Effectiveness of needle and syringe Programmes in people who inject drugs : an overview of systematic reviews

RESEARCH ARTICLE Open Access E ec i eness o needle and sy inge P og ammes in people who injec d ugs – An o e iew o sys ema ic e iews Rica do M Fe nandes 1,3 , Ma ia Ca y 2 , Gonçalo Dua e 1 , Gonçalo Jesus 1 , Joana Ala cão 1 , Ca la To e 2 , Suze e Cos a 2 , João Cos a 1,3 and An ónio Vaz Ca nei o 1,3* Abs ac Backg ound: Needle and sy inge p og ammes (NSP) a e a c i ical componen o ha m educ ion in e en ions among people who injec d ugs (PWID). Ou p ima y objec i e was o summa ize he e idence on he e ec i eness o NSP o PWID in educing blood-bo ne in ec ion ansmission and injec ing isk beha iou s (IRB). Me hods: We conduc ed an o e iew o sys ema ic e iews ha included PWID (excluding p isons and consump ion ooms), add essed communi y-based NSP, and p o ided es ima es o he e ec ega ding incidence/p e alence o Human Immunode iciency Vi us (HIV), Hepa i is C i us (HCV), Hepa i is B i us (HBV) and bac e emia/sepsis, and/o measu es o IRB. Sys ema ic li e a u e sea ches we e unde aken on ele an da abases, including EMBASE, MEDLINE, and PsychINFO (up o May 2015). Fo each e iew we iden i ied ele an s udies and ex ac ed da a on me hods, and indings, including isk o bias and quali y o e idence assessed by e iew au ho s. We e alua ed he isk o bias o each sys ema ic e iew using he ROBIS ool. We ca ego ized e iews by epo ed ou comes and use o me a-analysis; no addi ional s a is ical analysis was pe o med. Resul s: We included hi een sys ema ic e iews wi h 133 ele an unique s udies published be ween 1989 and 2012. Repo ed ou comes ela ed o HIV (n=9),HCV(n= 8) and IRB (n= 6). Me hods used a ied a all le els o design and conduc , wi h ou e iews pe o ming me a-analysis. Only wo e iews we e conside ed o ha e low isk o bias using he ROBIS ool, and mos included s udies we e e alua ed as ha ing low me hodological quali y by e iew au ho s. We ound ha NSP was e ec i e in educing HIV ansmission and IRB among PWID, while he e we e mixed esul s ega ding a educ ion o HCV in ec ion. Full ha m educ ion in e en ions p o ided a s uc u al le el and in mul i-componen p og ammes, as well as high le el o co e age, we e mo e bene icial. Conclusions: The he e ogenei y and he o e all low quali y o e idence highligh s he need o u u e communi y-le el s udies o adequa e design o suppo hese esul s. T ial egis a ion: The p o ocol o his sys ema ic e iew was egis e ed in P ospec i e Regis e o Sys ema ic Re iews (PROSPERO 2015:CRD42015026145). Keywo ds: Needle and sy inge p og ammes, Ha m educ ion in e en ions, People who injec d ugs, HIV/Aids-Hc -Hb * Co espondence: [email p o ec ed] 1 Cen e o E idence-Based Medicine, Facul y o Medicine, Uni e si y o Lisbonl, A . P o . Egas Moniz, 1649-028 Lisbon, Po ugal 3 Po uguese Collabo a ing Cen e o he Ibe oAme ican Coch ane Ne wo k-Coch ane Po ugal Facul y o Medicine, Uni e si y o Lisbon, A . P o . Egas Moniz, 1649-028 Lisbon, Po ugal Full lis o au ho in o ma ion is a ailable a he end o he a icle © The Au ho (s). 2017 Open Access This a icle is dis ibu ed unde he e ms o he C ea i e Commons A ibu ion 4.0 In e na ional License (h p://c ea i ecommons.o g/licenses/by/4.0/), which pe mi s un es ic ed use, dis ibu ion, and ep oduc ion in any medium, p o ided you gi e app op ia e c edi o he o iginal au ho (s) and he sou ce, p o ide a link o he C ea i e Commons license, and indica e i changes we e made. The C ea i e Commons Public Domain Dedica ion wai e (h p://c ea i ecommons.o g/publicdomain/ze o/1.0/) applies o he da a made a ailable in his a icle, unless o he wise s a ed. Fe nandes e al. BMC Public Heal h (2017) 17:309 DOI 10.1186/s12889-017-4210-2 Backg ound People who injec d ugs (PWID) expe ience high le els o mo bidi y and mo ali y. D ug- ela ed ha ms include o e - dose, d ug- ela ed dea hs, and blood-bo ne in ec ions such as Human Immunode iciency Vi us (HIV), Hepa i is C (HCV), Hepa i is B (HBV), and bac e emia/sepsis. HCV is cu en ly he mos p e alen in ec ious disease a ec ing PWID, while HIV p e alence a es a e lowe . In an es i- ma ed o al o 12.7 o 16 million PWID wo ldwide, i is belie ed ha 1.2 million a e in ec ed wi h HIV [1] and 10 million a e in ec ed wi h HCV [2]. Sha ing needles and sy inges, as well as o he injec ing pa aphe nalia, is