Oral substitution treatment of injecting opioid users for prevention of HIV infection

Article (PDF Available)inCochrane database of systematic reviews (Online) 10(8):CD004145 · August 2011with60 Reads
DOI: 10.1002/14651858.CD004145.pub4 · Source: PubMed
Abstract
Injecting drug users are vulnerable to infection with HIV and other blood borne viruses as a result of collective use of injecting equipment as well as sexual behaviour. This review looks at original studies that reported the frequency or prevalence of risk behaviours, or information on HIV infection related to substitution treatment of opioid dependence to assess the extent to which oral substitution treatment prevents the transmission of HIV infection. It was not possible to accurately estimate the extent of reduction, but it is clear that oral substitution treatment reduces risk behaviours and also the probability of HIV infection amongst injecting drug users in substitution treatment.
Substitution treatment of injecting opioid user s for
prevention of HIV infection (Review)
Gowing L, Farrell M, Bornemann R, Sullivan LE, Ali R
This is a reprint of a Cochrane review, prepared and maintained by The Cochrane Collaboration and published in The Cochrane Library
2008, Issue 3
http://www.thecochranelibrary.com
Substitution treatment of injecting opioid users for prevention of HIV infection (Review)
Copyright © 2008 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.
T A B L E O F C O N T E N T S
1HEADER . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1ABSTRACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2PLAIN LANGUAGE SU MMARY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2BACKGROUND . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3OBJECTIVES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3METHODS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6RESULTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
23DISCUSSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
25AUTHORS’ CONCLUSIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
26ACKNOWLEDGEMENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
27REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
33CHARACTERISTICS OF STUDIES . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
54DATA AND ANALYSES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
56Analysis 1.1. Comparison 1 Drug use and risk outcomes (f ol low-up studies), Outcome 1 Proportion reporting injecting
use. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
58Analysis 1.2. Comparison 1 Drug use and risk outcomes (follow-up studies), Outcome 2 Proportion sharing injecting
equipment. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
60Analysis 1.3. Comparison 1 Drug use and risk outcomes (follow-up studies), Outcome 3 Opioid use (proportion
reporting use or with positive urine samples). . . . . . . . . . . . . . . . . . . . . . . .
62Analysis 1.4. Comparison 1 Drug use and risk outcomes (follow-up studies), Outcome 4 Cocaine use (proportion using
or with positive urine screens). . . . . . . . . . . . . . . . . . . . . . . . . . . . .
63Analysis 2.1. Comparison 2 Sex-related risk outcomes (follow-up studies), Outcome 1 Proportion reporting multiple
sex partners. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
64Analysis 2.2. Comparison 2 Sex-related risk outcomes (follow-up studies), O utcome 2 Proportion reporting unprotected
sex (or use of condoms in half or less of occasions). . . . . . . . . . . . . . . . . . . . . .
65Analysis 2.3. Comparison 2 Sex-related risk outcomes (follow-up studies), Outcome 3 Proportion reporting exchanges
of sex for drugs or money. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
66Analysis 3.1. Comparison 3 Overall risk assessments (follow-up studies), Outcome 1 Proportion at risk. . . . .
67Analysis 4.1. Comparison 4 Drug use and risk outcomes - substitution treatment versus no substitution treatment,
Outcome 1 Proportion reporting injecting use. . . . . . . . . . . . . . . . . . . . . . .
69Analysis 4.2. Comparison 4 Drug use and risk outcomes - substitution treatment versus no substitution treatment,
Outcome 2 Frequency of injecting use. . . . . . . . . . . . . . . . . . . . . . . . . .
70Analysis 4.3. Comparison 4 Drug use and risk outcomes - substitution treatment versus no substitution treatment,
Outcome 3 Proportion sharing injecting equipment. . . . . . . . . . . . . . . . . . . . .
72Analysis 4.4. Comparison 4 Drug use and risk outcomes - substitution treatment versus no substitution treatment,
Outcome 4 Frequency of opioid use. . . . . . . . . . . . . . . . . . . . . . . . . . .
73Analysis 4.5. Comparison 4 Drug use and risk outcomes - substitution treatment versus no substitution treatment,
Outcome 5 Frequency of cocaine use. . . . . . . . . . . . . . . . . . . . . . . . . .
74Analysis 4.6. Comparison 4 Drug use and risk outcomes - substitution treatment versus no substitution treatment,
Outcome 6 Drug risk scores. . . . . . . . . . . . . . . . . . . . . . . . . . . . .
75Analysis 5.1. Comparison 5 Sex-related risk outcomes - substitution treatment versus no substitution treatment,
Outcome 1 Sex-related risk score. . . . . . . . . . . . . . . . . . . . . . . . . . . .
76Analysis 5.2. Comparison 5 Sex-related risk outcomes - substitution treatment versus no substitution treatment,
Outcome 2 Number of sex partners. . . . . . . . . . . . . . . . . . . . . . . . . . .
77Analysis 5.3. Comparison 5 Sex-related risk outcomes - substitution treatment versus no substitution treatment,
Outcome 3 Proportion reporting unprotected sex (or use of condoms in half or less of occasions). . . . . .
78Analysis 5.4. Comparison 5 Sex-related risk outcomes - substitution treatment versus no substitution treatment,
Outcome 4 Frequency of condom use. . . . . . . . . . . . . . . . . . . . . . . . . .
79Analysis 5.5. Comparison 5 Sex-related risk outcomes - substitution treatment versus no substitution treatment,
Outcome 5 Proportion reporting exchanges of sex for drugs or money. . . . . . . . . . . . . . .
iSubstitution treatment of injecting opioid users for prevention of HIV i nfection (Review)
Copyright © 2008 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.
80Analysis 5.6. Comparison 5 Sex-related risk outcomes - substitution treatment versus no substitution treatment,
Outcome 6 Frequency of exchanges of sex for drugs or money. . . . . . . . . . . . . . . . . .
81Analysis 5.7. Comparison 5 Sex-related risk outcomes - substitution treatment versus no substitution treatment,
Outcome 7 Number reporting multiple sex partners. . . . . . . . . . . . . . . . . . . . .
82APPENDICES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
83WHAT’S NEW . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
84HISTORY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
84CONTRIBUTIONS OF AUTHORS . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
84DECLARATIONS OF INTEREST . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
84SOURCES OF SUPPORT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
84INDEX TERMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
iiSubstitution treatment of injecting opioi d users for prevention of HIV infection (Review)
Copyright © 2008 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.
[Intervention review]
Substitution treatment of injecting opioid users for
prevention of HIV infection
Linda Gowing
1
, Michael Farrell
2
, Reinhard Bornemann
3
, Lynn E Sullivan
4
, Robert Ali
5
1
Department of Clinical and Experimental Pharmacology, University of Adelaide, Adelaide, Australia.
2
Institute of Psychiatry, National
Addiction Ce ntre, London, UK.
3
Fakultaet fuer Gesundheitswissenschaften (School of Public Health), Universitaet Bielefeld, D-33501
Bielefeld, Germany.
4
School of Medicine, Yale University, New Haven, USA.
5
Clinical and Experimental Pharmacology, Drug and
Alcohol Services Council, Parkside, Australia
Contact address: Linda Gowing, Department of Clinical and Experimental Pharmacology, University of Adelaide, DASC Evidence-
Bsed Practice Unit„ Adelaide, 5005, Australia.
linda.gowing@adelaide.edu.au. (Editorial group: Cochrane Drugs and Alcohol Group.)
Cochrane Database of Systematic Reviews, Issue 3, 2008 (Status in this issue: Edited)
Copyright © 2008 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.
DOI: 10.1002/14651858.CD004145.pub3
This version first published online: 23 April 2008 in Issue 2, 2008. Re-published online with edits: 16 July 2008 in Issue 3, 2008.
Last assessed as up-to-date: 2 December 2007. (
Dates and statuses?)
This record should be cited as: Gowing L, Farrell M, Bornemann R, Sullivan LE, Ali R. Substitution treatment of injecting
opioid users for prevention of HIV infection. Cochrane Database of Systematic Reviews 2008, Issue 2. Art. No.: CD004145. DOI:
10.1002/14651858.CD004145.pub3.
A B S T R A C T
Background
Injecting drug users are vulnerable to infection with HIV and other bloo d borne viruses as a result of collective use of injecting equipment
as well as sexual behaviour.
Objectives
To assess the effe ct of oral substitution treatment for opioid dependent injecting drug users on rates of HIV infections, and high risk
behaviours.
Search strategy
We searched the Cochrane Central Register of Controlled Trials, MEDLINE, EMBASE and PsycINFO to March 2007. We also
searched reference lists of articles, reviews and conference abstracts We also searched reference lists of articles, reviews and conference
abstracts
Selection criteria
Studies were required to consider th e incidence of risk behaviours, or the incidence of HIV infection related to substitution treatment
of opioid dependence. All types of original studies were considered. Two reviewers independently assessed studies for inclusion.
Data coll ection and analysis
One reviewer extracted data from included studies, assessed quality and confirmed decisions by consulting with all other reviewers.
Main results
Thirty-three studies, involving 10,400 participants, were included. The majority were not randomised controlled studies and there were
problems of confounding and bias. The studies varied in several aspects limiting the extent of quantitative analysis. Studies consistently
show that oral substitution treatment for opioid-dependent injecting drug users with methadone or buprenorphine is associated with
1Substitution treatment of injecting opioid users for prevention of HIV infection (Review)
Copyright © 2008 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.
statistically significant reductions in illicit opioid use, injecting use and sharing of injecting equipment. It is also associated with
reductions in the proportion of injecting drug users reporting mul tiple sex partners or exchanges of sex for dr ugs or money, but has
little effect on condom use. It appears that the reductions in risk behaviours related to drug use do translate into reductions in cases of
HIV infection.
