The Impact of Treatment on 3 Years

7/29/2019 The Impact of Treatment on 3 Years

1/9

The impact of treatment on 3 years outcome for

heroin dependence: findings from the Australian

Treatment Outcome Study (ATOS)

Maree Teesson1, Katherine Mills1,2, Joanne Ross1, Shane Darke1, Anna Williamson3 &Alys Havard1

National Drug and Alcohol Research Centre, University of New South Wales,Australia,1 Centre for Mental Health Studies, University of Newcastle, Australia2 and

Sax Institute, University of Sydney, Australia3

ABSTRACT

Aim To examine the impact of treatment for heroin dependence on drug use, injection-related risk-taking, health

problems, criminality and general physical and mental health over 3 years among heroin-dependent Australians.

Design Longitudinal prospective cohort study. Participants A total of 615 heroin users enrolled in the AustralianTreatment Outcome Study; 94.5% of the sample completed at least one follow-up interview over 36-month follow-up.

Findings The proportion who reported using heroin in the preceding month continued to decrease significantly from

baseline to 24-month follow-up (99% versus 35%), with this rate remaining stable to 36-month follow-up. The

reduction in heroin use was accompanied by reductions in other drug use. There were also substantial reductions in

risk-taking, crime, injection-related health problems and improvements in general physical and mental health. Positive

outcomes were associated with more time in maintenance therapies and residential rehabilitation and fewer treatment

episodes. Time spent in detoxification was not associated with positive outcomes. Major depression was also associated

consistently with poorer outcome. Conclusions At 3 years, there were impressive reductions in drug use, criminality,

psychopathology and injection-related health problems following treatment exposure.

Keywords Criminality, health problems, heroin dependence, injection risk, prospective cohort study, treatment

outcome.

Correspondence to: Maree Teesson, National Drug and Alcohol Research Centre, University of New South Wales, NSW 2052, Australia.

E-mail: [email protected]

Submitted 19 February 2007; initial review completed 14 June 2007; final version accepted 11 September 2007

INTRODUCTION

Over the past decade heroin use and dependence has

emerged as an issue of increasing public health

concern. Indeed, irrespective of the increasing use ofother drugs, opiates continue to be the main problem

drug world-wide [1]. Heroin is overwhelmingly the pre-

dominant opioid used by illicit drug users in Australia

and world-wide [1]. At the beginning of this century it

was estimated that 0.2% of the worlds population aged

15 and above used heroin, translating to over 10 million

people [1]. Of these approximately 25% will be depen-

dent on the drug [2].

Heroin dependence is a remarkably persistent, and in

many cases, a life-long condition [3]. In order to improve

the prognosis for those with this condition it is necessary

to determine the factors which impact on treatment out-

comes for this group. Although a number of studies have

examined the impact of treatment on outcomes for

heroin dependence, these have been restricted largely to

randomized controlled trials or prepost observationalstudies [4]. Large-scale, naturalistic prospective longitu-

dinal, multi-site treatment outcome studies of heroin

dependence are rare. These studies allow for the exami-

nation of the impact of treatment over time, as it is deliv-

ered in the real world.

The few large-scale prospective studies conducted in

the United States and Europe have demonstrated that

individuals in the major treatment modalities can

substantially reduce their use of illicit drugs and criminal

activity [59]. The UK-based National Treatment

Outcome Research Study (NTORS [10]) found a

RESEARC H REPO RT doi:10.1111/j.1360-0443.2007.02029.x

2007 The Authors. Journal compilation 2007 Society for the Study of Addiction Addiction, 103, 8088
mailto:[email protected]:[email protected]


2/9

substantial reduction in heroin use at 45 years post-

treatment entrance [11].Consistently, the US Drug Abuse

Treatment Outcome Study (DATOS) found weekly or

more frequent heroin use had reduced from 89% to 28%

by 5 years post-treatment entrance [8]. NTORS also dem-

onstrated significant improvements in physical and

mental health across treatment modalities [11].

Positive outcomes have been associated consistently

with spending more time in treatment maintenance

therapies or residential rehabilitation [7,8,1116] but

enrolling in fewer treatment episodes [13]. To our knowl-

edge, no study has examined the association between

time in detoxification and outcome. Additionally, none of

the studies conducted to date have included a non-

treatment group.

