)

Any health setting

NSEP

OAT

• HCV incidence, all studies: aOR 0.52 [0.32–0.83]

• HCV incidence, OAT studies: aOR 0.41 [0.21–0.82], I2 = 48%

• HCV incidence, NSEP studies: aOR 0.48 [0.25–0.93], I2 = 0%

• Injection frequency, all studies:

− 20.8 [− 27.3, − 14.4] injections/month

Moderate

OAT and high coverage NSEP substantially reduced the risk of HCV transmission among injecting drug users (full harm reduction of needle sharing by 48% and mean injecting frequency by 20 injections per month)

Yes

Wright [56]

SLR

18 interventional and observational studies

Any health setting

NSEP

OAT

Behavioral interventions

• HCV incidence, NSEP studies: significant positive effect NSEP

• HCV incidence, OAT studies: significant positive effect of OAT

• Limited evidence evaluating behavioural interventions

–

NSEP or OAT primary interventions are marginally effective in reducing HCV prevalence. Further combined interventions should be assessed

Yes