The effect of baseline cocaine use on treatment outcomes for heroin dependence over 24 months:...
-
Upload
anna-williamson -
Category
Documents
-
view
212 -
download
0
Transcript of The effect of baseline cocaine use on treatment outcomes for heroin dependence over 24 months:...
Journal of Substance Abuse Tre
Regular article
The effect of baseline cocaine use on treatment outcomes
for heroin dependence over 24 months:
Findings from the Australian Treatment Outcome Study
Anna Williamson, (B.Psych. (Hons), Ph.D.)4, Shane Darke, (B.A. (Hons), Ph.D.),
Joanne Ross, (R.N., B.Sc. (Hons), Ph.D.), Maree Teesson, (B.A. (Hons), Ph.D.)
National Drug and Alcohol Research Centre, University of New South Wales, New South Wales 2052, Australia
Received 22 August 2006; received in revised form 19 December 2006; accepted 25 December 2006
Abstract
Aims: The aim of this study was to determine the effects of baseline cocaine use on treatment outcomes for heroin dependence over a
24-month period. Design: A longitudinal cohort (24 months) study was carried out. Interviews were conducted at baseline, 3, 12, and
24 months. Setting: The study setting was Sydney, Australia. Participants: Six hundred fifteen heroin users were recruited for the Australian
Treatment Outcome Study. Findings: Cocaine use was common at baseline (40%) but decreased significantly over the study period. Even
after taking into account age, sex, treatment variables, current heroin use, and baseline polydrug use, baseline cocaine use remained a
significant predictor of poorer outcomes across a range of areas. Baseline cocaine users were more likely to report heroin use, unemployment,
needle sharing, criminal activity, and incarceration over the 24-month study period. Conclusions: Cocaine consumption among heroin users
has repercussions across a range of areas that persist far beyond the actual period of use. Consequently, treatment providers should
regard cocaine use among clients as an important marker for individuals who are at risk of poorer treatment outcome. D 2007 Elsevier Inc.
All rights reserved.
Keywords: Heroin; Cocaine; Treatment; Cohort
1. Introduction
Cocaine use among heroin dependent individuals has
long been a major problem in the United States (Chambers,
Taylor, & Moffett, 1972; Platt, 1997) and, more recently, the
United Kingdom (Beswick et al., 2001). In Australia,
cocaine use among heroin-dependent individuals has tradi-
tionally occurred at low levels due to its prohibitive cost and
poor availability (Hando, Flaherty, & Rutter, 1997). In 2001,
however, heroin became substantially more expensive and
harder to obtain than usual, whereas the opposite was true of
cocaine (Darke, Kaye, & Topp, 2002). This led to a large-
0740-5472/07/$ – see front matter D 2007 Elsevier Inc. All rights reserved.
doi:10.1016/j.jsat.2006.12.009
4 Corresponding author. Dr. Anna Williamson, The Sax Institute, PO
Box 123 Broadway, New South Wales 2007, Australia. Tel.: +61 2 9514
5971; fax:+ 61 2 9514 5951.
E-mail address: [email protected] (A. Williamson).
scale increase in heroin and cocaine co-use, with unknown
implications for the efficacy of existing treatments for
heroin dependence.
The majority of research examining the efficacy of
treatment for concurrent heroin and cocaine use has centered
on methadone maintenance. Although cocaine use at treat-
ment entry has been associated with poorer outcomes
(Downey, Helmus, & Schuster, 2000), some studies have
found methadone maintenance leads to reductions in
cocaine use (Kosten, Rousanville, & Kleber, 1988; Magura,
Siddiqi, Freeman, & Lipton, 1991; Shaffer & La Salvia,
1992), whereas others indicate that significant proportions
of people maintain their pretreatment levels of use (Hartel
et al., 1995) or even commence or increase cocaine use
while on methadone (Kosten et al., 1988; Magura et al.,
1991; Van Beek, Dwyer, & Malcom, 2001). A small number
of studies have also examined the effects of concurrent
atment 33 (2007) 287–293
A. Williamson et al. / Journal of Substance Abuse Treatment 33 (2007) 287–293288
cocaine and heroin use on detoxification and residential
rehabilitation outcomes (Hubbard, Craddock, & Anderson,
2003), but there is a clear dearth of research in this area.