a key ou e o ansmission o hese in ec ions [3]. Needle and sy inge p og ammes (NSP) a e hough o be a c i ical componen o ha m educ ion in e en ions among PWID [3, 4]. The i s NSP was es ablished in he 1980s [5], in esponse o he global HIV epidemic, wi h he goal o p o iding access and encou aging he use o s e ile injec ion pa aphe nalia by PWID. Since hen, p o ision o hese se ices has g own apidly. Impo an ly, a shi in pa adigm has a ou ed NSP as componen s o ha m educ ion o ha m minimiza ion policies, which ocus on educing all d ug- ela ed ha ms, i.e. p e en ing HIV, HBV and HCV in ec ion, minimizing needle and sy inge sha ing and euse, educing he olume o dis- ca ded needles and sy inges in he en i onmen , and a- cili a ing access o s e ile pa aphe nalia. Fu he mo e, NSP may also p omo e he use o condoms and p o ide oppo unis ic ele an heal h in o ma ion and se ices [3]. NSP a e complex heal h in e en ions wi h se e al in e ac ing componen s, such as beha iou al changes in PWID and p o ide s, a complex ope a ing amewo k (use s, p o ide s, se ing, heal h sys ems), and some deg ee o lexibili y o in e en ions [6]. The e is consid- e able a iabili y among egions and coun ies in se ice p o ision, co e age and ange o ha m educ ion in e - en ions o e ed by NSP. In pa icula , a a ie y o measu es ha e been de eloped o imp o e access o and use o s e ile injec ing equipmen and o inc ease use s’ choice. These include se e al me hods o dis ibu ion o sale such as con en ional NSP in ixed-si es, pha macy- based dis ibu ion, dispensing machines and ou each p og ammes –o en using a mobile an o bus, o h ough home- isi s [3]. Gene ally, pha macy and specialis needle exchange p o ides a wide ange o ha m educ ion in o ma ion and ad ice, along wi h clean needles and sy inges and possibly injec ing pa aphe na- lia. The legal amewo k in which NSP ope a e also a ies a coun y le el [7]. Pha macy-based NSP a e known o be in place in a leas Aus alia [8], Belgium, F ance, I eland, Ky gyzs an, he Ne he lands, Po ugal, Spain, Slo enia, Uk aine [7, 9], UK [10], and New Zealand [11]. Many sys ema ic e iews on he e ec i eness o in e - en ions p o iding injec ing equipmen ha e been conduc ed o da e. O e iews o e iews which compile in o ma ion om hese mul iple sys ema ic e iews, ha e also been published. The aim o hese o e iews is o p o ide end-use s wi h a comp ehensi e and c i ical summa y o he a ailable e idence on his in e en ion. O e iews a e o pa icula in e es in his ield because exis ing sys ema ic e iews ha e ocused on dispa a e ques ions, ega ding ei he speci ic popula ions (e.g. coun y-speci ic e idence), speci ic ypes o in e en ions (e.g. di e en ypes o NSP p o ision) o , mos commonly, speci ic ou comes (e.g. HIV o HCV ans- mission) [12–14]. Fu he mo e, he e is a iabili y in me hods used in hese sys ema ic e iews, including he assessmen o possible biases and he use o quan i a i e syn hesis (me a-analysis). Ne e heless, o he bes o ou knowledge, he published o e iews in his ield ha e ocused only on speci ic ou comes, i.e. ansmission o HIV and/o HCV. Objec i es Ou p ima y objec i e was o conduc an o e iew o sys ema ic e iews ha e alua ed he e idence o he e ec i eness o NSP o PWID ac oss a ange o di e - en ele an ou comes, i.e. blood-bo ne in ec ion ans- mission and injec ing isk beha iou s (IRB). Ou seconda y objec i e was o assess how di e en aspec s o NSP p o ision, including p o ide , se ing, co e age and any ela ed componen deli e ed in pa allel, such as ha m educ ion se ices and opia e subs i u ion he apy, modi ied he e ec o NSP, wi h a pa icula ocus on pha macy-d i en NSP. Me hods We ollowed cu en guidance on he conduc o o e iews o e iews, including ecommenda ions om he Coch ane Collabo a ion [15] and guidance on public heal h in e en ion e iews by he Cen e o Re iews and Dissemina ion o he Uni e si y o Yo k [16]. We also ollowed he ecommenda ions om he PRISMA-P s a emen ega ding epo ing i ems ha we conside ed applicable o his o e iew [17]. The p o ocol o his o e iew was egis e ed a PROSPERO (CRD42015026145) a ailable a h p://www.c d.yo k.ac.uk/ PROSPERO/display_ eco d.asp?ID=CRD42015026145. Eligibili y c i e ia Fo inclusion in his o e iew, s udies had o mee he ollowing c i e ia: 1. S udy design: sys ema ic e iews, ope a ionally de ined as s udies epo ing a clea ly s a