In a new article from the NIH Pragmatic Trials Collaboratory Biostatistics and Study Design Core, the authors share analytic considerations for cluster randomized trials with hierarchical nesting of participants within clusters. The authors illustrate the problem using theoretical derivations, a simulation study, and data from the STOP CRC NIH Collaboratory Trial as an example.

“We conclude that an analysis using both an exchangeable working correlation matrix and weighting by inverse cluster size, which may be considered the natural analytic approach, can lead to incorrect results. That is, two weights make a wrong. The bias is minimal when there is homogeneity of treatment effects according to cluster size but unacceptable when there is heterogeneity of treatment effects according to cluster size. In addition, we show that only an analysis with an independence working correlation matrix and weighting by inverse cluster size always provides valid results for the UATE [unit average treatment effect] estimand.”

Read the full article.