Increasingly, medical overuse—including the ordering of unnecessary tests for patients—is being characterized as a patient safety problem. In this retrospective study involving inpatients treated at three academic medical centers, researchers found that ordering laboratory tests with low diagnostic yield was common. Implementing machine learning models, as they did in this study, has the potential to promote high-value care by characterizing the diagnostic utility of information gained from ordering tests.

This retrospective audit of electronic health record orders assessed the performance of bundled computerized provider order entry into order sets. Researchers examined how often order sets were supplemented with additional orders, retraction of orders within sets, infrequent ordering of specific components of order sets, and use of individual orders when sets were available. These workarounds occurred frequently and prompted the authors to conclude that existing electronic order sets do not align with frontline clinician needs.

Providing insights from a physician regarding the complexity of prescribing opioids safely, this commentary discusses several elements of the care environment that can affect safe pain management, including patient-centeredness, uncertainty, bias, collaborative decision-making, and hierarchy. A WebM&M commentary described best practices for opioid prescribing.