ed se o objec i es, eligibili y c i e ia, a sys ema ic sea ch using wo o mo e sou ces, a sys ema ic p esen a ion o he cha ac e is ics and indings o he included s udies, Fe nandes e al. BMC Public Heal h (2017) 17:309 Page 2 o 15 and es ima es o he size and di ec ion o he e ec o in e en ions p esen ed as nume ical da a, on an indi idual s udy-le el basis and/o wi h quan i a i e syn hesis (me a-analysis); 2. Pa icipan s: PWID, de ined as people who injec some o m o d ug a he beginning o he s udy. We excluded s udies ocusing exclusi ely on pa icipan s whose consump ion was con ined o p isons and consump ion ooms, since hese a e popula ions wi h dis inc cha ac e is ics om ou a ge popula ion; 3. In e en ions: sys ema ic e iews had o e alua e communi y-based NSP, de ined as he supply o a leas needles and sy inges, wi h o wi hou o he injec ing pa aphe nalia o he p epa a ion and consump ion o d ugs; 4. Ou comes: equi ed epo ed ou comes included he incidence and/o p e alence o blood-bo ne in ec ions (HIV, HCV, HBV and bac e emia/sepsis), and/o measu es o IRB (including bu no limi ed o sy inge e-use, bo owing, sha ing, en ing and lending). When mo e han one e iew included exac ly he same s udies, he e iew ha epo ed he mos comple e p esen- a ion o esul s was selec ed o inclusion in he o e iew. Sea ch s a egy and sc eening Sea ches we e unde aken on MEDLINE®In-P ocess & O he Non-Indexed Ci a ions ( om 1946 o 12 h o May 2015), EMBASE ( om 1974 o 12 h o May 2015) and PsycINFO ( om 1806 o 12 h o May 2015) ia he OVID SP in e ace. In addi ion, he ollowing da abases we e sea ched: he NHS Economic E alua ion Da abase (NHS EED), he Da abase o Abs ac s o Re iews o E ec s (DARE), he Coch ane Da abase o Sys ema ic Re iews (CDSR), he Na ional Ins i u e o Heal h and Clinical Excellence (NICE), he Campbell Lib a y o Sys ema ic Re iews and he Da abase o P omo ing Heal h E ec i eness Re iews (DoPHER). The sea ch s a egies o each da abase a e shown in a supplemen a y ile (Addi ional ile 1). No language o o he ypes o es ic- ions we e applied. In addi ion o he da abase sea ches, handsea ching o he e e ences o he included e iews was unde aken o iden i y u he ele an s udies. Ti les and abs ac s o a icles iden i ied we e sc eened independen ly by wo au ho s (MC, JA), and classi ied as include, unclea o exclude. The ull epo s o all a icles ha classi ied as include o unclea we e hen ob ained, and wo au ho s (MC, GD) examined compli- ance o e iews wi h eligibili y c i e ia, wi h a hi d au ho ac ing as an a bi e (RF). Da a ex ac ion Da a om epo s o all included sys ema ic e iews we e ex ac ed by wo au ho s (MC, GD) and alida ed by a hi d au ho (RF), using a da a ex ac ion o m designed and p e-pilo ed o his o e iew. The ollow- ing gene al cha ac e is ics we e ex ac ed om each sys ema ic e iew: publica ion de ails; s udy objec i es; eligibili y c i e ia (popula ion, in e en ion, compa a o s, ou comes, s udy designs); any epo ed p o ocol; and me hods used o sea ch, sc eening, da a ex ac ion and syn hesis. Fo each sys ema ic e iew, we lis ed all in- cluded s udies and e alua ed whe he hey ma ched he eligibili y c i e ia o his o e iew ega ding popula ion, in e en ions and ou comes. We hen ex ac ed he ollowing esul s om he eligible g oup o s udies, whene e epo ed a he sys ema ic e iew le el: s udy design; coun ies in ol ed; cha ac e is ics o included pa icipan s (demog aphics, p e alence o HIV/HCV); desc ip ion o in e en ions and any co-in e en ion, du a ion o in e en ion and ollow-up; e ec es ima es om me a-analysis (i a ailable) o a indi idual s udy- le el, o each ele an ou come; and any subg oup o sensi i i y analyses. The au ho s’conclusions o each ele an ou come we e also collec ed. We con ac ed he au ho s o e iews o ele an missing da a. Assessmen o me hodological quali y A a s udy le el, we ex ac ed da a on any isk o bias assessmen s o p ima y s udies when pe o med and e- po ed by e iewe s in each sys ema ic e iew, including ools used and summa ized esul s. We also collec ed da a on any epo ed e alua ions o he quali y o e idence conce ning ou ou comes o in e es in included e iews, pa icula ly hose using he G ading o Recommenda ions Assessmen , De elopmen and E alu- a ion (GRADE) ool [18], as well as da a on assessmen s o publica ion bias. A a sys ema ic e iew le el, we assessed he me