Sample size calculation for a stepped wedge trial

Table 2 Comparison of the simulation-based approach with the analytical formulae of HH. The cells in the table are the estimated number of clusters as a function of the ICC and outcome type (continuous, binary and rate) to obtain 80 % power

ICC	Analytical power based on HH	Simulation-based calculations
	K=20,J=6	K=20,J=6
Continuous outcome^a
0	9	9
0.1	13	13
0.2	14	13
0.3	14	14
0.4	14	14
0.5	14	14
Binary outcome^b
0	11	15
0.1	17	16
0.2	18	17
0.3	18	18
0.4	18	18
0.5	18	18
Count outcome^c
0	8	8
0.1	13	12
0.2	13	12
0.3	13	12
0.4	13	11
0.5	13	11

^aIntervention effect = −0.3785; σ _e=1.55.
^bBaseline outcome probability = 0.26; OR = 0.56.
^cBaseline outcome rate = 1.5; RR = 0.8.
Notation: K= number of subjects per cluster; J= total number of time points, including one baseline

ISSN: 1745-6215