Table 4 Number (%) of trials at risk of an inflated type I error rate
Trials (n = 99a) | |
|---|---|
Assuming a minimum of 40 clusters is required for individual-level analysis with no correction (primary analysis) | |
At risk of an inflated type I error rate – number (%) | |
Yes | 64 (65) |
No | 35 (35) |
Reason for being at risk – number (%) | |
Did not report adjustment for clustering in analysis | 14/64 (22) |
Individual-level analysis on <40 clusters without correction | 50/64 (78) |
Reason for not being at risk – number (%) | |
Cluster-level analysis | 8/35 (23) |
Individual-level analysis with a large number of clusters (40+) | 27/35 (77) |