Transcript Responder - CTN Dissemination Library

Adaptive Treatment Strategies
in the Addictions:
Current Examples and Future Directions
James R. McKay, Ph.D.
Professor of Psychology in Psychiatry
University of Pennsylvania
CTN Meeting
3.22.07
Overview of Presentation
• Major problems in providing addiction treatment
and how we’ve tried to address them
• Adaptive treatment models and how they are
developed
• Examples of adaptive treatment in specialty care
• Examples of adaptive treatment in other treatment
settings
• Challenges in designing and implementing
adaptive treatment protocols
Problems in Addiction Treatment
• High rates of dropout and continued alcohol
and drug use
– In community-based programs
– In research protocols
• Even with evidence-based treatments,
considerable response heterogeneity
Attempts to Address Nonresponse?
• Improve existing treatments
• Develop new treatments
• Tailoring, or “matching” treatments to
subgroups of patients
Results???
Still left with variable response…..
• Even when treatment delivery is standardized and
high adherence to manual is achieved, some
patients do well and others do not.
• Very hard to predict who will do well in a
particular treatment
• Some patients do well at first, but then deteriorate
• Nonresponse often blamed on the patient, but that
is likely not the whole story.
Another Possible Approach?
Adaptive Treatment
In Adaptive Treatment Protocols…
• Treatment is tailored or modified on the basis of measures
of response (e.g., symptoms, status, or functioning)
obtained at regular intervals during treatment
• Goal is to deliver the treatment that is most effective for a
particular patient at a particular time.
• Rules for changing treatment are clearly operationalized
and described…..
“If……..Then”
• Temporal issues important– when has sufficient time
elapsed to indicate “non-response”?
How Do You Put Together an
Adaptive Protocol?
Experimental Design for Developing
Adaptive Protocols
• Use randomization to develop optimal adaptive treatment
strategies
– Example: What to do with early non-responders?
• Switch treatment?
• Augment treatment?
• Determine the set of decision rules and interventions that
produce the highest percentage of responders
THEN…….
• Compare the optimal adaptive protocol to TAU or other
treatments in standard RCT
The alternative approach….
• Devise adaptive protocol on the basis of:
–
–
–
–
Expert clinical judgment
Feedback from patients
Prior research findings
Face validity
• Compare that adaptive protocol to TAU or other
treatment in standard RCT
• Pros and Cons: Faster than experimental
approach, but protocol may be flawed
Examples of Adaptive Protocols
from Addiction Specialty Care
Recovery Management Checkups
• Protocol developed by Dennis, Scott et al.
– Interview patients every quarter for 2 years
– If patient reports any of the following……
•
•
•
•
•
Use of alcohol or drugs on > 2 weeks
Being drunk or high all day on any days
Alcohol/drug use led to not meeting responsibilities
Alcohol/drug use caused other problems
Withdrawal symptoms
….. ….Patient transferred to linkage manager
RMC
• Linkage Manager provides the following:
–
–
–
–
–
Personalized feedback
Explore possibility of returning to treatment
Address barriers to returning to treatment
Schedule an intake assessment
Reminder cards, transportation, and escort to
intake appointment
Results: RMC vs. TAU
• Time to return to treatment
376 vs. 600 days (p< .05)
• Total days of treatment
62 vs. 40 days (p< .05)
• In need of treatment at 24 months
43% vs. 56% (p< .01)
• In need of treatment in at least 5 quarters
23% vs. 32% (p< .05)
Dennis et al. (2003) Evaluation and Program Planning, 26, 339-352
Adaptive Methadone Treatment
• Brooner & Kidorf (2002) protocol
– Methadone patients start in low intensity
psychosocial condition
– Missed session or dirty/missing urine leads to
increases in psychosocial counseling
– Providing additional contingencies for
participation further improves outcomes
• More/less convenient dosing times
• Methdone taper and possible discharge
Penn Telephone Continuing Care Study
• Patients:
– 359 graduates of 4-week IOP programs
– Cocaine (75%) and/or alcohol (75%) dependent
• Continuing care treatment conditions (12
weeks):
– Standard group counseling (STND)
– Individualized relapse prevention (RP)
– brief telephone-based counseling (TEL)
McKay et al., 2004, Journal of Consulting and Clinical Psychology
Continuing Care Conditions
• Telephone Monitoring and Counseling
– Weeks 1-4, patients make a 15 minute call and attend a
“transition” group (1x/week @)
– Weeks 5-12, patients have telephone contact only
(1x/week)
– During calls, patients report results of self-monitoring
and progress toward 1-2 goals, and plan goals for next
week
– Patients use a workbook that structures intervention for
each week.
