On the Comparison of Methods in Analyzing Bounded Outcome Score Data

Area of Science:

• Pharmacometrics
• Clinical Trial Design
• Biostatistics

Background:

• Clinical trial endpoints frequently use bounded outcome scores (BOS), which have discrete values within a finite range.
• Treating BOS data as continuous can lead to range violations and skewness, complicating analysis.
• Existing analysis methods for BOS data cause confusion among pharmacometricians regarding proper application and comparison.

Purpose of the Study:

• To review and compare primary BOS data analysis methods used in pharmacometrics.
• To discuss the theoretical and practical aspects of different BOS analysis approaches.
• To provide guidance on selecting appropriate BOS analysis methods for clinical trials.

Main Methods:

• Review of existing literature on BOS data analysis in pharmacometrics.
• Theoretical comparison of statistical properties of different methods.
• Practical considerations and performance expectations in various scenarios.

Main Results:

• Analysis of BOS data as continuous variables can violate data ranges and mishandle skewness.
• Several dedicated methods for BOS data exist, but their selection and comparison are unclear.
• The commentary evaluates the strengths and weaknesses of different BOS analysis techniques.

Conclusions:

• Proper analysis of BOS data is crucial for accurate clinical trial interpretation.
• Selecting the correct method depends on the specific characteristics of the BOS data.
• Guidelines are provided to aid pharmacometricians in choosing the most suitable BOS analysis method.