Authors conclusions
Oral substitution treatment for injecting opioid users reduces drug-related behaviours with a high risk of HIV transmission, but has less
effect on sex-related risk behaviours. The lack of data from randomised controlled studies limits the strength of the evidence presented
in this review.
P L A I N L A N G U A G E S U M M A R Y
Oral substitution treatment for injecting opioid users reduces drug- related behaviours with a high risk of HIV transmission,
but has less effect on sex-related risk behaviours.
Injecting drug users are vulnerable to infection with HIV and other blood borne viruses as a result of collective use of injecting
equipment as well as sexual behaviour. This review looks at original studies that reported the frequency or prevalence of risk behaviours,
or the prevalence of HIV infection related to substitution treatment of opioid dependence to assess the extent to which oral substitution
treatment prevents the transmission of HIV infection. It was not possible to accurately estimate th e extent of reduction, but it is
clear that oral substitution treatment reduces risk behaviours and also actual cases of HIV infection amongst injecting drug users in
substitution treatment.
B A C K G R O U N D
Opioid use is a global phenomenon. Almost 16 million people
in the world, or 0.4% of the world’s population aged 15-64, are
abusers of opiates. On a regional level estimated rates range from a
low of 0.2% in South-East Europe and Africa to a high of 1.6% in
Eastern Europe (
World Drug Rep. 2006). Although opioid users
constitute only a very small propor tion of the population, because
opioids are predominantly used by injection, the contribution of
opioid use to the transmission of Human Immunodeficiency Virus
(HIV) is significant: nearly one-third of new HIV infections, out-
side of sub-Saharan Africa, are due to injecting drug use; in East-
ern and Central Asia, use of contaminated injecting equipment
accounts for more than 80% of all HIV cases (
World AIDS Rep.
2006
).
The significance of HIV lies in the high rates of premature mor-
bidity and the high cost of treatment and care. The introduc-
tion of combination anti-retroviral therapy for those infected with
HIV has delayed progression to Acquired Immune Deficiency Syn-
drome (
Volberding 1999), but these drugs are expensive and are
not available in all countries. It is not only the spread of HIV/A IDS
amongst injecting drug users that is relevant. Injecting drug users
infected with HIV can become a means of transmission to the
general population through sexual contacts with people who are
not drug users, as well as transmission to unborn children by in-
fected mothers. Links between drug use and commercial sex work
are also significant to the spread of HIV beyond the population of
injecting drug users.
Injecting drug users are vulnerable to infection with HIV and
other blood borne viruses as a result of collective use of injecting
equipment as well as sexual behaviour. The risk of virus transmis-
sion through injecting drug use is determined by:
epidemiological factors (viral l oad of the infected contact
(
Mortimer 2001); the way in which injecting equipment is
shared (
Rutherford 1989); individual susceptibility);
psychosocial factors (drug craving; intoxication; risk knowl-
edge and attitude towards risk); and
environmental factors (hurried use in public places; access to
clean injecting equipment; settings, such as prisons that are
hostile to drug use).
Individuals who are new to injecting are particularly vulnerable
to sharing injecting equipment as they may not know how, or be
afraid, to inject themselves. They th us often ask a more experi-
enced user (who may be infected) for assistance. Women also have
increased vulnerability as they are more likely to be asked to share
injecting equipment by their partner and ofte n find it difficult to
negotiate low-risk sexual practices and condom use.
Since sharing or use of contaminated syringes and needles is a
very efficient way of spreading HIV, it can spread very rapidly
amongst injecting dr ug users (
Des Jarlais 1992; Stimson 1995).
For example, in th e Russian Federation the number of registered
drug abusers (mostly opioid users) rose from 65,000 in 1995 to
2Substitution treatment of injecting opioid users for prevention of HIV infection (Review)
Copyright © 2008 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.
270,000 in 2000. In 1996, seven persons per million inhabitants
were registered for the first time as having been infected by HIV
as a consequence of injecting drug use. This rate increased to 248
persons per million inhabitants in 2000 (
UNDCP 2002).
The data reported by United Nations AIDS Program (
World AIDS
Rep. 2006) indicate the extent of regional variation in the propor-
tion of HIV infections that are attributed to injecting drug use. A
number of factors are thought to underlie this regional variabil-
ity, including differences in the prevalence of HIV amongst the
population of injecting drug users, the extent of programs provid-
ing counselling in harm reduction and clean injecting equipment,
and the availability of treatment for injecting drug users (
NIDA
2002
). There is evidence that, with attention to these factors, it is
possible to contain or e ven reverse the epidemic of HIV infection
related to injecting drug use (
Des Jarlais 2000).
There is a widely held view that the provision of substitution treat-
ment for opioid dependent, injecting drug users is an important
component of efforts to contain HIV infections. The e ffectiveness
of substitution treatment in preventing HIV infections may arise
in several ways:
reduced craving for and use of illicit opioids;
reduced frequency of injecting drug use;
reduced need to obtain illicit supplies thereby reducing high
risk behaviours such as the exchange of sex for money or
drugs;
greater control of th e cycle of craving, intoxication and with-
drawal, making injecting drug users more receptive to pre-
ventive messages;
reduced frequency of sharing of injecting equipment arising
from reduced f requency of injecting use and greater awareness
of risk reduction strategies; and
regular contact with treatment agencies increasing opportu-
nities for medical and psychosocial interventions.
The aim of this review is to assess the effect of substitution treat-
ment on injecting and sexual risk behaviours, and thereby assess
its capacity to contain HIV infections.
Worldwide, methadone administered orally as syrup is the phar-
macological agent that is most commonly used for substitution
treatment of opioid dependence. Buprenorphine administered
sublingually is increasingly being used, and in France is the pre-
ferred agent for substitution treatment of opioid dependence. Levo
alpha acetyl methadol (LAAM) was used to a limited extent in
the USA, but was removed from the market because of concerns
about potential cardiovascular e ffects. In Germany, in the early
1990s, codeine was the most commonly used substitution agent.
Other slow release oral preparations of morphine are emerging as
alternative substitution agents, and there is some use of injectable
preparations for substitution treatment, notably heroin in Switzer-
land, the Netherlands, the UK, and Germany. Evaluations of the
latter approaches remain limited. Studies of the effectiveness of
substitution treatment were therefore expected to largely relate
to methadone, but evidence relating to other oral preparations
(buprenorphine, LAAM, codeine and slow release morphine) was
also considered for this review.
This review seeks to assess the effectiveness of oral substitution
treatment f or opioid dependent injecting drug users in ter ms of
rates of HIV infection, and the prevalence and frequency of be-
haviours associated with a high risk of HIV transmission (injecting
drug use; collective use of injecting equipment; unprotected sex;
number of sex partners).
O B J E C T I V E S
To assess the effect of oral substitution treatment for opioid de-
pendent injecting drug users on the prevention of human immun-
odeficiency virus (HIV) infections or the incidence of behaviours
associated with a high risk of transmission of HIV.
M E T H O D S
Criteria for considering studies for this review
Types of studies
The primary criterion for selection of studies was that they consider
the incidence of behaviours with high risk of HIV tr ansmission, or
the incidence of HIV infection (seroprevalence or seroconversion,
including further infection with different phenotypes of HIV),
related to substitution treatment for opioid dependence.
Studies with retrospective collection of data on risk behaviour
were excluded. Such studies were defined as those with one time
point of data collection with participants being asked to recall their
risk behaviour before and after commencement of substitution
treatment.
For the purposes of this review studies were identified as being one
of the following design types:
(1) Controlled clinical trials
Randomised and quasi-randomised studies comparing substitu-
tion treatment with another form of treatment, placebo or no
treatment.
(2) Cohort studies
Longitudinal studies in which the HIV status and risk behaviour of
injecting drug users is assessed prior to, and after commencement
of an intervention or contact point. Within the cohort a study
group (for whom the intervention is oral substitution treatment)
and a comparison group (receiving another form of treatment
3Substitution treatment of injecting opioid users for prevention of HIV infection (Review)
Copyright © 2008 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.
or no treatment) are identified. The determination of group is
not random and hence these studies are at risk of allocation bias,
potentially resulting in differences in participant characteristics at
baseline.
(3) Controlled before after studies
Data on HIV seroprevalence and risk behaviour for samples of
injecting drug users from identified sites before and after the in-
troduction of oral substitution treatment and samples from com-
parison sites where substitution treatment is not available.
(4) Interrupted time series studies
Data on HIV seroprevalence and risk behaviour is col lected for a
single sample of injecting drug users before and after commence-
ment of oral substitution treatment. The time of commencement
of oral substitution treatment must be clearly defined and there
must be two or three data points before and after commencement
of treatment. Studies that include only one data point before or
after treatment are considered to be descriptive studies.
(5) Case control studies
Studies that entail identification of one group of participants on
the basis of infection with HIV and selection of a comparison
group of participants not infected with HIV. These two groups
would then be compared in terms of exposure or non-exposure
to substitution treatment. As case control studies are retrospective
in nature, it can be difficult to control for differences in potential
confounding factors, such as risk behaviours, access to sterile in-
jecting equipment or preventive education, or exposure to differ-
ent background rates of HIV prevalence. However, case control
studies provide an indication of the capacity of substitution treat-
ment to protect against HIV infection, and thus they have been
included.
