Skip to main content

Table 4 Power estimation when the true distributions deviate from the assumed distributions

From: Power and sample sizes estimation in clinical trials with treatment switching in intention-to-treat analysis: a simulation study

shape

 

no switching

switching

s.dist = “beta”

switching

s.dist = “gamma”

switching

s.dist = “unif”

switching

s.dist = “indepExp”

   

Switching probability p = 0.2

 

n

130

166

165

168

169

1.0

Power

0.80

0.80

0.80

0.80

0.80

 

E1

104.0

130.8

130.1

132.5

132.8

 

E2

89.2

113.8

113.2

115.1

115.8

0.5

Power

0.24

0.26

0.27

0.25

0.25

 

E1

96.2

108.7

108.1

110.1

110.3

 

E2

77.0

98.4

97.5

99.4

99.9

0.75

Power

0.53

0.53

0.54

0.53

0.53

 

E1

96.1

121.0

120.2

122.4

122.6

 

E2

83.3

106.2

105.5

107.5

108.1

1.25

Power

0.95

0.94

0.95

0.95

0.95

 

E1

109.5

137.9

140.0

139.5

139.7

 

E2

94.4

120.6

119.7

122.0

122.7

 

Switching probability p = 0.6

 

n

130

294

301

297

344

1.0

Power

0.80

0.80

0.80

0.80

0.80

 

E1

104.0

225.0

230.0

227.1

259.4

 

E2

89.2

201.7

206.4

203.9

236.0

0.5

Power

0.24

0.26

0.25

0.26

0.27

 

E1

96.2

188.1

192.2

190.2

218.0

 

E2

77.0

173.8

178.2

175.8

203.5

0.75

Power

0.53

0.54

0.52

0.53

0.55

 

E1

96.1

208.0

212.7

210.3

240.4

 

E2

83.3

188.2

192.8

190.2

220.1

1.25

Power

0.95

0.95

0.94

0.95

0.94

 

E1

109.5

237.3

242.8

239.8

274.4

 

E2

94.4

213.4

218.6

215.6

249.7

  1. E1 and E2: the expected numbers of events in the control group and the experimental group