Optimal sample size determinations from an industry perspective based on the expected value of information

Area of Science:

• Decision Sciences
• Biostatistics
• Pharmaceutical Industry

Background:

• Traditional sample size calculations rely on arbitrary factors like type I and II errors.
• Bayesian decision theory offers an alternative using expected value of information.
• Existing methods focus on societal net gain or industry profit maximization.

Purpose of the Study:

• Propose a Bayesian approach for industry-specific sample size calculations.
• Determine sample size that maximizes expected profit for pharmaceutical trials.
• Align trial size with industry's financial objectives and return on investment.

Main Methods:

• Develop a model for expected total profit, incorporating per-patient profit, disease incidence, time horizon, trial duration, market share, and regulatory approval probability.
• Relate the expected value of information to increased expected profit via enhanced regulatory approval probability.
• Apply methods to a case study, examining robustness and extending the model for market share as a function of treatment effect.

Main Results:

• Expected value of information methods yield robust sample size solutions from an industry perspective.
• These methods maximize the difference between expected trial costs and the expected value of gained information.
• The approach aids in selecting trials that offer the highest return on investment for a fixed research budget.

Conclusions:

• The proposed method's accuracy depends on the expected total profit model's assumptions.
• Market share, per-patient profit, and incidence are assumed insensitive to trial outcomes.
• Industry stakeholders should consider expected value of information methods for trial planning and portfolio optimization.