Conditional power calculations for clinical trials with historical controls

Area of Science:

• Biostatistics
• Clinical Trials
• Pharmacoeconomics

Background:

• Comparing new therapies to standard treatments requires robust statistical methods.
• Accurate sample size calculations are essential for the power of clinical trial hypothesis tests.
• Previous trial data is often used to inform new trial design.

Purpose of the Study:

• To evaluate the accuracy of sample size calculation methods for new therapy trials.
• To identify and correct a popular but flawed method for power and sample size calculations.
• To provide alternative, valid statistical approaches for trial design across different outcome types.

Main Methods:

• The study critically analyzes a commonly used method for sample size and power calculations.
• It proposes and discusses alternative statistical methodologies.
• The methods are examined within the context of normal, binary, and time-to-event outcome data.

Main Results:

• A widely adopted method for sample size and power calculations in comparative therapy trials was found to be incorrect.
• Alternative methods are presented that provide accurate calculations for various outcome types.
• The findings highlight the importance of using validated statistical approaches in clinical trial planning.

Conclusions:

• Incorrect sample size calculations can compromise the reliability and power of clinical trial results.
• Validated statistical methods are necessary for accurate power and sample size determination in new therapy trials.
• The study offers practical guidance for researchers designing comparative clinical trials with different data types.