Governing Equations

Enrollment (Poisson Process):
  E(t) = rate · t · (1 - drop/100)

Effective Sample Size:
  Neff = N_target · (1 - drop/100)

Power (Two-sample t-test):
  Power = Φ(Zα + |δ|·√(Neff/2) / σ)
  where δ = effect size, σ = pooled SD

Interim Analysis Trigger:
  Trigger when E(t) ≥ N_interim

Expected Completion:
  T_complete = N_target / (rate · (1 - drop/100))

Dropout-Adjusted Power:
  Power_adj = Power · (1 - 0.5 · drop/100)

Adaptive Design (futility):
  If Z_interim < Z_futility → stop for futility
  Z_futility typically = 0 (conditional power < 50%)