Accuracy of patient-reported opioid use to verified prescription fills before and after surgery.

Abstract

Objective: To determine the correlation of patient-reported opioid use as compared to fills verified in a prescription drug monitoring program (PDMP) before and after surgery.

Background: Correctly determining prescription opioid use before and after surgery is critical to develop effective policies and care pathways for opioid stewardship perioperatively. While many surgical teams, health systems, and patient registries rely on patients to report their use of prescription opioids, the accuracy of this reporting is unclear.

Methods: Patient-reported data on opioid use from 12 225 adult patients who underwent surgical procedures between 1 January 2018 and 31 October 2019 across 70 participating hospitals in Michigan Surgical Quality Collaborative were compared to the reference standard of the state PDMP. The primary outcome was patient-reported opioid fill in the 30 days after surgery compared to PDMP-verified prescription fill, and the secondary outcomes were patient-reported use in the 30-day and 1-year period prior to surgery compared to PDMP data. Accuracy, sensitivity, specificity, and other predictive values were calculated.

Results: Among 12 225 patients (58% women, 19.2% non-white, mean (SD) age 15.6 (16.3)), the patient report had acceptable accuracy (83.2%, 95% CI 82.5% to 83.9%) and sensitivity (91.2%, 95% CI 90.6% to 91.8%) compared to 30-day postoperative opioid fills in the PDMP, while less than half of patients without PDMP-verified fills accurately reported not filling an opioid prescription (specificity 43.1%, 95% CI 40.9% to 45.3%). For 30-day and 1-year periods before surgery, patient-reported opioid use had acceptable accuracy (86.8% (95% CI 86.1% to 87.4%) and 77.9% (95% CI 77.2% to 78.7%), respectively) and specificity (90.0% (95% CI 89.4% to 90.5%) and 90.8% (95% CI 90.2% to 91.4%), respectively), while only slightly above half of patients with PDMP-verified opioid fills reported opioid use before surgery (sensitivity 58.9% (95% CI 56.2% to 61.7%) and 47.3% (95% CI 45.7% to 49.0%), respectively).

Conclusions: For perioperative opioid fills and use, the patient report appears to align with PDMP data. The patient report may also complement PDMP data by identifying persons who may report opioid prescription fills or use without corresponding data in the PDMP, especially in the postoperative period.