Real-world data can provide external controls, but insufficient accrual affects all arms of a study, not just controls. Studies that used generative models to simulate more patients were limited in ...