What Clinical Decision Support Can Offer
Use case #3: Safety in diagnostic procedures
In addition to supporting more accurate or complete diagnoses, CDS can assist with improving safety in diagnostic procedures, especially related to radiology and the overuse of tests.
For example, if a patient presents with shortness of breath, the clinician may be inclined to order a lung scan to rule out pulmonary embolism. The CDS can calculate the patient’s risk level, based on a combination of presentation details and lab results. If the patient is at low risk of pulmonary embolism, the CDS can issue a prompt suggesting a D-dimer—a noninvasive blood test that is very reliable in prediction value—before advancing to the lung scan. If the clinician opts to continue with the scan, the CDS can prompt the clinician to document the reason for overriding the alert.
In many cases, promoting the recording of a reason for ordering radiology, combined with a prompt that suggests a safer procedure, can steer clinicians toward a safer diagnostic course. Doing something simply because it has always been done that way may seem reasonable—at least, in some cases, until CDS-presented best practices suggest otherwise.
A word about alert fatigue
Discussions of CDS systems often turn toward concerns about their potential contribution to alert fatigue, in which clinicians ignore alerts that activate too often and are sometimes irrelevant. While any system that issues alerts seemingly has the potential for contributing to alert fatigue, that’s not an issue for CDS systems that work properly. There are a couple of requirements to ensure proper functioning. The first is that the system must act upon all patient information as in the above examples. The second is that CDS programming should be an ongoing process in which the system itself provides assistance.
For optimal helpfulness in alerts, the system should record clinicians’ responses to prompts, identifying instances in which alerts are followed, ignored, or aren’t followed expeditiously, indicating a lack of cohesive timing between the prompt and the clinicians’ workflow. If the system then relays this information automatically to the CDS vendor, the vendor has the opportunity to improve CDS usefulness for a more satisfying user experience.
When it all works accurately, as in averting potentially unsafe LPs, detecting developing sepsis, and halting unnecessary radiology—and in situations as simple as alerting the clinician to a mother who has been in a latent stage of labor with ruptured membranes for too long without appropriate intervention, or to a patient who has been sitting in triage for two hours—the number of prompts corresponds to the number of mistakes about to happen or opportunities for care improvements. For most clinicians, that means very few prompts, with the ones that do arrive being highly welcome.
Nancy Zimmerman is chief nursing officer for medCPU. She is a leader in the field of nursing technology, has extensive clinical experience, serving as a consultant to the school health board in Atlanta, where she created and established a clinical school health program. She has worked as a nurse for high-risk labor and delivery patients and as a certified perinatal educator at Northside Hospital in Atlanta. Zimmerman may be contacted at nzimmerman@medcpu.com.