(6) Other types of studies or secondary data analysis
Study designs with data on HIV seroprevalence and risk behaviour
and either a change in those outcomes is related to oral substitu-
tion treatment, or those outcomes are compared within different
groups of injecting drug users exposed to oral substitution treat-
ment, another form of treatment, or no treatment. If controlled
clinical trials included the collection of such data as secondary
analyses (for example, as part of a controlled trial comparing meth-
adone and buprenorphine maintenance), the se data were regarded
as being derived from descriptive studies.
We considered the risk of confounding and bias for all included
studies, with particular attention to:
baseline characteristics of study and comparison groups;
methods for recruiting and assessing eligibility of study par-
ticipants;
identification of confounders and use of statistical methods
to adjust for confounding;
extent of loss to follow-up, assessment of differences between
those contacted and those lost to foll ow-up, and approaches
used to adjust for losses to follow-up;
differences in exposure to interventions other than oral sub-
stitution treatment; and
timing and methods of data coll ection.
Types of participants
Included studies involved opioid dependent drug users identified
as injecting users, or with a recent history of injecting drug use
at the time of entry into the study. Studies involving participants
using other injectable drugs in addition to opioids were included.
Types of interventions
Interventions involved the oral administration of full or partial
opioid agonists (methadone, buprenorphine, L AAM, codeine or
oral morphine) for substitution treatment of opioid dependence.
The dose and duration of substitution treatment are factors that
influence the outcomes of substitution treatment (
Kreek 2000;
Ward 1998). Hence we considered it desirable that studies in-
cluded in the review reported details of the dose and duration of
substitution treatment at the time outcomes were assessed. Many
studies met this requirement but only to a l imited extent. Lack
of detailed information on the substitution treatment on its own
was not considered grounds for exclusion, except where such data
was essential to relate outcomes to substitution treatment. Studies
that did not report detailed information of substitution treatment
and also did not fully meet another criterion were excluded.
Types of outcome measures
Behaviours that are known to be associated with a high risk
of tr ansmission of HIV include sharing of injecting equipment
(
Rutherford 1989), and unprotected sexual intercourse, particu-
larly anal se x (
Kaldor 1994). These behaviours were rarely specif-
ically reported, but high frequency injecting drug use, f requency
of condom use, multiple sexual par tners, and providing sex in ex-
change for money or drugs are all indicators of high risk activities.
This review seeks to examine the impact of substitution treatment
on all of these behaviours. Instruments have been developed for
scoring the level of HIV risk with such instruments generally con-
sidering both injecting drug use and sexual behaviour. Studies re-
porting outcomes based on the use of such instruments were also
included.
The review also compares the prevalence and incidence of HIV
infection for populations of injecting opioid users who receive oral
substitution treatment, and those not participating in treatment
or receiving other forms of treatment. Since prevalence and inci-
dence rates of HIV infection are influenced by a number of factors
there is potential for these data to be confounded. Consequently
this review places most emphasis on behaviours with high risk for
transmission of HIV for assessing the effect of substitution treat-
ment on the prevention of HIV infection, but considers the preva-
lence and incidence of HIV as secondary measures.
An area of debate has been whether any effect of substitution treat-
ment on HIV risk is achieved primarily through changes in overall
drug use, changes in levels of injecting use, or changes in high risk
4Substitution treatment of injecting opioid users for prevention of HIV infection (Review)
Copyright © 2008 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.
injecting practices (Caplehorn 1995; Gossop 2000). This review
sought to relate these outcomes to substitution treatment with a
view to determining the mechanism by which substitution treat-
ment has an ef fect on HIV risk. The e xtent to which substitution
treatment can reduce injecting drug use will be influenced by con-
comitant use of drugs such as cocaine which may also be injected.
Hence this review will also consider the extent of injecting use
of non-opioid drugs during methadone or another substitution
treatment.
Where studies consider the frequency of high risk behaviours be-
fore and after substitution treatment, it was required that simi-
lar data on these behaviours was collected for equivalent periods
of time before commencement of treatment, and after a specified
period of substitution treatment. Studies that compare risk be-
haviours in different time periods (for example one month prior to
treatment compared to six months after treatment) were excluded.
Search methods for identification of studies
We searched The Cochrane Central Register of Controlled Tri-
als (The Cochrane Library issue 3, 2006) which includes the
Cochrane Drugs and Alcohol Group trials register, MEDLINE
(January 1996-March 2007), EMBASE (January 1996-March
2007), and PsycINFO (January 1996-March 2007). For this up-
date databases were searched from 1996 to March 2007. We de-
signed a search strategy to retrieve references that address opi-
oid dependence, HIV transmission and substitution treatment.
The strategy was adapted to each of the major databases (accessed
through OVID Online) as indicated in
Appendix 1; Appendix 2;
Appendix 3; Appendix 4
We handsearched the reference l ists of retrieved studies and re-
views and considered unpublished data in the form of conference
abstracts identified through database searches. However, no spe-
cific action was undertaken to locate unpublished data as it was
considered that the y ield of unpublished data was likely to be low
relative to the time required to locate it. The design of this review
exposes analyses to some risk of bias, and we consider that any
unpublished data we might have missed would not shift the risk
of bias to any major degree.
All searches included non-English language literature. The non-
English studies that we located were assessed by one of us (RB)
making formal translation into English unnecessary.
An initial sort of studies retrieved by th e search strategy excluded
those that, on the basis of the title and abstract, clearly were not
original studies or did not relate HIV infections or risk behaviour
to substitution treatment of opioid dependent drug users. We
assessed the remaining studies in more detail against the inclusion
and exclusion criteria for the review. Studies that are well known
and frequently cited, and studies that appeared on first assessment
to be relevant to the review, but which were subsequently found
to not meet the criteria, are listed as excluded studies.
Data collection and analysis
Study Selection
Two reviewers (LG and RB) independently assessed each poten-
tially relevant study obtained by the search for inclusion accord-
ing to the identified inclusion and exclusion criteria (except non-
English studies which were assessed only by RB).
Data extraction
One reviewer (LG) extracted key information from studies that
met the inclusion criteria, and this was confirmed by consultation
with the other reviewers. Key findings from included studies were
reviewed descriptively in the first instance. Quantitative meta-
analysis was undertaken only for groups of studies of similar design.
Study quality
We assessed the methodological quality of the included studies us-
ing a scale developed specifically for this review, drawing on guide-
lines for quality assessment of trials developed by the Cochrane
Drugs and Alcohol Group. In these guidelines the criteria for non-
experimental studies effectively identify six aspects as impor tant
indications of potential bias and/or confounding. For each of these
six aspects, a score of either 0 or 1 was all ocated, as outlined be-
low, with 1 indicating risk was present, and 0 indicating that no
specific risk was identified. An overall score was derived for each
study by adding the scores for each aspect. Hence the higher the
overall score, the greater the risk of bias and/or confounding in
the study.
The six aspects and the criteria for allocating a score of 1 were as
follows.
Description of the population base - “1” allocated for lack
of information on participants drug use history (duration of
use, route of administration, frequency of use) at baseline, or
significant differences present in demographics or dr ug use
history of comparison groups.
Recruitment - “1” allocated if eligibility criteria not specifi-
cally reported or if processes used could be expected to differ-
entially se lect particular groups of drug users (e.g. those with
high levels of de pendence, or different treatment histories).
Confounding - “1” allocated if there was no consideration
of possible confounding in the study, or if the possibility of
confounding was identified but not adjusted for statistically.
Follow-up - “1” allocated where there was greater than 20%
loss to follow-up and significant differences were identified
between those contacted and those lost to follow-up, or there
was no assessment of possible differences between th ose con-
tacted and those lost to follow-up.
Interventions received - “1” allocated where participants in
the groups being compared were reported to have received,
or could have received, different interventions apart from
substitution treatment.
Data collection - “1” allocated where inconsistencies (eg. in
assessment period) were reported to have occurred, or may
have occurred, for groups or time points being compared.
5Substitution treatment of injecting opioid users for prevention of HIV infection (Review)
Copyright © 2008 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.
One reviewer (LG) allocated ratings and confirmed these by con-
sultation with other reviewers. No attempt was made to validate
the rating scale. Hence scores should be taken only as a general
indication of the risk of confounding and/or bias.
Data analysis
Much of the data for this review was in the form of follow-up
data, with HIV risk behaviour assessed before and after a period of
substitution treatment. This provided a statistical pr oblem in that
the “before” and after samples were not independent, and hence
standard statistical methods embedded in Review Manager 4.2.10.
were not appropriate for analysis of continuous data. Many of the
original studies used statistical approaches, such as ANOVA, that
are appropriate for assessing change over time involving one or
more groups of participants. Without access to individual patient
data, it is not possible to apply more accurate tests of statistical
significance. We have therefore taken the approach of summaris-
ing continuous, follow-up data in additional tables. These tables
include the level of statistical significance determined during the
original studies.
We undertook statistical analyses using Review Manager 4.2.10.
For dichotomous data from f ollow-up and other studies, we cal-
culated relative risks, and for continuous data from independent
samples, we calcul ated standardised mean differences, both with
95% confidence intervals and using a random e ffects model. For
the reasons given in the results section, we did not calculate subto-
tals and totals.