Longitudinal studies of treatment outcome conducted

to date have also been limited in that the analyses have

been restricted to complete case analyses. That is, cases

were included in the analysis only if they were inter-viewed at all time-points. Hence, the results may be

biased towards those who are more likely to be inter-

viewed and are well functioning. Additionally, despite

being longitudinal studies, most utilize data from only

two time-points and fail to take into account longitudinal

changes. Psychiatric measures have also been lacking

from previous research. Thus the impact of conditions

such as major depression and post-traumatic stress disor-

der, which co-occur commonly in people with heroin

dependence [1719], is unclear.

The AustralianTreatment Outcome Study(ATOS) was

the first Australian prospective longitudinal study oftreatment outcomes for heroin dependence conducted in

Australia. A total of 535 heroin users were recruited

upon entry to maintenance therapies (MT: methadone

and buprenorphine), residential rehabilitation (RR) and

detoxification (DTX), in addition to a sample of 80 heroin

users who were not currently in treatment (NT). The

present paper details the 3 years findings of this study.

Specifically, the present paper aimed to examine the

impact of treatment for heroinuse on drug use, injection-

related risk taking, health problems, criminality and

general physical and mental health over 3 years. The

impact of psychiatric comorbidity on these outcomes isalso examined.

METHOD

Procedure

Data were collected as part of the New South Wales com-

ponent of ATOS, the first naturalistic prospective longitu-

dinal study of treatment outcome for heroin dependence

to be conducted in Australia [17]. Baseline data were col-

lected between February 2001 and August 2002, with

follow-up interviews conducted at 3, 12, 24 and

36 months post-baseline.

Participants were recruited from 19 agencies treating

heroin dependence in the greater Sydney region. Agen-

cies were selected randomly from within treatment

modality and stratified by regional health area.The agen-

cies represent the major treatment modalities available

for heroin dependence in Australia and comprised 10

methadone/buprenorphine maintenance agencies, four

drug-free residential rehabilitation agencies and nine

detoxification facilities. Four agencies provided both

maintenance and detoxification services. A comparison

group of people with heroin dependence who were not

currently in treatment were recruited from needle and

syringe programmes. Eligibility criteria were: (i) no treat-

ment for heroin dependence in the preceding month; (ii)

no imprisonment in the preceding month; (iii) aged

18 years or over; and(iv) agreedto givecontact details for

follow-up interviews.The cohort consisted of 615 people with heroin

dependence: 201 entering maintenance therapy, 201

entering detoxification, 133 entering residential reha-

bilitation and 80 non-treatment participants. A total of

549, 495, 469 and 429 individuals were re-interviewed

at 3-, 12-, 24- and 36-month follow-ups, respectively,

representing 89.3%, 80.5%, 76.3% and 69.8% of those

enrolled at baseline. Only 5.5% of the sample was not

interviewed at all following baseline interview. Back-

ward stepwise logistic regressions were conducted in

order to determine factors associated with retention in

the cohort at 3-, 12-, 24- and 36-month follow-ups.Baseline variables entered into the models included

index treatment modality, age, sex, previous treatment

history, number heroin use days in the preceding

month, number of drug classes used in the preceding

month, major depression, current post-traumatic stress

disorder (PTSD) symptomatology, antisocial personality

disorder (ASPD) and borderline personality disorder

(BPD). The samples re-interviewed at each follow-up

were broadly representative of the initial sample of 615

enrolled in ATOS, the only notable difference being that

females were more likely than males to be retained at

3 months [94.7% versus 86.5%; odds ratio (OR) 2.8895% confidence interval (CI): 1.425.85] and

24 months follow-up (81.3% versus 73.7%; OR 1.66

95% CI: 1.062.59).

All participants were paid A$20 for completing each

interview. All interviews were conducted by trained

Research Officers employed by the National Drug and

Alcohol Research Centre and independent of the treat-

ment agencies. Ethical approval was granted by the

Human Ethics Review Committees of all participating

area health services and the University of New South

Wales.

3-year treatment outcomes for heroin dependence 81



3/9

Structured interview

Participants were administered a structured interview at

baseline and each follow-up. The components of the

follow-up interviews are detailed in Ross et al. [17].

The baseline and follow-up interviews addressed

demographic characteristics, treatment history, drug use

history, general physical and mental health and psy-chopathology. Demographic characteristics measured

included participants age, sex and their main source of

income in the preceding month. Heroin and other drug

use (including other opiates, alcohol, marijuana, benzo-

diazepines, amphetamines, cocaine, hallucinogens and

inhalants) in the precedingmonth was assessed using the

Opiate Treatment Index [20].

At each follow-up interview, participants were asked

the number of times they had commenced the various

treatment options for heroin dependence since their last

interview, the length of stay for each episode and how

recently they had attended each type of treatment. The

number of days in each of the main treatment modalities

was calculated and divided by the number of days to

follow-up to give the proportion of time spent in each

treatment modality over the follow-up periods.

