Presentation (CTI)

OBJECTIVES: To evaluate variability across commonly referenced drug-drug interaction databases (DDIDBs) in identifying and classifying DDIs in prescriptions from a tertiary care hospital and assess for clinical decision support (CDSS).
METHODS: A cross-sectional study of 80 patients' prescriptions was conducted at a tertiary care hospital in India. Prescriptions containing ≥2 systemic medications were screened, identifying 185 DDI pairs. These pairs were evaluated using most common DDIDBs (e.g., Micromedex®, UpToDate®, DrugBank®, Drug.com®, Medscape®). Presence, severity, and documentation of each interaction were recorded. Discrepancies across databases were analyzed descriptively, and clinical relevance was interpreted through expert review. Severity consistency was calculated for each resource as the proportion of DDI pairs with matching severity classifications.
RESULTS: Among the 185 DDI pairs, detection varied across databases: Micromedex® 73, UpToDate® 85, DBIC 105, DDIC 78, and Medscape® 56. Severity consistency scores were: Micromedex® 27.4% (20/73), UpToDate® 28.2% (24/85), DBIC 43.8% (46/105), DDIC 62.8% (49/78), and Medscape® 55.4% (31/56). Inter-database agreement was low to moderate, especially for supportive care and pharmacokinetic interactions. Expert review indicated that reliance on a single database could miss clinically significant interactions, affecting patient safety and decision-making.
CONCLUSIONS: Significant variability exists among DDIDBs in detecting and classifying DDIs, with notable differences in severity consistency. This inconsistency may limit CDSS reliability and compromise clinical decisions. Context-specific guidance or hospital adapted protocols are needed to improve safe medication management in India.