Statistical Analysis

Blinding protects the validity of the comparison between intermittent fasting and calorie restriction. Higher training-hour categories were associated with larger increases in capacity score. Investigators expect active subperiosteal drainage to yield an 11.2% composite endpoint rate through a 30% relative risk reduction. Bayesian methods are preferred for flexible model parameterisation and intuitive interpretation through posterior tail probabilities. The augmented binary method reduces the required sample to 48 per arm while preserving 90% power. Meta-analysis will use random-effects models when studies are sufficiently homogeneous. The trial includes superiority analyses for on-demand PPI versus fixed taper and alginate versus fixed taper, plus non-inferiority of on-demand PPI versus alginate. Unplanned readmission at 90 days will be compared using logistic mixed regression with hospital as a random effect and randomisation arm and time period as fixed factors. Treatment groups will be compared using variability in smartphone-based patient-reported mood. The study will use nearest-neighbour propensity score matching with a planned 3:1 ratio of non-PPI to PPI studies. Hierarchi…