General physical and mental health were measured at

each interview using the Short Form-12 (SF-12), in

which lower scores indicate poorer health [21]. DSM-IV

diagnoses of major depression (past month) were

obtained at each interview using the Composite Interna-

tional Diagnostic Interview (CIDI) version 2.1 [22]. At

baseline, DSM-IV diagnoses of ASPD were obtained using

a version of the Diagnostic Interview Schedule modified

to give DSM-IV diagnoses, and participants were screened

for ICD-10 BPD using the International Personality Dis-

orders Examination Questionnaire [23] used in the

National Survey of Mental Health and Wellbeing. Life-

time DSM-IV diagnoses of PTSD were obtained at base-

line using the CIDI version 2.1 [22]. Individuals with

current PTSD were classified as those who received a

life-timediagnosis and who hadexperienced symptoms of

PTSD in the preceding 12 months.

Statistical analyses

Categorical and continuous measures of outcomes were

examined using generalized estimating equation analysis

(GEE: [24]). GEE analyses are able to take into account

longitudinal design and correlation among repeated

observations. GEE allows the relationships between the

variables of the model at differenttime-points to be analy-

sed simultaneously, so that the estimated regression coef-

ficient reflects the relationship between the longitudinal

development of the outcome variable and the longitudi-

nal development of corresponding predictor variables

[25]. These analyses also provide an estimate of the time-

averaged difference between groups.

Fixed and time-varying covariates were included in

the models. Fixed covariates included age, sex, the pres-

ence or absence ASPD and BPD and current PTSD at

baseline; variables which may impact on treatment

outcome. Time-varying covariates included the propor-

tion of time spent in maintenance therapies, detoxifica-

tion and residential rehabilitation between interviews,

the number of treatment episodes commenced between

interviews and the presence or absence of major depres-

sion. Significant results are reported as the estimate of

the difference in least-squares means for continuous

outcome measures, and as OR with 95% CI for dichoto-

mous outcome variables. Results are presented sequen-

tially as is consistent with reports from the large-scale

prospective studies conducted in this area [5,710]. All

GEE analyses utilized an unstructured correlation matrix

and were conducted using SAS (version 8.02).

RESULTS

Baseline cohort characteristics

Themean age was 29.3 years(SD 7.8, range 1856), and

66% were male. Participants completed a mean of

10.0 years of school education (SD 1.7, range 212),

29% had completed a trade/technical course and 6% a

university degree. The main sources of income reported

were: government allowances (46%), criminal activity

(24%) and wage/salary (18%). Over half had committed

a crime in the preceding month, and 41% had a prison

history. The mean length of heroin using career was

9.6 years (SD 7.4, range


4/9

sample had commenced a median of three treatment epi-

sodes (range 018) and had undertaken a median of

428.5 days in treatment (range 01308). The median

proportion of time spent in treatment over the follow-up

period was 0.39 (range 01). Only three people had not

commenced any treatment over the 36 months.

Treatment outcomes

Heroin use

The proportion who reported using heroin in the preced-

ing month decreased significantly from baseline to

3-month follow-up (OR 0.02, 95% CI: 0.010.04), from3-month to 12-month follow-up (OR 0.53, 95% CI:

0.410.66) and from 12-month to 24-month follow-up

(OR 0.79, 95% CI: 0.630.98). This rate remained stable

to 36-month follow-up (Table 2).

One-month abstinence from heroin was associated

with having spent more time in maintenance therapies

(OR 1.72, 95% CI: 1.332.22) and residential rehabilita-

tion (OR 16.67, 95% CI: 9.0933.33) and fewer treat-

ment episodes (OR 0.87, 95% CI: 0.830.91). It was

unrelated to time spent in detoxification. Major depres-

sion was the only other variable associated significantly

with 1-month abstinence and was associated with anincreased likelihood of use (OR 1.79, 95%CI: 1.312.45)

(Fig. 1).

Heroin dependence

The proportion who were heroin-dependent in the pre-

ceding month decreased significantly from baseline to

3-month follow-up (OR 0.02, 95% CI: 0.010.03), and

again from 3-month to 12-month follow-up (OR 0.60,

95% CI: 0.460.79), with this rate remaining stable to

36-month follow-up (Table 2).