R E S U L T S
Description of studies
See:
Characteristics of included studies; Ch aracteristics of excluded
studies
.
As was expected, there were very few controlled studies that met
the criteria for inclusion in this review - this is likely to be due,
at least in part, to ethical difficulties associated with allocating
injecting opioid users to control groups that do not receive sub-
stitution treatment. Consequently, th is review includes all types
of original studies that provide data relating HIV risk behaviour
to oral substitution treatment. Studies are grouped according to
study design, and potential sources of confounding and bias are
identified. This approach enables assessment of whether there is a
consistent effect across all study types on HIV risk behaviours or
the actual transmission of HIV.
(1) Excluded studies
The literature searching process identified a large number of re-
ports of potential relevance to the review. The reports listed as
excluded studies are those with multiple citations in relation to
substitution treatment and the prevention of HIV, or which were
only excluded following a detailed assessment against the criteria
for inclusion in the review. (see
Characteristics of included studies)
(2) Included studies
In total, 33 studies involving around 10,400 participants met
the inclusion criteria for this review see
Characteristics of included
studies
.
In a number of instances multiple reports were identified as having
been derived from the same study. Where possible these reports
were considered together with a view to extracting a single set of
data, thereby avoiding bias due to double-counting of participa nts.
Abbott 1998, Avants 1998, Baker 1995, Chatham 1999, Gossop
2000
, Grella 1996 and Teesson 2006 were all handled in this way.
Camacho 1996, Chatham 1999, and Simpson 1995 are all related
to the DATAR study in Texas, USA but the nature of the analyses
and subsamples of participants are such that it was not possible to
consider the reports together. The possibility of double-counting
of participants is considered in discussion of analyses.
(3) Study types
The studies included in this review were grouped into six cate-
gories, as defined in ’Criteria for considering studies for this re-
view’, ’Types of studies’.
Only two studies (
Dolan 2003 and Sees 2000), involving 561
participants, were identified as controlled trials. Four studies
(
Kwiatkowski 2001, Maddux 1997, Metzger 1993 and Teesson
2006
), involving 2005 participants, were identified as cohort stud-
ies. There were no controlled before-after or interrupted time se-
ries studies. Two studies (
Moss 1994 and Serpelloni 1994), in-
volving 761 participants, were identified as case control studies.
The remaining 25 studies, involving 7073 participants, were other
types of descriptive studies. It should be noted that some stud-
ies were initiated as controlled trials - for example, Abbott 1998,
Avants 1998, Finch 1995, Grella 1996, Iguchi 1998 , Lott 2006,
Marsch 2005, Schroeder 2006 and Strang 2000. In general, the
comparison for which these controlled trials were initiated was not
relevant to this review. The data used from these studies are pri-
marily comparisons of HIV risk behaviour at baseline and follow-
up after a period of substitution treatment. Hence, while the se
studies commenced as controlled trials, for the purposes of this
review, the data are considered to be from descriptive studies.
(4) Nature and setting of substitution treatment
Marsch 2005 used buprenorphine, in Teesson 2006 either
buprenorphine or methadone were used, and
Lott 2006compared
methadone, buprenorphine and LAAM for substitution treat-
ment. In the remaining studies that met the criteria for inclusion
in this review, methadone was the only drug used for substitution
treatment (see
Table 1). Doses of 60mg/day or more are generally
considered necessary for methadone maintenance treatment to be
effective (
NIH 1997). Of the 32 studies using methadone, 12
studies reported average doses of 60mg/day or more, and eight
studies reported average doses between 40 and 60mg/day. In 12
studies the meth adone dose was not reported. While it was gen-
erally not stated, these doses are assumed to relate to racaemic d,l-
methadone. In Germany l-methadone has been used for substi-
tution treatment of opioid dependence but in other countries the
6Substitution treatment of injecting opioid users for prevention of HIV infection (Review)
Copyright © 2008 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.
available preparation is the racaemic mixture.
Table 1. Nature of substitution treatment
Study Substitute agent Av. dose/day Setting
Abbott 1998 Methadone About 70mg Specialist drug & alcohol program
Avants 1998 Methadone About 80mg Specialist drug & alcohol program
Baker 1995 Methadone About 66mg Not reported
Batki 1989 Methadone Not reported Specialist AIDS program
Britton 1994 Methadone Not reported Specialist drug & alcohol program
Brooner 1998 Methadone 70mg Specialist drug & alcohol program
Camacho 1996 Methadone About 40mg Specialist drug & alcohol program
Chatham 1999 Methadone About 40mg Specialist drug & alcohol program
Dolan 2003 Methadone 61mg Prison methadone program
Finch 1995 Methadone About 50mg Specialist drug & alcohol program
Gossop 2000 Methadone About 50mg Specialist drug & alcohol program
& primary health
Grella 1996 Methadone Not reported Specialist drug & alcohol program
Iguchi 1998 Methadone 40mg/day (taper) Specialist drug & alcohol program
King 2000 Methadone 50mg/day Specialist drug & alcohol program
Kwiatkowski 2001 Methadone 70mg/day Specialist drug & alcohol program
Lott 2006 (1) Methadone (2) buprenorphine
(3) LA AM
(1) 60-100mg (2) 8-16mg (3) 36-
56mg
Specialist drug & alcohol program
Maddux 1997 Methadone Not reported Specialist drug & alcohol program
Magura 1991 Methadone Not reported Specialist drug & alcohol program
Margolin 2003 Methadone ~80mg/day Specialist drug & alcohol program
Mark 2006 Methadone Not reported Specialist drug & alcohol program
Marsch 2005 Buprenorphine, sublingual solution 7.2-7.5 mg/day Specialist drug & alcohol program
7Substitution treatment of injecting opioid users for prevention of HIV infection (Review)
Copyright © 2008 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.
Meandzija 1994 Methadone Not reported Not reported
Metzger 1993 Methadone About 44mg Specialist drug & alcohol program
Sees 2000 Methadone Not reported Not reported
Moss 1994 Methadone About 85mg Not reported
Schroeder 2006 Methadone 70-80 mg/dy Specialist drug & alcohol program
Serpelloni 1994 Methadone Not reported Specialist drug & alcohol program
Simpson 1995 Methadone Not reported Specialist drug & alcohol program
Stark 1996 Methadone 80-100mg/day Not reported
Strang 2000 Methadone 57mg/day Specialist drug & alcohol program
Teesson 2006 Methadone, buprenorphine Not reported Not reported
Thiede 2000 Methadone Not reported Specialist dr ug & alcohol program
Williams 1992 Methadone 70mg/day Specialist drug & alcohol program
In two studies methadone was provided, for at least one group
of participants, in the context of detoxification rather than main-
tenance. In Iguchi 1998 the detoxification was scheduled to be
completed in 90 days. Data on HIV risk behaviour was reported
for weeks one and two of treatment while participants were sta-
bilised on methadone (40mg/day) and weeks five and six at the
commencement of the dose taper. Hence the study provides some
insight into the immediate impact of methadone prescription on
risk behaviours.
Sees 2000 was de signed as an RCT comparing
methadone maintenance and 180-day methadone detoxification.
This review focuses on data for the maintenance group, but given
the long duration of the detoxification intervention, data from this
group is also relevant to consideration of the effect of substitution
treatment on HIV risk behaviour.
The setting for inter vention (general primary health care, specialist
drug and alcohol ser vices or specialist infectious diseases services)
is a factor that could potentially influence substitution treatment
outcomes relating to HIV prevention. In the majority of studies
(24/33) substitution treatment was pr ovided to participants in the
context of a specialist drug and alcohol treatment program (see
Table 1). In one study (Batki 1989) treatment was provided in the
context of a specialist AIDS program and in another study (
Dolan
2003) treatment was provided in a prison setting. In Gossop 2000
treatment was provided in both primary health care and specialist
drug and alcohol treatment settings with these two settings being
compared in one report derived from this study. In the remaining
five studies the treatment setting was eithe r not reported or partici-
pants were recruited from various sources making identification of
setting impossible. The lack of such data prevented consideration
of the effect of treatment setting through subgroup analysis.
To some extent the predominance of specialist drug and alcohol
settings may reflect the circumstances of the countries in which
the studies occurred. The majority of the studies (25/33) were
undertaken in the USA, where primary health care providers are
not able to prescribe methadone. (Of the remaining studies, three
were undertaken in the U K, three in Australia, one in Italy and
one in Germany).
Risk of bias in included studies
All of the included studies were assessed for the degree of risk of
bias and confounding, as described in the Methods section. With
this scale, the higher the overall score the greater the risk of bias and
confounding in the study. As indicated in
Table 2, only two studies
(
Abbott 1998; Avants 1998) were allocated an overall score of 0,
indicating a low level of risk of bias or confounding. An overall
score of 1 was allocated to 14 studies (
Camacho 1996; Chatham
1999
; Dolan 2003; Gossop 2000; Iguchi 1998; King 2000; Lott
2006
; Maddux 1997; Magura 1991; Marsch 2005; Metzger 1993;
Simpson 1995; Stark 1996; Strang 2000). Eleven studies were al-
located a score of 2 (
Baker 1995; Batki 1989; Finch 1995; Grella
1996
; Kwiatkowski 2001; Meandzija 1994; Schroeder 2006; Sees
2000; Serpelloni 1994; Teesson 2006, Thiede 2000), three stud-
ies received a score of 3 (
Britton 1994; Brooner 1998; Margolin
2003
) and three studies (Mark 2006, Moss 1994; Williams 1992)
received a score of 4, indicating a high risk of bias and confound-
8Substitution treatment of injecting opioid users for prevention of HIV infection (Review)
Copyright © 2008 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.
ing.
