Extreme restriction design as a method for reducing confounding by indication in pharmacoepidemiologic research

Abstract

Restriction to a single indication for treatment may reduce confounding by indication in studies conducted in distributed data networks and other large databases.