Dear Listmembers.
I would like to draw your attention to the following seminar to be
held in Leicester.
SEMINAR IN LEICESTER
There will be a seminar at 4.30 pm on Thursday,
9th December 1999 in Room G20, Department of
Epidemiology and Public Health, 22-28 Princess
Rd West, Leicester.
Dr Karl Claxton from the Department of Economics
and Related Studies, University of York,
will speak on:
'THE EFFICIENT DESIGN OF CLINICAL TRIALS:
AN APPLICATION TO THE EVALUATION OF TREATMENT
STRATEGIES FOR ALZHEIMER'S DISEASE.'
All are welcome to attend.
If you are travelling from outside Leicester, please contact Nick
Taub (email: [log in to unmask] 0116-252-5416) or Paul Lambert (email:
[log in to unmask] 0116-252-5407) for directions.
This seminar is part of the Nottingham and Leicester
Universities Statistical Seminars series.
There will be tea & coffee in the coffee room from 4.00 pm.
Abstract
=======
Objectives: To decide whether additional experimental research
is required to "substantiate" an economic claim for a new drug
for Alzheimer's Disease, and establish the efficient design of
any future clinical trial which may be required.
Methods: It has been estimated that the expected value of
perfect information for the choice between treatment strategies
in Alzheimer's Disease is substantial ($184 million). This
maximum potential benefit of research is likely to exceed the
cost of additional investigation. However this observation is
only a necessary condition for deciding to conduct another
trial. The sufficient condition requires estimates of the
marginal benefit and the marginal cost of sampling. Bayesian
decision theory is applied to a probabilistic model of
Alzheimer's Disease to establish the expected value of sample
information (EVSI). The difference between EVSI and the cost
of sampling is the expected net benefit of sampling (ENBS)
or the societal value of additional research. Using these
methods it is possible to answer questions such as: is an
additional clinical trial required before an economic claim
for the drug can be "substantiated"; if so, should an economic
evaluation be conducted alongside the new trial; and what is
the optimal follow-up, sample size and allocation of patients
between the arms of the trial?
Results: If the ENBS is positive then an additional trial
will be efficient (existing evidence cannot "substantiate"
the claim). The optimal sample size will be where the ENBS
reaches a maximum, given that patients are allocated
efficiently between the arms of the trial (based on the
marginal benefits and costs of alternative allocations).
ENBS can be established for designs that include different
endpoints. If the societal payoff to proposed research is
higher when only clinical endpoints are included then it
will be inefficient to conduct economic evaluation alongside
the trial. Whether a clinical trial should be "large and
simple" or have economic content becomes an empirical
question. The ENBS can also be established for a range of
possible follow-up periods. The optimal follow-up for the
trial will be where the ENBS reaches a maximum.
Conclusions: Bayesian decision theory provides a framework
for the efficient design of clinical trials which overcomes
the arbitrary nature of traditional power calculations.
It also establishes the societal value of proposed clinical
research which can be used to set research priorities and
inform the FDA decision about whether an economic claim for
a new technology has been "sufficiently substantiated".
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|