Table 2. Risk of confounding and bias
Study Characteristics Recruitment Confounders Follow-up Intervention Data coll ection Overall risk
Abbott 1998 0 0 0 0 0 0 0
Avants 1998 0 0 0 0 0 0 0
Baker 1995 1 0 1 0 0 0 2
Batki 1989 0 1 0 1 0 0 2
Britton 1994 1 1 1 0 0 0 3
Brooner 1998 1 1 1 0 0 0 3
Camacho 1996 0 0 0 1 0 0 1
Chatham 1999 0 0 0 1 0 0 1
Dolan 2003 0 0 0 1 0 0 1
Finch 1995 0 1 0 1 0 0 2
Gossop 2002 0 0 0 1 0 0 1
Grella 1996 0 1 0 0 0 1 2
Iguchi 1998 0 0 0 1 0 0 1
King 2000 1 0 0 0 0 0 1
Kwiatkowski 2001 0 0 1 1 0 0 2
Lott 2006 0 0 0 1 0 0 1
Maddux 1997 0 0 1 0 0 0 1
Magura 1991 0 0 0 1 0 0 1
Margolin 2003 1 0 0 1 0 1 3
Mark 2006 1 1 1 0 1 0 4
Marsch 2005 0 0 0 1 0 0 1
Meandzija 1994 1 0 1 0 0 0 2
9Substitution treatment of injecting opioid users for prevention of HIV infection (Review)
Copyright © 2008 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.
Metzger 1993 0 0 1 0 0 0 1
Moss 1994 1 1 1 0 1 0 4
Schroeder 2006 0 1 0 1 0 0 2
Sees 2000 0 0 1 1 0 0 2
Serpelloni 1994 1 0 0 0 1 0 2
Simpson 1995 0 0 0 1 0 0 1
Stark 1996 0 0 1 0 0 0 1
Strang 2000 0 0 0 1 0 0 1
Teesson 2006 0 0 1 1 0 0 2
Thiede 2000 0 0 1 1 0 0 2
Williams 1992 1 0 1 1 0 1 4
Total studies 9 6 12 15 2 3 -
The final row of
Table 2 indicates the number of studies identified
as being exposed to each of the six sources of bias or confounding.
As indicated in
Table 2, the most common source of risk is a high
rate of loss to follow-up, with actual or possible differences in the
baseline characteristics of study participants contacted at follow-
up and those who dropped out. The possibility of confounders
(such as diffe rences in gender balance, or levels of cocaine use
in groups being compared) or the identification of confounding
without the use of statistical adjustment is th e next most com-
mon source of risk. Inadequate reporting of baseline characteristics
(making it difficult to identify the possibility of bias or confound-
ing) and potential or actual bias in recruitment processes were also
significant sources of risk. However, actual or potential differences
in intervention were identified for only three studies (
Mark 2006;
Moss 1994; Serpelloni 1994) and actual or potential differences in
data collection in a further three studies (
Grella 1996; Margolin
2003; Williams 1992).
These ratings of risk of bias and confounding have been used to
enter a user-defined order in the data tables. This allows analyses
to be displayed in the order of lowest to highest risk of bias and
confounding.
Effects of interventions
The data tables included in this review are arranged firstly by th e
nature of the studies. Hence, the first three sets of comparisons
present follow-up data, meaning comparisons of HIV risk be-
haviour at baseline prior to entry into substitution treatment, and
at follow-up, a period of time after entry to substitution treatment.
Hence the data tables are set up to compare outcomes at follow-up
to baseline. The first set of comparisons cover drug use and drug-
related HIV risk behaviours; the second set of comparisons cover
sex-related HIV risk behaviours, and the third set of outcomes is
based on scores of overall risk behaviour, combining both drug-
related and sex-related risk behaviours. For reasons of statistical
validity, only dichotomous data is presented in these tables of fol-
low-up data. Associated continuous data is contained in Tables 06
to 12.
The fourth and fifth sets of comparisons present data from par-
ticipants receiving substitution treatment, with that from partici-
pants receiving no or limited substitution treatment (labelled “no
substitution or control”). The fourth set of comparisons covers
drug use and drug-related HIV risk behaviours, while the fifth set
of comparisons covers sex-related HIV risks.
In all these tables sub-categories have been created for the three
types of studies that met the inclusion criteria for this review,
namely, controlled studies, cohort studies, and descriptive studies.
However, most data were derived from descriptive studies with
the result that frequently the former two sub-categories do not
contain any data.
Combined totals have not been calculated for any of the analyses
included in this review as the studies varied in a number of aspects
making the validity of combined totals doubtful. Aspects that were
variable included:
the interval between baseline and follow-up interviews
10Substitution treatment of injecting opioid users for prevention of HIV infection (Review)
Copyright © 2008 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.
(seeTable 3) which ranged from one month to 18 to 24
months;
Table 3. Follow-up period
Study Interval
Abbott 1998 6 months
Avants 1996 6 months after completion of 12-week program
Baker 1995 No follow-up - single interview
Batki 1989 12 months - 3 months also available
Britton 1994 12 months
Brooner 1998 1 month
Camacho 1996 6 months - 3 months also available
Chatham 1999 12 months after leaving study
Dolan 2003 4 months
Finch 1995 2 months
Gossop 2000 1 year (primary data)
Grella 1996 18-24 months
Iguchi 1998 5-6 weeks (1-2 weeks also available)
King 2000 6 months
Kwiatkowski 2001 6 months
Lott 2006 18 weeks
Maddux 1997 12 months
Magura 1991 12 months
Margolin 2003 6 and/or 9 months
Mark 2006 No follow-up - single interview
Marsch 2005 24 weeks
Meandzija 1994
Metzger 1993
Moss 1994
Sees 2000
Schroeder 2006
Serpelloni 1994
Simpson 1995
Stark 1996
Strang 2000
Teesson 2006
Thiede 2000
Williams 1992
the proportion
baseline (see
T
number of studies
Table 4. Pr
Study
Abbott 1998
Avants 1998
Baker 1995
Batki 1989
Britton 1994
Brooner 1998
Camacho 1996
Chatham 1999
Dolan 2003
11Substitution treatment of injecting opioid users for prevention of HIV infection (Review)
Copyright © 2008 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.
Finch 1995 98% injecting drug users
Gossop 2000 62% injecting at intake
Grella 1996 Unclear
Iguchi 1998 Unclear
King 2000 86% injected in 30 days prior to intake
Kwiatkowski 2000 Injecting use in prior 30 days a requirement
Lott 2006 All daily heroin users at entry
Maddux 1997 Unclear
Magura 1991 All injecting users at entry
Margolin 2003 Injecting drug use selection criterion
Mark 2006 Injecting drug use in pr ior 6 months selection criterion
Marsch 2005 Injection preferred route for 65%
Meandzija 1994 Injecting drug users selected
Metzger 1993 87% injected in 6 months prior to baseline
Moss 1994 Injecting drug use selection criterion
Schroeder 2006 96%
Sees 2000 Unclear
Serpelloni 1994 Injecting drug users sel ected
Simpson 1995 Unclear
Stark 1996 Injecting drug use in prior 3 months required
Strang 2000 All injecting users
Teesson 2006 80% injecting daily
Thiede 2000 All injecting users
Williams 1992 Unclear (history of injecting use)
the reporting period for assessment of HIV risk behaviours
(see
Table 5) which r anged from two weeks to six months.
Table 5. T
Study
Abbott 1998
Avants 1998
Baker 1995
Batki 1989
Britton 1994
Brooner 1998
Camacho 1996
Chatham 1999
Dolan 2003
Finch 1995
Gossop 2000
Grella 1996
Iguchi 1998
King 2000
Kwiatkowski
Lott 2006
Maddux 1997
Magura 1991
Margolin 2003
Mark 2006
Marsch 2005
Meandzija 1994
Metzger 1993
12Substitution treatment of injecting opioid users for prevention of HIV infection (Review)
Copyright © 2008 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.
Moss 1994 Not applicable
Schroeder 2006 Prior week
Sees 2000 Prior 6 months
Serpelloni 1994 Not applicable
Simpson 1995 Prior 30 days
Stark 1996 Prior 6 months
Strang 2000 Prior 30 days
Teesson 2006 Prior month
Thiede 2000 1 month prior
Williams 1992 Not applicable
Table 6. Frequency of injecting use at baseline and follow-up
Study and group Group size Measure Baseline status Follow-up status Significance
Batki 1989 42 Days in past 30 27.5 6.3 significant
Brooner 1998,
referred from (1)
needle exchange (2)
standard sources
(1) 66 (2) 203 Days in prior 30 (1) 24.08 (2) 6.68 (1) 13.70 (2) 3.82 Tukey’s HSD (1)
0.01 (2) 0.01
Camacho 1996 326 Frequency in prior
30 days
111 59.5 6 19.3 P < 0.001
Chatham 1999 435 Frequency score for
prior 6 months
7.3 1.0 3.6 3.1 P < 0.001
Dolan 2003
(1) methadone
maintenance (2)
wait-list control
(1) 129 (2) 124 Times heroin
injected in prior
month
(1) 9 (2) 15 (1) 1.3 (2) 8.5 P < 0.001
Kwiatkowski 2001,
(1) methadone
maintenance & risk
reduction (2) risk
reduction only
(1) 99 (2) 216 Heroin injections
in prior month
(1) 77.1 39.9 (2)
60.1 37.4
(1) 22.9 35.9 (2)
36.3 44.5
P < 0.01
13Substitution treatment of injecting opioid users for prevention of HIV infection (Review)
Copyright © 2008 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.
