Preventing Bias in Cluster Randomised Trials

In randomised trials, internal validity is defined as ‘‘the extent to which the design and conduct of a study are likely to have prevented bias’’ [1,2]. In conducting such trials, trialists try to prevent selection bias through randomisation and allocation concealment (defined as ‘‘the process used to ensure that the person deciding to enter a participant into a randomised controlled trial does not know the comparison group into which that individual will be allocated’’ [2]) and attrition bias through an intention-to-treat (ITT) analysis. ITT analysis has indeed been defined as one of the cornerstones of the analysis strategy of randomised trials because it allows for preserving the benefits of randomisation [3]. With an ITT analysis, data for all randomised participants are analysed in the groups to which they were originally randomly allocated ‘‘regardless of their adherence with the entry criteria, regardless of the treatment they actually received, and regardless of subsequent withdrawal from treatment or deviation from the protocol’’ [4]. ITT analysis entails the use of ad hoc statistical methods to handle missing outcome data when participants withdraw from the trial or are lost to follow-up [4,5]. The ITT principle is widely used in analysing data from individually randomised trials but is much more difficult in cluster randomised trials (CRTs), and this issue is not clearly covered in the main methodological textbooks on the topic [6,7]. Here, we describe the difficulties in preventing selection bias and applying ITT analysis in CRTs (other biases are discussed in Puffer et al. [8]) and propose some solutions to deal with these issues in this trial design.