Correcting for multiple-testing in multi-arm trials: is it necessary and is it done?

Area of Science:

• Clinical Trials
• Biostatistics
• Pharmaceutical Development

Background:

• Multi-arm trials enhance clinical development efficiency by evaluating multiple treatments concurrently.
• Regulatory bodies often require statistical corrections for multiple hypothesis testing in multi-arm trials to control Type-1 errors.
• Conflicting literature exists regarding the necessity and application of multiple-testing corrections.

Purpose of the Study:

• To examine the frequency of multiple-testing correction usage in published multi-arm trials.
• To discuss the varying viewpoints on when these statistical corrections are appropriate.
• To review current practices in applying multiple-testing correction methods.

Main Methods:

• A systematic review of multi-arm clinical trials published in 2012 across four major medical journals.
• Extraction of trial design data, including trial type (exploratory/confirmatory) and use/type of multiple-testing correction.

Main Results:

• 49% of reviewed multi-arm trials reported using a multiple-testing correction.
• Correction usage was higher (67%) in trials with multiple doses/regimens of the same treatment.
• Higher correction rates in exploratory trials were attributed to the testing of multiple doses/regimens.

Conclusions:

• A significant proportion of multi-arm trials do not implement multiple-testing corrections.
• There is a need for clearer regulatory and methodological guidance on multiple-testing correction for trials with separate treatments against a common control.