Table 1 Summary of studies included in review following appraisal

Hagan et al. 2000 [19]

Cohort

Seattle, USA

Recruitment:

1994 - 1997

Random sample of IDU recruited from locations other than NSP including prisons, street outreach, social services and drug treatment programmes. 2,814 IDU aged 14 years or older who had injected elicit drugs in the last year of whom 2,208 (78.5%) completed a follow-up interview. NSP users (n = 1236) had 'ever' used the NSP in the period prior to enrolment; non-users (n = 346) had 'never' used NSP prior to. IDU paid to participate. The NSP provided drug cookers, cottons and rinse water.

Exposure: NSP use ('Ever' vs. 'Never') defined at study enrolment.

Outcome: Self-reported sharing of 'cookers or cottons' in the month and six months prior to study enrolment. Thereafter behaviours in the month prior to follow up.

Data collected by interview administered questionnaire at baseline and follow up visit at 1 year.

Dichotomous measure of NSP use 'ever' vs. 'never' does not allow for gradations in use. NSP use at baseline may not represent NSP use at follow up.

Self-reported outcome measures.

Statistical analyses were carried out on 71.6% (n = 1,582) participants who completed a follow up visit, representing 56.2% of all eligible participants enrolled at baseline. The authors report no association between study retention and NSP use or injecting risk behaviour, however high loss to follow up may have reduced statistical power.

Huo et al 2007 [20]

Cohort

Chicago, USA Recruitment: 1997 - 2000

Convenience sample of IDU that had injected in the last 6 months and were aged 18 years or older. Verification of age and recent IDU required. Regular NSP users (n = 729) recruited at NSP (multisite). Non NSP users (n = 172), individuals who had not used a NSP in the 6 months prior to study enrolment, were recruited from a 'control' area without local NSP services via street outreach and peer referral. The NSP supplied cotton, cookers, alcohol wipes and water. Non-users had access to cookers.

Exposure: NSP use (Regular vs. non NSP users) defined at study enrolment.

Outcome: Self-reported prevalence of sharing 'other' injecting paraphernalia in the 30 days prior to interview. This composite endpoint was used to denote the sharing of any single item of non-N/S paraphernalia.

Data collected by interview administered questionnaire at baseline and three annual follow up visits.

Convenience sampling

Differential sampling method NSP users and non-users.

Dichotomous measure of NSP use does not allow for gradations in use. NSP use defined at baseline may not represent NSP use at follow up.

Both NSP users and non-users had access to sterile injecting paraphernalia thereby limiting comparison between groups.

Self-reported outcome measures

Follow up data available on 83.7% (n = 610) NSP users and 75.6% (n = 130) non NSP users (did not differ according to risk behaviour at baseline).

Sears et al. 2001 [21]

Cohort

San Francisco, USA.

Recruitment:

July 1993

NSP users recruited using systematic sampling (every nth person) of IDU attending NSP in three locations. Non-users were recruited using a targeted sampling method through an ongoing multi-wave cross-sectional behavioural study that included HIV testing and counselling (the Urban Health Study). Participation rate for NSP users 88%, for non-users 84%. NSP status subsequently reclassified according to reported NSP use at baseline rather than recruitment site. NSP users (n = 132) reported at least one use of NSP in the last 30 days and at least one additional use of NSP in the preceding 6 months. Non-users (n = 97) reported no use of NSP in the past 30 days or the preceding 6 months.

IDU paid to participate. The NSP provided cottons and alcohol swabs.

Exposure: NSP use (NSP use vs. non use) defined at study enrolment.

Outcome: Self-reported sharing of cotton, cooker or water in the preceding 30 days.

Data collected during structured interviews at baseline, 6 month and 12 month follow up.

Differential sampling methods in recruitment of NSP users and non-users (possible selection bias).

Although the outcome measure was sharing of cottons, cookers or water, participants only had access to cottons and alcohol swabs from NSP.

Dichotomous measure of NSP use excludes gradations in use. NSP use at baseline may not reflect NSP use at follow-up.

Self-reported outcome measures.

Follow up data collected on 49% and 57% of participants at 6 and 12 month visits respectively. Differential follow up according to NSP use; in the non-users group those lost to follow up reported higher risk injecting behaviours. Bivariate analyses carried out on complete cases only (n = 101, 44% of participants). Data from full sample included in multivariate analysis taking into account random missing data. AOR stratified by NSP status not reported.

P-values not provided for bivariate analysis of reported prevalence of sharing 'cooker, cotton, or water' according to NSP use over time.

Stoltz et al. 2007 [22]

Cohort

Vancouver, Canada. Recruitment: 2004 - 2005

Random sample of IDU attending a medically supervised injection facility (SIF). 760 IDU who completed a follow up survey between 1^st July 2004 and 30^th June 2005. Frequent SIF users (n = 433) were defined as those who used the SIF for some, most or all of their injecting episodes (i.e. > 25% of all injecting episodes) in the month prior to interview. Infrequent SIF users (n = 327) used the facility for < 25% of injecting episodes. SIF use determined at baseline. Participation rate not reported.

