Quantification of the prentice criteria for surrogate endpoints

Area of Science:

• Biostatistics
• Clinical Trials
• Epidemiology

Background:

• The Prentice criteria (1989) are foundational for evaluating surrogate endpoints in clinical research.
• Meta-analyses play a crucial role in synthesizing evidence from multiple studies.

Purpose of the Study:

• To refine a proposed measure for assessing how well surrogate endpoints meet the Prentice criteria.
• To address limitations in the original proposal by Alonso et al. regarding censoring mechanisms.

Main Methods:

• The study examines a likelihood reduction factor suggested by Alonso et al. for evaluating surrogate endpoints.
• Modifications are proposed to correct the measure for the presence of censoring, ensuring its general applicability.
• The proposed measure is integrated into established statistical theory for robust validation.

Main Results:

• The modified likelihood reduction factor provides a reliable measure of surrogate endpoint validity, independent of censoring mechanisms.
• A standard SAS program can generate a coefficient directly transformable into the desired population-level estimate.
• The corrected measure can be confidently employed in meta-analyses evaluating surrogate endpoints.

Conclusions:

• The refined likelihood reduction factor offers a statistically sound method for assessing surrogate endpoints.
• This approach enhances the utility of meta-analyses by providing a more accurate evaluation of surrogate endpoint performance.
• The availability of SAS programs facilitates the practical application of this improved methodology in biomedical research.