hodo- logical quali y o each included sys ema ic e iew using he Risk O Bias In Sys ema ic Re iews (ROBIS) ool [19]. Two e iew au ho s (MC, RF) pe o med quali y assess- men s independen ly, using pilo ed decision ules. Dis- ag eemen s ega ding o e all assessmen s we e esol ed h ough discussion, wi h a hi d e iewe se ing as he inal a bi a o (AVC). The a ionale behind assessmen s was documen ed. We calcula ed measu es o ag eemen and eliabili y be ween a e s o each ROBIS domain. Da a syn hesis We s a i ied he included sys ema ic e iews by: (i) ype o ou comes assessed (blood-bo ne in ec ions, IRB), and (ii) ype o analysis (wi h o wi hou me a-analysis). We summa ized da a om he included e iews bo h in ex and in summa y ables and igu es. When me a-analysis was pe o med, we epo pooled es ima es using he models and measu es o e ec epo ed by sys ema ic e- iew au ho s, wi h 95% con idence in e als (95% CI); Fe nandes e al. BMC Public Heal h (2017) 17:309 Page 3 o 15 we did no pe o m any addi ional s a is ical analysis. When epo ed, he accompanying I 2 alues, which desc ibe he pe cen age o o al a ia ion ac oss s udies ha is due o he e ogenei y a he han chance, we e collec ed [20]. Resul s The sea ch and sc eening p ocess is summa ised in Fig. 1. A o al o 667 ci a ions we e iden i ied h ough he a ious da abase sea ches. Th ee addi ional e- co ds we e iden i ied in he e e ence lis s o sc eened s udies. A e 37 duplica es we e emo ed, we ob ained 633 ci a ions, which we e sc eened by i le and abs ac . We excluded 582 ci a ions as hey did no mee he inclu- sion c i e ia, and he emaining 51 we e sc eened ull ex . Thi y- i e ci a ions we e u he excluded, and wo epo s [21, 22] we e unob ainable. Fou een epo s we e hus included, co esponding o 13 sys ema ic e iews, as wo eco ds e e ed o he same s udy [12, 23]. Wi hin hose publica ions h ee we e epo s om Na ional Ins i u es/Expe Commi ees in he US [24], UK [12] and Canada [25], while 10 we e egula pape s published in scien i ic jou nals. Desc ip ion o included e iews Table 1 lis s he key cha ac e is ics o he 13 included sys ema ic e iews ha e alua ed he e idence o he e ec i eness o NSP o PWID in he communi y se - ing. We classi ied e iews by ype o ou comes epo ed and ype o da a analysis (Fig. 2). The numbe o da abases sea ched pe e iew a ied be ween wo and 15, wi h he h ee mos common sou ces being MEDLINE, EMBASE and PsycINFO. The da es o las sea ch a ied be ween 1995 [26] and 2012 [27]. O 287 s udies included in hese 13 e iews, 200 s udies ma ched he eligibili y c i e ia o his o e iew, co esponding o 133 unique s udies (a ma ix wi h included s udies by e iew is a ailable in a supplemen- a y ile: Addi ional ile 2). We included he comple e se o s udies om ou e iews, as in he emaining e iews a leas one s udy did no ul ill ou eligibili y c i e ia [27–30]. The numbe o ele an s udies included pe e iew anged om h ee [31, 32] o 43 [24]. While he e was some o e lap in he included s udies be ween e iews, he majo - i y (68%) was included in one e iew only. We ound subs an ial a iabili y in he c i e ia used by e iew au ho s o de ine and classi y s udy designs, as well Fig. 1 Sc eening decisions up o da e as o 30Jun2015. O iginal sea ch da e 12May2015 Fe nandes e al. BMC Public Heal h (2017) 17:309 Page 4 o 15 Table 1 Key cha ac e is ics o he included sys ema ic e iews and Risk o bias o each e iew Au ho s, da e o publica ion a PICOS b Da es o las sea ch Da abases sea ched (n) Risk o Bias and quali y o e idence ools To al numbe o s udies/ Numbe o s udies ele an o he o e iew (designs) c,d Risk o bias o he e iew (assessed using ROBIS ool) Quan i a i e syn hesis- me a-analysis Aspinall, 2014 [27] P: PWID I: NSP (Specialis si es, pha macies o ou each se ices) C: No, a e, o less exposu e o NSP O: HIV incidence S: any 2012 CINAHL, Coch ane Lib a y, EMBASE, IBSS, MEDLINE and PsycINFO (6) Newcas le-O awa each s udy & GRADE o assess quali y o e idence as a whole 12/12 (11 CC/Co; 1 OTH) Low Hagan, 2011 [34] P: indi iduals who could concei ably ha e acqui ed HCV in ec ion ia adminis a ion o an illegal d ug (PWID among hem) I: in e en ions ha could p e en HCV in ec ion C:NR O: HCV p e alence o incidence a es, and measu es o associa ion wi h p e alence o incidence S: any 2010 Cu en Con en s, Disse a ion Abs ac s, ERIC, NIH CRISP, MEDLINE, PsycINFO, Sociological