Courses given by
Medical and Pharmaceutical Statistics Research Unit
The University of Reading
Meta-analysis of Clinical Trials
23-24 September
Presenters Anne Whitehead and John Whitehead
Cost 510 pounds if booked 6 weeks in advance
(with discounts for academic/public sector/students)
Statistical Methods for Ordered Categorical Data
25-27 September
Presenters Kim Bolland and John Whitehead
Cost 765 pounds if booked 6 weeks in advance
(with discounts for academic/public sector/students)
Further details are given below.
To register please contact
Mrs Barbara Dodds
MPS Research Unit
The University of Reading
PO Box 240
Earley Gate
Reading
RG6 6FN, UK
Tel: 0118 931 6662
Fax: 0118 975 3169
Email: [log in to unmask]
or see http://www.rdg.ac.uk/mps/mps_home/courses/scheduled_c.htm
Meta-analysis of Clinical Trials
Presenters Anne Whitehead and John Whitehead
The rapid increase in the number of meta-analyses being conducted in the medical
world during the last twenty years has been mainly due to greater emphasis on
evidence-based medicine and the need for reliable summaries of the vast and
expanding volume of clinical research. At the same time there have been great
strides in the development and refinement of the associated statistical
methodology. This course concerns the planning, conduct and reporting of a
meta-analysis as applied to a series of randomised controlled clinical trials.
The course presents a general framework for meta-analysis, applicable to binary,
ordinal, time to event or normally distributed patient responses. Fixed and
random effects models for combining study summary statistics or individual
patient data will be presented. Methods of dealing with heterogeneity as well
as Bayesian approaches are examined, together with preparation of a
meta-analysis protocol and reporting of results. In addition to lectures, there
will be practical sessions and group discussions based on published examples of
meta-analyses.
Programme
* A general framework for meta-analysis
* Binary, ordinal, time to event and normal data
* Combining study summary statistics or individual patient
data using fixed and random effects models
* Protocol development
* Presentation and interpretation of results
* Dealing with heterogeneity and selection bias
* Retrospective vs prospective meta-analyses
* Cumulative meta-analysis
* Bayesian methods
Statistical Methods for Ordered Categorical Data
Presenters Kim Bolland and John Whitehead
Ordered categorical data arise from many clinical trials and epidemiological
studies. Commonly used scales include the Glasgow Outcome Scale in head injury,
where the five response categories are Good Recovery, Moderate Disability,
Severe Disability, Vegetative State and Death; the Barthel Index and modified
Rankin score in stroke; the Expanded Disability Status scale in multiple
sclerosis as well as various measures of quality of life. This course examines
the design and analysis of studies in which the primary response is of this
form.
Binary data are a special case of ordered categorical data and the course first
reviews the statistical methodology for binary data, with an emphasis on
logistic regression analysis. The generalisation from binary to ordinal data is
based on the assumption of proportional odds, and for a simple comparison of two
homogenous groups the method leads to the Mann-Whitney test. Linear modelling
and model checking are given prominence. Calculation and review of sample sizes
are examined, together with other design issues. Methods for the analysis of
repeated ordinal data based on the marginal model and on the subject-specific
model are presented and compared.
The course is extensively illustrated using examples drawn from the presenters'
consultancy experience. Many practical sessions are included, most of which
involve using SAS(.
Programme
* Review of binary data analysis
* The proportional odds model
* Fitting linear models using PROC LOGISTIC
* Model checking
* Alternative models
* Ordinal data and clinical trials
* Sample size and power
* Repeated ordinal data
|