– Total minutes of contact with therapist 50% of minutes
in other conditions
Total Abstinence Rates
80
70
% Abstinent
60
50
STND
RP
40
TEL
30
20
Tx Main Effect
TEL > STND
p< .05
10
0
3
6
9
12
15
Month
McKay et al., 2005, Archives of General Psychiatry
18
21
24
Adaptive Treatment Strategy:
Using Progress in Initial Phase of
Treatment to Select Optimal
Continuing Care Models
7-Item Composite Risk Indicator
• Failure to achieve key goals while in IOP:
–
–
–
–
–
–
Any alcohol use in prior 30 days
Any cocaine use in prior 30 days
Attendance at < 12 self-help meetings in prior 30 days
Social support < median for the sample
Does not have goal of absolute abstinence
Self-efficacy < 80%
• Current dependence on both alcohol and cocaine
(each item: yes=1, no=0)
McKay et al., 2005, Addiction, Archives of General Psychiatry
Distribution of Scores on the
Composite Risk Indicator
Number of Participants
120
100
Mean score= 2.50
80
60
40
20
0
0
1
2
3
4
5
Composite Risk Indicator Score
6
7
TEL vs. STND contrast X Risk Index Score: p < .05
Figure #1 Predicted Aggregated and Compressed Log "Bad" Costs
Model: FGRBadcostlog = Txcond Sequence FGRBLBadcostlog Program
Txcond*FGRBLBadcostlog
10
STD
RP (Not Significant)
TR
3-month Follow-up "Bad" Costs (log)
8
≥ $830; 27%
(N = 98)
Crossover point STD vs. TR:
"Bad" costs $830
(6.72 on log scale)
6
< $830; 73%
(N = 261)
4
2
Most favorable interventions
Most favorable
baseline level
Least favorable
baseline level
0
0
2
4
6
Baseline "Bad" Costs (log)
8
10
Extended Telephone-Based
Adaptive Protocol for the
Management of Cocaine
Dependence
Design
• Patients: Cocaine dependent IOP participants
recruited after achieving early engagement
• Treatment conditions:
– Treatment as usual (TAU)
– TAU plus adaptive protocol (24 mo.)
– TAU plus adaptive protocol (24 mo.), plus incentives
for participation and cocaine-free urines (12 mo)
• Outcomes assessed over 24 months
The Telephone Calls
• Frequency: weekly at first, titrated to bimonthly
• Each call starts with a brief “risk assessment” that assesses
negative and positive factors and yields overall risk score
(low, moderate, high)
• Similar protocol to prior study for telephone counseling:
1.
2.
3.
4.
5.
6.
7.
Provide feedback on risk level
Review progress/goals from last call
Identify upcoming high-risk situations
Select target for remainder of call
Brief problem-solving regarding target concern(s)
Set goal(s) for interval before next call
Suggest change in level of care if warranted
Adaptive Protocol
• Increases in services triggered when risk
reaches moderate level
– First: increase frequency of phone calls
– Second: bring patient in for 1-2 face-to-face
evaluation and motivational interviewing (MI)
sessions
– Third: provide 8 CBT relapse prevention
sessions
– Fourth: refer back to IOP
Examples of Adaptive Protocols
from Non-Specialty Addiction
Care
Adaptive Primary Care Protocols for
Heavy Drinkers
• Kristenson et al. (1983, 2003)
– Patients randomized to visits with a nurse (every
month) and physician (every 3 months), vs. TAU
– Both provided for up to 4 years
– GGT levels monitored, and treatment/drinking goals
modified on basis of scores
– Results: fewer sick days, fewer hospital days, lower
mortality over 6 and 16 years than TAU
Adaptive Continuing Care
Naltrexone Protocol
• O’Malley et al. (2003) study of NTX treatment
comparing primary care (PC) and specialty care
(CBT) approaches
• First, pts given NTX and randomized to PC or
CBT for 10 weeks
• Responders (57%) further randomized:
– PC plus extended NTX vs. placebo (24 wks)
– CBT plus extended NTX vs. placebo (24 wks)
Alcohol Use Results and Interpretations
• Findings:
– Initiation phase: PC=CBT
– Extended PC phase: NTX > placebo
– Extended CBT phase: NTX= placebo
• Resulting treatment algorithm
– If patient responds to PC and NTX in first 10 weeks,
continue both for at least 24 more weeks
– If patient responds to CBT and NTX in first 10 weeks,
continue CBT but stop NTX
• Note: no guidance regarding nonresponders
Adaptive Naltrexone Study
(David Oslin, PI)
• Experimental design to determine optimal algorithms for
naltrexone responders and nonresponders
• All patients begin with 8 week trial of open label
naltrexone, plus weekly medication management session
• During the 8 week trial, patients self-select into Responder
and Non-responder groups
• First randomization: Different definitions of “nonresponse”
– More than 1 heavy drinking day
– More than 4 heavy drinking days
Adaptive Naltrexone, cont.
