Disproportionality Methods (DP) for Observational Data
Disproportionality methods such as MGPS, PRR, or BCPNN, are the main tools for mining spontaneous reporting databases for associations (between drugs and conditions). These methods are widely used by FDA, FAERS, and the Uppsala Monitoring Centre. Here is an adaptation of those methods for longitudinal observational data. The first version contains obvious imperfections and simplifications, although the code itself is capable of handling very large databases if it is run on appropriate hardware. One may test DP method on a simulated data that is available from here.
Download link: DP_METHOD