Facility provided a location to inject drugs intravenously under the supervision of a nurse, advice on safe injecting practices, referral to health care and addiction services and medical intervention in the event of an overdose. IDU paid to participate. Paraphernalia provided included alcohol swabs, sterile water and cookers.

Exposure: SIF use (frequent vs. infrequent) defined at study enrolment.

Outcome: Self-reported change in using clean water for injecting since using the SIF.

Data collected by interview administered questionnaire at baseline and 6 month follow-up.

Baseline classification of SIF use may not represent SIF use at follow up.

Dichotomous measure of SIF use does not allow for gradations in use.

Losses to follow up not documented although authors report no difference in the baseline characteristics of the cohort according to follow up status.

Self-reported outcome measures.

Vlahov et al. 1997 [23]

Cohort

Baltimore, USA.

Recruitment:

1994 - 1995

Systematic sampling of all self-reported IDU enrolled in the NSP during the first year of operation. 422 IDU completed baseline interviews (14.2% of all individuals enrolled in the NSP). IDU paid to participate. NSP provided injection kits containing alcohol wipes, cotton and cookers.

Exposure: NSP use (before and after establishment of NSP).

Outcome: Self-reported sharing of cookers or cottons in the preceding 2 weeks.

Data collected by interview administered questionnaire at baseline (NSP inception), 2 weeks and 6 months

Self-reported outcome measures.

Unadjusted analyses only.

No comparison group.

Follow up 79.4% (n = 335) at 2 weeks, 52.4% (n = 221) at 6 months. Individuals lost to follow up differed with respect to injecting risk behaviours from those with complete data.

Bluthenthal et al. 1998 [24]

Serial Cross- Sectional Studies

California, USA.

Recruitment:

1992 - 1995

Targeted sample of IDU aged 18 years or older reporting injecting drug use in the preceding 30 days, recruited by street outreach and snowballing techniques. Verification of recent IDU required. Individuals who participated were encouraged to return on subsequent semi-annual waves. Total of 1034 IDU interviewed once with 53% (n = 684) returning for one or more follow up visit. Setting: Illegal NSP. NSP users had used the NSP in the 30 days prior to interview to obtain injecting equipment and NSP was the usual source of syringes in the 6 months prior to interview. NSP provided alcohol and cotton filters. Individuals who had received other risk reduction supplies such as cottons and cookers from any HIV provider in the 30 days prior to interview were identified.

Exposure: NSP use (NSP users vs. non-users and trends over time).

Outcome: Self-reported sharing of 'cookers, cottons or rinse water' in the 30 days prior to interview. A composite outcome of sharing injection supplies was used to indicate the self-reported sharing of any single item of non- N/S paraphernalia.

Data collected by interview administered questionnaire in seven semi-annual waves.

Participants only had access to cottons and alcohol wipes from NSP thereby limiting the ability to interpret results.

Convenience sampling.

Self-reported outcome measures.

Morissette et al. 2007 [25]

Cross-sectional

Montreal, Canada. Recruitment: 2004 - 2005

Convenience sample of IDU (n = 275) aged 18 years or older who had injected at least once in the last 6 months recruited by flyers, peer referral and onsite personnel at three large NSP. Verification of recent injecting required. Participation rate not reported.

IDU paid to participate. NSP distributed sterile water and Securicup kits containing cooker, filter and swabs.

Exposure: Use of sterile non-N/S injecting equipment (frequent vs. infrequent).

Outcome: Self-reported sharing of drug preparation equipment (cookers, filters, water) in the preceding 6 months (yes/no), and frequency of use of sterile cookers, filters and water in the preceding month, HCV status. Source of sterile injecting equipment and reasons for not using sterile equipment in the past 6 months.

Data collected by interview-administered questionnaire.

Convenience sample.

Self-reported outcome measures.

Longshore et al. 2001 [26]

Cross-sectional

Rhode Island, USA.

Recruitment:

1997 - 1998

Convenience sample of IDU over the age of 18 yrs old attending NSP or MMT.

248 IDU (participation rate 70%). IDU paid to participate. NSP supplied cotton, cookers, alcohol wipes and rinse water.

Exposure: NSP use (frequent vs. infrequent use).

Outcome: Self-reported sharing of cooker or other items of injecting paraphernalia in the 6 months prior to interview and frequency of visits to NSP. Responses treated as categorical variable.

Data collected by interview administered questionnaire.

Convenience sample.

Self-reported outcome measures.

No comparison group.

