Untangling Workflow
Workflow affects everything. In the race to squeeze every last bit of efficiency from a radiology department, administrators often stand before the Gordian knot of their workflow process wondering which dangling thread to pull first. For some, gains come from the introduction of a new technology, while others find success with an intervention focused on staff processes.

More tools than ever are available to help administrators with workflow projects. Vendors offer workflow consultations when a new technology is implemented and auditing or analytics tools tapping into RIS data can flag pain points. Each workflow project comes with its own revelations, benefits and challenges.

Consulting the experts

In 2006, University of Kentucky HealthCare (UK) in Lexington wanted to increase capacity and procedures with a new CT system, and the organization tapped its CT vendor for a workflow consultation to assess the current state of operations. The analysis identified multiple opportunities for improvement, including a shift to a two-technologist workflow model.

CT scanners have reduced scan time to mere seconds, so additional workflow optimization may be mined from the processes leading up to a procedure, rather than during the procedure itself. Having two technologists per scanner, where one is focused on preparing the patient while the other manages the scanner, significantly helped workflow at UK.

“Having a dedicated technologist or a nurse prepare the patient for the scan is critically important,” says Anil K. Attili, MD, chief of the division of cardiothoracic radiology at UK.

The provider had three target metrics with the project:
  • 10 percent reduction in process steps;
  • 25 percent increase in procedures per CT technologist; and
  • 10 percent increase in procedure volume.

All goals were met or exceeded, and the department is currently running at 100 percent capacity. Wait times are key for patient satisfaction, says Attili. UK strives to schedule patients in three days and many are scheduled within 24 hours; inpatients are scheduled within an hour or two.

Another facility that got a major assist from an outside consultant is the Cleveland Clinic Breast Center. When Cleveland Clinicwas implementing a new PACS in 2009, it tapped its vendor for PACS training and workflow analysis.

At first, the consultants just watched, says Alice S. Rim, MD, head of breast imaging in the diagnostic radiology department at Cleveland Clinic. Every once in a while, the consultants would ask a staff member to explain a particular action, but they initially remained non-intrusive.

Rim had her doubts until the vendor visually laid out the department workflow by sticking notes on a whiteboard in a meeting with administrators. “They put the Post-its up there and showed ‘Alice’s normal day’ and it looked ridiculous,” she says, describing the workflow map as more of a maze suitable for lab rats. “They knew our workflow better than we did.”

The workflow plan was tailored to the PACS implementation, and clinical, technical and administrative staff was trained extensively, says Rim. To maximize the effectiveness of training, the team at Cleveland Clinic worked with the vendor to schedule training sessions two weeks prior to the go-live date and have them led by the same team that conducted the original workflow analysis. Training was conducted away from the reading environment to limit distractions, and all organizational levels were included.

“The key was making sure everybody from the radiologists to the technologists to the front desk staff to the imaging library staff was trained,” she says.

The idea that everybody should be trained to use a new system might seem obvious, but if staff is not completely comfortable with a technology, serious workflow obstructions can develop.

It’s important to have vendors train in-house staff who are then capable of training users in the department, says Katherine P. Andriole, PhD, director of imaging informatics in the department of radiology at Brigham and Women’s Hospital in Boston. A “train the trainer” model works well for new implementations, says Andriole.

The training program at Cleveland Clinic eventually paid off, because when the PACS went live, the transition was seamless, says Rim. Prior to implementation, turnaround times for screening mammography could stretch to three weeks, but after implementation, results were available within 24 hours or less. In fact, Rim says the department received some complaints from patients who were “spooked” that the letter with results was mailed so quickly.

“It was strange. I received a deluge of calls from women thinking when they saw the letter it must be bad news. I explained that we are just trying to make it so patients don’t have to wait so long.”

