Florence Busschbach

Cost-effectiveness
in Personality Disorder
• Dr. J.J.V. Busschbach
• Psychotherapeutic centre ‘De Viersprong’
– PO Box 7
4660 AA Halsteren
+31 164 632200
+31 164 632220 (fax)
[email protected]
• Erasmus MC, Rotterdam
– Department of Medical Psychology & Psychotherapy
• www.xs4all.nl/~jannetvb/busschbach
– Contains the slides of this presentation
1
The usual convention….
• Doubt about the cost-effectiveness
– Treatment of personality disorder is expensive
» Treatment is long
– Effect is low
– Cost-effectiveness is unfavourable
• How to deal with such stigma?
2
Stigma is not unique
• Typical for new interventions
– Especially new pharmacy
• Prozac is example
• Prozac was said to be
– More expensive
– As effective as old medication
» As established in RCT
– Therefore not a cost-effective alternative
3
Stigma versus science
• Reaction of Ely Lilly…
– Manufacturer of Prozac
– Two main arguments
• They questioned the randomised trial results
– The generalisability of results for clinical practice
– Introducing ‘Outcome Research’
• They questioned the assumption about higher costs
– Medication cost may be higher, but total cost may be lower
– Introducing ‘Health Economics’
4
Outcome Research
• Clinical research
– Does it work?
– Efficacy
– Perfect patient
» No co morbidity
• Randomized Clinical Trial
– Controlled conditions
• Outcome research
– Does it work in practice
– Effectiveness
– Every day patient
» Normal co morbidity
• Trials in a naturalistic
setting
– Real life conditions
5
In RCT no differences in efficacy…
• Between Prozac and old medication
– No differences between TCA and SSRI
• Citation British Medical Journal:
– “Randomised, controlled clinical trial (RCTs ) generally show equal
efficacy among antidepressants”
– Song F et al. BMJ, 1993;306:683-7
6
But in outcome research…
• In practice: much better effectiveness
• Drop out ration TCA : SSRI = 3 : 1
– Lobowitz, JAMA 1997;278:1186-90
• After drop out, recurrence depression 2 to 4 time higher
• Minimal effective dose
– SSRI 98% (Prozac)
– TCA 61%
– N = 23000, General Practitioner
» De Waal et al, NTVG 1996;140:2131-4
• Randomised trials mask differences compliance!
– Outcome research reveals remarkable results
7
Health economics
• Simon et al, JAMA 1996;275:1897-902
• Six-month health care expenditures
– Total cost, not just medication costs
• Compared
– Desipramine: N = 181
» Old TCA
» $ 2361
– Imipramine: N = 182
» Old TCA
» $ 2105
– Fluoxetine N = 173
» New SSRI: Prozac
» $ 1967
• No statistical significant differences
8
Regression in quasi-experiment
controlled for sex, age, prior-period expenditures etc.
$400
313
Regression weights TCA vs SSRI
$300
208
187
$200
$100
41
36
2
$$-100
$-200
Pre
sc
Ph
rip
t
i on
ysi
c
Ps
i an
ych
La
i at
ris
t
bo
rat
o
Ho
sp
ry
ita
Ps
l
ych
i at
-162
Sclar et al, 1994
N = 701
To
ta
ric
al
ho
co
sp
sts
ita
l
9
What can we learn?
• Randomised trials are not the holy grail
– They do serve in efficacy
– But there are higher order measurements
• Effectiveness
– Outcome research
» Randomised trials AND naturalistic studies
» Quasi experimental design
• Cost-effectiveness
– Health economics
» Randomised trials AND naturalistic studies
» Quasi experimental design
10
Where do we stand?
• Favourable results in (randomised) trials
– Psychotherapy versus usual care
– 6 Reviews en 1 meta analysis
» Perry et al, Am J Psychiatry 1999;57:1312-21
• What about cost effectiveness….?
– …is psychotherapy in personality disorder worth the costs?
1.4
1.2
Effect size
1
0.8
0.6
0.4
0.2
11
0
Self-report
Observer-rating
Existing evidence suggests considerable
savings
• New investigations
– Bateman, Fonagy, Am
J Psychiatry
2003;160:169-71
• Reviews
– Gabbard et al. Am J
Psychiatry
1997;154:147-50
$60,000
$50,000
Experimental group
$40,000
General Care
$30,000
$20,000
$10,000
$0
6 months
before
treatment
18 months of
treatment
18 month
follow-up
12
Problem in cost effectiveness results
• Cost estimates made in trial environment
– No ‘real’ cost estimates
– No adjustment made for trial situation
• No formal cost-effectiveness study designs
• Typical elements are missing
– Discounting
» Costs and effects in the future are valued lower
– Generic outcome measures
» Quality adjusted life years (QALYs)
» Disease specific outcome do not allow for comparisons between
different allocations in health care
13
What do we need?….
• Naturalistic trial
– To prove the effects in practice
– To estimate costs in practice
• Formal cost-effectiveness study
– Following international guidelines
14
Sceptre hopes to fulfil these demands
• Quasi experimental trial in a naturalistic setting
– Introducing outcome research
• The design follows standards in health economics
– Introducing health economics
• But even more than Sceptre we need….
15
Confidence
• Good treatment will be cost effective
– If a treatment works in practice, it will almost certainly be costeffective
– Like Prozac
• In that conviction we need to put our treatments to the
test….
16