Different patterns of cocaine use have been noted in
different areas. In Australia, where cocaine powder domi-
nates, injection is the primary route of administration (Darke
et al., 2002), whereas in America and the United Kingdom,
where crack cocaine accounts for much of the cocaine-
related harm, smoking cocaine is common (Gossop,
Marsden, Stewart, & Kidd, 2002; Platt, 1997). To date, no
research into the effect of cocaine use on treatment outcome
for heroin dependence has been conducted in Australia.
The Australian Treatment Outcome Study (ATOS), the
first large-scale longitudinal study of treatment entrants for
heroin dependence in Australia, provided an opportunity
to examine the effects of concurrent cocaine use on
treatment outcomes for heroin dependence. Baseline
analyses showed that cocaine was being used by a large
proportion of treatment entrants and that these individuals
displayed particularly poor clinical profiles (Williamson,
Darke, Ross, & Teesson, 2006). Indeed, despite not
differing in drug use history or psychopathology, heroin
users who also used cocaine exhibited a greater degree of
drug-related harm, being more likely to report home-
lessness, criminality, needle sharing, and injection-related
health problems.
The current study examined the effects of baseline
cocaine use on treatment outcomes for heroin dependence
in all the major treatment modalities over a 24-month period.
In order to utilize all data collected and control for possible
confounders, generalized estimating equation (GEE) model-
ing was employed (Liang & Zeger, 1986). One of the major
advantages of GEE modeling is that it allows the inclusion
of subjects with incomplete data (Twisk, 2003).
The current study thus aimed to examine the effects of
baseline cocaine use on treatment outcomes for heroin
dependence over a 24-month period. The specific aim of the
study was to examine the relationship between baseline
cocaine use and outcome while taking into account the
effects of age, sex, baseline polydrug use and heroin use,
and treatment exposure at each time point.
2. Materials and methods
2.1. Procedure
The data were collected from the New South Wales
(NSW) component of ATOS. Data from the Victorian and
South Australian arms of ATOS do not form part of this
article, as the cohorts from these states were not followed
up at 24 months. Baseline interviews were conducted in
NSW between February 2001 and August 2002. ATOS is a
24-month longitudinal study of entrants to treatment for
heroin dependence and a comparison group of nontreat-
ment heroin users. Subjects were recruited from 19
agencies treating heroin dependence in the greater Sydney
region, randomly selected from within treatment modality
and stratified by regional health area. The agencies
comprised 10 outpatient methadone/buprenorphine main-
tenance agencies (MT), four drug-free residential rehabil-
itation agencies (RR), and nine detoxification facilities
(DTX). In addition, a comparison group of heroin users
not currently in treatment were recruited from needle and
syringe exchange programs.
MT programs in Sydney are outpatient and run for an
indefinite period provided that rules are adhered to.
Expulsion as a consequence of drug use during treatment
is rare. The DTX programs examined in this study were
primarily inpatient programs lasting 5–7 days. These
programs aim to address all drug use. A reducing schedule
of buprenorphine was sometimes used in order to facilitate
opiate detoxification. The RR programs in this study target
all drug use. Participants were recruited from two short
(28-day) and two long (3–12 months) inpatient programs.
Drug use while enrolled in an inpatient DTX or RR is
grounds for expulsion.
Participants were interviewed at baseline, 3, 12, and
24 months. Eligibility criteria were the following: (i) no
treatment for heroin dependence in the preceding month, (ii)
no imprisonment in the preceding month, (iii) agreed to give
contact details for follow-up interviews, (iv) had a good
understanding of English, and (v) were 18 years or older.
The total baseline sample consisted of 615 heroin users: MT
(n = 201), DTX (n = 201), RR (n = 133), NT (n = 80). All
subjects were reimbursed A$20 for travel expenses on
completion of each interview.
2.2. Structured interview
At baseline, participants were administered a struc-
tured interview that addressed demographics, treatment
history, drug use history, heroin overdose history, suicide
history, prison history, and a range of psychopathology.
