The Society for Imaging Informatics in Medicine (SIIM) is trying to make it easier to compare quality between radiology sites by developing a lexicon of common terms for describing the workflow of a patient moving through an imaging facility. But how well does the SIIM Workflow Initiative in Medicine (SWIM) work in the real world?
Pretty well, although SWIM would benefit from some new workflow steps describing the patient experience, as well as better definitions of some existing steps, according to University of Maryland researchers who presented their experiences with the lexicon at the recent SIIM annual meeting in Grapevine, TX. A little more assistance from vendors in capturing data points would also help the cause.
The Maryland researchers followed 100 patients during the imaging process to see how well SWIM worked in a real-world environment. They found that nearly half of the workflow steps used to manage patients appeared in the lexicon.
Sink or SWIM?
First unveiled as a pilot at the 2011 SIIM annual meeting, SWIM attempts to define a common set of terms, or lexicon, to describe the workflow of clinical imaging departments. The SWIM project has drawn interest for its potential to enable true comparison, measurement, and analysis of quality data from different departments -- something that's difficult to do when facilities use different terms to describe the same process and actions involved in managing patients.
Indeed, one of the biggest healthcare challenges is how to measure data across various platforms to understand and improve quality and performance, said Dr. Joseph Chen from the University of Maryland School of Medicine.
As part of a SIIM-funded study, a team led by Chen sought to validate the lexicon developed by the SWIM committee in a real clinical setting using a combination of physical observations and digital time stamps.
Patient tracking
The research team followed 100 patients throughout their imaging experiences. Two observers tracked 50 patients each and noted the workflow steps used, while a data analyst examined IT records for the appearance of lexicon terms in electronic medical record (EMR)/RIS, PACS, and dictation data. Half of the patients were examined at the in-hospital imaging facility, while half were seen at a standalone outpatient facility.
The observers used a checklist to track the events involved in managing patients and how often the events matched terms in the SWIM lexicon. They also assessed whether there were any workflow steps that weren't included or adequately described by SWIM.
The observers identified 126 steps involved in managing patients that could potentially be tracked. Using both the observational data and the information gathered via IT systems, they found that 47.6% of the workflow steps used to manage patients were described by terms in the SWIM lexicon. The IT data alone included lexicon terms only 26% of the time, Chen said.
Some of the workflow steps could be found in multiple IT systems. Two percent of the steps could be found in all three IT systems, while 21% of the steps could be found in two IT systems.
Of the workflow steps that were unique to a particular IT system, 33% were found only on the PACS, 30% were found only on the EMR/RIS, and 9% were found only on the dictation system.
In addition, the researchers identified 17 new workflow steps that were not described in the SWIM lexicon. None of these steps were captured in the IT data, and most related to aspects of the patient experience such as registration and preparation processes (including changing for the imaging test and IV insertions), he said.
The patient experience should be one of the key parts of tracking, as those are our customers, Chen said.
"It's great that we get the report out on time, and it's great that the physicians see the report on time," he said. "But what does the patient feel at the end of the day, because they are the ones who are going to come back [or not]."
The researchers faced a number of IT system challenges in their project. Databases are difficult to discover, and there are also mapping challenges with different names across different systems, Chen said.
In addition, there can be multiple accession numbers for a single exam in the case of grouped procedures, making it a challenge to track these exams.
Some parts of imaging workflow do not currently generate time stamps. There can also be many time stamps for one event, such as when an imaging study is opened. For tracking purposes, it's important to know what all of the user roles are, he said.
Lessons learned
From their study, the researchers learned that new steps need to be created in the SWIM lexicon, such as contrast complications and early termination of exams because of claustrophobia and other issues, Chen said.
It's valuable to track early stoppage of the imaging study and to determine why the study was stopped, as trends can then be observed, he noted.
In addition, the researchers discovered that some steps need to be better defined. For example, "arrive" was in both the University of Maryland's system and in the lexicon list, but "registration" was not.
"Is 'arrive' or 'registration' the same thing or not?" Chen asked. Patient arrival and registration may be handled simultaneously at one site and differently at another.
"We need to better define what these steps are, and we need to create more steps," he said.
Some steps should be reordered for more logical flow, and the researchers suggested the grouping of workflow steps such as prearrival, arrival, exam, diagnosis, etc.
More vendor collaboration
The team also learned that the percentage of data-point captures could be improved with additional vendor collaboration and support, Chen said. After talking with their PACS vendor, the researchers were able to realize a 3% improvement in the capture of data points from PACS.
Given that only 26% of lexicon terms were captured on IT systems, vendors could adopt some additional steps to build into their workflow.
"Or [sites should] build it in-house if the vendors are not willing to," he said.
Critical results notification is a vital issue, and a system should be able to provide data such as when a physician or nurse practitioner has been notified of the critical results and when the doctor actually sees the results.
"These are features that have clinical implications, and if we're not capturing these in your IT systems, I would encourage that we do so because they are pretty important," Chen said.
He acknowledged limitations of the study, including the need to investigate additional IT systems such as a 3D server. Also, there were some vendor-specific features utilized in the study, and the research only encompassed outpatients.
Chen encouraged attendees to participate in the SWIM initiative by joining the SWIM community forum, getting involved in the monthly meetings, and helping to refine the model. Also, users can map the terms they are using to the SWIM lexicon, he said.
"We would love to see more people participate," he added.
