Multc Lean
Multc Lean implements a special case of the single-arm safety monitoring
method of Thall, Simon, and Estey [1]. This
"lean" version of Multc is very easy to use, retaining only the most commonly
used features of the
original Multc99
program. This version monitors only two
outcomes: efficacy and toxicity. However, Multc Lean adds some features not
found in the original Multc:
- a Windows user interface,
- extensive documentation,
and
- the ability to simulate trial duration.
Prior to version 1.1, Multc Lean only supported continuous monitoring. Now
Multc Lean now also allows the user to specify monitoring in cohorts. Version
1.2 added the ability to specify a minimum number of patients to treat before
evaluating the stopping rules.
The method of Thall, Simon, and Estey is a general family
Bayesian designs for monitoring phase II trials. The method can be used to
monitor any number of outcomes and can accommodate any number of interim
evaluations of the stopping rules. (The Multc Lean software only monitors two
outcomes and currently supports only continuous monitoring. These are
limitations of the software implementation, not the statistical method. The
Multc99
software allows one to monitor any practical number of events and allows varying
cohort sizes.)
An advantage of Bayesian methods is that they allow arbitrary sample sizes.
One starts with an uninformative probability distribution on the parameters of
interest (such as the probabilities of toxicity and response) and ends with a
more informative posterior distribution after the trial. There is no
all-or-nothing threshold where n patients are too few but n+1
patients are plenty. The uncertainty in the posterior parameter estimates
decreases continuously as the number of patients in the trial increases. Small
studies are not disallowed; they simply have more posterior uncertainty than
larger studies.
An extensive statistical tutorial
is provided with the software. This tutorial gives
guidelines for how to use the method and includes exercises and solutions.
See also the
Multc Lean user's guide included with the software.
Conducting a trial designed by this method does not require software since
the stopping conditions can be tabulated before the trial begins. Here
is a document explaining the logisitics of running a Multc trial
for the benefit of the person responsible for monitoring the stopping rules.
Hoang Nguyen developed the user interface using
C#
and the Microsoft .NET framework version 1.1.
John Cook and Hoang Nguyen implemented the numerical algorithms using Visual C++.
References
[1]
Peter Thall, Richard Simon, and
Elihu Estey in "Bayesian
sequential monitoring designs for single-arm clinical trials with multiple
outcomes",
Statistics in Medicine, vol
14, 357-379 (1995)