Heroin dependence was associated with having spent

less time in maintenance therapies (OR 0.46, 95% CI:

0.340.62) and residential rehabilitation (OR 0.04, 95%

CI: 0.020.09), more time in detoxification (OR 15.58,

95% CI: 1.13215.16) and more treatment episodes (OR

1.12, 95% CI: 1.031.22). Those with major depression

in the last month (OR 1.88, 95% CI: 1.322.61) and

those who screened positive for BPD (OR 1.49. 95% CI:

1.141.94) were also more likely to be heroin-dependent.

Other substance use

The proportion who used other drugs in the preceding

month decreased significantly from baseline to 3-monthfollow-up (OR 0.47, 95% CI: 0.310.71), again from

3-month to 12-month follow-up (OR 0.68, 95% CI:

0.510.91) and from 12-month to 24-month follow-up

(OR 0.77, 95% CI: 0.591.00). This rate then remained

stable to 36-month follow-up (Table 2).

Other drug use was associated with having spent less

time in residential rehabilitation(OR 0.02, 95%CI: 0.01

0.04) and more time in maintenance therapies (OR 1.68,

95% CI: 1.210.71) and detoxification (OR 51.94, 95%

CI: 2.271176.15). Number of treatment episodes was

not associated with other drug use. Other drug use was

also associated with being younger (OR 0.98, 95% CI:0.961.00), male (OR 1.65, 95% CI: 1.192.29) and the

presence of major depression (OR 1.62, 95% CI: 1.14

2.30).

Needle sharing

The proportion who had shared needles in the preceding

month decreased significantly from baseline to 3-month

follow-up (OR 0.35, 95% CI: 0.230.53) and again from

3-month to 12-month follow-up (OR 0.67, 95% CI:

0.460.98). This rate then remained stable to 36-month

follow-up (Table 2).

Table 1 Treatment exposure over 36 months.

Index group

All (n =429)MT (n =148) DTX (n =147) RR (n =92 ) NT ( n =42)

Three-year treatment exposure (%)

MT 100 66.0 44.6 76.2 74.1

DTX 10.8 100 35.9 50.0 50.6RR 5.4 37.4 100 21.4 38.1

Other 19.6 38.8 39.1 14.3 29.8

Any treatment (%) 100 100 100 92.9 99.3

Post-index treatment (%) 70.3 94.6 87.0 92.9 84.4

Median treatment episodes since baseline interview 2 4 3 3 3

Median cumulative treatment days since baseline

interview

887.5 305 270 288 428.5

MT = maintenance therapies; DTX = dextoxification; RR = residential rehabilitation; NT = not currently in treatment

3-year treatment outcomes for heroin dependence 83



5/9

Needle sharing was associated with having spent less

time in maintenance therapies (OR 0.46, 95% CI: 0.07

0.29) and residential rehabilitation (OR 0.10, 95% CI:

0.020.47). Time spent in detoxification and number

of treatment episodes was not associated with needle

sharing. Those with major depression in the last month

(OR 1.55, 95% CI: 1.142.12) and those who screened

positive for BPD (OR 1.93, 95% CI: 1.422.61) were also

more likely to have shared needles (Fig. 2).

Injection-related health problems

The proportion who had injection-related health prob-

lems in the preceding month decreased significantly frombaseline to 3-month follow-up (OR 0.22, 95% CI: 0.16

0.30) and again from 3-month to 12-month follow-up

(OR 0.58, 95% CI: 0.450.75). This rate then remained

stable to 36-month follow-up (Table 2).

Injection-related health problems were associated

with having spent less time in maintenance therapies (OR

0.62, 95% CI: 0.460.84) and residential rehabilitation

(OR 0.07, 95% CI: 0.030.14) and more treatment epi-

sodes (OR 1.16, 95% CI: 1.071.26). Time spent in

detoxification was not associated with injection-related

health problems. Injection-related health problems were

Table 2 Generalized estimating equation results: the impact of

time, proportion of time in treatment and number of treatment

episodes on treatment outcomes.*

Outcome b SE P

% Heroin use

Time

3 months -3.85 0.37 < 0.000112 months -4.49 0.37 < 0.0001

24 months -4.74 0.37 < 0.0001

36 months -4.74 0.37 < 0.0001

Proportion of time in treatment

MT -0.55 0.13 < 0.0001

DTX 0.80 0.99 0.4216

RR -2.82 0.29 < 0.0001

No. of treatment episodes 0.14 0.04 0.0001

% Heroin-dependent

Time

3 months -3.98 0.29 < 0.0001

12 months -4.49 0.29 < 0.0001

24 months -4.74 0.29

The Impact of Treatment on 3 Years

Documents

Transcript of The Impact of Treatment on 3 Years