Drug use, needle risk taking, injection-related health
problems, and criminal behaviors over the month
preceding interview were measured with the Opiate
Treatment Index (Darke, Hall, Wodak, Heather, & Ward,
1992). General physical and mental health were measured
with the Short-Form 12 (Gandek et al., 1998; Ware,
Kolinski, & Keller, 1996). A Diagnostic and Statistical
Manual of Mental Disorders, Fourth Edition diagnosis of
past month Major Depression was obtained by using the
Composite International Diagnostic Interview (Andrews,
Hall, Teesson, & Henderson, 1999).
Follow-up interviews incorporated all of the elements
of the baseline interview outlined above. In addition,
participants were asked how many times they had
commenced treatment, in any modality, for heroin depend-
ence since the most recent interview and the time spent in
each treatment episode. At 12 and 24 months, participants
were also asked if they had used heroin, overdosed on
A. Williamson et al. / Journal of Substance Abuse Treatment 33 (2007) 287–293 289
heroin, attempted suicide, or been incarcerated in the
preceding 12 months.
2.3. Statistical analyses
McNemar chi-square tests were used to measure
changes in prevalence across time. A logistic regression
was conducted to determine treatment status at baseline or
cocaine use status at 24 months. In order to examine the
effects of baseline cocaine use over the 24-month study
period taking into account all data collected, GEE
modeling (Liang & Zeger, 1986) was employed. The
GEE method estimates regression coefficients and their
standard errors taking into account the correlation between
subjects’ scores on the outcome measure in question across
the study period (Twisk, 2003). This method yields valid
and robust estimates of variance, even when there is a
known positive correlation between multiple outcome
measures within subjects (Liang & Zeger, 1986). GEE
modeling analyzes the relationships between the variables
of a model at different time points simultaneously (Twisk,
2003). This is an iterative process, using quasi-likelihood
to estimate regression coefficients (Liang & Zeger, 1986).
One of the major advantages of GEE is that it allows the
inclusion of subjects with incomplete data (Twisk, 2003).
GEE offers important advantages over other regression
approaches used to measure change across time (Twisk,
2004). It permits simultaneous modeling of the relation of
specific factors (e.g., baseline cocaine use) with perform-
ance on outcome measures at baseline, 3, 12, and
24 months.
Odds ratios (ORs) and 95% confidence intervals (CIs)
are reported for all GEE analyses. Each model consisted of
the fixed covariates age, sex, baseline cocaine use (yes/no),
and the time-varying covariates past month heroin use
(yes/no), proportion of time spent in treatment since last
Table 1
Prevalence of past-month and 12-month outcomes at baseline, 3, 12, and 24 mon
Variable Baseline (n = 615)
Heroin use (past month), n (%) 584 (95)
(OR, 95% CI)a
Needle sharing (past month), n (%) 208 (34)
(OR, 95% CI)
Unemployment (past month), n (%) 108 (18)
(OR, 95% CI)
Criminal activity (past month), n (%) 336 (55)
(OR, 95% CI)
Prison (12 months), n (%) 102 (17)
(OR, 95% CI)
Severe physical disability (past month), n (%) 53 (9)
(OR, 95% CI)
Heroin overdose (12 months), n (%) 155 (25)
(OR, 95% CI)
Severe psychological disability (past month), n (%) 303 (49)
(OR, 95% CI)
ns = not significant.a Odds of use at previous interview point compared to current time point (G
interview, and number of treatment episodes commenced
since last interview. Treatment is not analyzed in terms of
modality entered at baseline because of the substantial
movement of the cohort both within and between treatment
modalities over the 2-year period. The unstructured corre-
lation option was used, as it has been demonstrated to be the
most accurate working correlation for estimating GEE
models (Twisk, 2004). The results presented here are based
on all the available data, including cases in which
information was not obtained at all follow-up points (n =
615). All analyses were conducted using SAS version 8.02
(SAS Institute Inc., 1999).
3. Results
3.1. Cohort characteristics
The mean age of subjects was 29.3 years (SD = 7.8,
range = 18–56), and 66% were men. Men were significantly
older than women (30.0 vs. 27.8 years), t(613) = �3.34,p b .01. The sample had completed a mean number of 10.0
years (SD = 1.7, range = 2–12) school education. Twenty-
nine percent had completed a trade/technical course, 6% a
university degree, and 65% had no tertiary qualifications.
