Perioperative Dexmedetomidine and 30-Day Outcomes After Adult Cardiac Surgery

This observational study uses de-identified electronic health record data from the TriNetX Global Collaborative Network (2010-01-01 to 2025-07-01) to examine whether perioperative dexmedetomidine (DEX) is associated with 30-day outcomes after adult cardiac surgery. Adults aged 18-100 years undergoing first-time coronary artery bypass grafting or heart valve surgery are included. Exposure is any DEX administration from 24 hours before to 48 hours after the index operation; comparators receive no DEX in this window. The primary outcome is delirium within 30 days. Secondary outcomes are 30-day all-cause mortality, acute kidney injury, pneumonia, sepsis, red blood cell transfusion/major bleeding, myocardial infarction, ischemic stroke/transient ischemic attack, atrial fibrillation, mechanical ventilation \>96 hours (days 4-30), and 30-day readmission (days 1-30). No treatments are assigned by investigators and no identifiable information is used. Findings aim to inform perioperative sedation strategies in routine cardiac surgery care.

This is a retrospective cohort study using de-identified data from the TriNetX Global Collaborative Network (2010-01-01 to 2025-07-01). Adults aged 18-100 years undergoing first-time coronary artery bypass grafting or heart valve surgery will be identified. Perioperative exposure is defined as any dexmedetomidine administration from 24 hours before to 48 hours after the index surgery; the comparison cohort consists of patients without dexmedetomidine in this window. Clinical outcomes within 30 days after surgery include delirium as the primary outcome and a range of major postoperative complications and mortality. Propensity scores based on demographics, comorbidities, baseline medications and laboratory tests will be used to match or adjust between exposure groups, with balance assessed using standardized mean differences. Time-to-event analyses will use Cox proportional hazards models and Kaplan-Meier curves, with multiplicity controlled using the Benjamini-Hochberg false discovery rate. All analyses use only de-identified data within the TriNetX platform, with no direct contact with patients or access to identifiers.