How does research inform best practice in treatment of the addictions

1
How does research inform best practice in
treatment of the addictions?

• Prof Michael Gossop
• National Addiction Centre
• Maudsley Hospital/Institute of Psychiatry

2
A History of European Art in the 15th and 16th
Centuries
OR ..
3
German woodcut, 1549
4
Hieronymus Bosch
5
A vessel whose deranged passengers neither know
nor care where they are going
The Ship of Fools
6
OR .
What does addiction research tell addiction
treatment?
7
What is it that we are treating?

• -

8

• Acute physical dependence
• Psychological, behavioural dependence
• Relapse prevention
• Rehabilitation

9
What do we think happens to dependence during and
after effective treatment?
10
What do we think happens to dependenceduring and
after effective treatment?
Treatment Entry
Severity of Dependence
Time
11
What do we think happens to dependenceduring and
after effective treatment?
Treatment Entry
Severity of Dependence
Time
12
What do we think happens to dependenceduring and
after effective treatment?
Threshold for relapse
Treatment Entry
Severity of Dependence
Time
13
What do we think happens to dependenceduring and
after effective treatment?
Acquisition of behavioural and other coping
skills to avoid relapse
Treatment Entry
Severity of Dependence
Time
14
If the treatment of addiction requires the
acquisition of coping skills that enable the
individual to avoid relapse to illicit drug use
what if the factors that lead to relapse are
largely independent of the processes that occur
during treatment?
15
How should we assessthe effectiveness of
treatment?
16
A simple paradox of treatment evaluation

• Suppose a person receives a specific treatment
  and clearly and immediately fails.
• Suppose that they immediately re-enter the same
  treatment, and clearly and immediately show a
  full recovery.
• Was the treatment (i) ineffective, (ii)
  effective,
• (iii) 50 effective?.

17
There are certain general principles that apply
to the evaluation of treatment effectiveness. One
of the most important is that the research
designs should be matched to the research
question.
18
Sometimes it is claimed that some methodologies
are better or more scientific than others.
The proposition that randomised double-blind
prospective quantitative group trials are better
than .... observational studies is of course
absurd.
19
Conducting useful and informative research into
treatment effectiveness does not necessarily
depend upon using the advanced methods, but
rather depends upon using the most appropriate
methods. The key issue is how to choose the
best and most appropriate methodology to answer
the specific research question. Using the wrong
method is unhelpful, no matter how systematically
the method is applied.
20
The ideal RCT, a treatment efficacy study uses an
experimental design with

• Random allocation to treatment and control groups
• Rigorous controls (eg a no-treatment condition)
• Treatments are controlled (eg manualised)
• Patients receive treatments of fixed duration
• Ideally design should be double blind with
  clinicians and patients blind to tx condition
• Patients meet criteria for single disorder
  (multiple disorders are excluded)

21
RCT - Fetish or Science?

• -

22

• Fetish - an object believed to have magical
  powers
• - irrationally reverenced in an obsessive manner

The Oxford English Dictionary
23

• Fetish - a non-sexual object acting as
• a focus for sexual desire

The Oxford English Dictionary
24
If RCTs are they only acceptable evidence of
efficacy why has no-one done an RCT to test
whether naloxone reverses opiate overdose?
25
RCTs can be a powerful methodology for the
evaluation of simple or well-defined treatments
with single or well-defined outcomes.
26
Randomised trial comparing 10 day methadone and
lofexidine detoxification
Lofexidine
Total withdrawal score
Methadone
Days
Bearn et al., 1996
27
Accelerated (5 day) lofexidine vs standard (10
day) detoxification and (10 day) methadone
detoxification
Total withdrawal score
Lof-10
Lof-5
Days
Bearn et al., 1998
28
RCTs may be less appropriate (or inappropriate)
where interventions are complex and not
well-defined, and where outcomes are influenced
by complex interactions of factors that may or
may not be related to treatment.
29
Drug users seeking treatment are an extremely
heterogeneous group

