Sample sizes based on exact unconditional tests for phase II clinical trials with historical controls

Area of Science:

• Biostatistics
• Clinical Trial Design
• Pharmacoeconomics

Background:

• Phase II clinical trials commonly employ two-sample binomial tests to compare experimental and control group response rates.
• Traditional methods often rely on asymptotic inference, which may not be accurate for all sample sizes.
• The use of historical control data is a practical consideration in trial design.

Purpose of the Study:

• To develop and present a method for determining sample size in phase II clinical trials using historical control data.
• To compare the exact unconditional inference method with the traditional asymptotic method for sample size calculation.
• To provide practical sample size tables for researchers.

Main Methods:

• The study investigates the two-sample binomial problem (H0: pe = pc vs. H1: pe > pc).
• It utilizes exact unconditional inference, avoiding asymptotic approximations.
• Sample size determination is performed with consideration for pre-existing control group data (historical controls).

Main Results:

• Sample size tables comparing exact and asymptotic methods are provided.
• While asymptotic results often approximate exact results, notable discrepancies were observed.
• The exact unconditional method offers a more precise approach to sample size determination.

Conclusions:

• Exact unconditional inference provides a more accurate method for sample size calculation in phase II trials with historical controls.
• Researchers should be aware of potential differences between exact and asymptotic methods.
• The findings support the use of precise statistical methods for robust clinical trial design.