Optimising the analysis of clinical trials with missing data



This CSO-funded fellowship for Shona Fielding (University of Aberdeen) addressed the issue of missing quality of life (QoL) outcomes in the analysis of clinical trials.   Missing data are a problem for any outcome but are particularly prominent for QoL outcomes, as the reason why the data are missing is likely to be related to the QoL itself.  Ignoring this could lead to biased results and ultimately impact on clinical practice.  The issues surrounding missing data and methods to deal with it were explained.  Seven trial datasets were used for illustration. Despite their being extensive literature on this subject this work used a novel approach by making use of data collected through postal reminders.

Different analysis strategies were shown to have an impact on the result of the trial. The most suitable imputation method depended on the missing data mechanism which was inherent in the trial. Repeated measures provided a framework for analysis which could include every patient for which at least one QoL assessment was available. In conclusion there is no single best way of dealing with missing data. The choice between different options must be made taking into account any assumption about the missing data mechanism.


Craig Ramsay; c.r.ramsay@abdn.ac.uk




Fielding S, Fayers P, Ramsay CR. Investigating the missing data mechanism in quality of life outcomes: A comparison of approaches. Heath and Quality of Life Outcomes 2009; 7:57.