The extent and complexity of individual
differences among clients (many of which are not
properly understood) makes it highly problematic
to construct matched treatment and control
groups. This problem is amplified where
multiple outcomes are being assessed.
30

There is growing recognition that program
evaluation is more than just a technical
enterprise. Moos, Finney and Cronkite, 1990
31

The starting point for many substance abuse
treatment evaluations is a comparison of
treatment and control conditions, or . of
alternative treatment programs. Unfortunately,
that is too often the ending point. John
Finney, 1995.
32
(No Transcript)
33
(No Transcript)
34
Potential problems with the RCT design

• Random allocation does not reflect the processes
  by which patients actually seek treatment.

35
Potential problems with the RCT design

• The multitude of uncontrolled, potentially
  relevant independent variables means that random
  assignment will not equate groups for all
  possible confounds.

36
Potential problems with the RCT design

• Where within-cell variation exceeds measurement
  error, there are likely to be reliable
  differences among patients within treatments.
• Some patients are responding differently to the
  same treatment.

37
Potential problems with the RCT design

• Effect sizes are rarely large enough to eliminate
  overlap between the outcome of patients in
  different treatment groups.

38

A common but mistaken model of treatment outcomes

Heavy drinking
Treatment response
Outcome
39

(Pre-treatment problems affect outcomes)
Heavy drinking
Treatment response
Outcome
40

Heavy drinking
Treatment response
Outcome
Therapists respond to presenting problems by
providing targeted interventions
41

Presenting problems lead to altered treatment
interventions as well as affecting post-treatment
outcomes.

-


Heavy drinking
Treatment response
Outcome
42
How to choose the most appropriate treatment for
the individual patient?

• The US Institute of Medicine (1990)
  recommended that the simplistic question of
  whether a treatment works should be redefined as
  
• "Which kinds of individuals, with what kinds
  of ..... problems, are likely to respond to what
  kinds of treatments by achieving what kinds of
  goals when delivered by which kinds of
  practitioners?" (p.143).

43
Inter-agency variation in heroin useresidential
programmes
44
Inter-agency variation in heroin use methadone
programmes
45
Poor response to methadone maintenance
Low rate use (IV)
Improved (I)
59 of sample improved
Poor response (III) 19 of sample
Improved (II)
46
Heroin use at intake and 1 year follow-up
Poor response
Improved response
47
Stimulant use at intake and 1 year follow-up
Poor response
Improved response
48
Benzo use at intake and 1 year follow-up
Poor response
Improved response
49
Illicit drug use at intake and 1 year
follow-upPoor response group

50
One concern expressed repeatedly has been that
treatment is largely a "black box". The people,
events, and interactions subsumed by labels such
as "therapy, "counseling," "referral for
services," and "remediation" remain largely
undescribed in drug programs. As a result,
variations in delivery and efficacy, which are
crucial to treatment evaluation, are also
unspecified. (Lipton and Appel, 1984).

51
There are also many treatment factors that may be
related to outcome, although few have been
identified...... There is an even greater need
for basic and systematic descriptions,
classifications, and measurements of treatment
factors than of patient characteristics. (Gotthei
l et al., 1981)

52
Although most behavioral scientists endorse the
idea that both personal and environmental factors
determine behavior, evaluation researchers have
typically conceptualized the treatment program as
a "black box" intervening between patient or
staff inputs and outcomes. Thus, these programs
often are assessed only in terms of broad
categories. (Moos, 1997)

53

The complexity of treatment is difficult to
conceptualize and even harder to define and
quantify. (Hubbard et al., 1989)
54

Many studies have used a black-box paradigm to
evaluate treatment programs . Such evaluations
can gauge the overall outcome of a treatment
program but they reveal little about the process
of treatment or how to improve it. Moos, Finney
and Cronkite, 1990
55
(No Transcript)