Table 6. Frequency of injecting use at baseline and follow-up (Continued)
Lott 2006 (1) 35 baseline, 24 Injections of any 38.4 2.5 2.5 2.9 P < 0.01
Lott 2006 (2)
buprenorphine
30 baseline, 18
follow-up
Injections of any
drug per week in
prior month, mean
SEM
30.1 2.7 2.2 3.3 P < 0.01
Lott 2006 (3)
LAAM
23 baseline, 11
follow-up
Injections of any
drug per week in
prior month, mean
SEM
34.1 3.1 2.2 4.2 P < 0.01
Simpson 1995 521 Injections in last 30
days
108 57.6 9.5 47.4 P < 0.001
Strang 2000 15 baseline, 11
follow-up
Days in past 30 20.1 9.9 11.9 11.7 P < 0.001
An additional source of h eterogeneity for these studies was the
means of reporting the frequency of injecting use.
Camacho 1996,
Simpson 1995, Dolan 2003, Kwiatkowski 2001 and Lott 2006
reported the actual number of injections during the assessment
period.
Chatham 1999 translated the number of injections into a
frequency score, while
Strang 2000, Batki 1989 and Brooner 1998
reported the days of injecting use.
Despite th e variability, all studies show a statistically significant
decrease in injecting behaviour (either as the proportion of partic-
ipants injecting, or the frequency of injecting drug use, or both)
after entry into methadone treatment. The relative risk of injecting
drug use at follow-up compared to baseline ranged from 0.40 at
12 month follow-up for
Magura 1991 to 0.75 at 6 month follow-
up for
King 2000 (corresponding to reductions in relative risk of
60% and 25%, respectively).
Relatively few data were available on injecting behaviour amongst
cohorts of drug users in substitution treatment, compared to those
receiving no or limited substitution treatment. In
Metzger 1993
injecting drug users in or out of treatment were recruited for a
longitudinal study, with data on HIV risk behaviour collected at
baseline. At interview 27 out of 152 in methadone maintenance
treatment and six out of 103 not in treatment reported no inject-
ing drug use in the previous six months. To be consistent with data
reported from other studies, these data were translated into the
proportion injecting (see
Analysis 4.1), by assuming that all other
participants had injected in the prior six months. This shows a
statistically significant difference with those in substitution treat-
ment less likely to be injecting on entry to the study (relative r isk
(RR) 0.87, 95% confidence inter val (CI) 0.80 to 0.95).
Teesson 2006 recruited 745 individuals entering treatment for opi-
oid dependence (methadone or buprenorphine maintenance treat-
ment, detoxification, or residential rehabilitation) and 80 heroin
users not seeking treatment. Follow-up interviews were completed
with 728 at three months and 657 individuals at 1 year. At three
months 61% of the whole sample and 72% of the maintenance
group were in treatment. At 1 year, 59% were in treatment (79% of
the maintenance group, 49% of the detoxification group, 49% of
the residential rehabilitation group, and 47% of the no-treatment
group). At baseline 80% were injecting daily (78% of the main-
tenance group, 84% of the detoxification group, 75% of the resi-
dential rehabilitation group and 83% of the no-treatment group).
At the 3-month follow-up, 19% were injecting daily (9% mainte-
nance, 26% detoxification, 9% residential rehabilitation, 50% no
treatment). At 1-year follow-up 17% were injecting daily (7% of
the maintenance group, 23% of the detoxification group, 15% of
the residential rehabilitation group and 36% of the no-treatment
group). The differences between the groups in prevalence of daily
injecting at baseline, and variability in the extent of exposure to
treatment after recruitment into the study make these data diffi-
cult to interpret. Nonetheless the figures are suggestive of a greater
decrease in injecting behaviour with maintenance treatment rela-
tive to no treatment, and probably also other forms of treatment.
Dolan 2003 compared outcomes for opioid users in prison ran-
domly allocated to receive methadone maintenance treatment, or
a wait-list control, with access to methadone maintenance guar-
anteed after four months. This study provided data both on the
change in injecting drug use associated with a period of substitu-
tion treatment and a comparison between substitution treatment
and standard care not involving substitution treatment, albeit in
the somewhat unique setting of a prison. The proportion of the
methadone maintenance group reporting injecting drug use de-
creased from 64% at baseline to 34% at follow-up (
Analysis 1.1).
In contrast, there was no significant change in the proportion of
wait-list controls reporting injecting drug use (70% at baseline,
75% at follow-up). Similarly those in the meth adone maintenance
group reported a greater decrease in the number of heroin injec-
14Substitution treatment of injecting opioid users for prevention of HIV infection (Review)
Copyright © 2008 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.
tions in the month prior to follow-up, compared to baseline (Table
6
). As a result, at th ree months, significantly fewer par ticipants
in the methadone maintenance group reported injecting drug use
(
Analysis 4.1).
Two studies (
Baker 1995 and Meandzija 1994) provided data on
frequency of injecting use for cohorts of drug users receiving or
not receiving substitution treatment at the time of a single in-
terview.
Kwiatkowski 2001 compared HIV risk behaviour at in-
take and six-month follow-up in cohorts of injecting drug users
receiving risk reduction plus methadone maintenance treatment
or risk reduction treatment only. Hence the six-month follow-up
data represents a comparison of behaviours in cohorts receiving
or not receiving substitution treatment at th e time of interview.
Data from these three studies are presented together (see
Analysis
4.2
). Baker 1995 translated the number of injections reported by
participants into a frequency score, while
Meandzija 1994 and
Kwiatkowski 2001 reported the actual number of injections in the
prior 30 days. Data from these three studies are not therefore di-
rectly comparable, but all studies indicate a statistically significant
difference with participants in substitution treatment at the time
of interview injecting less frequently.
Four studies provided additional information on th e effect of sub-
stitution treatment on injecting drug use by comparing HIV risk
behaviour in cohorts of drug users continuing in or ceasing sub-
stitution treatment (
Britton 1994; Maddux 1997; Thiede 2000).
Due to the diversity of study design and outcomes assessed these
data have not been entered into the table of comparisons, but are
reviewed here narratively.
Maddux 1997 compared HIV risk behaviour at intake and 12-
months follow-up for cohorts of opioid users who remained in or
had left methadone maintenance treatment at the time of follow-
up. O verall the cohort retained in methadone treatment (n = 319)
received a mean 11.6 months of treatment between intake and
follow-up, while the cohort not in treatment at follow-up (n =
291) received a mean of 5.2 months of methadone treatment. At
intake the mean days of intravenous drug use in the month prior
to interview was similar for the two cohorts (29.4 and 29.3 days
respectively) but at follow-up the frequency of intravenous drug
use was higher for the cohor t that had left treatment (8.7 versus
3.7 days). Furthermore, the authors noted a significantly increased
level of incarceration at follow-up for the cohort off methadone
treatment, suggesting that this contributed to the reduction in
intravenous drug use in this cohort.
Thiede 2000 similarly compared HIV risk behaviour at intake and
12-month follow-up for cohorts of injecting drug users who had
left, disrupted or stayed in methadone treatment at the time of fol-
low-up. A t follow-up the proportion of participants who reported
injecting in th e prior month was significantly less for the cohort
of users retained in methadone treatment, compared to either of
the comparison cohorts. Participants retained in methadone treat-
ment, and those who disrupted treatment (left and returned) re-
ported a decreased frequency of injecting use at follow-up com-
pared to intake, whereas the frequency of injection was unchanged
from intake for the cohort who had left treatment at the time of
follow-up.
In
Britton 1994 two cohorts of injecting drug users were followed.
At baseline participants in both cohorts were receiving methadone
maintenance treatment. At follow-up, one year later, one cohort
remained in treatment while the other cohort had ceased treatment
as a result of a subsidised treatment no longer being available. The
data reported shows that for the cohort remaining in methadone
treatment the frequency of injecting use was similar at baseline
and follow-up, while the cohort who ceased methadone treatment
reported a significant increase in the days of her oin injecting at
follow-up. However, at baseline the cohort who would be ceasing
treatment reported more days of heroin injecting in the prior six
months. Wh ile the difference was not statistically significant, it
raises questions about the selection of participants to have funding
ceased and also the possibility of some exaggeration of levels of
injecting use reported, perhaps with the intent of reversing the
decision to cease funding.
Insummary, it is clear th at substitution treatment is associated with
a significant decrease in the proportion of participants reporting
injecting drug use, and in th e frequency of injection. However, the
diversity of study design and the means of assessing and reporting
on injecting drug use prevent calculation of an overall quantitative
estimate of the extent of th e decrease. Furthermore, the benefit
may not be sustained following ce ssation of methadone treatment,
particularly if cessation of treatment is not voluntary.
1.2 Sharing of injecting equipment
Substitution treatment provides an opportunity for delivery of
information about means of reducing th e transmission of HIV.
Hence substitution treatment may reduce risk behaviour, even
amongst those who continue to inject.
Data on the propor tion of participants reporting sharing of inject-
ing equipment, before and after a period of substitution treatment,
were able to be extracted from nine studies (
Analysis 1.2).