Abs ac s (7) De eloped hei own quali y scale 26/7 (7 CC/Co) High Tu ne , 2011 [32] P: PWID in he UK I: NSP o OST (o combina ion) C: OST s non OST; high s low co e age o NSP O: HCV in ec ion and IRB S: any NR PubMed and Web o Science (2) None 4/3 (1 CC/Co; 2 OTH) High C oss, 1998 [26] P: PWID I: NSP o educa ional p og ammes C: NR O: IRB S: expe imen al o quasi-expe imen al design 1995 AIDSLINE, MEDLINE, PsycINFO, Social Science Ci a ion Index (4) None 26/10 (1 RCT; 9 OTH) High Quali a i e syn hesis Abdul-Quade , 2013 [29] P: PWID I: S uc u al in e en ions ha add ess ex e nal ac o s o educe HIV and HCV C: any quan i a i e compa ison O: Changes in HIV/HCV in ela ion o in e en ion S: con olled o be o e & a e 2011 Coch ane Cen al Regis e o Con olled T ials, EMBASE, LILACS, PsycINFO, PubMed, and he Web o Science/Web o Social Science (6) WHO-Johns Hopkins 9-Poin Rigou Scale. 15/15 (1 CC/Co; 14 OTH) Unclea Des Ja lais, 2013 [28] P: PWID in low-and middle income and ansi ional coun ies (LMICs) I: NSP C: any O: changes in p e alence o incidence o HIV, HCV, HIV/ HCV co-in ec ion o changes o e ime in newly epo ed HIV o HCV cases o in ec ion S: any 2011 EMBASE, NLM Ga eway, PubMed, Web o Science (4) None 13/13 (13 OTH) High Fe nandes e al. BMC Public Heal h (2017) 17:309 Page 5 o 15 Table 1 Key cha ac e is ics o he included sys ema ic e iews and Risk o bias o each e iew (Con inued) Hong and Li, 2009 [31] P: People wi h HIV/ AIDS in China I: Any beha iou al in e en ion C: Any O: Ra e o needle-sha ing S: Any 2008 AIDSLine, EBSCO, Fi s Sea ch, PsycINFO, PubMed (4) None 25/3 (1 RCT; 1 CC/Co; 1 OTH) High Jones, 2008 [12] P: PWID I: NSP C: any O: IRB, HIV, HVC S: RCT, con olled non- andomised s udies, con olled and uncon olled be o e and a e s udies, c oss-sec ional s udies, coho s udies, case-con ol s udies and ecological s udies 2008 ACP Jou nal Club, ASSIA, CINAHL, Coch ane Lib a y, EMBASE, EPPI-Cen e da abases, Heal h In o ma ion Managemen Conso ium, IBSS, MEDLINE, Na ional Resea ch Regis e A chi e, OpenSIGLE, P ojec CORK, PsycINFO, Sociological Abs ac s, SOMED (15) Quali y assessmen ools de eloped by NICE Cen e o Public Heal h Excellence, England and Wales and he E ec i e Public Heal h P ac ice P ojec , Canada 24/15 (2 RCT; 1 CC/Co; 12 OTH) Low Kall, 2007 [30] P: PWID I: NSPs C: con ol o compa ison g oups O: HIV incidence o p e alence S: any 2005 MEDLINE and PsyclNFO (2) None 16/16 (8 CC/Co; 8 OTH) High W igh , 2006 [33] P: PWID I: p ima y p e en ion in e en ion a ge ing injec ing d ug use C: any O: HCV p e alence and incidence S: in e en ion o obse a ional s udies 2003 Coch ane Lib a y, CINAHL, EMBASE, MEDLINE and PsycINFO (4) De eloped hei own quali y scale 18/12 (9 CC/Co; 3 OTH) High Tilson, 2006 [24] P: PWID I: NSP C: Wi h and wi hou OST; Placebo. O: IRB, HCV and HIV S: NR 2006 Coch ane Lib a y, G ey Li e a u e Repo , EMBASE, PubMed, PsycINFO, Social Science Abs ac s, Web o Science, Wo ldCa (8) GRADE 45/43 (28 CC/Co; 15 OTH) Unclea Gibson, 2001 [35] P:PWID I:NSP C: any O:HIV, HBV, HCV and IRB S:any 1999 Medline and PsycINFO (2) None 42/32 (19 CC/Co; 13 OTH) High Leona d, 1999 [25] P: PWID I: NSP C: Any compa a o g oup (bu none was no allowed) O: HIV, HCV and IRB S: NR 1999 AIDSLINE, CINAHL, Coch ane Lib a y, EMBASE, MEDLINE, PHE ec (6) Quali y assessmen ools de eloped by he E ec i e Public Heal h P ac ice P ojec eam, Canada 21/19 (15 CC/Co; 4 OTH) Unclea a Ca ego ized by ype o syn hesis and lis ed by publica ion da e b PICOS: ac onym o Popula ion, In e en ion, Compa a o , Ou come and S udy Design c Classi ied by au ho s o he e iew d CC/Co: Case-Con ol/Coho ; RCT: Randomized Con ol T ial; OTH: o he ( ime-se ies designs, be o e-a e s udies, ecological s udies) Fe nandes e al. BMC Public Heal h (2017) 17:309 Page 6 o 15 as ypes o designs o included s udies. Only ou e iews speci ied he ypes o s udy design in hei eligibili y c i e ia [12, 26, 29, 33]. Th ee e iews included andom- ized con olled ials ( ou RCTs o e all) [12, 26, 31], while he emaining e iews included coho s and case-con ols (101 o e all), as well as o he s udy designs ( ime se ies, be o e/a e s udies o ecological s udies) (95 o e all). Th ee o he included e iews we e es ic ed o s udies conduc ed in speci ic coun ies o economic e- gions (China, UK, and low- and middle-income coun ies) [28, 