The three most commonly reported primary sources of
income for the preceding month were a government
allowance (46%), criminal activity (24%), and employment
(18%). A prison history was reported by 41% of the sample,
with 17% having been incarcerated in the 12 months
preceding interview. The mean age at first heroin use was
19.7 years (SD = 5.3, range = 9–43) and the average length
of heroin use career at baseline was 9.6 years (SD = 7.4,
range = 0–35). Follow-up rates at the three time points were
3 months (89%, n = 549), 12 months (80%, n = 495), and
24 months (76%, n = 469).
ths
3 Months (n = 549) 12 Months (n = 495) 24 Months (n = 469)
198 (36) 139 (28) 108 (23)
(4.43, 3.72–5.15) (1.55, 1.25–1.90) ns
59 (11) 43 (9) 37 (8)
(2.36, 1.72–3.22) ns ns
91 (17) 114 (23) 159 (34)
ns (1.58, 1.22–2.05) (1.60, 1.28–1.99)
147 (27) 120 (24) 79 (17)
(2.51, 1.93–3.25) ns ns
58 (12) 71 (15)
Not applicable ns ns
31 (6) 29 (6) 24 (5)
(1.67, 1.07–2.59) ns ns
61 (12) 41 (9)
Not applicable (2.92, 2.20–3.82) ns
134 (24) 110 (22) 88 (19)
(2.10, 1.63–2.69) ns ns
EE).
Table 2
Prevalence of past month and 12-month outcomes at baseline, 3, 12, and 24 months among CUs and NCUs
Variable
Baseline 3 Months 12 Months 24 MonthsGEE analyses:
CU vs. NCU,
ORa (95% CI)
CUb
(n = 246)
NCUc
(n = 369)
CU
(n = 211)
NCU
(n = 337)
CU
(n = 196)
NCU
(n = 299)
CU
(n = 183)
NCU
(n = 286)
Heroin use 99 99 80 43 49 37 39 32 1.51 (1.17–1.95)
Needle sharing 40 30 17 7 14 5 13 5 1.67 (1.13–2.23)
Employment 12 21 12 20 22 23 31 36 0.68 (0.50–0.92)
Criminal activity 65 48 35 22 23 17 21 14 1.63 (1.26–2.10)
Prison (12 months) 22 13 N/A N/A 17 8 18 13 1.73 (1.31–2.29)
Severe physical disability 7 10 6 5 6 6 6 5 ns
Heroin overdose (12 months) 28 24 N/A N/A 14 8 8 8 ns
Severe psychological disability 48 50 27 23 24 22 20 18 ns
N/A = not applicable; ns = not significant.a Odds of use at previous interview point compared to current time point (GEE).b Cocaine user at baseline.c Non-cocaine user at baseline.
A. Williamson et al. / Journal of Substance Abuse Treatment 33 (2007) 287–293290
3.2. Treatment exposure over the 24-month follow-up period
At 24-month interview, 99% of participants reported
having received treatment for heroin dependence at some
time over the study period. The cohort had spent a median
of 41% of the time since baseline enrolled in treatment for
their heroin dependence and had commenced a median of
three treatment episodes. Approximately half (54%) of those
followed up at 24 months were enrolled in treatment at the
time of interview.
3.3. Prevalence of cocaine use
At baseline, 39% of the cohort reported cocaine use in
the month prior to interview. For analytic purposes, these
subjects (n = 246) were classified as cocaine users (CUs),
whereas those who had not used cocaine in the month prior
to baseline (n = 369) were classified as non-cocaine users
(NCUs). The prevalence of past month cocaine use had
decreased significantly among the cohort at 3 months, v2(1) =
53.57, p b .001, and declined again to 10% at 12 months,
v2(1) = 70.88, p b .001. At 24 months, approximately 7% of
the cohort reported cocaine use in the month prior to
interview, a proportion not significantly different from that
noted at 12 months, v2(1) = 85.32, p = .07.