The department learned to make the most of the new technology, says Rim. For example, the team modified how mammography worklists were handled. Within the Cleveland Clinic system, there are a number of geographically dispersed sites and the original workflow was set so that each site would have its own worklist. Radiologists would be assigned to a site where they would be responsible for their set of diagnostic reads, but once that work was finished, they were to look at the list of digital screening mammograms from other sites that needed to be read. Lists at each site would need to be at zero before anyone at any site could go home.

The problem with the previous system, says Rim, is that as more sites went digital, more lists were created and it started to become overwhelming. Labeling a list for each site is cumbersome, and people could be lulled into a false sense of security thinking they were done with the day after they cleared their site’s worklist, when in fact many more studies were waiting on another site’s list.

“You want to make it as simple as possible for whoever is reading to know what work has to be done that day,” explains Rim. This was accomplished by populating a single list for screening mammograms so radiologists from multiple sites were pulling studies from the same virtual list. With everybody on the same page pulling from the same list, it can be dwindled down in minutes.

Signature Time Trend Analysis
A workflow project at Brigham and Women’s Hospital in Boston used a mix of technological interventions and financial incentives to reduce report signature time. Phase 1, which introduced a paging system and speech recognition, reduced the median times, but the outliers in the 80th percentile saw the most substantial reduction only after the incentives introduced in Phase 2.
Comparison PeriodReduction for Median Signature TimeReduction for 80th Percentile Signature Time
Pre-Project vs. Phase 185.5%37.8%
Phase 1 vs. Phase 237.5%81.3%
Pre-Project vs. Post-Project90%88.3%
Source: J Am Coll Radiol 2010;7:198-204
 

Individualized workflow

Another benefit of the new system at Cleveland Clinic is that radiologists don’t have to stop what they are doing to search the medical record for prior studies, because they are automatically launched when a new study is opened.

A potential hitch is that not all radiologists want to see all prior studies. That is, there are benefits to having standard procedures and global worklist strategies, but not every radiologist works the same way. It’s important to find the balance between an individualized workflow and standard templates that apply to a whole enterprise.

Different radiologists have different tendencies, says Bruce I. Reiner, MD, of the department of radiology at the Veterans Affairs (VA) Maryland Health Care System’s Baltimore VA Medical Center. When looking at a head CT, for example, one physician might want to see the current study and any in the last six months, while another might want all existing priors. Cancelling out unwanted studies takes precious time and displaying too many prior studies can cause sensory overload.  

“Wouldn’t it be nice if you could track the individual workflow and the technology adapts to the patterns of the end user?” asks Reiner. Currently, there are auditing tools that can track individual tendencies and locate commonalities and differences. This can help departments to build workflow templates.

Templates can be based on groups of radiologists divided by common characteristics such as specialty or experience, says Reiner. One template could be for radiologists working in a rural environment and another could be better suited for those without advanced training, and so on. These profile groups allow context to be taken into account when optimizing workflow.

But there’s still a catch. “The reality is, even if you have these individual profile groups, there’s still going to be variability between one user and another,” says Reiner. “There are going to be different tendencies that one user has in terms of how he or she accesses different tools or functionality vs. another.”

The only way to truly customize workflow templates for each user is to have a technology that learns from the user, adapting automatically to the various workflow tendencies. Although auditing tools exist, Reiner says no vendor has yet devised a system that has this level of automatic adaptability. “It seems like an intuitive, natural progression for the technology, but nobody’s doing it.”

Show me the money

Although departments often turn to technology to save the day with regard to workflow, sometimes technological interventions just aren’t enough.

A recent workflow project at Brigham and Women’s Hospital illustrates the point. When staff analyzed turnaround time, they found the time spent waiting for reports to be signed was a major driver of the overall turnaround time in the report dictation to transcriptionist workflow. From 2005 to 2008, they implemented several interventions to reduce signature time.

The first two interventions were technological, says Andriole. First, a paging portal was created to notify radiologists that reports had been transcribed and were ready to be reviewed and signed. Next, speech recognition was gradually rolled out, which allowed for the instant creation of reports without the need for transcriptionists.

