How AI can aid those battling the drug diversion crisis in healthcare

Advanced, predictive computing and data management are helping to transforming the fight against medication misappropriation by clinicians.

Drug overdose deaths continue to rise, claiming about 100,000 American lives every year. But lost lives represent only one of the risks created when prescription drugs are consumed outside the healthcare system, as when employees divert medication from their workplaces.

It happens too often. Although nurses are the backbone of health care, an estimated 18 percent of nurses, or more than 900,000, have problems with substance use. Opioids – including fentanyl, hydrocodone, hydromorphone, morphine and oxycodone – are the most commonly diverted drugs from hospitals, and they flow through healthcare organizations in large volumes every day. Anyone with access could be at risk of misappropriating or diverting controlled substances from hospitals, pharmacies, clinics, nursing homes and emergency care centers.

Patients, hospitals and staff all suffer if a healthcare worker diverts controlled substances for their own use and cares for patients while under the influence of these addictive substances. Potentially serious medical errors can occur. Patients may receive inadequate pain control if they are denied the medication they were prescribed or become infected with a blood-borne pathogen after receiving medication that was tampered with.

The individual diverting these substances faces consequences such as disciplinary action from licensure agencies, criminal charges and, in the worst-case scenario, loss of life because of an accidental overdose. The organization suffers when there are heavy federal fines incurred because of a lack of controls preventing diversion events.  These financial penalties drive up healthcare costs and damage the organization's reputation in the community.

Addressing the problem

Healthcare organizations have tackled the drug diversion problem for years, but they have experienced varying levels of success. Special teams of employees have worked long and hard – often using manual auditing methods – to detect, prevent and deter diversion.

Medication intelligence solutions can help healthcare organizations become safer, more efficient and enhance compliance with controlled substance regulations. Ideally, hospital systems could use these to automatically and conclusively track every controlled substance dose through the entire medication use process from purchase to vault, dispensing cabinet, patient administration or waste stream. This closed-loop, dose-reconciliation capability also would be able to identify each individual handling a drug and the medication usage behavior of individuals handling the drugs. As a result, an organization would be immediately alerted to any single misappropriated dose of medication so supervisors could investigate and close the gap. 

Fortunately, sophisticated controlled substance diversion-prevention technology solutions like these are available to ensure safety for all.

The data challenge

Although automation sounds easy, these solutions need to be sophisticated to work. Suspicious activity is the needle in the haystack, and detecting it is a data problem. Here are the major data technology improvements that the drug diversion technology industry has employed to improve detection.

Second-generation artificial intelligence. First-generation drug diversion prevention technology captured anomalies based on standard deviations from normal volume or activity. Coming after the fact, an alert could take weeks or months to be identified and investigated. With artificial intelligence, a diversion prevention solution can detect any missing dose in near real-time and provide workflow tools to close the gap. It should be able to predict – revealing patterns, conditions and vulnerabilities that could presage a diversion. And it should flag high-risk individuals, assessing factors like poor documentation, medication trends, time lags, movement and wasting practices to help teams target, focus and manage investigation efforts. Second-generation AI is here and increasingly providing these capabilities. Newer solutions now digest multiple data streams, enabling actionable analytics and less siloed data.

Better compliance. Gaps in medication documentation have been common in healthcare organizations, complicating responses to pharmacy boards and DEA auditors. However, the newest generation of diversion-prevention logic enables one-click daily audits that report in near-real-time on the cradle-to-grave approach to ensuring controlled substance accountability. This complete audit trail helps organizations successfully respond to pharmacy boards and DEA inquiries and avoid hefty fines.

Less dependence on humans. Diversion-detection solutions increasingly use artificial intelligence and machine learning to proactively identify individuals most at risk for diversion. These solutions also can drive workflow efficiencies for those who interact with the software. Newer diversion prevention technologies are integrating “unsupervised” AI, which continuously learns and relearns the model of what constitutes normal circumstances, so that users who stand out as high risk for diversion are identified. With unsupervised machine learning, humans do not need to spend time updating algorithms. Instead, they can focus their time on investigating clinicians who have high diversion risk scores.   

Near-real time updates. Until recently, there has been a lack of diversion software with near-real time data to provide controlled substance accountability at the unit dose level and feed transactional data into their algorithms. But that’s absolutely necessary to optimize pattern recognition and keep the machine continuously learning. Controlled substance accountability and readily retrievable documentation is more important than ever as hospitals experience more unannounced audits from the DEA.

Increasingly, major healthcare organizations are acquiring software solutions and establishing processes to address these concerns. The ideal diversion detection tool uses metric-based dashboards to automatically detect more diverters earlier, enabling those running programs to spend less time wading through spreadsheets or processing old, misleading data.

Although drug diversion still occurs in healthcare facilities, the deed finally has a worthy nemesis – a combination of technology, good data management and people committed to doing the right thing. Everyone benefits from this safety net, starting with the patient.

Sandy Still, Pharm.D., is the director of clinical strategy at Bluesight.

More for you

Loading data for hdm_tax_topic #care-team-experience...