Transcript EBP Presidential Task Force

Evidence-based Practice in Psychology:
Epistemological Diversity*
Steven D. Hollon
Member, Presidential Task Force on EBPP
Vanderbilt University
Email: [email protected]
*Based upon: Report of the 2005 Presidential Task Force on Evidence-Based Practice:
Policy statement: http://www.apa.org/practice/ebpstatement.pdf
Complete report: http://www.apa.org/practice/ebpreport.pdf
2005 Presidential Task Force on
Evidence-Based Practice (APA)
Ronald F Levant EdD (Chair)
Carol D Goodheart EdD (Chair)
David H Barlow PhD
Frederick L Newman PhD
Jean Carter PhD
John C Norcross PhD
Karina Davidson PhD
Doris K Silverman PhD
Kristofer J Hagglund PhD
Brian D Smedley PhD
Steven D Hollon PhD
Bruce E Wampold PhD
Josephine D Johnson PhD
Drew I Westen PhD
Laura C Leviton PhD
Brian T Yates PhD
Alvin R Mahrer PhD
Nolan W Zane PhD
APA Staff:
Geoffrey M Reed PhD
Lynn F Bufka PhD
Ernestine Penniman
Basic Definition and Process
 Based on IOM definition that emphasized integration of
research evidence with clinical expertise and patient values
 Drew on diverse group with range of expertise and interests
from research scientists through clinical practitioners
 Produced draft policy statement and position paper that was
then posted for comments and subsequently revised
 Approved by vote of council at the 2005 APA convention
EBPP Defined
 Evidence-based practice in psychology (EBPP) is the
integration of the best available research with clinical
expertise in the context of patient characteristics, culture,
and preferences.
 Closely parallels the definition adopted by the Institute of
Medicine (2001) as adapted from Sackett and colleagues
(2000, p. 14): “Evidence-based practice is the integration of
best research evidence with clinical expertise and patient
values.”
Best Research Evidence
 Evidence drawn from basic and applied research
 Hierarchy from clinical observation through
randomized controlled trials with respect to efficacy
 Address efficacy and effectiveness (utility)
 Absence of evidence not evidence of absence
 Untested does not mean ineffective
 Claims for efficacy should be tested
*American Psychological Association (2002). Criteria for evaluating treatment
guidelines. American Psychologist, 57, 1052-1059.
Clinical Expertise
 Encompasses number of competencies



positive therapeutic relationships
integration of diverse information
recognizes own bias and limitation
 Derived from clinical and scientific training
 Used to integrate research evidence with
clinical data in context of patient preference
Patient Characteristics, Values, and
Context
 Services most effective when responsive to
patient problems, strengths, and preferences
 Important variations in age, gender, race and
ethnicity, and culture (among others)
 EBPP seeks to maximize patient choice
among effective alternative interventions
Integration
 Psychologist determines applicability of research
evidence to particular patient
 Application of research to given patient always
involves probabilistic inferences
 Continuous monitoring of patient progress and
adjustment of treatment as needed
 Clinical decisions made in collaboration with
informed patient and in consideration of costs,
benefits, and options available (never by untrained
persons unfamiliar with specifics of the case)
In Defense of RCTs
 RCTs best way to detect causal influence

far from perfect but still the best we have



need not do therapy like a scientist to evaluate effects


uncontrolled trials confound patients and procedures
hormone replacement therapy just latest example
Carl Rogers one of the first to do controlled trials
good data best way to keep the critics at bay



no controlled trials before Eysenck’s critique
hundreds of subsequent trials show that psychotherapy works
some leading therapies still not adequately tested
In Defense of ESTs
 ESTs one reasonable way to see what works


look for well done studies that show effects
need not sacrifice external validity for internal validity



do not mandate specificity but can detect it



need not exclude representative patients (and no longer do)
can be used to test long-term treatments (and starting to do so)
treatment needs to work but not for reasons specified
special case for medications not for psychotherapy
treatment manuals neither necessary nor sufficient


need not constrain clinicians unduly if integrity maintained
merely useful aid for training and dissemination
Psychotherapy and Medications in the
Treatment of Unipolar Depression
(AHCPR)
60%
50%
40%
Dynamic?
IPT
CBT
30%
20%
Medications
10%
Placebos
0%
Site Differences in the TDCRP (High Severity
Completers N=42)
16
14
Placebo (n=13)
12
Drug (n=15)
CT (n=14)
10
Post HRSD
CT by Site:
Site 1 (n=3)
8
6
Site 2 (n=6)
Site 3 (n=5)
4
2
0
Treatment Condition
(CT by Site)
In Defense of ESTs
 ESTs one reasonable way to see what works


look for well done studies that show effects
need not sacrifice external validity for internal validity



do not mandate specificity but can detect it



need not exclude representative patients (and no longer do)
can be used to test long-term treatments (and starting to do so)
treatment needs to work but not for reasons specified
special case for medications not for psychotherapy
treatment manuals neither necessary nor sufficient


need not constrain clinicians unduly if integrity maintained
merely useful aid for training and dissemination
Must RCTs Exclude
Representative Patients?
 805 patients evaluated
 240 (30%) randomized
 565 (70%) excluded


235 (29%) low severity
240 (30%) diagnostic







96 (12%) psychosis
63 (08%) sub abuse
17 (02%) axis I
19 (02%) axis II
45 (06%) medical
95 (10%) med refusal
08 (01%) suicide risk
10%
6%
1%
2%
30%
2%
8%
12%
29%
randomized
low severity
psychosis
sub abuse
axis I
axis II
medical
med refusal
suicide risk
From DeRubeis et al., 2005
Must RCTs Exclude
Complicated Patients?
 240 patients randomized
 40 (16%) depressed only
 200 (84%) comorbid

