Data mining may be a promising method to search for unknown drug-drug interactions in cardiovascular medicine.
Cardiovascular medications have the potential to interact badly.
Presenting a poster abstract at the ESC 2016 Congress in Rome, Italy, Peter Waede Hansen of the Danish Heart Foundation, Copenhagen, Denmark and colleagues reported on their attempt to see if data-mining could be used to find undetected interactions with warfarin.
The drug was selected because of its biomarker (INR) and well-known drug interaction profile.
They looked at records of 10,219 patients who had a total 2,478 events of high INR and 4,232 events of low INR. All patients had a stable INR prior to a novel prescription and at least one measured INR value in the next 45 days.
Events were defined as an INR value outside the therapeutic range after a new drug was started.
They then noted at how the patients fared after they got another drug with the warfarin. Initation of total of 331 different medications was nvestigated.
That enabled them to predict which medications were more likely to be more effective with warfarin and which were less likely.
"Data mining may therefore be a prom ising supplement in the search for unknown drug-drug interactions in cardiovascular medicine," they concluded.