Study links illicit fentanyl exposure to higher overdose risk than medical use

Researchers used electronic health records from Epic Cosmos, a U.S. database covering more than 300 million patients in inpatient and outpatient care. They studied records from January 1, 2015, through December 31, 2023.

The final study cohort included 295,728 patients with evidence of fentanyl exposure. Of these, 85,535 people, or 28.9%, were placed in the medical-source cohort. Another 210,193 people, or 71.1%, were placed in the illicit-source cohort.

The researchers did not directly observe where every person obtained fentanyl. Instead, they linked urine drug test results with medication records using timing-based screening methods. A positive urine test alongside relevant fentanyl medication records was used to identify medical-source exposure, while other positive-test patterns were used to infer illicit-source exposure.

That distinction matters because a urine test can show fentanyl exposure, but it cannot on its own identify the source. The study’s labels are therefore research classifications based on available clinical records, not direct confirmation of each patient’s drug source.

The team tracked opioid-related harmful outcomes over time. These included nonfatal opioid overdose, opioid abuse, and opioid dependence. Dependence means the body has adapted to a drug and a person may have withdrawal symptoms when use stops.

Nonfatal overdose was consistently more common in the group classified as having illicit-source exposure. By the end of the study period, overdose prevalence reached 18.6% in that group, compared with 4.3% in the medical-source group. The illicit-source group also had higher prevalence of opioid abuse and dependence throughout the study period, although the prevalence in that group declined after 2020, especially for dependence.

The researchers also examined outcomes during the 30 days after fentanyl initiation. After adjusting for demographic factors, social vulnerability and 37 existing health conditions, illicit-source initiation was associated with a higher risk of each outcome than medical-source initiation.

For nonfatal overdose, the adjusted hazard ratio was 2.99. In plain language, the estimated short-term overdose risk was nearly three times as high in the illicit-source group as in the medical-source group. The adjusted hazard ratio was 1.96 for opioid abuse and 3.05 for opioid dependence.

These are associations from a retrospective study, meaning the researchers looked back at existing records rather than assigning people to different types of fentanyl exposure. Even with statistical adjustment, other differences between the groups may help explain part of the gap in outcomes.

The study also cannot measure fentanyl use that never entered the health system or was not captured in the database. It did not report fatal overdoses because cause-of-death information was not part of the dataset.

The findings point to a large burden of harm among patients classified as having illicit-source fentanyl exposure. But more research is needed to improve ways of identifying exposure sources, understand how medical and illicit fentanyl use may overlap over time, and examine factors that may contribute to these risks.