Transcript Which is More Important: Therapy or Therapist?

William R. Miller, Ph.D.
The University of New Mexico
The Eileen Pencer Memorial Lecture

Have you noticed:
that in multisite clinical trials (like CTN)
of treatments that are already evidence-based
even under highly controlled, supervised,
manual guided delivery conditions
the expected main effects are often quite small?
and that efficacy often varies by therapist and by site?


This even happens with pharmacotherapies
In practice, EBTs are not homogeneous entities
Project MATCH
COMBINE Study
90
90
80
80
70
70
60
60
50
50
40
40
30
30
20
20
-2
-1
0
4
5
6
CBT
7
8
9
10 11 12 13 14 15
MET
TSF
Intake
4 Months
CBI + Placebo
Naltrexone, No CBI
Placebo, No CBI
CBI + Naltrexone
16 Months
Percent Days Abstinent
50
42.3 43.2
45.4 46.6
40
30
29.5 29.5
20
10
0
Baseline
3 Months
MI
12 Months
SBNT
MI = 3 Sessions vs. SBNT = 8 Sessions
TREATMENT TESTED
004
005
013
021
009
011
015
006
007
MET
MI
MET
MET
Smoking cess.
Tele calls
Seeking Safety
Incentives
Incentives
PRIMARY OUTCOME
Drug use & retention days - nsd
Use: nsd 5 vs. 4 sessions p<.05
Retention nsd (pregnant users)
Drug use & retention nsd
Cessation nsd by 13 weeks
Retention nsd;
PTSD symptoms & drug use nsd
8.6 vs 5.2 sessions abstinent p<.001
5.5 vs 2.3 sessions abstinent p <.001



The specific effect size for manual-guided EBT
is typically small
Even small effects may be clinically meaningful
With large multisite samples, statistical
significance can be found for small effects, but
do clinicians care?
Clinically significant improvement
= large enough to be interested in
learning a new treatment method.
Approximately 10 point
increase in % doing well
doubling or halving of
continuous measures
Miller, W. R., & Manuel, J. K. (2008). How large must a treatment effect be before it matters to
practitioners? An estimation method and demonstration. Drug and Alcohol Review, 27, 524-528.
or
Learning More from Clinical Trials
In practice, there is.
In practice, evidence-based
treatments
are not homogeneous entities
So
Where are the bigger effects?
100
90
% Days
Abstinent
80
70
60
50
Traditional
40
CRA
30
20
10
0
Azrin 1973
Azrin 1976
Azrin 1982
% Days
Abstinent



Psychologists tested patients in three different
treatment programs
They identified patients with particularly high
alcoholism recovery potential (HARP)
HARP vs. non-HARP patients did not differ
from each other on prior treatment history or
severity of alcoholism







More motivated for counseling
More punctual in meeting appointments
Showing greater self-control
Neater and more attractive in appearance
More cooperative
Trying harder to stay sober
Showing better recovery




Higher rates of abstinence
Longer spans of abstinence
Fewer slips
More employment
“HARPs” had been selected at random.
Counselor expectations matter a lot.
Average 12-Month Drinking Outcomes for 8,389 Clients





Survival
Percent Continuous Abstinence
Percent Days Abstinent:
Average Reduction in Consumption
98.5%
24.1%
81.4%
87.0%
Overall outcomes were fairly similar for treatment as
usual (RAND, VAST, RREP), pharmacotherapy trials
(lithium, disulfiram), and controlled trials of
psychosocial treatments (MATCH)
Miller, W. R., Walters, S. T., & Bennett, M. E. (2001). How effective is alcoholism treatment?
Journal of Studies on Alcohol, 62, 211-220.
Percent Days Abstinent
0.9
0.8
0.7
0.6
MATCH
0.5
0.4
0.3
Pre
Months 1-3
Months 4-6
Westerberg, V. S., Miller, W. R., & Tonigan, J. S. (2000). Comparison of outcomes for clients in randomized
versus open trials of treatment for alcohol use disorders. Journal of Studies on Alcohol, 61, 720-727
Percent Days Abstinent
0.9
0.8
0.7
0.6
MATCH
0.5
TAU
0.4
0.3
Pre
Months 1-3
Months 4-6
Westerberg, V. S., Miller, W. R., & Tonigan, J. S. (2000). Comparison of outcomes for clients in randomized
versus open trials of treatment for alcohol use disorders. Journal of Studies on Alcohol, 61, 720-727
Luborsky, McLellan, Woody, O’Brien & Auerbach, 1985
Archives of General Psychiatry 42:602-611



