Digital Dashboards & Business Intelligence: Defined at Last

	The RADIANCE dashboard uses open-source software to extract CT dose data and help estimate radiation exposure. Image courtesy of the Hospital of the University of Pennsylvania.

Business intelligence. Digital dashboards. While the terms are often bandied about, radiologists and imaging stakeholders are striving toward more precise definitions and applications to help their practices gain maximum efficiencies. In search of a consensus, we asked leaders in the field to understand current applications and future potential.

"[B]efore outcome data can protect patients, drive payment, sanction providers, inform the public or spur improvement, the healthcare industry must determine acceptable levels of validity and reliability and develop a mechanism to collect these data," wrote Peter J. Pronovost, a professor of anesthesiology and critical care medicine, surgery, and health policy and management at the Johns Hopkins University, in Baltimore, in the April edition of Health Affairs.

Such mechanisms hinge on standardized language and metrics. In radiology, leaders are revving their engines to standardize imaging management reports. Bradley J. Erickson, MD, PhD, from the department of radiology at the Mayo Clinic in Rochester, Minn., and chair of the TRIP (Transforming the Radiology Interpreting Process) Committee within the Society for Imaging Informatics in Medicine (SIIM), says that one of the more recent efforts for the TRIP Committee is to develop agreed-upon terms of nomenclature of processes for imaging department workflow.

According to Erickson, the project has identified 100 different names of points in the workflow of an imaging department. "Once we agree on what certain terms mean, we can have ways to electronically capture those events and then display them on a tool such as a digital dashboard," says Erickson. "If you want to compare effectiveness across departments then we need to agree on what different terms mean in a workflow."

Everyone wants to know their turn-around time but those times can be measured in different ways resulting in overcooked, good-looking data, says Erickson. Another useful metric in need of standardization is room utilization. Does the end user consider an imaging room being used for 24 hours a day or when the doors to a hospital open? The discussion to standardize these metrics to forward quality care is needed, says Erickson.

Erickson and colleagues, who plan to demonstrate the concept and gather feedback at the SIIM meeting in June, have created a database with anatomical examinations of various times and identified key metrics and those events present in the database.

"Standardizing the definition of clinical information will be an important advance in dashboarding tools," Erickson explains.

Internal development

Just because digital dashboard tools haven't hit the streets in waves doesn't mean a hospital should ignore the option to develop its own solution. For instance, Woojim Kim, MD, assistant professor of radiology and associate director of imaging informatics at the Hospital of the University of Pennsylvania (UPenn) in Philadelphia, created a Google-esque web-based search engine application and analytics dashboard for the University of Pennsylvania Hospital's RIS in 2009. Containing report text and metadata on more than 13 million reports dating back to 1988, radiologists can easily search and data mine imaging reports.

"Most institutions have been, in terms of radiology reports, digital for a long time and are sitting on a mountain of data," says Kim. It used to take hours to days to request all exams relating to a condition such as pneumonia, he notes. It's now accomplished with the click of a radiologist's mouse, typically in less than a second.

Rounded out with interactive demographics and graphics, a radiologist can access individual report results and a bird's eye view of search results.  The metadata can be broken down to display summaries like gender distribution, study years and ordering physicians.  Because the application is not tied to a specific vendor, the information is presented in a uniform and consistent manner regardless of the source, and the application can tap into other specialties in medicine like pathology.

For administrators, the dashboard provides a high-level tool to understand the state of the hospital. For example, Kim notes that turn-around times at the Hospital of UPenn have decreased since the dashboard was deployed. "The dashboard allows administrators to track volume filtered by modality and is able to drill down into referral patterns and clinicians' individual performance," says Kim.

At the department level, Kim notes that the rate of laterality errors has decreased by 45 percent since the introduction of a laterality checker dashboard. Other dashboards can be created  as well. Meanwhile, the 3D Lab uses its web-based worklist and dashboards to keep radiologists aware of the status of each case, reducing unnecessary phone calls to the lab.

With the help of fellow physicians William Boonn and Tessa Cook, the Hospital of the University of Pennsylvania also created RADIANCE, an open-source software designed to automatically extract and archive CT dose-related parameters from image-based dose sheets. The RADIANCE dashboards have been used to decrease estimated doses for common chest CT exams, with dose savings of 30 to 40 percent for single-phase chest CTs and of up to 70 percent for multi-phase high-resolution parenchymal chest CTs.

Radiology & beyond

The Radiology Department of Brigham and Women's Hospital (BWH) in Boston has been harnessing dashboard and business analytical tools to expose key data elements from a variety of radiology systems to improve efficiency, safety and quality of care for four years.  

Initially, a semi-automated dashboard view of carefully selected metrics was configured to be continually measured. The system converted and formatted RIS/CPOE data from raw data into a graphical page to be published on the intranet and queried PACS data in real time for operational metrics. "Macro-scripting of repetitive tasks were built and scheduled to be routinely updated," according to Luciano M. Prevedello, MD, from the Center for Evidence-Based Imaging at BWH. "Once data are gathered and graphs are generated, they are automatically uploaded in a final presentation state to a password-protected website."

Ninety-seven metrics are currently monitored on the departmental quality dashboard (29 on quality, 18 on operational efficiency and nine on safety). Quality metrics include inpatient MRI performance, outpatient MRI appointment access and referring physician order signing. Safety metrics include communication of critical test results, allergy documentation in the longitudinal medical record (LMR) by department and allergy documentation in the LMR by radiologists. Operational efficiency metrics include exam reporting complete to final, unread exams by section and exam reporting complete to preliminary.

The metrics allows Prevedello and company to get a better understanding of organizational performance. For key metrics in the first quarter of 2010, the organization noted that while inpatient MRI performance exceeded target metrics, comparison to the last cycle showed a downward performance trend. Additionally in the same quarter, the "Unread Exams by Section" metric was below the intended target for performance but was improving in a trend comparison to the last cycle.

A dashboard view of the metric allows early recognition of potential problems so they can be addressed in a timely fashion, Prevedello says. "It provides a good view of how your department is running but also gives you granularity of the data. You have the ability to compare and contrast and that is an advantage of having the dashboard available to everyone in the organization so you can be transparent and motivate people to work on an area you'd like to improve."

Brigham and Women's department of radiology uses an analytical tool to consolidate and display computerized physician order entry (CPOE) data. "This allows us to track our adherence to particular evidence-based guidelines and further understand the appropriateness of our radiology orders," says Prevedello. "Another use case of our analytical tool is in recognizing radiologist adherence to our local policies. Our communication of critical test results policy, for example, is currently being tracked this way and can report adherence at the departmental, section or even at the physician level."

Prevedello stresses that dashboard and business intelligence tools are not magical but rather require effort from a cast of characters that need to understand each other's workflow and jobs. "If you gather data incorrectly, the data and data display will ultimately prove unbeneficial," concludes Prevedello.