Kipke et al. 1997 [27]

Cross-sectional

California, USA. Recruitment:

1994

Convenience sample of IDU aged 16-24 years who had injected at least once during the last 30 days. 89 NSP users recruited through NSP (mobile van). 109 non-users (poorly characterised) recruited by street outreach and using snowballing techniques. Participation rate not provided. IDU paid to participate. NSP provided water, cotton, cookers, alcohol wipes.

Exposure: NSP use (NSP users vs. non-users).

Outcome: Self-reported sharing of 'cookers, cottons or rinse water' in the preceding 6 months.

Data collected during semi-structured interview.

Convenience sample.

Differential sampling methods between NSP users and non NSP users.

Flawed assessment of exposure to NSP and access to paraphernalia.

NSP non-users poorly characterised.

Self-reported outcome measures.

Sears et al. 2001 [28]

Cross-sectional

San Francisco, USA.

Recruitment:

July - Sept. 1997

Convenience sample of IDU aged 15-25 years who had injected drugs in the last 30 days and had experienced homelessness within the last 30 days. Recruitment in two areas of city: one in which a secondary exchange programme operated (intervention site) and a comparison site near which two NSP operated. 67 IDU recruited from the intervention site and 55 from the comparison site (100% participation). IDU paid to participate. Secondary syringe exchange programme led by 4 peer exchangers provided cookers, filters, water and alcohol swabs. Items of paraphernalia provided by other NSP unclear.

Exposure: Access to secondary syringe exchange vs. access to other NSP (comparison based site of recruitment).

Outcome: Self-reported sharing of cookers in the preceding 30 days.

Data collected during structured interview.

Convenience sample.

Almost 90% of study population used NSP. In the comparison group almost 40% reported obtaining sterile paraphernalia from secondary exchange, half from a recognised formal NSP, a quarter from an underground NSP and three quarters via satellite distribution.

Self-reported outcome measures.

Results of adjusted analysis not presented.

Heimer et al. 2002 [29]

Cross- sectional

Multi-site study Connecticut, California, Illinois, USA.

Recruitment: 1998 - 2000

Targeted sample of IDU aged 18 years or older reporting injecting drug use in the preceding 30 days, recruited by street outreach and modified snowballing.

493 IDU (n = 122 NSP users, n = 371 NSP nonusers). Participation rate not provided.

NSP provided water but unclear whether other items of injecting paraphernalia such as cookers or filter were provided.

Exposure: NSP use (NSP users vs. non-users).

Outcome: Self-reported sharing of cookers, cottons, rinse water or drug water in the 30 days prior to interview. In addition information on HIV and hepatitis knowledge was collected.

Data collected during semi-structured interview

NSP users and nonusers not clearly defined.

Paraphernalia provided by NSP poorly defined.

Convenience sampling.

Self-reported outcome measures.

Cohort study but cross-sectional data only presented.

Unadjusted analyses comparing sharing of injecting paraphernalia among NSP users and nonusers only.

Guydish et al. 1998 [30]

Cross-sectional

San Francisco, USA.

Recruitment:

October 1994

Systematic sampling of all self-reported IDU attending multi-site NSP. Study population poorly characterised.

143 IDU recruited of whom 114 (80%) participated. IDU paid to participate.

NSP distributed cotton and alcohol wipes.

Exposure: NSP use

Outcome: Self-reported sharing of rinse water in the 30 days prior to interview. The sharing rinse water selected to represent all 'indirect' sharing. Also collected information on the proportion of all needles used by a client that came from the NSP and frequency of visits to NSP in last 30 days.

Data collected by interview administered questionnaire.

Rinse water was selected to represent all 'indirect' sharing; NSP did not distribute rinse water thereby limiting the ability to interpret results.

Study population poorly characterised.

Self-reported outcome measures.

Only unadjusted analyses presented.

No comparison group.

Colon et al. 2009 [31]

Non-randomised intervention study (pilot)

Puerto Rico, USA.

Recruitment:

16 week intervention period of recruitment not described

Targeted sample of IDU aged 18 years or older reporting injecting drug use in the preceding 7 days recruited by street outreach. 37 IDU (70% participation).

16 week community level intervention aimed at modifying injecting risk behaviours. IDU provided with injection kits that included hand sanitizer, sterile water in a water bottle and syringe filter and advised on use.

Exposure: Community level intervention (see study population)

Outcome: Self-reported change in injecting practices in the preceding 7 days: use of hand sanitizer, use of water bottle and use of syringe filter provided in intervention, use of non-study materials including water and cookers. Self-reported sharing of cooker during last day of injection.

Data collected by interview administered questionnaire. Pre and post intervention data collected by structured interview. Analysis of systematically sampled water bottles and cookers for the presence of red blood cells. Samples collected from shooting galleries 4 and 1 weeks prior to intervention and 8, 14 and 18 weeks following commencement of the intervention.

Non-randomised, uncontrolled study design

Convenience sample.

Pilot study with small numbers therefore unable to generalise findings to other populations.