Second Randomization
• Nonresponders:
– Add CBI and drop NAL (i.e., “switch”)
– Add CBI and continue NAL (i.e., augment”)
• Responders:
– NAL script plus no further care
– NAL script plus telephone disease management
Adaptive Intervention Strategies
Embedded in Oslin Trial
Adaptive
intervention
Definition of
nonresponder
Decision rules for
responders
Decision rules for
nonresponders
1
> 1 heavy drinking day
Stay with NTX alone
NTX with CBI
2
> 1 heavy drinking day
Stay with NTX alone
Change to CBI alone
3
> 1 heavy drinking day
NTX with TDM
NTX with CBI
4
> 1 heavy drinking day
NTX with TDM
Change to CBI alone
5
> 4 heavy drinking day
Stay with NTX alone
NTX with CBI
6
> 4 heavy drinking day
Stay with NTX alone
Change to CBI alone
7
> 4 heavy drinking day
NTX with TDM
NTX with CBI
8
> 4 heavy drinking day
NTX with TDM
Change to CBI alone
Comparing Definitions of Response
Adaptive
intervention
Definition of
nonresponder
Decision rules for
responders
Decision rules for
nonresponders
1
> 1 heavy drinking day
Stay with NTX alone
NTX with CBI
2
> 1 heavy drinking day
Stay with NTX alone
Change to CBI alone
3
> 1 heavy drinking day
NTX with TDM
NTX with CBI
4
> 1 heavy drinking day
NTX with TDM
Change to CBI alone
5
> 4 heavy drinking day
Stay with NTX alone
NTX with CBI
6
> 4 heavy drinking day
Stay with NTX alone
Change to CBI alone
7
> 4 heavy drinking day
NTX with TDM
NTX with CBI
8
> 4 heavy drinking day
NTX with TDM
Change to CBI alone
Comparing Augment vs. Switch for
NonResponders
Adaptive
intervention
Definition of
nonresponder
Decision rules for
responders
Decision rules for
nonresponders
1
> 1 heavy drinking day
Stay with NTX alone
NTX with CBI
2
> 1 heavy drinking day
Stay with NTX alone
Change to CBI alone
3
> 1 heavy drinking day
NTX with TDM
NTX with CBI
4
> 1 heavy drinking day
NTX with TDM
Change to CBI alone
5
> 4 heavy drinking day
Stay with NTX alone
NTX with CBI
6
> 4 heavy drinking day
Stay with NTX alone
Change to CBI alone
7
> 4 heavy drinking day
NTX with TDM
NTX with CBI
8
> 4 heavy drinking day
NTX with TDM
Change to CBI alone
Summary of Possible Adaptations
• Non-responders
–
–
–
–
Step up (e.g., OP to IOP or residential)
Lateral move (e.g., CBT to TSF)
Modality change (e.g., CBT to medication)
Step down (e.g., IOP to telephone monitoring)
• Responders
– Reduce frequency of intervention (e.g., IOP to OP)
– Change to lower burden intervention (e.g., OP to
periodic check-ups, or e-treatment)
Adaptive Treatment and the
CTN:
Difficult Problems…………..
But Big Opportunities
and Potential Benefits
Challenges in Adaptive Treatment
Clinical
• Keeping patients engaged, especially when
deterioration occurs
• Increasing compliance with adaptive changes,
especially “step ups”
• Identifying alternative treatments for nonresponders
– Lack of a variety of effective medications
– Are different types of “talk” therapy really different
enough?
– How important is patient preference/choice?
Challenges, cont.
Research
• Incorporating choice in algorithms
– Comparing heterogeneous condition to other
interventions
• Sequential randomization designs
– Randomizing patients 2+ times
– Analytic issues (first decision)
• Power
– Primary vs. secondary comparisons
– New methods under development
Focus of Efforts in Treatment
Development
• Emphasis in field has been on improving efficacy
and adherence to manuals, and coming up with
more cost-effective approaches.
• Shift emphasis to making participation more
attractive to the patients to improve retention:
– Greater weight to patient choice– at intake, and for nonresponders
– Use of more convenient forms of care whenever
possible
– Incentives for participation?
Possible Research Designs
• Adaptive strategies to address early dropout
– Test providing a menu of treatment options vs. efforts
to re-engage in standard care
“So you don’t like IOP. How about…….?”
• Adaptive medication algorithms
– Start with promising med– augment with or switch to
additional medication for nonresponders
Research Designs, cont.
• Adaptive studies that combine behavioral and
pharmacological interventions:
– Start with medication and low intensity behavioral
treatment, step up to more intensive treatment if no
response
– Offer non-responders sequential package that first
involves switching meds, but then includes
augmentation with stepped up behavioral treatment if
response still not achieved.
Acknowledgments
• Colleagues:
– NIDA CTN algorithms group
– Dave Oslin, Kevin Lynch, Tom TenHave
– Susan Murphy, Linda Collins
• Grant support:
– NIDA: K02-DA00361, R01-DA14059, R01DA20623
– NIAAA: R01AA14850