In order to determine the effect of treatment status at
baseline on cocaine use status at 24 months a logistic
regression was conducted. Factors entered into the equation
were age, sex, cocaine use status at baseline, and treatment
status at baseline (treatment/nontreatment). The final model
was significant, v2(4) = 29.11, p b .001, and a good fit:
Hosmer–Lemeshow v2(8) = 4.80, p = .78. Baseline cocaine
use was the strongest predictor of cocaine use at 24 months
(14% vs. 3%, OR = 4.92, 95% CI = 2.14–11.33). Those who
were not enrolled in or seeking treatment at baseline were
also more likely to have used cocaine at 24 months (19% vs.
6%, OR = 0.40, 95% CI = 0.17–0.93). There were no gender
differences in the prevalence of last month cocaine use at
any follow-up point.
3.4. Cohort changes over time
GEE analyses were used to examine changes in the
prevalence of key outcome measures among the cohort over
24 months. At 3 months, significant declines were noted in
the prevalence of past month heroin use, needle sharing,
criminal activity, severe psychological disability and severe
physical disability (Table 1). At 12 months, the prevalence
of past month heroin declined still further, whereas the
prevalence of heroin overdose was significantly less than
that noted at baseline. The proportion of the sample that
reported being currently employed at 12 months was signif-
icantly greater than that noted at 3 months. No significant
changes in the prevalence of major outcome variables
occurred from 12 to 24 months. The prevalence of past year
incarceration did not change during the study period.
3.5. Effects of baseline cocaine use on outcomes
over 24 months
GEE analyses were used to examine the effects of
baseline cocaine use on major outcome variables across the
study period. After controlling for other factors, baseline
cocaine use was found to be a significant, independent
predictor of higher levels of heroin use (OR = 1.51, 95% CI =
1.17–1.95), needle sharing (OR = 1.67, 95% CI = 1.125–
2.23), unemployment (OR = 0.68, 95% CI = 0.50–0.92),
crime (OR = 1.63, 95% C.I. = 1.26–2.10), and incarcer-
ation (OR = 1.73, 95% CI = 1.31–2.29) over the 24-month
study period (Table 2). Baseline cocaine use did not
significantly predict heroin overdose or severe physical or
psychological disability.
4. Discussion
The substantial increase in cocaine use among heroin-
dependent individuals in NSW in 2001(Darke et al., 2002)
appears to have had long-lasting implications for treatment
A. Williamson et al. / Journal of Substance Abuse Treatment 33 (2007) 287–293 291
outcome. Although the prevalence of cocaine use among
heroin users in NSW had returned to previous levels by
2003 (Roxburgh, Degenhardt, & Breen, 2004) as reflected
by the reduced prevalence of cocaine use in the cohort from
3 months onward, baseline cocaine use was associated with
poorer treatment outcomes for the 2 years after study entry.
Indeed, although the cohort showed significant improve-
ments across most domains during the study period,
baseline cocaine use was a significant, independent pre-
dictor of poorer outcome across a range of important
measures including heroin use, employment, needle sharing,
criminal activity, and incarceration. More importantly, this
effect was present even after controlling for baseline
polydrug use.
Cocaine was unusually affordable and easy to obtain in
2001 (Roxburgh, Degenhardt, Breen, & Barker, 2003), a fact
reflected by the high prevalence of cocaine use in the cohort
at study entry (39%). As a result of changes in the
availability of cocaine in Sydney coupled with the effects
of treatment, a significant decrease in cocaine use was noted
from baseline to 3 months and from 3 to 12 months. At
24 months, the proportion of the cohort using cocaine was
not significantly different from that at 12 months, partly
reflecting the stabilization of Sydney’s cocaine market
(Black, Degenhardt, & Stafford, 2005). Baseline cocaine
use continued to be the strongest predictor of cocaine use at
24 months, whereas those in the nontreatment group at index
were also more likely to have used cocaine at this time.
Across the study period, baseline cocaine use predicted
continuing heroin use, even with treatment, demographic,
and general polydrug use factors taken into account. This
suggests that cocaine use at treatment entry may indicate
poorer treatment outcome in the medium term, even after
cocaine use has ceased for the majority of clients.
Those who reported cocaine use at baseline were
significantly more likely to have shared needles over
24 months. Although numerous studies have linked current
cocaine use to needle sharing (Bux, Lamb, & Iguchi, 1995;
Joe & Simpson, 1995), in the current study baseline
cocaine use was an independent predictor of needle sharing
even 2 years later. This is especially important given that
the majority of baseline cocaine users had ceased use by
3-month follow-up. It appears that treatment providers
should be alert to cocaine use among their clients as its
potential implications in terms of exposure to and trans-
mission of blood borne viruses may far exceed the actual
cocaine use period.
