Completeness of reporting of simulation studies on responder analysis methods and simulation performance: a...

Area of Science:

• Biostatistics
• Statistical Methodology
• Simulation Studies

Background:

• Responder analysis is crucial for evaluating treatment effects.
• Simulation studies are used to compare different responder analysis methods.
• Assessing simulation performance is key to understanding method reliability.

Purpose of the Study:

• To evaluate the completeness of reporting in simulation studies.
• To assess the performance of various responder analysis methods.
• To identify optimal simulation practices for responder analysis.

Main Methods:

• Systematic methodological survey of simulation studies.
• Searched Embase, MEDLINE, PubMed, and Web of Science Core Collection.
• Included studies comparing responder analysis methods and assessing simulation performance metrics.

Main Results:

• Seven simulation studies were identified, exploring augmented binary, distributional, and model-based methods.
• Reporting quality was suboptimal, with missing details on simulation parameters (seed, failures, RNG, number of simulations) and accuracy.
• Augmented binary methods showed increased power and precision compared to standard binary methods; distributional methods were adaptable to skewed data.

Conclusions:

• Simulation studies on responder analysis methods exhibit suboptimal reporting.
• Augmented binary, distributional, and model-based methods may be preferable to standard binary methods.
• No single responder analysis method is universally the best; method choice depends on specific data characteristics.