Antidepressant Drug-Drug-Drug Interactions and Unintentional Traumatic Injury: Pharmacoepidemiologic Screening to Detect Three-Way Drug Interaction Signals

Background: Antidepressants are associated with traumatic injury and are widely used together with other medications. The high rate of polypharmacy among antidepressant users raises the concern for potential drug interactions leading to unintentional traumatic injury. While several studies have investigated pairwise antidepressant interactions associated with injury, higher-order interactions involving more than two drugs (such as drug-drug-drug interactions [3DIs]) remain understudied.

Objectives: To generate hypotheses regarding potential antidepressant 3DI signals associated with elevated injury rates.

Methods: We conducted high-throughput pharmacoepidemiologic screening of 2000-2020 Optum's de-identified Clinformatics® Data Mart, identifying 3DI signals by performing self-controlled case series studies for drug triads consisting of an antidepressant + co-dispensed drug (base pair) with a candidate interacting precipitant. For each triad, we included persons aged ≥16 years who: 1) received the base pair; 2) experienced an injury during base pair therapy; and 3) used the candidate interacting precipitant concomitantly with the base pair. We compared injury rates during observation time exposed to the drug triad vs. the base pair only, adjusting for time-varying covariates. We calculated adjusted rate ratios (RRs) and 95% confidence intervals using conditional Poisson regression and minimized type 1 errors via semi-Bayes shrinkage.

Results: We identified 147,747 eligible antidepressant users with an injury. Using this sample, we studied 120,714 antidepressant triads, of which 334 (0.3%) were associated with elevated rates of injury and thus considered potential 3DI signals. Adjusted RRs for signals ranged from 6.60 (3.23-13.46) for escitalopram + simvastatin with aripiprazole (vs. escitalopram + simvastatin without aripiprazole) to 1.31 (1.04-1.65) for sertraline + levothyroxine with tramadol (vs. sertraline + levothyroxine without tramadol) . Nearly half of the signals (134, 40.1%) had adjusted RRs > 2.0, suggesting relatively strong associations between these drug triads and elevated injury rates.

Conclusions: Using real-world data, we identified 334 antidepressant triads associated with elevated injury rates. Our findings represent 3DIs of potential clinical concern and may warrant future etiologic studies aimed to test these hypotheses.