Comparing Three Approaches For Handling A Fourth Level Of Nesting Structure In Cluster-Randomized Trials