**SSE/EFI Working Paper Series in Economics and Finance**
# No 459:

Cost-utility analysis of Interferon Beta-1b in the treatment of different types of Multiple Sclerosis

*Gisela Kobelt, Linus Jönsson, Sten Fredriksson *()

* and Bengt Jönsson *()

**Abstract:** Background

Economic evaluation of treatments in
multiple sclerosis (MS) presents a challenge. The disease affects a number
of different body functions and leads to severe disability over time,
without however a strong effect on mortality. At onset, the majority of
patients will have relapsing-remitting disease (RRMS) and will then convert
to secondary-progressive disease (SPMS) overtime. However, the course of
the disease is unpredictable, and the conversion to SPMS can take place at
different times since onset and at different levels of disability for
different patients. Relapses appear to occur with the same frequency at all
levels of disability, but will diminish over time. The effectiveness of
treatments can be measured in different ways such as disease activity, the
number and the severity of relapses or the progression of functional
disability, regardless of the type of MS.

However, improvements in
outcome achieved over a short term may have an effect on the disease in the
longer term, and effectiveness data from clinical trials must therefore be
extrapolated to the longer term, using modelling techniques. This requires
good epidemiological data on the natural course of the disease, where
disease progression is expressed with the same measures as in the clinical
trials. Also to perform economic evaluations, a global outcome measure is
required to capture the impact of treatments on the disease and the most
frequently used such measure is quality-adjusted life years (QALYs).
However, for QALYs to be used in cost-effectiveness analysis of MS, they
must be related to a measure of the disease and disease progression. The
Expanded Disability Status Scale (EDSS) provides a good measure of the
disease and has been widely used in epidemiological studies and clinical
trials, in all types of MS. Lastly, detailed economic data that can be
related to the different levels of disability (EDSS) are required.

Objective

We have earlier proposed a basic framework for
cost-effectiveness modelling in MS, and the original model has been
updated, as new data have become available. The current study proposes a
further development of the modelling technique and estimates the
cost-effectiveness of treatment with interferon b-1b (IFNB-1b) in a defined
patient population with active disease, both RRMS and SPMS, from a societal
perspective in Sweden.

Methods

The framework of the earlier Markov
model is used, where states are defined according to EDSS. Transition
probabilities for the first years in the model are calculated from clinical
trial data, and for the extrapolation from a large epidemiological database
on the natural history of MS. In view of the fact that the number of
relapses at given levels of disability did not differ between patients with
RRMS or SPMS in any of the three datasets used in this analysis, and that
conversion from RRMS to SPMS did not occur at well defined levels of
disability, we combined data from two large clinical trials in RRMS and
SPMS. Patients were selected on whether or not they had active disease at
enrolment, defined as an increase in the EDSS by at least 1 point (0.5
points for scores between EDSS 6 and 7) or at least 2 relapses in the
preceding 2 years. This allows simulating treatment start at any stage and
for any type of the disease and estimating long-term consequences within
the same model. The combination of the two types of MS is further supported
by the fact that it has been shown in 3 observational studies that costs
and quality of life at given EDSS levels are not different for patients
with different types of the disease.

Transition probabilities between
the Markov states are estimated for both the clinical trial and the natural
history cohorts using an ordered probit model. Transitions thus depend on
several factors, including what state a patient is in, whether or not
she/he has a relapse, age, age at onset of the disease, time since onset of
the disease, age at treatment start.

The base case simulations use mean
costs and mean utilities in each state from a large observational study in
Sweden. However, the model allows calculating acceptability curves, i.e.
the probability with which the cost effectiveness ratio of a treatment
scenario is below given levels of willingness-to-pay for a QALY, using the
entire distribution of costs and utilities at each EDSS level. Costs and
benefits are discounted with 3%.

Results

The base case assumes
treatment with IFNB-1b during 36 months, with no further effect when
treatment is stopped, and includes both patients with active RRMS and SPMS.
Sensitivity analysis is presented for treatment during 54 months. The
annual cost of IFNB-1b treatment was 102 587 SEK plus 1600 SEK for special
monitoring, and was adjusted for compliance in the clinical trial. In the
base-case treatment adds 13 000 to costs over 10 years, and the cost per
QALY gained is 71 400 SEK. When the time horizon is increased to 15-25
years, treatment dominates no treatment (higher utility and lower cost).
With treatment during 54 months, the cost per QALY is 353 800 SEK, all
costs included. When treatment is started early, the cost-effectiveness
ratio is higher, e.g. 643 100 SEK in state 2, as patients in these states
progress only very slowly.

In the net benefit approach, there is a 80%
probability that the treatment initiated in states 3 or 4 (EDSS 4.0-5.5) is
cost-effective, if the willingness to pay for a QALY is 400 000 to 600 000
SEK. At that level of willingness to pay, the probability in state 2 is
45%.

Conclusions

With this new model, which combines active RRMS
and SPMS, the effect of early treatment on the long-term outcome can be
estimated for the first time using patient-level clinical data for RRMS and
SPMS, as well as natural history data. The combination of the two types of
MS into one model is supported by the finding that, at given levels of
EDSS, there was no difference in the number of annual relapses in the three
clinical datasets used, nor in the mean cost and mean utilities in the
observational study. The model is more flexible than previous models, as it
includes individual patient demographics and the entire distribution of
costs and utilities in the different states. It thus represents a valuable
tool to estimate the cost-effectiveness of treating different patient
groups with IFNB-1b.

**Keywords:** MS; Cost-effectiveness; (follow links to similar papers)

**JEL-Codes:** I10; I19; (follow links to similar papers)

41 pages, July 28, 2001

Before downloading any of the electronic versions below
you should read our statement on
copyright.

Download GhostScript
for viewing Postscript files and the
Acrobat Reader for viewing and printing pdf files.

**Full text versions of the paper:**

hastef0459.pdf (432kB)

Download Statistics

Questions (including download problems) about the papers in this series should be directed to Helena Lundin ()

Report other problems with accessing this service to Sune Karlsson ()
or Helena Lundin ().

Programing by

Design by Joachim Ekebom