Is it CostEffective to Pay People to Stop Using Illicit Drugs

1
Is it Cost-Effective to Pay People to Stop Using
Illicit Drugs?

• Jody Sindelar, PhD, Yale
• Todd Olmstead, PhD, Yale
• Nancy Petry, PhD, UCONN
• October 25, 2005
• We acknowledge financial support from the
  National Institute on Drug Abuse (NIDA
  RO1-DA14471). Thank CTN esp Dr. Stitzer.

2
Overview

• Cost effectiveness analyses (CEA) of a CTN
  effectiveness trial
• Multi-site trial
• Low cost, prize based Contingency Management (CM)
• Focus on policy implications
• Note- DRAFT not for circulation

3
How we add to the literature.

• One of the first CEA of CM,
• CM an important intervention, we provide critical
  policy relevant info- (LR goals of our research
  agenda is to further study CM)
• CM Perfect for CEA
• Add acceptability curves- new method for SAT-
  gives policy relevant info, accounts for
  uncertainty
• Other strengths, large sample, multi-site

4
Background NIDA Clinical Trials Network (CTN)

• Network of researchers and providers conduct
  trials to assess effectiveness of promising TX
• Community based settings
• Multiple sites, geographically disperse, more
  generalizable
• CM was one of the first selected
• Two CM trials- this DF, companion MM

5
Background CM reinforces behavior with tangible
incentives

• CM has been found to be effective in previous
  literature
• Reinforcing desirable behavior- abstaining.
• Escalating payments
• Often pay vouchers (eg 2)
• Previous TX for illicit drug use were relatively
  expensive, adds costs on to usual care
• As high as 1000 paid to successful patients

6
Intermittent CM used in these CTN studies

• Draw prizes from an urn if drug free
• Is a relatively low cost CM by using intermittent
  reinforcement (Petry)
• Not all draws earn a prize
• Can vary the expected value of prizes earned
  conditional on being drug free
• Can ask, how much do you need to pay
• (Petry et al effectiveness Sindelar, Petry,
  Elbel, CEA)

7
Effectiveness Study in brief

• (Drs. Petry, Pierce, Stizer and CTN)
• Design Random assignment of 412 stimulant
  abusers to UC or UCCM for 12 weeks
• Setting 8 community-based outpatient
  psychosocial SAT programs
• Test for- stimulants
• Primary outcome measures- Retention, counseling
  attendance, neg urines, longest duration
  abstinent

8
Incentives Intermittent reinforcement

• Chance to win prizes for stimulant-negative
  samples (2X per week)
• draws earned increases with continuous
  abstinence
• draws resets to zero with positive or missing
  sample
• 500 chips in urn
• 250 (50) good job gt 0
• 209 (41.8) small prize gt 1
• 40 (8) large prize gt 20
• 1 (0.2) jumbo prize gt 100

9
Effectiveness Study Findings

• Those in CM arm have better outcomes
• LDA, retention, counseling attendance, neg
  urines
• ( But, percentage negative was low overall but
  not different by UC and CM we think that this is
  a poor measure of success!)

10
It is effective, but

• Is it cost-effective?
• Do you get your moneys worth?
• Should society pay for adding CM?

11
Our CEA study

• Objective Evaluate the cost-effectiveness of
  the prize-based intervention (CM) added to usual
  care (UC)
• Secondary- to look at site differences, but not
  report on here.

12
Methods

• Calculate incremental cost-effectiveness ratios-
  (ICERs)
• Change in costs of adding CM/ change in
  effectiveness gained
• Use trial data on effectiveness
• Outcomes
• Longest duration abstinent in study (LDA weeks)
• of negative urine samples in study
• Length of stay in study (LOS weeks)
• Collect data on unit costs, calculate incremental
  costs unit cost resources used Conduct a
  survey.

13
Cost categories

• Counseling (session time admin time)
• Individual
• Group
• Family
• Testing (materials time)
• Urinalysis
• Breathalyzer
• Prizes
• Drawing time
• Value of prizes themselves
• Administration time to run the prizes system , eg
  stocking time

14
Clinic cost survey

• Survey 8 clinics (14 between the two trials)
• Aim to obtain unit prices of counseling, testing,
  and prize admin
• RA administered the survey, asked key people eg
  CEO, CFA, Medical director
• Paid clinic 100 for completed survey

15
Cost data and calculations

• Ask questions such as, hourly wage rate of
  counselors, fringe benefits no. of clients in a
  group session admin time re session admin time
  of running the prize system, etc
• Calculate unit costs
• Multiply number of units of inputs by unit costs
• Derive cost of variable inputs to UC and CM
• Calculate incremental costs (and effects), ICERs

16
Results Overall
p-value lt .1 p-value lt.05 p-value lt.01
17
Interpretation of results

• Find testing costs are high.
• Prizes add costs too. Is it worth it?
• How to interpret 231 more per additional week of
  consecutive abstaining (LDA)? Worth it?
• No thresholds available.

18
Acceptability curve

• Provides policy relevant interpretation of
  results
• Provides measure of uncertainty (Difficult to
  calculate s.e. of ICER denominator may be 0)

19
Acceptability curve-how to

• Bootstrap 1000 replicates from sample
• Consider correlation of changes in incremental
  effects and outcomes
• Scatterplot of ICERs of 1000 replicates
• Plot acceptability curves

20
Acceptability curves

• Plot of
• Probability that the ICER that is found is
  acceptable at a range of societys willingness to
  pay (WTP)
• Problem is that we do not have a measure of
  societys maximum WTP for a given outcome in SAT
• (use QALYs in other areas not good for SAT as
  not include extranalities- crime, spread of
  disease, work)

21
Results Acceptability Curve LDA
Overall-percent and WTP

22
Interpretation

• If society is WTP about 270 per extra
  consecutive week abstaining, then it is 90
  likely that society should accept the additional
  expense of CM- used in this way
• As society is WTP more for the outcome, the
  probability of acceptance increases.

23
Next steps

• Derive some bounds of WTP per extra week, eg.
  consider values of reduced crime, spread of
  AIDS/HIV due to longer abstaining. Is societys
  WTP 270 or more?
• Sensitivity analysis (eg price of testing,
  running the prize system at full levels)

24
Also,

• Examine difference by site with goal of
  understanding how to interpret for policy
  purposes.
• Glad to have comments, suggestions, CM is a cont.
  interest of our research.

25
Strengths and weaknesses

• Strengths-
• Large sample, multi-site thus generalizable,
  community based, trial implies causality, one of
  the first CEA of CM, CM an important intervention
• Weaknesses-
• Missing obs, Need longer follow-up, patient costs
• CBA instead?, need more compete data, crime,
  spread of AIDS/HIV, etc

26
Future analyses

• Sensitivity analyses robust to different
  assumptions- tests are dropping in price
• What if it were operating at full capacity- costs
  would decline
• What to make of site differences- possible policy
  conundrums

27
Further work

• Analyze CEA of other CM trials, determine what
  accounts for variability/ develop thresholds
• Analyze policy options
• More comprehensive outcomes- crime, drugs

28
Results Acceptability Curves by Site
29
p-value lt .1 p-value lt.05 p-value lt.01