Cluster Randomised Trials (CRTs) are the experiments where clusters of individuals such as villages, schools, medical practices, hospital wards-rather than independent individuals- are randomly allocated to each of intervention and control groups, while individual-level responses are collected within each cluster. Such trials are being increasingly used in the fields of health promotion and health service research. Attrition is common in CRTs which to leads to missing data that often create a problem in the analysis of such trials. Not only do they cause a loss of information and as a result usually reduce the power credibility of a study, but also they might be a potential source of bias in the parameter estimates. Inadequate handling of missing data may result in misleading inferences. Parameter estimates obtained from a data with substantial amount of missing values are usually biased and less efficient. The aim of our research is to investigating the applicability of existing methods for handling missing data, including multiple imputation, to the context of missing outcome data in CRTs, and where necessary, the development of new statistical methods.
Jonathan Bartlett: http://www.lshtm.ac.uk/aboutus/people/bartlett.jonathan