Adaptive Clinical Trial Sample Size Re-estimation Engine

Biostatistics & Regulatory Science: Adaptive Clinical Trials

In conventional fixed-sample clinical trials, underestimating effect size variance causes study failure. Adaptive Sample Size Re-estimation (SSR) allows mid-trial corrections:

• The Type I Error Inflation Problem: Simply increasing sample size based on unblinded interim data without statistical adjustments inflates the false positive rate (α > 0.025), leading to regulatory rejection.

• The Cui-Hung-Wang (CHW) Weighting Principle: Pre-fixes the statistical weights of Stage 1 and Stage 2 according to the originally planned information fraction: . This guarantees Z_CHW N(0, 1) under the null hypothesis regardless of how much Stage 2 sample size is increased.

• The Mehta-Pocock Promising Zone:

• Unfavorable Zone (<50% CP): Effect size is too small; increasing sample size wastes millions on a likely futile drug.

• Promising Zone (50%–85% CP): The drug works, but needs more statistical power; sample size expansion rescues the trial.

• Favorable Zone (≥85% CP): The trial is already on track for statistical significance; no expansion needed.

How to Use This Calculator

Enter your Original Planned Total Sample Size (N_plan) (e.g. 300 (300 patients across both treatment arms)), Interim Analyzed Sample Size (n_1 [Stage 1]) (e.g. 150 (150 patients evaluated at 50% interim)), Interim Observed Test Statistic (Z_1 Score) (e.g. 1.50 (Standardized interim z-score)), Target Significance Level (α [1-Sided Alpha]) — choose from α = 0.025 (1-Sided | Standard FDA / EMA Phase III), α = 0.05 (1-Sided / 0.10 2-Sided Exploration), α = 0.01 (Strict Multiplicity Control), Target Re-estimated Statistical Power (%) (e.g. 90 (80% or 90% conditional power target)), and Maximum Total Sample Size Budget Ceiling (N_max) (e.g. 600 (2x planned sample size budget cap)). Results update instantly as you type — adjust any value to compare scenarios.

The Formula

Cui-Hung-Wang (1999) \& Mehta-Pocock (2011) Adaptive Trial equations:

Information Fraction (t_1) = (n_1) / (N_plan)

Conditional Power (CP) = Φ((Z_1 - z_α √(t_1)) / (√(1 - t_1)) + Z_1 √((1 - t_1) / (t_1)))

Promising Zone Criterion: 50\% CP < 85\%

Cui-Hung-Wang Weighted Test Statistic: Z_CHW = √(t_1) Z_1 + √(1 - t_1) Z_2 z_α

Type I Error Control: P_H_0(Z_CHW z_α) = α

Worked Example

For example, evaluating a Phase III double-blind trial planned for \ at a \ with an observed \ under and targeting 90.0\% power): Interim Conditional Power is 59.0\%, placing the trial directly in the Mehta-Pocock Promising Zone (50% to 85% CP). The trial increases Stage 2 enrollment to \. Final analysis uses the Cui-Hung-Wang statistic strictly preserving the 2.5% false positive rate.

Understanding Your Result

Evaluate your result in the context of your specific use case and relevant industry standards or reference values. A single calculated result is a starting point — test two to three scenarios to understand how sensitive the outcome is to your key input assumptions. For high-stakes decisions, verify your result against a secondary source or subject-matter expert before acting.