Approximate multinormal probabilities applied to correlated multiple endpoints in clinical trials

Area of Science:

• Biostatistics
• Clinical Trial Design
• Statistical Methods

Background:

• Clinical trials often involve multiple endpoints, increasing the risk of familywise type I errors.
• The Bonferroni correction is a standard method for adjusting p-values but can be overly conservative, especially with correlated endpoints.

Purpose of the Study:

• To present a novel statistical procedure for adjusting p-values in the presence of correlated endpoints in clinical trials.
• To offer a less conservative alternative to the Bonferroni correction, thereby enhancing statistical power.

Main Methods:

• The proposed method utilizes an approximation derived from multinormal probability calculations.
• It allows for the direct incorporation of correlation between endpoints into the p-value adjustment process.

Main Results:

• The new procedure provides a more accurate adjustment for multiple testing when endpoints are correlated compared to the Bonferroni method.
• It demonstrates reduced conservatism, particularly in scenarios with high endpoint correlation.

Conclusions:

• The developed procedure offers a more statistically rigorous approach to handling multiple endpoints in clinical trials.
• This method can lead to more accurate conclusions and potentially increased power in studies with correlated outcomes.