At a Tennessee hospital, a nurse stole fentanyl and AI missed it, state records say
Tennessee Hospital Nurse Stole Fentanyl, AI Missed It
At a Tennessee hospital a nurse - At a Tennessee hospital, a nurse’s actions went unnoticed for months, according to state records. The incident occurred at Erlanger Medical Center in Chattanooga, where a nurse was found to have systematically stolen fentanyl from surgical procedures. Despite the use of an advanced AI system, the thefts remained undetected, raising questions about the software’s effectiveness. The Tennessee Board of Nursing later confirmed the nurse’s misconduct, citing a consent order that detailed the oversight.
AI System Fails to Detect Drug Misuse
Drug diversion cases are common in U.S. hospitals, but this one gained attention due to the AI-driven technology in place. Erlanger Medical Center implemented Sentri7, a medication-monitoring software designed to flag discrepancies in drug usage. The system was expected to catch inconsistencies, such as missing doses or unusual patterns, before they became significant. However, state documents reveal that the software failed to alert staff about the nurse’s repeated thefts.
The nurse, identified in the records as John Stevenson, exhibited symptoms of drowsiness and speech slurring during shifts. These signs were noted by anesthesia staff, who initially suspected the issue was related to the nurse’s drug use. After further investigation, Stevenson admitted to stealing fentanyl from surgical cases over several months. The AI system, which should have flagged these anomalies, did not trigger alerts, allowing the theft to continue unnoticed.
Transparency and Accountability Concerns
Experts argue that the lack of transparency in AI systems like Sentri7 could lead to repeated errors. David Rastall, a neurologist and AI researcher, highlighted how proprietary algorithms often obscure the decision-making process. “When AI is involved, mistakes can be hidden rather than corrected,” he said. “This means errors might go unaddressed in other hospitals without anyone realizing.”
Rastals emphasized that transparency is crucial for accountability. “If an AI system misses something obvious, it should be made public,” he explained. “That way, healthcare providers and patients can understand its limitations.” However, current reporting mechanisms allow hospitals to keep AI’s role in such incidents confidential. The Drug Enforcement Administration requires confidential reports of drug thefts, but there is no mandate to detail how AI contributed to the oversight.
Terri Vidals, founder of Rxpert Solutions, questioned whether the AI system was the sole factor in the missed detection. “This is the most basics of basics for the system,” she noted. “It’s intriguing that the software didn’t catch the nurse’s behavior. There might be more to this story.” Vidals’ comments reflect broader concerns about the reliability of AI in drug security, especially when used without thorough validation.
The Tennessee Department of Health inadvertently exposed the AI failure in a routine release of disciplinary records in December. Among the documents was a Board of Nursing order detailing the investigation into the nurse. Stevenson, a nurse anesthetist, signed the order in November, settling the case without facing criminal charges. His license was placed on probation, and he attended drug counseling as part of the resolution. Erlanger Medical Center has not commented on the use of Sentri7 or the specific drug thefts.
A spokesperson for Wolters Kluwer, the Dutch firm behind Sentri7, defended the software’s capabilities. “We believe in the effectiveness of our system,” the representative stated, though they did not address the incident. This response has fueled debates about the need for greater oversight of AI technologies in healthcare. Jacob Smith, a pharmacist at Johns Hopkins Medicine, added, “It’s surprising to see an AI system miss something so obvious. More testing is needed to ensure it’s reliable.”