CUs were significantly less likely than NCUs to report
employment across the study period. This is an important
finding, as unemployment has been linked to premature
mortality (Harlow, 1990) and relapse (Llorente del Pozo,
Gomez, Fraile, & Perez, 1998) among heroin users. In
addition, this finding further illustrates the psychosocial
dysfunction that has previously been noted among cocaine
users (Bux et al., 1995; Condelli, Fairbank, Dennis, &
Rachal, 1991; Kosten et al., 1988). It is also indicative of the
fact that baseline cocaine users were less likely than others
to cease heroin use, and hence perhaps to be capable of
commencing full-time work.
Consistent with heavier levels of heroin use and higher
unemployment, baseline cocaine use significantly predicted
criminal activity throughout the study. Both heroin and
cocaine use (Gossop, Marsden, Stewart, & Rolfe, 2000;
Grella, Anglin, & Wugalter, 1995; Hando et al., 1997) alone
and in combination (Best, Sidwell, Gossop, Harris, &
Strang, 2001; Gossop et al., 2000; Kosten et al., 1988)
have been strongly linked to criminal activity, primarily
acquisitive crime conducted in order to raise the necessary
capital to purchase these drugs. It is noteworthy, however,
that the effect of cocaine use on crime persisted for the
duration of the study despite the low levels of cocaine use
noted at 12 and 24 months. In contrast to the decreasing
prevalence of criminal activity reported, the rates of past
year incarceration did not change over 24 months. In
keeping with the greater prevalence of criminal activity
noted among baseline cocaine users throughout the study,
CUs were significantly more likely than NCUs to report past
year incarceration at all interview points.
In contrast to drug use and crime, baseline cocaine use
was not found to be significantly related to general physical
health over the study period. Past year heroin overdoses
were significantly less common at 12 months than at
baseline and this reduction was maintained at 24 months.
Although there has been some suggestion in the literature
that the combination of heroin and cocaine may be
particularly dangerous in relation to heroin overdose (Coffin
et al., 2003), baseline cocaine use at least, was not
associated with a greater risk of overdose in the cohort.
In contrast to the findings of some other studies (Garlow,
Purselle, & D’Orio, 2003; Van Beek et al., 2001), no
relationship was demonstrated between baseline cocaine use
and mental health in this cohort. Several possible explan-
ations for this difference exist. For example, unlike the
cocaine users in much of the literature surrounding the
effects of cocaine use on mental health, the cocaine users in
the ATOS cohort were, in fact, primary heroin users. This
may have implications in relation to potential differences in
the duration and extent of cocaine use careers between study
groups. Moreover, the elevated levels of mental health
problems known to exist among heroin users (Ward,
Mattick, & Hall, 1999) may create a ceiling effect that
obscures any additional harms attributable to cocaine use.
In considering the long-term association between base-
line cocaine use and poorer treatment outcome evident in
this cohort, a question of causality arises. Is this poorer
treatment outcome a consequence of cocaine use itself? Or
rather, does cocaine use serve as a marker for a particularly
briskyQ group of heroin users? Although it is likely that both
factors contribute to some degree, it should be noted that
within-group comparisons of CUs and NCUs on the basis of
3-month cocaine use status revealed that decreased perform-
ance on outcomes including heroin use, needle sharing, and
A. Williamson et al. / Journal of Substance Abuse Treatment 33 (2007) 287–293292
injection-related health problems was associated with the
commencement or continuation of cocaine use, whereas
cessation of cocaine use resulted in significant improve-
ments on these measures (Williamson et al., 2006). Thus, in
the short term at least, cocaine use appeared to exacerbate
dysfunction, rather than serving as a marker for a more
dysfunctional group of individuals.
