2019 Imaging Innovation Award Winner: An Initiative to Power Clinical Decision Support with Structured Reporting of Positive Ultrasound Exams for Deep Vein Thrombosis

By UT Southwestern Medical Center at Dallas and Parkland Health & Hospital System

Doppler ultrasound for suspected deep vein thrombosis (DVT) is a fairly targeted exam with limited outcome options, which made it ideal for the first use of a concept we’d been exploring: triggering events in the EHR based on how we report exams when using structured reporting. Although delayed treatment for identified DVT was not considered a leading concern, our institution is a training environment with numerous handoffs, each of which raises the potential for losing information. In discussions with the hospitalist service, we decided this would be a useful tryout of a helpful safety system—as long as it only alerted when apparent delays in treatment initiation occurred. We engaged a lead hospitalist during the conceptualization and implementation of this project. This, in turn, facilitated communication with the treating services.

Aims and objectives. We wanted to use the report created by the radiologist to trigger clinical decision support (CDS) within the EHR for inpatient and emergency department patients. Using structured reporting transmitted discretely as a data element, the EHR could identify when a Doppler ultrasound of the extremity was reported as positive for DVT. This then initiated automated assessment of the patient state by the electronic system. If the patient did not have anticoagulation medication orders two hours after a positive ultrasound exam, and did not have an International Normalized Ratio of greater than 2, the EHR’s CDS system would trigger a popup window for the next provider viewing the patient’s chart. This interruptive alert would display the date and result of the imaging exam, recent pertinent labs and links to orders for anticoagulation medications.

There also are options to acknowledge receipt of treatment guidance, such as a patient’s contraindication to anticoagulation, which would close out the alert. The primary aim was to ensure timely treatment for patients testing positive for DVT. Secondary aims were to be able to track the positive rate for ultrasound testing for DVT at an enterprise level and lay the foundation for a standard framework for structured reporting to impact other clinical scenarios.

Leadership and project management. This was a true team effort. The radiology informaticist and a structured reporting champion championed the project. A lead hospitalist and deputy medical informaticist for the hospital also contributed key input. Radiology department leadership and the hospital’s chief medical information officer approved the project, and the CMIO provided access to the technical support services of the EHR for the CDS build. The imaging IT team managed the interface with the reporting application. The structured reporting champion, radiology division chiefs and radiology department chair promoted use of the structured reporting templates. The division chiefs monitored and reinforced adherence to use of the templates. Communication of this system to the hospital and emergency department teams was handled through the standard enterprise communication systems for the hospital used for all technical updates. Outcomes of the project were monitored through the EHR reporting systems. The lead hospitalist embedded with the care teams in the hospital and could “hear” the voice of the customer.

Key steps. One of the linchpins to the success of this project was the enterprise-wide adoption of system standard report templates by the radiology practice. This took a few years and many discussions. In fact, the radiology department had accepted the process as a standard best practice by the time of this project’s inception. The report template for Doppler ultrasound for suspected DVT had already gone through several iterations during active clinic use. Prior to the initiation of this project, the impression field for this template was fixed into a set picklist with six categories of interpretive outcomes separating out negative, positive (four variant options) and superficial thrombus-only results. With a now-codified impression field in hand, we activated transmission of this report element into the EHR as a discrete data element separate from the report text. We then engaged the hospitalist informaticist in discussions on how this data element could be utilized. From these conversations we decided the EHR’s CDS would be a good target.

Knowing that people generally dislike interruptive alerts, we outlined logic to prevent any alerting where providers were practicing as expected: We would only alert for apparent delays in treatment. Project leaders presented the idea to radiology and hospital leadership, who approved and authorized use of technical support personnel time. We built the EHR alert to guidelines and activated it in “silent” mode, meaning it would trigger but no one would see it in production. We then reviewed the times for alerting. The radiology informaticist reviewed to make certain it alerted for positive exams and did not alert for negative exams. The hospital informaticist reviewed to make sure it only alerted in the correct clinical scenario. These reviews resulted in a few rounds of logic modification. Once alert triggering appeared accurate in production, we took the alert out of silent mode and made it visible to treating providers. The EHR reporting application monitored rates of alerting, and the hospitalist informaticist informally surveyed his coworkers with regards to the alerting. No additional modifications were identified as needed, and the system currently remains active.

Positive outcomes. In the first six months of 2018, we performed 4,024 Doppler ultrasound exams on inpatient and ER patients for suspected DVT. Of these exams, 3,982 (99%) were reported with the system structured report template. Of those, 455 (11.4%) were reported as positive for DVT. Of those patients with a positive imaging test, 358 (78.7%) received anticoagulation in less than two hours; no alert was triggered. Of the other 97 (21.3%) patients, 68 (70.1%) went on to receive anticoagulation after providers saw the DVT notification alert. To date, no treating provider has complained about this alert. When we asked the hospitalist how things went with the alert go-live, he responded: “I really haven’t heard anything about it.” That’s a pretty positive outcome when it comes to activating disruptive alerts for providers.

Innovative elements. The innovation of this project is to use radiology reports generated during routine clinical practice to power real-time clinical decision support for treating providers in the EHR based on the interpretation outcome and not simply the existence of an imaging exam or radiology report. We created the templates to be acceptable by the radiologists and actionable by providers accessing the EHR. Finally, because the alerting system only fires during apparent delays of treatment, it does not disrupt providers performing clinical care as routinely expected.

Submitted by Travis Browning, MD, director of quality in UTSW’s radiology department.