Jointly modeling multiple endpoints for efficient treatment effect estimation in randomized controlled trials.

Area of Science:

• Biostatistics
• Clinical Trials
• Tobacco Regulatory Science

Background:

• Randomized controlled trials (RCTs) are crucial for intervention efficacy evaluation.
• A common challenge in RCTs is the trade-off between scientifically relevant primary endpoints and more powerful, less relevant ones.
• Subgroup analyses are often underpowered, limiting their utility.

Purpose of the Study:

• To develop an improved estimator for treatment effects on primary endpoints by incorporating information from secondary endpoints.
• To enhance statistical power and robustness in clinical trial analyses.
• To provide a method that leverages multiple endpoints for increased confidence in trial results.

Main Methods:

• Developed a novel estimator for treatment effects based on a joint model for primary and secondary efficacy endpoints.
• Employed model averaging to ensure robustness against potential model misspecification.
• Applied the method to estimate the effect of very low nicotine content cigarettes on smoking abstinence.

Main Results:

• The proposed estimator demonstrated increased efficiency compared to standard methods when the joint model was correctly specified.
• The approach proved robust to model misspecification through model averaging.
• In a specific application, the method reduced the standard error by 27% when assessing smoking abstinence.

Conclusions:

• The developed joint modeling approach effectively leverages secondary endpoints to improve the estimation of primary endpoint treatment effects.
• This method offers a statistically robust and more powerful alternative for analyzing clinical trial data, particularly in areas like tobacco regulatory science.
• The findings suggest a valuable tool for enhancing the interpretation and reliability of evidence from clinical trials with multiple efficacy measures.