Royal Statistical Society Merseyside Local Group
Wednesday 17th October 2007 2.00pm
(Tea & Coffee afterwards)
Venue : Penthouse, Maths & Oceanography building University of Liverpool
Please contact Ashley Jones ([log in to unmask]) for further details.
---------------------------
Sarah Walker - Medical Research Council
When are factorial designs not the answer?
Experience from the DART trial of strategies for antiretroviral therapy in Africa
Factorial designs have a number of advantages and disadvantages, but may be particularly useful in a new indication or population where there are many more questions than would be feasible to address in separate trials. The Development of AntiRetroviral Therapy (DART) Trial in HIV-infected adults in Uganda and Zimbabwe is an example where for good reasons three factorial randomisations were included in one design, one a period of time after trial entry (conceptually the same as co-enrollment). Subsequent events highlighted particular problems with inclusion of this co-enrollment-like partial factorial randomisation. Possible statistical analysis methods are discussed and briefly compared in a simulation study.
|