The structured and unstructured data in an electronic health record can be harnessed together to improve medication reconciliation and detect discrepancies between medication orders and patient adherence, according to a new study in BMC Medical Informatics and Decision Making.
Medication discrepancy is a significant problem with the potential to have substantial negative impact on patients. However, medication reconciliation to detect matching and discrepancies is not adequately performed.
The researchers built a computerized algorithm for medication reconciliation using state of the art Machine Learning and Natural Language Processing. Their goal was to develop a program that could compare medication information from patient discharge orders (structured data) to clinicians' notes (unstructured data). To assess the algorithm, they used data on patients in the Complex Care Medical Home Program at Cincinnati' Children's Hospital Medical Center.
They found the algorithm to be effective. It achieved a "much improved recall" on matched and discrepant medications and in detecting discrepant medications in the discharge summaries. The algorithm wasn't perfect, though; there were some errors in detection and mismatching. For instance, the clinical notes may have used an abbreviation, had a misspelling or used an uncommon medication name that the algorithm did not match to the discharge summary.
"However, even at this early stage of development, automated medication discrepancy detection shows a promising outcome in assisting medication reconciliation," the researchers said. "Consequently, we hypothesize that the computerized algorithm, when transferred to the production environment, will have potential for significant impact in reduction of effort for conducting medication reconciliation in the clinical practice setting."
The industry continues to refine EHRs to better improve outcomes and patient safety. Strides have been made to tailor the systems to reduce errors. ONC has recently reported that health IT and the Meaningful Use of EHRs have had a "dramatic" impact on patient safety but noted that EHRs are not "magic bullets" and that further research is needed.
To learn more:
- read the study (.pdf)