31, 32]. Nine e iews epo ed da a on HIV, eigh on HCV, and six on IRB. Se en e iews e alua ed mo e han one o hese ou comes. Fou e iews used me a-analysis [26, 27, 32, 34], one o which using indi idual-pa icipan da a [32]. The emaining nine e iews used na a i e syn hesis [12, 24, 25, 28–31, 33, 35]. The easons epo ed by he au ho s o conduc ing o no conduc ing me a-analysis a ied be ween e iews, e en when he e was a deg ee o o e lap be ween included s udies. Me hodological quali y Se en e iews epo ed ha ing assessed he me hodo- logical quali y o included s udies [12, 24, 25, 27, 29, 33, 34]. In wo e iews he au ho s de eloped ad hoc ools [33, 34], while he emaining e iews used p e iously exis ing ins umen s, o en wi h modi ica ions ( ools used: Newcas le-O awa, GRADE, WHO-Johns Hopkins 9- Poin Rigou Scale, Quali y assessmen ools de el- oped by NICE Cen e o Public Heal h Excellence, England and Wales and he E ec i e Public Heal h P ac ice P ojec , Canada) [12, 24, 25, 27, 29]. Only wo ins umen s we e used mo e han once, and all o he ins umen s a ied. No all e iews epo ed he esul s o hese assessmen s; when epo ed, mos s udies we e conside ed o ha e low me hodological quali y, gi en me hodological weaknesses and biases ela ed o obse a ional s udy designs. Only one quan i a i e e iew [27] inco po a ed quali y in da a syn hesis, by conduc ing a sensi i i y analysis. In e iews wi hou me a-analysis, me hodological quali y and isk o bias was p esen ed desc ip i ely in he in e p e a ion o s udy esul s. Two e iews [24, 27] epo ed using he GRADE ool o assess he quali y o he body o e idence. One conside ed he o e all quali yo e idenceaslow[27],while heo he ,using an adap a ion o GRADE, e alua ed he quali y o Fig. 2 S udies selec ed o inclusion classi ied by ou come(s) epo ed and s a egy o da a syn hesis. *G ey shaded boxes ep esen he ou come epo ed in he e iews indica ed in he igh column. “HIV”,”HCV”and “Injec ing isk beha iou s”is used o classi y he e iews ha epo ed he impac o NSP in he numbe o HIV in ec ions, numbe o HCV in ec ions and change in injec ion beha io . The e iews a e also p esen ed in he igh column by s a egy used o da a syn hesis Fe nandes e al. BMC Public Heal h (2017) 17:309 Page 7 o 15 e idence as modes o mode a e o di e en ou - comes [24]. Using he ROBIS ool, only wo o he included e iews we e conside ed o ha e low isk o bias, whe eas he emaining we e conside ed o ha e ei he high (n=8)o unclea isk o bias (n= 3) (summa y o esul s a ailable in a supplemen a y ile: Addi ional ile 3). The majo i y o e- iews we e a ed as ha ing high o unclea isk o bias ac oss all ROBIS domains, i.e. s udy eligibili y c i e ia, iden- i ica ion and selec ion o s udies, da a collec ion and s udy app aisal, syn hesis and indings. Pe cen age o ag eemen be ween ROBIS a e s a ied be ween 77% and 92%, and weigh ed kappa (quad a ic) be ween 0.82 and 0.95. The esul s om he PRISMA-P s a emen applied o his o e iew a e p esen ed a Addi ional ile 4. E ec s o NSP HIV p e alence and/o incidence O nine e iews ha epo ed da a on HIV ou comes, wo we e a low isk o bias [12, 27], and only one pe o med me a-analysis [27]. The p opo ion o s udies included in only one o hese nine e iews was 71%. Re iews wi h me a-analysis The mos ecen e iew by Aspinall e al. included wel e s udies (10 coho , one case-con ol and one c oss-sec ional s udy), and was e alua ed as ha ing low isk o bias. The e iew epo ed a 34% “ isk” educ ion o HIV ansmission (pooled e ec es ima e o 0.66; 95% CI: 0.43 o 1.01; n=10 s udies; I 2 = 76%) in indi iduals exposed o NSP, com- pa ed wi h hose who we e no , o we e less equen ly, exposed o NSP. This es ima e was ob ained by pooling bo h adjus ed and unadjus ed e ec sizes ega dless o s udy design, in a andom e ec s model, and pooling all ypes o e ec measu es (odds a io, isk a io, haza d a io). Sensi i i y analyses suppo ed hese esul s, wi h a s a is ically signi ican educ ion in HIV ansmission associa ed wi h NSP exposu e in bo h highe quali y s ud- ies (0.42; 95% CI: 0.22 o 0.81), and s udies epo ing ela- i e isks o haza d a ios (0.60; 95% CI: 0.37 o 0.97). Subg oup analyses showed ha s udies using sequen ial ollow-up (i.e. g oups exposed o NSP and g oups no ex- posed ollowed sequen ially, and no concu en ly) and s udies ec ui ing pos -1990 had lowe e ec es ima es (0.21 [95% CI: 0.11 o 0.41] and 0.52 [95% CI: 0.28 o 0.95], espec i ely). Fu he , s udies compa ing 100% NSP co e age (i.e. clean needle and sy inge used o 100% o injec ions) wi h <100% NSP co e age gene a ed a pooled e ec es ima e o 0.58 (95% CI: 0.22 o 1.57). Quali y o e idence was conside ed low using GRADE, and mos o hese esul s had subs an ial he e ogenei y. Re iews wi hou me a-analysis The quan i a i e esul s p esen ed abo e a e suppo ed by mos e iews wi h quali a i e summa ies o he e idence only, all o which published be ween 1999 and 2013. Di e ences be ween e iews could be ound ega ding eligibili y c i e ia (e.g. ype o NSP, popula ion- s indi idual-le el ou comes, s udy design), sea ch and sc eening me hods, quali y assessmen s and in e p e a ion o s udy esul s. In mos o hese e iews, he au ho s desc ibed and con ex ual- ized esul s om each indi idual s udy o g oup o s udies ega ding HIV ou comes. Some e iews used o e coun ing o compa e he numbe o posi i e s udies wi h educ ion in HIV ansmission, wi h he numbe o nega i e s udies. In he ollowing pa ag aphs, we p esen he main esul s o hese e iews, s a ing wi h hose ocusing on he impac o di e en aspec s o NSP p o ision ollowed by esul s o e iews ocusing exclu- si ely on he compa ison be ween NSP andno NSP. Rega ding aspec s o NSP p o ision, Jones and colleagues epo ed esul s om a ecen sys ema ic e- iew classi ied as low isk o bias, ha ocused on le el o co e age, sy inge dispensa ion policies, ype o NSP, p o ision o addi ional ha m- educ ion se ices and o opia e subs i u ion he apy [12]. Au ho s judged ha he ange o s udy designs, in e en ion app oaches exam- ined and ou comes p ecluded he use o me a-analysis. Only h ee s udies ocused on HIV ou comes. A low-qu- ali y s udy showed no signi ican end o HIV p e a- lence when compa ing p ima y sou ces o needles (pha macies, ixed si e NSP and an-based NSP), al hough HIV p e alence was lowe among pha macy use s han in pa icipan s who epo ed using an o ixed si e NSP (16% s. 21% and 25%, espec i ely, p= 0.16) [36]. Ano he s udy [37] included in his e iew examined he impac o dispensa ion policies, and no ed a dec ease in HIV p e alence (based on es - ing o sel - epo ) be ween he pe iod o legal pha macy sy inge pu chase and when up o i e needles could be exchanged a newly es ablished NSP (35% o 22%; p< 0.05). Finally, one mode a e quali y coho s udy [38] also included in he e iew by Jones and colleagues e alua ed di e en le els o ha m educ ion and ound ha a ull ha m educ ion s a egy (combina ion o me hadone ea men and ull pa icipa ion in NSP) educed he incidence o HIV when compa ed o incom- ple e o no ha m educ ion (incidence a e a io: 0.32; 95% CI: 0.17 o 0.62). In line wi h hese esul s, a e iew wi h an o e all unclea isk o bias by Abdul-Quade e al. [29] iden i- ied s udies wi h s uc u al-le el NSP, ie in e en ions in which changes in policy and legal en i onmen ha e acili a ed an inc eased a ailabili y o s e ile sy inges, and ocused on popula ion-le el ou comes. The ope - a ional de ini ion o s uc u al-le el NSP was a minimum 50% co e age o PWID and dis ibu ion o 10 o mo e needles/sy inge pe PWID pe yea . Nine s udies Fe nandes e al. BMC Public Heal h (2017) 17:309 Page 8 o 15 included in his e iew epo ed dec eases in HIV p e a- lence, and h ee epo ed dec eases in HIV incidence. All s udies we e non- andomized be o e-a e compa isons o in e up ed ime se ies analyses, and mos showed e idence o po en ial selec ion biases. The au ho s con- cluded ha hese esul s suppo NSP as a s uc u al-le el in e en ion o educe popula ion-le el in ec ion. Fou olde published e iews wi h quali a i e syn hesis and an o e all unclea o high isk o bias epo ed mixed indings ega ding he impac o NSP on HIV in- cidence/p e alence. These e iews p o ide a his o ical pe spec i e on he e olu ion o NSP implemen a ion and e alua ion. The i s published e iew by Leona d e al. [25], was an upda e o a p e ious sys ema ic e iew. Gibson e al. [35] included six s udies published up o 1999 wi h a ange o s udy designs, pa icipan s and se ings, epo ing on HIV ou comes. The Ins i u e o Medicine’s e idence epo [24], iden i ied 12 ele an s udies, including coho , case-con ol and ecological designs, as well as s udies using ma hema ical models. Finally, he e iew by Kall e al. [30] included 16 s udies published up