All nine studies showed a reduction in the proportion of partici-
pants reporting sharing of injecting equipment at follow-up com-
pared to baseline. The difference was statistically significant for
eight of the nine studies. The exception,
King 2000 was a relatively
small study and the 95% confidence interval was large preventing
the relative risk from achieving statistical significance. The relative
risk of sharing of injecting equipment at follow-up compared to
baseline ranged from 0.14 at three months for
Teesson 2006 to
0.78 at 6 months for
Margolin 2003 (corresponding to reductions
in relative risk of 86% and 12%, respectively).
In addition,
Brooner 1998 reported days of sharing in 30 days
prior to baseline and one-month follow-up interviews for cohorts
recruited from needle exchange or standard referral sources. For
the cohort recruited from needle exchange sources (n = 66) there
was a statistically significant reduction in the days of sharing, from
a mean of 4.63 at baseline to 1.93 at follow-up. However, for
the cohort recruited from standard referral sources (n = 203) the
15Substitution treatment of injecting opioid users for prevention of HIV infection (Review)
Copyright © 2008 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.
frequency of sharing was much lower at baseline (a mean of 0.69
days in the prior 30) and there was no significant difference at
follow-up (a mean of 0.72 days in the prior 30).
A further five studies provided data on the proportion of par-
ticipants reporting sharing of injecting equipment for cohorts of
drug users in substitution treatment compared to those receiving
no or limited substitution treatment (
Analysis 4.3). In three of
these studies (
Dolan 2003; Metzger 1993; Stark 1996) the relative
risk of sharing of injecting equipment was significantly less in the
cohort receiving substitution treatment. In the other two studies
(
Mark 2006; Thiede 2000) the relative risk was less for the cohort
in substitution treatment, but the difference did not achieve sta-
tistical significance.
Kwiatkowski 2001 compared HIV risk behaviour at intake and
six-month follow-up in cohorts of injecting drug users receiving
risk reduction plus methadone maintenance treatment or risk re-
duction treatment only. Participants in both groups reported a
reduced frequency in use of dirty needles and in sharing of para-
phernalia in the 30 days prior to follow-up compared with intake,
with no significant difference between the cohorts in the extent of
reduction.
Two studies provided additional information on the effect of sub-
stitution treatment on the sharing of injecting e quipment by com-
paring HIV risk behaviour in cohorts of drug users continuing in
or ceasing substitution treatment (
Britton 1994; Maddux 1997).
Maddux 1997 compared HIV risk behaviour at intake and 12-
month follow-up for cohorts of opioid users who remained in or
had left methadone maintenance treatment at the time of follow-
up. Both cohorts reported reductions in days with use of an un-
cleaned needle in the prior month (from a mean of 5.9 to 0.7 for
the cohor t continuing in MMT and from 7.2 to 2.4 for the cohort
ceasing MMT) and in days in which they gave an uncleaned nee-
dle in the prior month (from a mean of 6.4 to 0.8 for the cohort
remaining in MMT and from 8.1 to 2.8 for the cohort ceasing
MMT).
In
Britton 1994 two cohorts of injecting drug users were followed.
At baseline participants in both cohorts were receiving methadone
maintenance treatment. At follow-up, one year later, one cohort
remained in treatment while the other cohort had ceased treatment
as a result of subsidised treatment no longer being available. The
data reported shows that for the cohort remaining in treatment
there was a reduction in the days of needle sharing in the six months
prior to interview (from 13.51 28.56 at baseline to 0.51 1.22
at follow-up). In contrast, for the cohort who ceased methadone
treatment, the days of needle sharing in the 6 months prior to
interview increased from 24.93 32.18 at baseline to 39.07 77.46
at follow-up.
These data suggest th at substitution treatment is usually associated
with a significant decrease in the sharing of injecting equipment.
The reduction in sharing behaviour may be due to a reduction in
injecting behaviour but it is not possible to determine th is from
the data available. It appears that similar reductions in the sharing
of injecting equipment can be achieved by forms of treatment
other than substitution, and reductions may be sustained even
after cessation of substitution treatment.
1.3 Scores of dru g -related risk
Rating scales are available for the assessment of HIV-risk be-
haviours. These include the Risk Assessment Battery (used by
Abbott 1998 and Avants 1998) and the HIV Risk-Taking Be-
haviour Scale (used by
Baker 1995). These instruments typically
provide separate r atings of drug-related and sex-related risk as well
as an overall score. This section considers only drug-related risk.
Data on drug-related risk scores of participants, before and after
a period of substitution treatment, were able to be extracted from
five studies (
Table 7).
Table 7. Drug risk scores at baseline and follow-up
Study and group Group size Measure Baseline status Follow-up status Significance
Abbott 1998
(1) methadone-
free at intake
(2) transferring
from methadone
maintenance
(1) 70 (2) 78 Risk Assessment
Battery, drug risk
score
(1) 9.35 7.49 (2)
4.82 5.14
(1) 2.80 4.21 (2)
2.09 3.90
P < 0.001
Avants 1998 307 Risk Assessment
Battery
2.14 3.06 1.25 1.83 P < 0.001
Chatham 1999 435 Risky Needle
Exposure Score for
5.3 1.6 P < 0.001
16Substitution treatment of injecting opioid users for prevention of HIV infection (Review)
Copyright © 2008 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.
Table 7. Drug risk scores at baseline and follow-up (Continued)
prior 6 months
Marsch 2005 Not reported HIV Risk Behavior
Score (mean SD) for
prior month
6.3 6.1 1.4 3.0 P < 0.001
Sees 2000 (1) MMT
(2) 180-day detox
(1) 91 (2) 88 Injection risk score
prior 6 months
(1) 6.51 6.69 (2) 6.0
6.44
(1) 3.04 4.35 (2)
4.07 5.79
(1) & (2) not
significant
Abbott 1998 and Avants 1998 used slightly different variants of
the Risk A ssessment Battery.
Sees 2000 used the Risk of AIDS Be-
haviour Scale.
Chatham 1999 reported a Risky Needle Exposure
Index for opioid users reflecting the number of persons with whom
injecting equipment had been shared prior to and 12 months after
a period of methadone maintenance treatment. In four of the five
studies there was a statistically significant decrease in drug-related
risk score from baseline to follow-up. The exception was Sees 2000
where the re was a reduction in the mean score for six months prior
to follow-up, compared to intake for both th e methadone main-
tenance and 180-day methadone detoxification groups, but the
associated standard de viations were large preventing the reduction
from achieving statistical significance.
Baker 1995 compared drug r isk scores for cohorts of injecting
drug users currently or previously in, or with no prior history of
methadone maintenance treatment. Data from the latter two co-
horts were combined and entered into analyses as a comparison
of current substitution and no substitution treatment (
Analysis
4.6
). Mark 2006 also provided a comparison of drug risk scores
for users currently in substitution treatment and users not in sub-
stitution treatment. In both studies the mean score is significantly
lower for the cohort receiving substitution treatment at the time
of interview.
These data of drug-related HIV risk scores are consistent with data
on prevalence and frequency of injecting and sharing of injecting
equipment in showing a significant reduction in risk associated
with substitution treatment.
2. Effect of substitution treatment on drug use
It is of interest to know whether the reduction in injecting be-
haviour associated with substitution treatment reflects a reduction
in overall levels of drug use, or a change to non-injecting routes
of administration, or both. This review focuses on substitution
treatment for opioid dependence. Hence it would be expected
that substitution treatment would primarily affect levels of use of
opioid dr ugs. However, other drugs may also be injected. For the
studies included in this review the non-opioid drug most likely to
be injected was powder cocaine, either alone or in combination
with heroin (speedballs). The refore consideration is also given in
this section to changes in cocaine use associated with substitution
treatment.
Data on the proportion of participants reporting illicit opioid use,
or the proportion of positive urine samples, before and after a
period of substitution treatment, were able to be extracted from
eight studies (
Analysis 1.3). A different set of eight studies provided
data on the frequency of opioid use, before and after a period of
substitution treatment (
Table 8).
Table 8. Frequency of opioid use at baseline and follow-up
Study and group Group size Measure Baseline status Follow-up status Significance
Batki 1989 42 Mean days in prior
30
27.5 3.8 sig
Brooner 1998,
referral from (1)
needle exchange (2)
standard sources
(1) 66 (2) 203 Days in prior 30 (1) 28.59 (2) 7.06 (1) 16.46 (2) 4.43 Tukey’s HSD (1)
0.01 (2) 0.01.
Chatham 1999 435 Frequency score for
past 6 months
7.27 3.37 P < 0.001
17Substitution treatment of injecting opioid users for prevention of HIV infection (Review)
Copyright © 2008 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.
Table 8. Frequency of opioid use at baseline and follow-up (Continued)
Gossop 2000, (1) 333 (2) 145 Days use in prior 90 (1) 57.4 36.7 (2) (1) 24.0 30.4 (2) P < 0.0001
Iguchi 1998 51 Use episodes per
week
13.8 9.8 2.5 2.8 Not reported
Lott 2006 (1)
methadone
35 baseline, 24
follow-up
Opioid injections
per week in prior
month, mean SEM
22.1 1.4 0.8 1.6 P < 0.01
Lott 2006 (2)
buprenorphine
30 baseline, 18
follow-up
Opioid injections
per week in prior
month, mean SEM
16.8 1.5 1.0 1.8 P < 0.01
Lott 2006 (3)
LAAM
23 baseline, 11
follow-up
Opioid injections
per week in prior
month, mean SEM
19.4 1.7 0.7 2.3 P < 0.01
Simpson 1995 521 Frequency score for
last 30 days
6.65 2.0 1.03 1.6 P < 0.001
Strang 2000 15 baseline, 10
follow-up
Days use in past 30 22.4 9.8 8.7 10.8 P < 0.001
Teesson 2006 227 Mean days in last
month
19.1 2.9 Not reported
18Substitution treatment of injecting opioid users for prevention of HIV infection (Review)
Copyright © 2008 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.
