Early Predictors of Long-Term Opioid Use After Traumatic Brain Injury: Insights From a High-Volume, Level-1 Trauma Center in the United States and State Prescription Data

OBJECTIVE: To characterize postdischarge opioid prescription patterns in the year after traumatic brain injury (TBI) hospitalization and examine whether early demographic, injury, and hospital-level factors predict long-term or high-risk opioid use. SETTING: Harborview Medical Center, a high-volume Level I trauma center serving the WWAMI region (Washington, Wyoming, Alaska, Montana, Idaho). PARTICIPANTS: A total of 9647 adults admitted to Harborview Medical Center with a TBI diagnosis between January 1, 2012 and December 31, 2019 who survived their hospitalization. DESIGN: Retrospective cohort study linking 3 data sources: the Harborview Trauma Registry, the University of Washington Medicine electronic health record system, and the Washington State Department of Health Prescription Drug Monitoring Program. MAIN MEASURES: Postdischarge opioid prescription fills were characterized using 4 Bree Collaborative metrics measured quarterly over 364 days after discharge: any opioid prescription, chronic opioids (≥60-day supply), chronic high-dose opioids (≥60-day supply and ≥50 mg/d morphine equivalent dose [MED]), and chronic opioids with concurrent sedatives. Demographic, injury, and inpatient opioid administration variables were examined as potential predictors. RESULTS: Approximately 12% of participants filled an opioid prescription within 90 days of discharge. Among those who did, 26% had refills reflecting chronic use (≥60-day supply), of whom 54% continued filling prescriptions through 1 year postinjury. Neither demographic characteristics nor inpatient opioid administration variables, including average MED, maximum MED, or average MED in the final 48 hours of hospitalization, were significant predictors of postdischarge opioid use across any metric or quarter. CONCLUSION: Although most individuals with TBI do not fill opioid prescriptions after discharge, among those who do, 1 in 4 follow trajectories consistent with chronic use. The absence of identifiable demographic or hospital-level predictors suggests that routine electronic health record-derived variables are insufficient for risk stratification in this population and highlights the risk of relying on unvalidated demographic proxies in clinical prescribing decisions.