Abstract
PURPOSE: To compare the co-prescription of metoprolol and potent CYP2D6-inhibiting antidepressants before and during a 10-year period after implementation of an optimized drug interaction database into clinical decision support systems in Norway. METHODS: The study was a retrospective, cross-sequential nationwide analysis of drug-dispensing data retrieved from the Norwegian Prescription Database over a 1-year period before (2007) and two 1-year periods after (2012 and 2017) implementation of a drug interaction database providing recommendations on non-interacting alternative medications. Primary outcome was changes in co-prescription rates of metoprolol and the potent CYP2D6-inhibiting antidepressants fluoxetine, paroxetine, or bupropion relative to alternative antidepressants with no or limited CYP2D6 inhibitory potential. To control for potential secular trend bias, a comparison group consisting of atenolol/bisoprolol users was included. RESULTS: The co-prescription rate of metoprolol with potent CYP2D6 inhibitors declined following implementation of the optimized database, by 21% (P < 0.001) after 5 years and by 40% (P < 0.001) after 10 years. Compared with atenolol/bisoprolol users, patients treated with metoprolol had significantly reduced likelihood of being prescribed a CYP2D6-inhibiting antidepressant in the two post-implementation periods (OR 0.61 (95% CI 0.54-0.69) and OR 0.45 (95% CI 0.40-0.51), respectively, versus OR 0.84 (95% CI 0.74-0.94) prior to implementation). Small and mostly insignificant differences in average daily metoprolol dosage were found between patients treated with the various antidepressants. CONCLUSION: The present study suggests that implementation of a drug interaction database providing recommendations on non-interacting drug alternatives contributes to reduced co-prescribing of drug combinations associated with potentially serious adverse effects.