New tool set for clinical testing at Cincinnati Children’s in 2018
CINCINNATI – Researchers at Cincinnati Children’s Hospital Medical Center have developed a novel real-time computerized system that automatically increases medication administration accuracy and rapidly alerts providers to correct any errors. The new patient safety enhancement system collects and processes information from several electronic health record (EHR) sources simultaneously to monitor medication administration with a high degree of sensitivity.
A recent pre-clinical test of the system described in the Journal of the American Medical Informatics Association found that it achieved 85.3 percent sensitivity in detecting errors, a 20-fold improvement over traditional methods. The new system also showed potential to reduce patient exposure to possible harm by as much as 86 percent. It is set to be tested in clinical practice during rollout in Cincinnati Children’s neonatal intensive care unit (NICU) in 2018.
Importantly, the new system also achieved 78 percent positive predictive value, suggesting that for every ten error notifications, only two were false positive alarms. Alert fatigue caused by false alarms is a major contributor to clinicians ignoring useful alerts. Minimizing false alarms is a major goal of such IT systems.
“Our system introduces a novel computerized paradigm for medication safety, to keep medication administration errors at the lowest possible level,” says lead study author Yizhao Ni, PhD, assistant professor of Biomedical Informatics. “Compared with most systems that use static formulas for one-off assessment, this system applies a set of complex algorithms including natural language processing technologies to continuously monitor information from multiple sources in the medication use process.”
The researchers have developed and fine-tuned both the algorithms and the system iteratively over a number of prior National Institutes of Health-funded studies. Following the current study, they made additional system refinements before its testing in an actual clinical setting this year.
“The system we’ve developed allows timely and efficient error detection. Very few other systems have the real-time capability to detect these errors and notify providers quickly with a high degree of accuracy,” says senior author Eric Kirkendall, MD, MBI, the Associate Chief Medical Information Officer at Cincinnati Children’s.
Complex medications, complex environment
The study focused on medication administrations in patients in the Cincinnati Children’s NICU over a four-month period in 2017, representing a total of 3,462 patient days. A NICU is a complex, sometimes hectic environment where the highest level of care is provided to complex and critically ill newborns. This kind of environment can make detection and prevention of errors difficult by traditional methods.
The researchers used the system to monitor data related to ten continuous infusion medications. Such infusions carry higher risks than other medication administrations because they often span multiple clinical shifts and involve frequent and complex dosage adjustments.
“Neonates are a particularly vulnerable patient population when it comes to medication administration. Their weights change rapidly and doses must be adjusted frequently. This system was created to make the already excellent care we give to neonates as perfect as possible,” says study coauthor Kristin R. Melton, MD, associate professor in the Division of Neonatology and Pulmonary Biology and a practicing NICU physician.
Harnessing computers to improve safety
Cincinnati Children’s already has a number of NICU medication safety interventions in place. These include the use of computerized provider order entry with embedded clinical decision support, a bar code medication administration system, smart infusion pump technology with a customized neonatal library of medications, daily prescription review by dedicated NICU pharmacists, and clinical guidelines for high-risk medications.
The new system provides a complementary solution to capture medication administration errors that might be missed by current safeguards. It gathers real-time medication use information from disparate sources using the institutional EHR system (Epic). It then identifies any medication use discrepancies using logic-based rules. In particular, it employs a set of regular expression-based natural language processing algorithms to analyze free-text physician-to-nurse communication orders that include complex dose/rate adjustments.
If medication errors do occur, they are rapidly detected and delivered to providers via a real-time messaging platform (using a secure, HIPAA-compliant implementation of Skype for Business). This allows medical staff to reduce the time it takes to make adjustments to dosing levels.
Cincinnati Children’s will continue to study the technology by clinically integrating the system in early 2018. The NICU will be the first clinical environment studied, but the research team has already begun evaluating the portability to other environments including other intensive care unit and adult settings at the University of Cincinnati Medical Center.
This work was supported by the National Institutes of Health (grant numbers 1R01LM012230, 1U01HG008666, 5U18DP006134, 1R21HS024983) and by internal Cincinnati Children’s funds.