Estimation of Treatment Effect Following a Clinical Trial with Adaptive Design

Abstract

Parameter estimation following an adaptive design or group sequential design has been extremely challenging due to potential random high from its face value estimate. In this paper, we introduce a new framework to model clinical trial data flow based on a marked point process (MPP). The MPP model allows us to use methods of stochastic calculus for analyses of any adaptive clinical trial. As an example, we apply this method to a two stage treatment selection design and derive a procedure to estimate the treatment effect. Numerical examples will be used to evaluate the performance of the proposed procedure.

Key Words:

• Adaptive design
• Clinical trial
• Parameter estimation
• Treatment selection

ACKNOWLEDGMENT

We thank an editor for valuable comments and suggestions that greatly improved the presentation.

Notes

*Normal distribution. m ₀ = 1, ρ = 0.1, n ₁ = 90, and N = 230. Bias and RMSE of estimators are defined in section 3. All values were obtained based on B = 10, 000 simulations.

*Log-normal distribution. m ₀ = 1, ρ = 0.1, n ₁ = 90, and N = 230. Bias and RMSE of estimators are defined in section 3. All values were obtained based on B = 10, 000 simulations.

*Binomial distribution. p ₀ = 0.1, ρ = 0.05, n ₁ = 150, and N = 250. Bias and RMSE of estimators are defined in section 3. All values were obtained based on B = 10, 000 simulations.