(Cost-)Effectiveness of Psychotherapy for Personality Disorders

Jan van Busschbach

Prof. Dr. J. van BusschbachDepartment of Medical Psychology and Psychotherapy Erasmus MC PO Box 2040  3000 CA Rotterdam  +31 10 7043807 (direct: 7044306)J.vanbusschbach@erasmusmc.nl

An different mindset: Comparing “dosages”

Usually ...– Comparison between theoretical orientation of therapy

Typically ...– Amount of therapy is keep constant

This assumes ...– amount of therapy is relevant

Little differences– Nonspecific factors seems to drive treatment success

Amount of therapy relates to costs Yet ...

– Relation between costs and effects is rarely investigated

SCEPTRE

Study on Cost-Effectiveness of Personality Disorder Treatment

Start: March 2003 6 clinics

SCEPTRE

About 900 patient with PD Followed over 3 years Dosages compared

– Outpatient, day-hospital and inpatient psychotherapy– Shorter than or equal to 6 months, longer than 6 months

Clusters– A; N = 58– B; N = 241– C; N = 466

Naturalistic design

In need of a super covariate

Question to clinician: – “What are the important variables for treatment allocation?”

Answer:– “Everything is important!”

How to control for everything?– “We are in need of a super covariate”

Correction for selection bias

Propensity score– A sophisticated co-variance analysis– Combines several co-variates– To correct for baseline differences

If successful– Results can be interpreted as an RCT

Several checks on validity Often used in

– (health) economics– Epidemiology

Super Covariate: the propensity score

Age Sex Diagnosis (SIDP-IV) Baseline GSI Motivation Measures of pathology

– DAPP-BQ; SIPP; OQ-45 Quality of life (EQ-5D)

Can super covariate fly?

Separate PhD, 2010

Medical Care, 2010

Psychotherapy and Psychosomatics, 2009

Cluster A: one of the largest studies ever

Bartak, et al. Effectiveness of outpatient, day hospital, and inpatient psychotherapeutic treatment for patients with cluster A personality disorder. Accepted for publication Psychotherapy and Psychosomatics

But assumptions are not met in cluster A

Assumptions met in:

3 groups in cluster B– Inpatient– Day-hospital– Outpatient

5 groups in cluster C– Short-term inpatient– Long-term inpatient– Short-term day-hospital– Long-term day-hospital– Long-term out-patient

Results cluster B

Bartak et al. Effectiveness of outpatient, day hospital, and inpatient psychotherapeutic treatment for patients with cluster B personality disorder. Psychotherapy and Psychosomatics, in press

But no significant results in cluster B…

Differences diminish till P = 0.06– After correction with the propensity score– Complicates conclusions

Assumptions of propensity score are met– Effect are reduces after correction

But costs could make the difference…

Results cluster C

Corrected rsults CGSI - Difference score

Treatment group

Long outpatient

Short day hospital

Long day hospital

Short inpatient

Short day hospital

-0.0770

Long day hospital

-0.1278 -0.0508

Short inpatient 0.3035 0.3805** 0.4313**

Long inpatient -0.0030 0.0740 0.1247 -0.3065*

* p < 0.05 ** p < 0.01 *** p < 0.001

Propensity escore in cluster C

Better effects of short-term inpatient psychotherapy remain significant

Assumptions propensity score are met– Results maintain

But costs could still make a difference…

Conclusions: effects

No comparison possible in cluster A– But psychotherapy seems to work– Inpatient / day hospital seems better

Non difference in B (after correction)– But costs can be decisive ….

Cluster C– Favorable results for short-term inpatient psychotherapy– Expect to dominates long in-patient– But is short-term inpatient worth the costs?

• Compared to long day hospital / short day hospital

QALY

Health economics addresses the efficient allocation of health care resources

For instance– Psychotherapy “long” versus “short”– “Psychotherapy in PD” versus “Care for diabetics”

Make effects comparable – Same effect parameter in diabetes as in PD

Survival and Quality of Life Combined: Quality Adjusted Life Years (QALY)

19

QALY

Quality Adjusted Life Years Area under the curve

0

0,1

0,2

0,3

0,4

0,5

0,6

0,7

0,8

0,9

1

0 10 20 30 40 50 60 70 80Life years

Adj

usm

ent f

acto

r QA

LY

Co-morbidityWith psychotherapyNo psychotherapy

20

EQ-5D MOBILITY

– I have no problems in walking about – I have some……. – I am confined to bed

SELF-CARE– I have no problems with self-care – I have some problems….. – I am unable…

USUAL ACTIVITIES– I have no problems with performing my usual

activities – I have some problems… – I am unable….

PAIN/DISCOMFORT– I have no pain or discomfort – I have moderate ….. – I have extreme……..

ANXIETY/DEPRESSION– I am not anxious or depressed – I am moderately……..– I am extremely…..

The EuroQol EQ-5D is specially designed to measure the quality of life index for QALYs

21

Burden of disease: EQ-5D

0 0,2 0,4 0,6 0,8 1

Major depression

Heamodialysis

Rheumatic disease

Personality disorder

Lung cancer

Parkinson

Diabetes II

Schizophrenia (treated)

HIV

Normal population

Soeteman et al. Journal of Personality Disorders, 2008;22:259-68 . Soeteman et al. Psychiatric Services, 56, 1153-1155, 2005

Markov model Cluster B

Soeteman et al. Cost-effectiveness of psychotherapy for cluster B personality disorders. British Journal of Psychiatry 2010;196:396–403.

Costs and effects in Cluster B

Much difference

Much difference

Little difference

Cost per QALY

Costs and effects Cluster C

25

Soeteman et al. Cost-effectiveness of psychotherapy for cluster C personality disorders. Journal of Clinical Psychiatry (In Press)

Cost effectiveness Cluster C

Conclusion

Cost-effective treatment strategies are: Cluster C PD:

– Short-term inpatient psychotherapy (first choice)– Short-term day hospital psychotherapy– Sub-optimal treatment options are:

• Long-term day hospital and long-term inpatient Cluster B PD:

– Outpatient psychotherapy (first choice)– Day hospital psychotherapy– Sub-optimal treatment option is:

• Inpatient psychotherapy

27