Skip to main content

Table 3 Proportion of doses being selected as ED90 of Bayesian Emax and NDLM model at different dose response curves in the Half Adaptive design settings (Scenario 3)

From: Design considerations and analysis planning of a phase 2a proof of concept study in rheumatoid arthritis in the presence of possible non-monotonicity

Ā 

Dose Level (mg/kg)

0.03

0.3

3

10

20

30

Bayesian Emax Model

ā€ƒProportion of doses being selected as ED90

ā€ƒā€ƒFlat placebo like Curvea

0%

0%

0%

0%

38%

0%

ā€ƒā€ƒ Emax like Curve

0%

0%

0%

0%

89%

11%

ā€ƒā€ƒLog Linear Curve

0%

0%

0%

0%

81%

19%

ā€ƒā€ƒU-Shape Curvea

0%

0%

0%

0%

64%

0%

Bayesian NDLM Model

ā€ƒProportion of doses being selected as ED90

ā€ƒā€ƒFlat placebo like curve

16%

17%

14%

14%

11%

12%

ā€ƒā€ƒ Emax like Curve

1%

1%

14%

25%

35%

26%

ā€ƒā€ƒLog Linear Curve

1%

1%

5%

11%

29%

54%

ā€ƒā€ƒU-Shape Curve

2%

42%

48%

5%

2%

0%

  1. aED90 is missing where the maximum dose was not estimated correctly