Technology adoption reduced median signature time from more than five hours to less than an hour, but Andriole says an interesting trend appeared. While the median time was significantly reduced by the new technology, there were still outliers. Signature times for radiologists in the 80th percentile were in the 15- to 18-hour range.

What Andriole and her colleagues noticed was that some radiologists were still working under an old workflow that didn’t take advantage of the new technology. Prior to speech recognition, transcriptionists would create reports based on the dictation. Attending radiologists would often sign reports in bulk at the end of the day, rather than as they came in. While this might have sufficed for efficient time management before speech recognition, the new technology made this workflow model obsolete.

The fact that some radiologists were still lagging behind with signature times despite the addition of speech recognition and the paging portal meant that a behavioral intervention, rather than just a technology intervention, would be needed to push turnaround time over the top.

That’s when the department turned to the powerful tool of financial incentives. Radiologists who met the median signature time goal of under eight hours or who had 80 percent of reports signed within 16 hours were rewarded with a semi-annual $4,000 bonus.

The results were dramatic. Signature times for radiologists in the 80th percentile were slashed to between four and eight hours. Gains persisted when the financial incentives were phased out after a year.

“Technology can help. But you need to understand the workflow into which you introduce that technology,” says Andriole. “If people are using the new technology in the pre-implementation workflow, they may not realize the benefits.”

Change management

No matter what type of workflow project a department embarks on, staff buy-in is key. “Making the staff feel that they are a team member is very helpful,” says Attili.

Staff must be involved early in the process, and though some may be hesitant to change the way they work, once the benefits are realized they typically understand the importance of modifications. Rim says change never really sinks in until staff has a chance to see and engage in the new workflow. It also helps to have a good cheerleading contigent, either in the department or with outside consultants, to get over the early challenges and to ensure staff remain open to changes.

Although it can seem easy to look to a new PACS upgrade or some other informatics investment to help untangle workflow issues, departments should not overlook interventions related to behavior or policy.

“You should not be required to completely re-engineer what you are doing to use a technology, but sometimes the benefits of a new technology will be even greater if you can modify your workflow in some ways,” says Andriole.

Technology, training and incentives all can play a role in a workflow optimization project. Success lies in making sure these distinct threads all are pulled in the same direction.

say verb's?, what pronoun'hwät ?
Before an in-depth conversation on workflow can take place, everybody must first speak the same language. Common definitions for workflow steps within medical imaging don’t yet exist, and key performance indicators or definitions of data elements can be ambiguous. For example, how can a department compare PACS/RIS when vendors might not mean the same thing when they say a study has been “completed?”

Enter the Society for Imaging Informatics in Medicine’s (SIIM) TRIP initiative. The committee behind TRIP—which stands for Transforming the Radiological Interpretation Process—has turned its focus to creating an industry standard for medical imaging workflow metrics. This way, data can be compatible across multiple business analytics dashboards, regardless of the vendors.

At SIIM’s annual meeting in June 2011, the TRIP initiative demonstrated a dataset that was vendor neutral and could be read by a variety of vendor-created or open-source dashboards.

SIIM has compiled a draft list of normalized workflow names and definitions open for viewing and comment, including:
  • Acquired: All raw image data are acquired and the patient may exit the imaging device. Does not include reconstruction time.
  • Complete: All images are acquired and reconstructed on theimaging device.
  • Prepared: All steps required for reporting have been completed,including the acquisition of images, transmission to reporting deviceand all post-processing.
  • Reported: Final electronic report signed by interpreting physician.
Evan Godt
Evan Godt, Writer

Evan joined TriMed in 2011, writing primarily for Health Imaging. Prior to diving into medical journalism, Evan worked for the Nine Network of Public Media in St. Louis. He also has worked in public relations and education. Evan studied journalism at the University of Missouri, with an emphasis on broadcast media.

Trimed Popup
Trimed Popup