CITIES: Clinical trials with intercurrent events simulator.

Area of Science:

• Biostatistics
• Pharmaceutical Research
• Clinical Trial Methodology

Background:

• Clinical trials face challenges with intercurrent events (ICEs) like treatment discontinuation, leading to missing data and confounding causal inference.
• Existing methods for analyzing trials with ICEs are difficult to compare due to data sharing limitations and customized simulation studies.
• The pharmaceutical industry needs standardized approaches to evaluate causal inference methods in the presence of ICEs.

Purpose of the Study:

• To develop a general simulation model and R-based Shiny application (CITIES) for generating data in clinical trials with ICEs.
• To provide a tool that facilitates the generation of realistic clinical trial data, mimicking real-world summary statistics without using original trial data.
• To enable comparison of different analytical methods for ICEs within identical, benchmark settings.

Main Methods:

• Development of a novel, general simulation model under the Rubin Causal Model framework.
• Implementation of the model into a user-friendly Shiny application in R, named CITIES.
• Formulation of treatment effects that account for ICEs in clinical trials with repeated measures.

Main Results:

• CITIES effectively generates data resembling real-life clinical trials regarding summary statistics.
• The simulation model allows for the creation of benchmark datasets for comparing analytical methods.
• Case studies demonstrate CITIES' utility in comparing methods for analyzing clinical trials with ICEs.

Conclusions:

• CITIES offers a comprehensive tool for pharmaceutical industry practitioners to compare analytical methods for clinical trials with ICEs.
• The application provides identical, benchmark settings for evaluating method performance, enhancing causal inference.
• This approach addresses the need for standardized data generation in clinical trial methodology research.