Skip to main content

Table 2 True and false positive rates in cumulative subgroup analysis using data from simulated trials

From: Cumulative subgroup analysis to reduce waste in clinical research for individualised medicine

No. of trials sequentially included

No. of total patients involved

True positive

False positive

X1

X2

X3

X4

1

800

9.8%

4.8%

5.4%

5.2%

2

1600

14.8%

4.7%

5.3%

4.5%

3

2400

19.7%

5.1%

5.5%

5.0%

4

3200

25.7%

5.1%

5.5%

5.3%

5

4000

30.7%

4.9%

5.4%

4.6%

6

4800

36.4%

5.0%

5.3%

4.9%

7

5600

41.3%

5.4%

5.5%

4.6%

8

6400

46.0%

5.5%

5.2%

4.7%

9

7200

50.7%

5.5%

5.4%

4.5%

10

8000

54.4%

5.2%

5.2%

4.9%

11

8800

58.3%

5.5%

5.3%

4.9%

12

9600

61.6%

5.2%

4.9%

4.5%

13

10,400

65.2%

5.2%

5.3%

4.7%

14

11,200

68.9%

5.5%

5.3%

4.9%

15

12,000

71.9%

5.3%

5.5%

4.9%

16

12,800

74.4%

5.5%

5.4%

4.6%

17

13,600

77.1%

5.7%

5.5%

4.5%

18

14,400

79.6%

5.7%

5.1%

4.8%

19

15,200

82.0%

5.9%

4.7%

4.7%

20

16,000

83.8%

5.7%

4.6%

4.7%

  1. Except for varying input values shown in the table, the following basic input parameters were used for other variables: No. of patients per arm = 400; OR = 0.7; event rate in the control arm = 0.3; heterogeneity variance = 0.01; ROR = 0.8; proportion of patients with X1, X2, X3 and X4 = 0.4. Positive rates were based on 5000 simulations