146 (73%) axis I




127 (53%) anxiety disorder
86 (36%) sub abuse
40 (16%) eating disorder
125 (52%) axis II




axis I+II
30%
10 (04%) cluster a
10 (04%) cluster b
84 (35%) cluster c
37 (16%) pd nos
dep only
17%
axis I only
30%
axis II only
23%
From DeRubeis et al., 2005
In Defense of ESTs
 ESTs one reasonable way to see what works


look for well done studies that show effects
need not sacrifice external validity for internal validity



do not mandate specificity but can detect it



need not exclude representative patients (and no longer do)
can be used to test long-term treatments (and starting to do so)
treatment needs to work but not for reasons specified
special case for medications not for psychotherapy
treatment manuals neither necessary nor sufficient


need not constrain clinicians unduly if integrity maintained
merely useful aid for training and dissemination
CPT III
Acute Treatment
(1-18 months)
1st
R
a
n
d
o
m
i
z
a
t
i
o
n
ADM
and
CT
Maintenance/Follow-up
(36 months)
Continuation
(6-18 months)
ADM
(N=90+)
(twice weekly/weekly)
(monthly)
(N=225)
Response
Relapse
2nd
R
a
n
d
o
m
i
z a
t
i
o
n
No ADM
(N=90+)
Recurrence
ADM
ADM
(N=225)
(weekly/biweekly)
(N=90+)
(monthly)
No ADM
(N=90+)
Remission
(monthly/
quarterly)
Recovery
(monthly/
quarterly)
Time to Remission as a Function of Condition
(July 2005)
100%
79%
64%
60%
Comb 191
Meds 194
40%
20%
Weeks in Treatment
72
78
60
66
48
54
36
42
24
30
12
18
6
0%
0
% Remitted
80%
Time to Recovery as a Function of Condition
(July 2005)
100%
79%
69%
60%
Comb 116
Meds 102
40%
20%
Months in Continuation
18
16
14
12
10
8
6
4
2
0%
0
% Recovered
80%
Time to Recurrence as a Function of Condition
(July 2005)
100%
71%
Comb Main 37
Comb With 33
Meds Main 23
Meds With 23
60%
40%
41%
20%
19%
09%
Months in Maintenance/Follow-up
36
30
33
24
27
18
21
12
15
9
6
3
0%
0
% Recurrence
80%
Sustained Recovery as a Function of Condition
(July 05)
100%
80%
60%
No
Yes
40%
20%
0%
Comb Main
Comb With
Meds Main
Meds With
Treatment Condition
Sustained Recovery = pRemit x cpRecover x 1-cpRecurrence
In Defense of ESTs
 ESTs one reasonable way to see what works


look for well done studies that show effects
need not sacrifice external validity for internal validity



do not mandate specificity but can detect it



need not exclude representative patients (and no longer do)
can be used to test long-term treatments (and starting to do so)
treatment needs to work but not for reasons specified
special case for medications not for psychotherapy
treatment manuals neither necessary nor sufficient


need not constrain clinicians unduly if integrity maintained
merely useful aid for training and dissemination
Response to Treatment as a Function of Condition
70%
Placebo (n=60)
ADM (n=120)
60%
Percentage
CT (n=60)
50.0%
50%
57.5% 58.3%
43.0%
40%
30%
25.0%
20%
10%
0%
8 Weeks
16 Weeks
Prevention of Relapse and Recurrence Following
Successful Treatment
Continuation
Followup
0.8
Placebo (n=35)
Drug (n=34)
Compliant (n=30)
Prior CT (n=35)
0.6
0.4
0.2
Months (following active
treatment)
24
22
20
18
16
14
12
10
8
6
4
2
0
0
% Survival
1
Sustained Improvement
for All Assigned to Treatment
45%
39%
40%
30%
35%
30%
25%
20%
16%
15%
10%
5%
0%
ADM followed by
Placebo
ADM followed by ADM
CT followed by 3
additional sessions
Cumulative Direct Costs of ADM and CT
3000
Cost in Dollars
2500
2000
1500
CT
1000
ADM
500
0
1
2
3
4
5
6
7
8
9
10 11 12 13 14 15 16
Months in Treatment
In Defense of ESTs
 ESTs one reasonable way to see what works


look for well done studies that show effects
need not sacrifice external validity for internal validity



do not mandate specificity but can detect it



need not exclude representative patients (and no longer do)
can be used to test long-term treatments (and starting to do so)
treatment needs to work but not for reasons specified
special case for medications not for psychotherapy
treatment manuals neither necessary nor sufficient


need not constrain clinicians unduly if integrity maintained
merely useful aid for training and dissemination
Temple Study (Sloane et al. 1975)
Target Symptoms
12
10
8
Behavioral
Dynamic
Wait List
6
4
2
0
Pre
Post
Follow-up
Summary and Conclusions
 Multiple components contribute to outcome
 Patient, therapist, relationship also matter
 Validate treatments to improve patient care
 New methods emerge over time
 Pursue other ways to improve care
 Emphasize that which we can reliably teach
Putting Things in Perspective
 No one pretends that democracy is perfect or allwise...indeed, it has been said that democracy is the
worst form of Government except all those other
forms that have been tried from time to time –
Winston Churchill
 The first principle for being a good psychologist is
to not kid yourself, the second principle is to not kid
anybody else – Paul Meehl