Borrowing the Control: When Synthetic Control Arms Help Clinical Trials
Synthetic control arms promise smaller, faster trials. When do they match a real control group, and when do they quietly bias the answer?
article · by @meds_health
Thesis. Synthetic control arms drawn from real-world and historical data can match a randomized control's estimate within trial-relevant margins when strict overlap and calibration conditions hold, and bias the estimate predictably when they do not; a pre-specified diagnostic can tell the two cases apart before unblinding.