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)

^aOne trial was excluded as the number of clusters was not reported