Exploring Risks of Adverse Drug Events From Single Drug to Three-Drug Combination

Use of three-drug combinations is common and could increase the risk of adverse drug events (ADEs). Signals of ADEs from three-drug combinations could be identified from real-world data, and the risks of ADEs from single drug exposure to three-drug combination exposure could be illustrated by a graphic model. We used a US nationwide insurance claim database. We explored the relationship between drug exposure (from single drugs to three-drug combinations) and risks of ADEs (acute kidney injury, gastrointestinal bleeding, and opioid-related ADE). We used a conditional logistic regression model to estimate the odds ratios (ORs), the posterior probability of null hypothesis to control false discovery rate (FDR), and a graphic model to illustrate the ORs from single drug exposure to three-drug combination exposure. We derived approximately 1.7 million covariate-matched ADE-case-control pairs and investigated approximately 0.25 million three-drug combinations. We identified 3412 signals of adverse three-drug combinations (FDR < 0.01 and OR > 2). For the signals, we observed the medians and interquartile ranges (IQRs) of ORs increased from single drug exposure (median = 1.03, IQR: 0.95-1.15), through two-drug combination exposure (median = 1.45, IQR: 1.23-1.75), to three-drug combination exposure (median = 2.95, IQR: 2.43-3.91). We built a webpage-based tool to illustrate the ORs from single drug exposure to three-drug combination exposure. In conclusion, certain three-drug combinations are associated with an increased risk of ADE, and the risks of ADE could increase from single drug exposure to three-drug combination exposure.