In summary, despite the small proportion of CUs who
continued cocaine use throughout the study, baseline
cocaine use was a significant predictor of poorer outcome
across a range of areas including heroin use, employment,
needle sharing, criminal activity, and incarceration. It is
especially striking that this effect was present even after
taking into account age, sex, treatment variables, current
heroin use, and baseline polydrug use. These findings
suggest that any increase in cocaine consumption among
heroin users may have repercussions far beyond the actual
period of use. Consequently, treatment providers should
regard cocaine use among clients as a marker for individuals
who are at greater risk of poor treatment outcome.
Acknowledgments
This research was funded by the National Health and
Medical Research Council and the Commonwealth Depart-
ment of Health and Ageing. The authors thank all
participating agencies.
References
Andrews, G., Hall, W., Teesson, M., & Henderson, S. (1999). The mental
health of Australians: National survey of mental health and wellbeing.
Canberra7 Commonwealth Department of Aged Care.
Best, D., Sidwell, C., Gossop, M., Harris, J., & Strang, J. (2001). Crime and
expenditure amongst polydrug misusers seeking treatment. British
Journal of Criminology, 41, 119–126.
Beswick, T., Best, D., Rees, S., Coomber, R., Gossop, M., & Strang, J.
(2001). Multiple drug use: Patterns and practices of heroin and crack
use in a population of opiate addicts in treatment. Drug and Alcohol
Review, 20, 201–204.
Black, E., Degenhardt, L., & Stafford, J. (2005). NSW drug trends 2004 —
Findings from the Illicit Drug Reporting System. Sydney7 University of
New South Wales.
Bux, D. A., Lamb, R. J., & Iguchi, M. Y. (1995). Cocaine use and HIV risk
behaviour in methadone maintenance patients. Drug and Alcohol
Dependence, 37, 29–35.
Chambers, C. D., Taylor, W. J. R., & Moffett, A. D. (1972). The incidence
of cocaine use among methadone maintenance patients. International
Journal of the Addictions, 7, 427–441.
Coffin, P. O., Galea, S., Ahern, J., Leon, A. C., Vlahov, D., & Tardiff, K.
(2003). Opiates, cocaine and alcohol combinations in accidental drug
overdose deaths in New York City, 1990–98. Addiction, 98, 739–747.
Condelli, W. S., Fairbank, J. A., Dennis, M. L., & Rachal, V. (1991).
Cocaine use by clients in methadone programs: Significance, scope and
behavioural interventions. Journal of Substance Abuse Treatment, 8,
203–212.
Darke, S., Hall, W., Wodak, A., Heather, N., & Ward, J. (1992).
Development and validation of a multi-dimensional instrument for
assessing outcome of treatment among opiate users: The Opiate
Treatment Index. British Journal of Addiction, 87, 733–742.
Darke, S., Kaye, S., & Topp, L. (2002). Cocaine use in New South Wales,
Australia, 1996–2000: 5 year monitoring of trends in price, purity,
availability and use from the Illicit Drug Reporting System. Drug and
Alcohol Dependence, 67, 81–88.
Downey, K. K., Helmus, T. C., & Schuster, C. R. (2000). Treatment of
heroin-dependent poly-drug abusers with contingency management and
buprenorphine maintenance. Experimental and Clinical Psychophar-
macology, 8, 176–184.
Gandek, B., Ware, J. E., Aaronson, N. K., Apolone, G., Bjorner, J., Brazier,
J., et al. (1998). Cross-validation of item selection and scoring for the
SF-12 Health Survey in nine countries: Results from the IQOLA
Project. International Quality of Life Assessment. Journal of Clinical
Epidemiology, 51, 1171–1178.
Garlow, S. J., Purselle, D., & D’Orio, B. (2003). Cocaine use disorders and
suicidal ideation. Drug and Alcohol Dependence, 70, 101–104.
Gossop, M., Marsden, J., Stewart, D., & Kidd, T. (2002). Changes in use of
crack cocaine after drug misuse treatment: 4–5 year follow-up results
from the National Treatment Outcome Research Study (NTORS). Drug
and Alcohol Dependence, 66, 21–28.
Gossop, M., Marsden, J., Stewart, D., & Rolfe, A. (2000). Reductions
in acquisitive crime and drug use after treatment of addiction
problems: 1-year follow-up outcomes. Drug and Alcohol Dependence,
58, 165–172.