o 2005, only wo o which we e no included in o he e iews om his o e iew. All hese e iews included landma k p ospec i e coho s udies conduc ed in Mon eal and Vancou e in he 1990s [39, 40], which ound an associa ion be ween NSP pa icipa ion and highe isk o HIV se ocon e - sion. This led Leona d e al. o conclude ha he e was me hodologically weak e idence ha NSP we e no as e ec i e as p e iously ound in modi ying HIV p e a- lence and incidence among PWID. Kall e al. used o e coun ing and epo ed ha on mos s udies assessing se oincidence he e ec o NSP was no signi ican , while ou s udies in es iga ing se op e alence a base- line we e un a ou able o NSP. The au ho s also s a ed ha in s udies ha ound posi i e e ec s, con ounde s had no been adequa ely con olled o . Howe e , based on epidemiological e idence ha accumula ed p og es- si ely, au ho s o he o he e iews highligh ed a numbe o selec ion biases ha could accoun o hese indings, including: he inclusion o high- isk cocaine injec o s, who injec ed mo e o en han he oin use s; he limi ed numbe o needles and sy- inges ha use s could ha e access o in ea ly NSP; and he eady a ailabili y o clean injec ing equipmen h ough pha macies which could ha e a ac ed ma ginalized, pa icula ly high- isk indi iduals. Follow-up s udies in he same se ings, conduc ed a e expanding and adjus ing NSP (e.g. by allowing unlimi ed dis ibu ion o needles/sy inges, inc easing he numbe o access poin s, and o e ing di e en dis ibu ion me hods), ound no such inc ease in isk, o a dec ease in HIV p e alence. Gibson e al. [35] used o e coun ing and con- side ed he e was subs an ial e idence ha NSP we e e ec i e in p e en ing HIV se ocon e sion. Tilson e al. [24] included ou ecological s udies ha ound an associ- a ion be ween HIV p e en ion p og ammes ha include NSP wi h educed p e alence o HIV in u ban se ings. Based on he weakness o hese s udies designs, his e i- dence was conside ed modes using a modi ied GRADE app oach. Fu he , mode a e e idence was ound ha mul i-componen HIV p e en ion p og ammes ha in- clude NSP educe in e media e HIV isk beha iou . The Ins i u e o Medicine epo highligh ed how almos all published s udies o igina e in No h Ame ica, Wes e n Eu ope, and Aus alia [24]. Two addi ional e iews we e es ic ed o speci ic popula ions by geog aphical o economical sou ce. Hong and Li summa ized e idence om wo s udies conduc ed solely in China [31], while Des Ja lais e al. ocused on 13 s udies o 11 NSPs wi h high-co e age conduc ed in low/middle-income coun ies [28]. In bo h cases, esul s om included s udies gene ally suppo ed he e ec i eness o NSP in educing HIV. Des Ja lais e al. epo ed a educ ion o HIV p e alence in ou s udies ( om −3% o −15%), o es ima ed HIV inci- dence in h ee s udies ( om −11/100 o −16/100 pe son- yea s a isk), and o newly epo ed na ionwide cases in h ee na ional epo s ( om −30% o −93.3%). Con e sely, inc eases in HIV p e alence we e ound in wo s udies ( om +5.6% o +15.8%) and one na ional epo (+37.6%) included in he e iew by Des Ja lais e al. The au ho s conside ed ha , i high co e age is achie ed, NSP appea o be as e ec i e in low/middle-income as in high-income coun ies [28]. HCV p e alence and/o incidence Eigh included e iews syn hesized he e idence on he use o NSP in p e en ing HCV p e en ion in PWID. One was a ed as being a low [12], h ee a unclea [24, 25, 29] and ou a high isk o bias [28, 32–34], including wo e iews ha used me a-analysis [32, 34]. The p opo ion o s udies included exclusi ely in one o hese e iews was 88%. Bo h e iews wi h quan i a i e and quali a i e syn hesis showed mixed esul s. Re iews wi h me a-analysis Two e iews [32, 34], bo h published in 2011 and a ed as being a high isk o bias, used me a-analysis. Hagan e al. [34] included 7 s udies ocusing on NSP and HCV ou comes (6 coho and 1 case-con ol s udy), all om No h Ame ica. The pooled analysis o all s udies, using andom e ec s models and wi h all measu es o e ec con e ed o ela i e isks, showed an inc ease in he isk o HCV acquisi ion wi h NSP ( ela i e isk, 1.62; 95% CI: 1.04 o 2.52). The e was conside able he e ogen- ei y (I 2 = 81%), bu no subg oup o sensi i i y analyses we e pe o med, and s udy quali y was no explici ly epo ed o conside ed in he analysis. Au ho s cau ioned Fe nandes e al. BMC Public Heal h (2017) 17:309 Page 9 o 15