Modelling and Predicting Patient Recruitment to Ensure Successful Completion of Clinical Trials
Vladimir Anisimov
Using Experimental Design and Randomisation to Construct a Mixed Model
Simon Bate and Marion Chatfield
In many areas of scientific research scientists routinely use complex experimental designs when conducting their experiments. With the advent of the modern statistical package, and a lack of trained statisticians, the scientist often carries out the analysis of data generated from such experiments. This can lead to incorrect and misleading results, especially if the scientist fails to correctly identify the experimental design they are using and the influence the design has on the statistical analysis.
In this paper we describe a procedure that would allow non-statisticians to identify the structure of the experimental design without an in-depth knowledge of design theory. Once this has been achieved it should then be possible for them to produce an appropriate model for the statistical analysis. By placing experimental design at the centre of the statistical process we are able to simplify the statistical model selection. This procedure should also make the benefits of good experimental design more accessible to both statisticians and non-statisticians alike.
Meta-analysis of clinical trials, particularly of rare adverse events
Peter Lane
Abstract: Meta-analyses are increasingly being used to summarize information across trials, often to publicize good or bad news. Public access to trial results on the Internet has made it especially easy to generate such meta-analyses, particularly of safety issues. Once the hurdles of acquiring and selecting data have been cleared, the task of analysis with some given technique is only too easy. The results can be strongly influenced, however, by the choice of technique and the approach to combining information when the operating details vary across individual trials. The analysis of rare events, particularly safety events, is prone to disagreement and misunderstanding. I will look specifically at the fixed-effects meta-analysis of a binary response, illustrated by publicly available data from last year’s high-profile analysis of Avandia with respect to cardiovascular safety. This raised issues including the choice of summary statistic to employ, the combination of trials with different control treatments, and the handling of trials with no events. And lurking in the background was the ever-present danger of being misled by Simpson’s Paradox.
