Rapidly evolving technology capabilities are reason enough to standardize imaging protocols across a health-care enterprise. Add the desire to reduce costs, increase efficiency, and eliminate unnecessary exams by using imaging appropriately, and it is easy to make the case for such a project, particularly in a health-care enterprise with multiple sites of acquisition and lots of mobility.
Nonetheless, the herding-cats metaphor comes to mind when one thinks about the challenge of gaining consensus among a large group of radiologists on imaging protocols, and some told Janice Honeyman-Buck, PhD, that it could not be done. In “Defining and Managing Standardized Imaging Protocols With Appropriateness Criteria,” a June 3 presentation that she gave in Minneapolis, Minnesota, at the 2010 annual meeting of the Society for Imaging Informatics in Medicine, Honeyman-Buck, editor-in-chief of the Journal of Digital Imaging, described how Veterans Integrated Service Network (VISN) 23, a VA hospital network headquartered in Minneapolis, accomplished that goal.
“The issues with imaging protocols—and this is not just for this institution, but any institution with multiple hospitals or sites of acquisition—is that radiologists often have to cover multiple sites and patients move between them,” she explains. “If you are not using consistent protocols, then you may have to redo the study. We need to maximize efficiency, and we need to promote appropriate use of the technology.”
Another aspect of the project, intended to eliminate unnecessary imaging, was linking each procedure with appropriateness criteria. Imaging procedures’ and protocols’ names can be vague and confusing to referring physicians, who might, for example, order a CT of the abdomen with and without contrast, rather than just with or just without, simply to cover their bases. This linking of procedures and criteria “is very important because Medicare is going to start requiring it soon,” Honeyman-Buck says.
“Another thing we really wanted to make sure of was that everything was consistently named, so that radiologists could make hanging protocols that would be consistent across the organization,” she adds.
With eight VA health systems covering more than a million veterans in all of five Midwestern states and part of five more, VISN 23 had a technology inventory of 15 CT units; 3T MRI systems (tuned to 1.5T) at nearly every site, all with different hardware, software, and firmware; multiple DR units at each site; and one or more ultrasound scanners at each site. These were the modalities of interest, according to Honeyman-Buck, who consulted on the project. The adoption of a computerized provider order entry system that would facilitate ordering the correct study precipitated the project.
The project was divided into three stages. First, protocols for each study were reviewed with radiologists and technologists, who arrived at consensus. Second, the list of approved protocols was documented and distributed to each site. Third, a process for reevaluating the protocols annually and planning for change was established.
To find out what protocols were being used, Honeyman-Buck downloaded the data directly from the imaging modalities when possible, but in many cases, the data were in proprietary formats, so access to the equipment vendors was necessary. She also examined all of the documented procedure manuals from each location; each manual was different.
Using a variety of sources to help define the initial set (see box), Honeyman-Buck sent a recommended list of protocols to the radiologists and technologists, asking whether they approved, and if not, how they would change the protocols. Each study was reviewed with radiologists and technologists; iteratively, they arrived at consensus on 238 protocols: 69 for CT, 64 for MRI, 45 for ultrasound, and 60 for DR.
For example, the CT protocols (see figure) included contrast specification, special instructions, acquisition specifications with timing start and end points, reconstruction specifications, orientation, thickness, series name, spacing, algorithm, direction of reconstruction, and indications. Currently, VISN 23 saves the survey images and the reconstructions only to PACS. “The reconstructions used a unique series name, so that they could implement standard hanging protocols,” Honeyman-Buck says. “The thickness, spacing, timing, and field of view were specified.”
MRI protocols included a description