Study Design

This is a historical matched cohort study conducted at a large academic medical center. Following approval by the Institutional Review Board (IRB; Mayo Clinic, Rochester, MN), we utilized existing electronic databases to identify patients ≥ 18 years old with an implanted IDDS (present at the time of surgery) delivering intrathecal opioid medication who underwent surgical procedures between January 1, 2007 – December 31, 2016 with associated postoperative hospital admission. Furthermore, patients must have received immediate postoperative care in the postoperative care unit (PACU), as our primary outcomes included intraoperative and PACU opioid consumptions. The requirement for written informed consent was waived by the Mayo Clinic IRB. Surgical procedures were defined as any surgical procedure that required general anesthesia or monitored anesthesia care. Patients were excluded if they had previously declined research authorization for review of medical records, classified as American Society of Anesthesiologists’ (ASA) Physical Status VI (brain death), if the IDDS pump was not administering an opioid (e.g. baclofen only), or if a subsequent surgical procedure occurred <1 week from the first surgical encounter, as these patients may have inherent care complexity that could influence perioperative management and postoperative complications.

Patients in the IDDS group were matched with up to two patients without an IDDS. Controls were matched by age (+/− 10 years), sex (exact), ASA physical status score (exact), anesthetic type (general anesthesia or monitored anesthesia care; exact), type of surgical procedure (exact), and length of surgery (+/− 60 min). Within surgical types (e.g. orthopedic surgery lower extremity), patients were matched to similar surgeries to ensure comparability of surgical insults, which may influence perioperative opioid utilization and clinical outcomes. Matching was performed using a computerized algorithm which identifies the “best” match for a particular case. This is done by identifying the pool of all potential matches based on the matching variables that need to be matched exactly and then selecting the control that minimizes the sum of the absolute differences across all matching variables that do not need to be matched exactly.¹⁶