Drug-drug interactions in an intensive care unit of a tertiary hospital in southern Chile: Evaluating databases agreement

Context: Patients in intensive care units have a high risk of experiencing a pharmacological interaction due to complex pharmacotherapy, severe disease, and comorbidities; increasing the risk of adverse effects of medications. Electronic databases are useful sources to identify drug-drug interactions (DDI), especially when new therapeutic alternatives are added to conventional treatments. Aim: To identify the frequency and severity of potential drug-drug interactions (pDDIs) in ICU patients using three electronic databases. Methods: Clinical pharmacists collected data on medication dosage and route of administration, sex, age, length of stay, comorbidities, and APACHE II score using patient records. Micromedex, Medscape, and Lexicomp databases were used to identify and categorize pDDIs. Intensivists confirmed if a pDDI was clinically present. kappa concordance test was utilized as a measure of agreement among databases. Results: Of the 93 ICU patients studied, pDDIs were identified in 89. A positive incremental relationship was found between number of medications, length of stay, and number of pDDIs. Patients with respiratory pathologies were most predisposed to presenting DDIs. Agreement among databases was mixed. Intensivists confirmed 5% of pDDIs. Conclusions: Discrepancies among databases and in intensivist judgment highlight a significant information gap in the identification of DDIs.