In addition to length of follow-up (seeTable 3) and reporting pe-
riod (see
Table 5) these studies al so varied in the nature of outcome
assessed and the means of reporting.
Abbott 1998 and Avants 1998
reported the proportion of opiate-free urine. These data were con-
verted to the proportion of opioid positive urine (assuming that
all other samples were positive) to be consistent with other data on
drug use.
Chatham 1999 reported the proportion of opioid posi-
tive urine.
Simpson 1995 and Grella 1996 reported the proportion
of participants self-reporting heroin use.
Dolan 2003 reported the
proportion of participants who either reported heroin use or their
hair sample tested positive for morphine. It should be noted that
the proportion of urine samples positive for opioids is not nec-
essarily the same as the proportion of participants using opioid
drugs, al though we have assumed that urine samples provide a
reasonable indication of this.
Simpson 1995 reported the propor-
tion of opioid positive urine samples as well as the proportion of
participants self-reporting heroin use. We have preferred the self-
report data for entry into the data tables because it is clearly related
to par ticipants and not to urine samples. (The results of urine
screening were consistent with the se lf-report data.) In terms of
means of reporting,
Gossop 2000 and Strang 2000 reported days
of use while
Iguchi 1998 reported the number of use episodes.
Despite the variability in studies, all showed statistically signifi-
cant decreases in opioid use (either as the proportion of opioid
positive urine samples or the proportion of participants or the f re-
quency of opioid use) from baseline prior to entry into methadone
treatment and follow-up between five to six weeks and 12 months
after entry. The relative risk of illicit opioid use at follow-up (six
month) compared to baseline ranged from 0.36 for
Abbott 1998
and Margolin 2003 to 0.60 for Avants 1998 (corresponding to
reductions in relative risk of 64% and 40%, respectively).
Baker 1995 reported data on the frequency of opioid use for co-
horts of drug users currently or previously in methadone main-
tenance treatment, or with no history of methadone mainte-
nance treatment. Data from the latter two cohorts were combined
and entered into analyses as a comparison of participants receiv-
ing or not receiving substitution treatment at the time of inter-
view (see
Analysis 4.4). Kwiatkowski 2001 reported the number
of h eroin injections by participants for 30 days prior to baseline
and follow-up interviews for cohorts receiving risk reduction plus
MMT or risk reduction only. Data from the six-month follow-
up has been taken as a comparison of current methadone mainte-
nance treatment with no methadone maintenance treatment and
included in
Analysis 4.4. For both these studies heroin use was sig-
nificantly less frequent for participants in substitution treatment.
Data on the proportion of participants reporting cocaine use, or
the proportion of cocaine-positive urine samples, were able to be
extracted from four studies (
Analysis 1.4). Five studies reported
data on the frequency of powder cocaine use before and after a
period of substitution treatment (
Table 9) and two studies reported
data on the frequency of use of speedball (
Table 10).
Table 9. Frequency of powder cocaine use at baseline and follow-up
Study and group Group size Measure Baseline status Follow-up status Significance
Batki 1989 42 Days in past 30 8.1 2.5 sig
Brooner 1998,
referral from (1)
needle exchange (2)
standard sources
(1) 66 (2) 203 Days in prior 30 (1) 15.29 (2) 5.97 (1) 4.84 (2) 3.02 Tukey’s HSD (1)
0.01 (2) 0.05
Chatham 1999 435 Frequency score for
last 6 months
3.13 2.15 P < 0.001
Gossop 2000,
(1) methadone
maintenance
(2) Methadone
reduction
(1) 333 (2) 145 Days use in prior 90 (1) 1.8 9.4 (2) 1.6
4.8
(1) 0.8 6.1 (2) 1.7
10.3
P < 0.001
Simpson 1995 521 Frequency score for
last 30 days
1.75 2.4 0.75 1.5 P < 0.001
19Substitution treatment of injecting opioid users for prevention of HIV infection (Review)
Copyright © 2008 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.
Table 10. Frequency of use of opiates and cocaine mixed (speedball), baseline & follow-up
Study and group Group s ize Measure Baseline status Follow-up status Significance
Chatham 1999 435 Frequency score in last 6 months 2.47 1.09 P < 0.001
Simpson 1995 521 Frequency score in last 30 days 2.62 3.1 0.58 1.2 P < 0.001
The data in
Table 9 and Table 10 indicate that in all studies a
statistically significant reduction in the frequency of use of cocaine
(as powder cocaine or combined with opiates as speedball) was
associated with methadone maintenance treatment. In the case
of
Gossop 2000, a significant effect of time was reported, but
it can be seen that reductions occurred only for the cohort allo-
cated to methadone maintenance treatment, and not to the metha-
done reduction cohor t. The data in
Analysis 1.4 are more variable.
Of the four studies providing data,
Margolin 2003 and Simpson
1995
reported a reduction in cocaine use at follow-up compared
to baseline that is clearly statistically significant.
Chatham 1999,
which like
Simpson 1995 was an analysis of data from the DATAR
study, reported a reduction tending towards significance. Data
from
Grella 1996 suggest no change in cocaine use associated with
substitution treatment for opioid dependence.
Only one study (
Baker 1995) reported data on the frequency of
cocaine use for cohorts of drug users receiving or not receiving
substitution treatment at the time of interview (see
Analysis 4.5).
The frequency score was lower for the cohort in substitution treat-
ment, but the difference was not statistically significant.
On the basis of these data it is clear that substitution treatment
is associated with a significant decrease in the proportion of par-
ticipants using illicit opioids and in the frequency of use. As with
data on injecting behaviour, the diversity of study design and the
means of assessing and reporting on opioid use prevent calculation
of an overall quantitative estimate of the extent of the decrease.
Furthermore, the studies providing data on opioid use were largely
distinct from the studies providing data on injecting behaviour. As
a result, although it is possible to conclude th at both opioid use
and injecting use decrease with substitution treatment, it is not
possible to relate the decrease in injecting use to opioid use.
The data on cocaine use is less consistent. Most studies found a
significant reduction in cocaine use associated with substitution
treatment for opioid dependence, but some studies found no sig-
nificant difference.
3. Effect of substitution treatment on sexual behaviour
Drug use is linked to sexual beh aviours with a high risk of HIV
transmission in several ways. The most obvious is through com-
mercial sex work to finance drug purchases, or the exchange of
sexual favours for drugs. Drug users frequently form relationships
with other drug users with the associated risk of sexual transmis-
sion of HIV. Intoxication is a barrier to negotiation of the use of
protection, as well as breaking down inhibitions further reducing
the likelihood of protected sex.
Substitution treatment, by reducing drug use, can reduce the need
for exchange of sex for drugs or money. The administration of
regular doses of an opioid drug as occurs in substitution treatment
reduces libido which may also reduce the number of sex partners.
Furthermore, substitution treatment provides the opportunity for
interventions including strategies for reducing HIV risk and ne-
gotiating the use of protection during sex.
3.1 Multiple sex partners or commercial sex work
Commercial sex work is generally assumed to be associated with
higher risk of HIV transmission, and multiple sex partners is prob-
ably indicative of commercial sex work. Hence these two outcomes
are considered together.
Camacho 1996 and Chatham 1999 reported data only on the
proportion of participants reporting multiple sex partners before
and after a period of substitution treatment (
Analysis 2.1). King
2000
and Grella 1996 reported both the proportion reporting
multiple sex partners and the proportion reporting exchanges of
sex for drugs or money (
Analysis 2.3).
Significantly fewer participants in
Camacho 1996, Chatham 1999
and Grella 1996 reported multiple sex partners or exchanges of sex
for drugs or money following substitution treatment, compared
to baseline prior to substitution treatment. Relatively few partici-
pants in
King 2000 reported these behaviours either before or after
substitution treatment resulting in a very wide confidence interval
encompassing a relative risk of 1 (i.e. no change).
Schroeder 2006
also had a wide confidence interval encompassing no significant
change.
In
Metzger 1993 injecting drug users in or out of treatment were
recruited for a longitudinal study, with data on HIV risk behaviour
collected at baseline. Significantly less participants in substitution
treatment reported exchanges of sex for drugs or money in the six
months prior to study entry (
Analysis 5.5). Mark 2006 compared
the HIV risk behaviours of injecting drug users recruited from
a needl e exchange pr ogram, methadone maintenance treatment,
or a detoxification program. Data from the needle exchange and
detoxification programs have been combined and compared with
data for the methadone maintenance group (
Analysis 5.5). This
shows significantly less participants in methadone maintenance
treatment reported exchanges of sex for drugs or money in the six
months prior to interview.
Meandzija 1994 reported the mean number of sex partners in
the 30 days prior to interview for cohorts of injecting drug users
currently in methadone maintenance treatment or not in treat-
ment (
Analysis 5.2). The cohort in methadone maintenance treat-
20Substitution treatment of injecting opioid users for prevention of HIV infection (Review)
Copyright © 2008 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.