Pharmacoepidemiology
Feasibility of applying pharmacoepidemiologic drug-drug interaction screening methods to a population of nursing home residents: an application to clopidogrel Adam M. D’Amico* Adam M D’Amico Charles E. Leonard Daniel A. Harris Yu-Chia Hsu Lori A. Daiello Douglas P. Kiel Andrew R. Zullo
Background: The self-controlled case series (SCCS) design has been used to screen potential drug-drug interactions (DDI) and inform etiologic research among community-dwelling older adults. The nature and extent of DDI among nursing home (NH) residents remains underexplored despite high rates of polypharmacy, frailty, and adverse events in this population.
Objective: To assess the feasibility of using SCCS-based screening methods to identify potential DDIs among NH residents.
Methods: We used Minimum Data Set and Medicare data from 2013 to 2020 to identify NH residents ≥65.5 years old who were enrolled in Medicare fee-for-service and received the object drug clopidogrel. New use of clopidogrel was identified using a 6-month washout. Among residents hospitalized for a major bleed, SCCS studies were conducted to test all common (30+ users) precipitant drugs for potential DDIs with clopidogrel. Rate ratios (RR) and 95% CIs for major bleed were calculated for periods of concomitant use versus clopidogrel use alone. We used semi-Bayes adjustment to reduce false positives from multiple estimation. An expert panel reviewed results for biological plausibility.
Results: We identified 987 major bleed events. Of 40 precipitant drugs analyzed, five (12.5%) had RR and 95% CI >1: amoxicillin-clavulanate [2.18 (1.19, 3.98)], sertraline [2.09 (1.14, 3.80)], potassium [2.02 (1.28, 3.21)], cephalexin [1.96 (1.11, 3.48)], and tramadol [1.90 (1.13, 3.19)]. Sertraline was the only signal considered biologically plausible.
Conclusion: Despite using a common object drug associated with adverse events and eight years of Medicare claims, this analysis yielded few DDI signals to inform future studies. Given the size of the NH population and rarity of many outcomes, SCCS-based screening methods face significant challenges when applied in the NH setting. Pairing SCCS findings with robust clinical review is important when triaging potential DDIs for subsequent etiologic work.