Two drug treatment counselors resigned
Their 62 cases were assigned randomly to the
four remaining counselors
There were dramatic differences in client
outcomes.
McLellan et al., 1988 Journal of Nervous and Mental Disease, 176, 423-430.
% Positive Urines
Methadone Dose
% Employed
% Arrested
McLellan et al. (1988). Journal of Nervous and Mental Disease, 176, 423-430.
120
All Clients got “the same” manual-guided treatment
in the NIAAA COMBINE Study
100
80
60
40
20
0
Mean PDA Baseline
Mean PDA Week 4
Mean PDA Week 26
Mean PDA Week 52
Mean PDA Week 68





Much emphasis is given to “common factors”
as an alternative to EBTs
So-called “common factors” may or may not
be all that common in practice
If they do exert a large effect, they should not
be hard to observe
“Nonspecific” just means that they have not
yet been adequately specified and tested
So why not both . . and?
Miller, Benefield & Tonigan (1993) JCCP 61: 455-461
Confront (r = .65)
42%
Miller, Benefield & Tonigan (1993) JCCP 61: 455-461

Accurate empathy is a well-specified,
learnable, reliably measurable therapist skill –
the ability to understand and reflect clients’
meaning

Originally defined and studied by Carl Rogers

It is not identification with your client





Were delivering the same manual-guided
behavior therapy (self-control training)
Were trained both in behavior therapy and
accurate empathy
Had sessions independently observed and
rated by three supervisors, including the
Truax & Carkhuff scale for accurate empathy
Were then rank-ordered (1-9) for empathic
skill while delivering behavior therapy
And when we examined 6-month client
outcomes . . . .
60
Miller, Taylor & West (1980) JCCP 48:590-601
6 months
12 months
24 months
r = .82
r = .71
r = .51
67%
52%
Miller & Baca (1983) Behavior Therapy 14: 441-448
26%
Patients in treatment for alcoholism were randomly
assigned to counselors with:
LOW levels of empathy and related skills
MEDIUM levels of empathy and related skills
or
HIGH levels of empathy and related skills
What percentage of patients relapsed?
Valle (1981) J Studies on Alcohol 42: 783-790
Invalidated
Resist
Withdraw
Not respected
Not understood
Not heard
Angry
Ashamed
Uncomfortable
Unable to change
Arguing
Discounting
Defensive
Oppositional
Denying
Delaying
Justifying
Disengaged
Disliking
Inattentive
Passive
Avoid/leave
Not return
Affirmed
Accept
Understood
Open
Accepted
Undefensive
Respected
Interested
Heard
Cooperative
Comfortable/safe
Listening
Empowered
Hopeful/Able to change
Approach
Talk more
Liking
Engaged
Activated
Come back
Open
Cooperative
Arguing
Engaged
Active
Empowered
Hopeful
Liking
Defensive
Oppositional
Listening
Disengaged
Passive
Powerless
Unable to change
Disliking
Bien et al 1993
Brown & Miller 1993
Aubrey 1998
20 minutes in ER
One handwritten letter
One telephone call
% Attending AA
Systematic Encouragement
100
90
80
70
60
50
40
30
20
10
0
100
0
Systematic Encouragement
Standard Procedures
Hire Empathic Therapists!
It is an evidence-based practice
to hire staff based on and to train staff in
the skill of accurate empathy
Percentage of variance in 12-month drinking
outcome determined by therapist factors
(random assignment designs)
There’s some beef!
Miller, Taylor & West 1980
52%
Empathy
Valle 1981
65%
Empathy+
Miller et al., 1993
42%
Confronting (-)
OR THE
BOARDS
and some of them are influenced by the therapist
Motivation for change
Self-efficacy
Hope/optimism
Attendance/adherence


Involving a spouse in treatment significantly
improves client substance use outcomes
And again it matters what you do


There are some specific treatment effects
generally supported by clinical trials
They are often relatively small compared to
Therapist effects
 Client effects and social context
 Overall impact of treatment
 And perhaps for these reasons they vary across sites


These larger “nonspecifics” (e.g., empathy) need
to be specified, tested, and trained as EBTs






Treatment-as-usual is a high standard to beat
Re-training staff in EBTs can be challenging
and expensive
Specific treatment effect size often shrinks
with dissemination into clinical practice
Therapist belief/enthusiasm / style matters
Testable question: Is it cost effective to retrain staff in an EBT?
In any event, it makes sense to train the next
generation of addiction professionals in EBTs
from the very beginning!



What is it about the therapists who are delivering
the treatment that affects outcome?
What is it about the treatment that really matters?
Understanding the underlying mechanisms of
efficacy of treatments and therapists will help us
to know:




Whom to hire
What is essential in training
What to focus on in fidelity monitoring
What can be changed in adaptations