Figure 5From: Assessing potential sources of clustering in individually randomised trials Loss in power from not accounting for ignorable clustering in the analysis. Back to article page