Grella, C. E., Anglin, M. D., & Wugalter, S. E. (1995). Cocaine and crack
use and HIV risk behaviours among high-risk methadone maintenance
clients. Drug and Alcohol Dependence, 37, 15–21.
Hando, J., Flaherty, B., & Rutter, S. (1997). An Australian profile on the
use of cocaine. Addiction, 92, 173–182.
Harlow, K. C. (1990). Patterns of rates of mortality from narcotics and
cocaine overdose in Texas, 1976–1987. Public Health Reports, 105,
455–462.
Hartel, D. M., Schoenbaum, E. E., Selwyn, P. A., Kline, J., Davenny, K.,
Klein, R. S., et al. (1995). Heroin use during methadone maintenance
treatment: The importance of methadone dose and cocaine use.
American Journal of Public Health, 85, 83–88.
Hubbard, R. L., Craddock, S. G., & Anderson, J. (2003). Overview of the
5-year follow up outcomes in the drug abuse treatment outcome studies
(DATOS). Journal of Substance Abuse Treatment, 25, 125–134.
Joe, G. W., & Simpson, D. D. (1995). HIV risks, gender, and cocaine use
among opiate users. Drug and Alcohol Dependence, 37, 23–28.
Kosten, T. R., Rousanville, B. J., & Kleber, H. D. (1988). Antecedents and
consequences of cocaine abuse among opioid addicts — A 2.5 year
follow-up. The Journal of Nervous and Mental Disease, 176, 176–181.
Liang, K. -Y., & Zeger, S. L. (1986). Longitudinal data analysis using
generalized linear models. Biometrika, 73, 13–22.
Llorente del Pozo, J. M., Gomez, C. F., Fraile, M. G., & Perez, I. V. (1998).
Psychological and behavioural factors associated with relapse among
heroin abusers treated in therapeutic communities. Addictive Behav-
iours, 23, 155–169.
Magura, S., Siddiqi, Q., Freeman, R. C., & Lipton, D. S. (1991). Cocaine
use and help-seeking among methadone patients. The Journal of Drug
Issues, 21, 617–633.
Platt, J. J. (1997). Cocaine addiction. Theory, research and treatment.
Cambridge7 Harvard University Press.
Roxburgh, A., Degenhardt, L., & Breen, C. (2004). Changes in patterns of
drug use among injecting drug users following changes in the
availability of heroin in New South Wales, Australia. Drug and Alcohol
Review, 23, 287–294.
Roxburgh, A., Degenhardt, L., Breen, C., & Barker, B. (2003). NSW drug
trends 2002. NDARC Tech. Rep. No. 144. Sydney7 National Drug and
Alcohol Research Centre.
SAS Institute Inc. (1999). SAS Version 8.02.
Shaffer, H. J., & La Salvia, T. A. (1992). Patterns of substance use among
methadone maintenance patients. Journal of Substance Abuse Treat-
ment, 9, 143–147.
A. Williamson et al. / Journal of Substance Abuse Treatment 33 (2007) 287–293 293
Twisk, J. W. R. (2003). Applied longitudinal data analysis for epidemiol-
ogy. Cambridge7 Cambridge University Press.
Twisk, J. W. R. (2004). Longitudinal data analysis. A comparison between
generalized estimating equations and random coefficient analysis.
European Journal of Epidemiology, 19, 769–776.
Van Beek, I., Dwyer, R., &Malcom, A. (2001). Cocaine injecting: The sharp
end of drug-related harm! Drug and Alcohol Review, 20, 333–342.
Ward, J., Mattick, R. P., & Hall, W. (1999). Psychiatric comorbidity among
the opioid dependent. In J. Ward, R. P. Mattick, & W. Hall (Eds.),
Methadone maintenance treatment and other opioid replacement
therapies (pp. 419–440). Amsterdam7 Harwood Academic Publishers.
Ware, J., Kolinski, M., & Keller, S. (1996). A 12-item Short-Form Health
Survey: Construction of scales and preliminary tests of reliability and
validity. Medical Care, 34, 220–233.
Williamson, A., Darke, S., Ross, J., & Teesson, M. (2006). The association
between cocaine use and short-term outcomes for the treatment of
heroin dependence: Findings from the Australian Treatment Outcome
Survey (ATOS). Drug and Alcohol Review, 25, 139–146.