An Efficient Method for Computing Expected Value of Sample Information for Survival Data from an Ongoing Trial

Area of Science:

• Health economics
• Decision analysis
• Biostatistics

Background:

• Health technology decisions are often made with early-stage trial data, leading to uncertainty in key outcomes like life expectancy.
• Collecting additional data can reduce uncertainty, and its value is quantified by the expected value of sample information (EVSI).
• EVSI is typically used for designing future trials, not for ongoing ones.

Purpose of the Study:

• To develop and present new methods for computing the EVSI of extending an existing trial's follow-up.
• To address both single-model assumptions and model uncertainty in survival analysis.
• To provide tools for decision-makers regarding ongoing health technology trials.

Main Methods:

• Developed a nested Markov Chain Monte Carlo (MCMC) procedure.
• Developed a nonparametric regression-based method.
• Compared these methods using single-model and model-averaged EVSI in synthetic case studies.

Main Results:

• The regression-based method showed good agreement with the MCMC procedure.
• The regression method was fast, easy to implement, and scalable for multiple survival models.
• The MCMC procedure was computationally intensive, especially with model uncertainty.

Conclusions:

• A straightforward regression-based method for computing EVSI of extended trial follow-up is presented.
• This method handles both single-model and model-uncertainty scenarios.
• EVSI for ongoing trials aids decisions on early patient access versus needing more mature evidence.