Background: Databases of prescription drug purchases are now widely used in pharmacoepidemiologic studies.
Several methods have been used to generate drug use periods from drug purchases to investigate various aspects;
e.g., to study associations between exposure and outcome. Typically, such methods have been fairly simplistic, with
fixed assumptions of drug use pattern and or dose (for example, the assumed usage of 1 tablet per day). This paper
describes a novel PRE2DUP method that constructs drug use periods from purchase histories, and verified by a
validation based on an expert evaluation of the drug use periods generated by the method.
Methods: The PRE2DUP method is a novel approach based on mathematical modelling of personal