|
Background/Aims: The two main approaches used to analyze cluster randomized trials are cluster-level and individual-level analysis. In a cluster-level analysis, summary measures are obtained for each cluster and then the two sets of cluster-specific measures are compared, such as with a t-test of the cluster means. A mixed model which takes into account cluster membership is an example of an individual-level analysis. The purpose of this study was to compare power and bias of a cluster-level analysis and an individual level analysis when data are complete, missing completely at random (MCAR) and missing at random (MAR).
Methods: We used a simulation study to quantify and compare power and bias of these two methods. Complete datasets were generated and then data were deleted to simulate MCAR and MAR data. A balanced design, with two treatment groups and two time points was assumed. Cluster size, variance components (including within-subject, within-cluster and between-cluster variance) and proportion missing were varied to simulate common scenarios seen in practice. For each combination of parameters, 1000 datasets were generated and analyzed.
Results: Results of our simulation study indicate that cluster-level analysis resulted in substantial loss of power (up to 26%) when data were MAR. Individual-level analysis had higher power and remained unbiased, even with a small number of clusters.
Conclusion: Individual-level modeling which takes into account cluster membership performs better in the presence of missing data in terms of power and bias.
|
