AI Drug Monitoring Software Overlooks Months of Fentanyl Theft at Tennessee Hospital, Records Reveal
Roughly a year ago at Erlanger Baroness—the largest hospital in Chattanooga—anesthesia staff noticed a nurse slurring his words and struggling to stay awake while on duty in the surgery center, according to a Tennessee Board of Nursing consent order. Days later, the nurse failed a drug test and was fired. He later admitted to state investigators that for months he had been stealing and misusing leftover fentanyl, sometimes daily.
Under most circumstances, this would be a routine case of what the industry calls "drug diversion"—the illegal taking of controlled substances from healthcare facilities, a problem so pervasive it is believed to occur in nearly every U.S. hospital. But the Erlanger case stands out because a high-tech watchdog was supposed to be on guard.
The hospital relied on Sentri7, medication-monitoring software powered by artificial intelligence and designed to detect missing drugs faster than any human could. Yet for months, the system failed to raise alarms, overlooking missing drugs and other inconsistencies that "should have been flagged," the nursing board's order states.
The case, which has not been previously reported, offers a rare glimpse into an apparent failure of AI drug-diversion software now used in hundreds of U.S. hospitals with little transparency or oversight. Healthcare facilities are not required to disclose their use of such software or report malfunctions, leaving no comprehensive record of how widely these systems are deployed or how often they fall short.
Erlanger Baroness declined to comment on its use of Sentri7 or the diverted drugs. André Rebelo, a spokesperson for Wolters Kluwer’s health division—the Dutch company behind Sentri7—also declined to discuss the incident but said the company remains "confident in our software."
David Rastall, a Johns Hopkins neurologist and AI researcher, said the lack of transparency around proprietary AI technology allows errors to go unexamined and potentially repeated at other hospitals. "The ideal for patients, caregivers, and hospital systems would be when an AI is found to be making some type of error, that becomes very transparent and public," he said.
While the Drug Enforcement Administration mandates that hospitals confidentially report lost or stolen drugs, those reports need not include details about any AI involvement, according to three drug-diversion prevention experts interviewed. All said they had never before seen an apparent AI failure documented as public as the one at Erlanger.
Jacob Smith, a pharmacist overseeing drug security at Johns Hopkins Medicine, called the Sentri7 failure puzzling. "It doesn't make sense to me how you could miss it," he said, noting that diverting leftover fentanyl is one of the most well-known methods of theft.
Some experts questioned whether the problem was user error rather than a software glitch. Terri Vidals, founder of Rxpert Solutions, said, "This is the most basics of basics for this software. I find it interesting that they're saying it wasn't flagged by the software. I think there's maybe more to that story."
The apparent failure came to light through a routine release of disciplinary orders by the Tennessee Department of Health in December. The board’s order summarizes an investigation into nurse anesthetist John Stevenson, who settled by signing the document in November. Stevenson declined to comment through his attorney and has not been charged with a crime. His license was put on probation while he undergoes drug counseling.
Kristy Drollinger, a Wolters Kluwer executive, said Sentri7 monitors about 60 "attributions of risk" to flag potential diversion. "Every health system, every health facility, has had diversion at some point—and probably has it now," she said.
Drug diversion remains a widespread challenge. The Centers for Disease Control and Prevention links it to at least 13 disease outbreaks since 1985, causing more than 200 infections, mostly hepatitis C. Hospitals have tried to track every pill and vial using electronic medication cabinets and patient records, but the task has largely shifted to AI-driven software like Sentri7 and Bluesight’s ControlCheck.
More than 1,500 hospitals use ControlCheck and 700 use Sentri7 Clinical Surveillance programs, according to the companies. Neither publishes pricing. Smith noted hospitals buy these expensive tools to avoid multimillion-dollar DEA fines, not because they promise a return on investment.
A 2022 NIH-funded study found that Sentri7—then called Flowlytics—could uncover diversion faster than humans. That study was led by a former employee of Invistics, the software’s previous owner. It tested the software on two years of data from 10 hospitals, finding all 22 known diverters faster by up to 18 months. But at Erlanger, humans spotted the trouble first.
According to the nursing board order, coworkers reported Stevenson appeared impaired on or around June 30, 2023. He admitted to diverting fentanyl waste from March 2023 onward, escalating to daily use by June. A hospital audit found roughly five instances where Sentri7 did not flag missing drugs, plus other inconsistencies that should have been caught.
The board noted Sentri7 was in its "initial learning phase" at Erlanger, though Wolters Kluwer says the software has no such phase—it trains on 9 to 12 months of historical data when first implemented. Smith of Johns Hopkins offered another theory: AI drug-diversion software may be less effective in operating rooms, where drugs are dispensed and charted differently than in emergency rooms or ICUs.
"We've got people whose entire job is to work with this software," Smith said. "The software is a piece of it, but if you rely on the software to give you all your signals, you'll miss stuff. It's just not 100%."