This is a follow-up article to a three-part ACR Bulletin series showcasing the work of the first cohort of ACR Learning Network quality improvement collaboratives. The Learning Network was launched in 2022 to improve diagnostic imaging care through a learning health systems approach. Each collaborative supports teams at the facilities of four to six sites working to solve similar problems. The first article in the series, Collaborating on Quality, focused on improving prostate MRI for clinically significant cancer detection and localization. The second, In It for the Long Haul, highlighted the importance of technologists and patient positioning in mammography. The third, Volume and Value, focused on the future of lung cancer screening. 

Patients often come to the emergency department (ED) with sudden pain caused by a kidney stone or another ailment, but when radiologists read the scans, they may notice incidental pulmonary nodules in the background. When this happened at Hudson Valley Radiologists P.C. (HVR), the imaging report would recommend a follow-up CT scan in a few months to investigate further. But by the time many patients came back, the nodule might have progressed into something worse.

“It’s a common thing that happens across the country,” explains Tony Abbate, director of clinical operations at HVR, a private radiology practice in Poughkeepsie, N.Y., that provides imaging services to several hospitals and outpatient imaging centers in the Hudson Valley. “When we find a nodule, we recommend follow-up imaging to the patient, but that’s the last thing they’re thinking about because they’re dealing with some other pain. The pain goes away, they totally forget about it, and unfortunately, these patients come back a few years later with potentially advanced cancer.”

Up to 10% of all radiology reports contain follow-up recommendations, but approximately half of those recommended follow-up exams are never performed. Lung nodules represent about half of all imaging follow-up recommendations, and noncompliance with recommended imaging can delay the diagnosis of lung cancer — making it more difficult to treat. 

As HVR started to notice more instances of missed follow-ups and incidental findings becoming cancerous, they realized they had to prevent their recommendations from falling through the cracks.

“This became a very important problem to solve,” says Laurie Wadsworth, chief operating officer at HVR. “We quickly realized this wasn’t just a problem for HVR; this is a problem throughout healthcare because there are so many gaps and inconsistencies between patient hand-offs.”

To solve the problem of bringing patients back on time for potentially lifesaving follow-up imaging, HVR looked beyond its own practice and turned to the ACR Learning Network’s Recommendations Follow-Up Improvement Collaborative, which focused on improving the early detection of lung cancer for incidentally detected pulmonary nodules. This shared learning collaborative provided the framework that HVR needed to tackle follow-up fulfillment, ultimately boosting its follow-up rate from 20% to over 90%.

This kind of structured learning is very hard to get in the real world. We were basically sending our team to a graduate school learning class, and they learned a methodology that we could apply anywhere throughout our practice.

Embracing Shared Learning 

Hudson Valley Radiology team members, from left to right: Wayne Wadsworth, chief data officer; Laurie Wadsworth, chief operating officer; Amanda Startup, clerical manager; Tony Abbate, director of clinical operations; Jeanine Villano, director of administrative operations; Melissa Xiques, director of clinical IT systems; and Richard J. Friedland, MD, FACR, president.

When the ACR Learning Network launched the Recommendations Follow-Up Improvement Collaborative, HVR President Richard J. Friedland, MD, FACR, instantly recognized it as an opportunity to discover new best practices.

“We thought working with the ACR would bring us a number of quality experts and a different kind of rigor to look at our program,” says Friedland, who had been analyzing the challenge of bringing patients back for follow-up exams with his team long before the program launched. “It presented us an opportunity to take the lessons learned by others and avoid making the same mistakes.”

HVR joined the Improvement Collaborative in January 2022 along with a cohort of three other imaging practices located around the country. Led by radiologist Ben C. Wandtke, MD, MS, vice chair of quality and safety in the department of imaging sciences at the University of Rochester Medical Center, teams from each participating site collaborated through the program to improve their follow-up tracking processes with the goal of improving patient outcomes. 

Friedland compiled a team of about nine of his 75 employees to participate in the Learning Network’s ImPower Program, a four-month team-based quality improvement (QI) training and project support platform. This program included 10 two-hour learning sessions over a six-month span, teaching participants how to apply key QI concepts, like A3 Thinking, to solve the problem of follow-up compliance.

“This kind of structured learning is very hard to get in the real world,” Friedland says. “We were basically sending our team to a graduate school learning class, and they learned a methodology that we could apply anywhere throughout our practice. It became a way of thinking for the whole team — questioning our data and trying to understand the root causes of our problems so we could impact our practice and our patients.” 

Taking a Team-Based Approach 

Convincing employees to participate in the Learning Network was easy, thanks to HVR’s already pervasive culture of continuous quality improvement. “There isn’t a member on the team who we had to talk into it,” Wadsworth says. “The excitement of participating in a quality project that would benefit patients drew people in.”

Friedland pioneered HVR’s QI focus more than a decade ago when he started attending the annual ACR Quality and Safety Conference in Boston. He began inviting other members of the HVR team along to ensure that they were all following best practices. Inspired by these meetings, the team pursued QI projects throughout the practice as part of their commitment to quality.

“By committing ourselves to quality, we knew we’d always be able to provide the best care possible for our patients, provide ongoing education and lifelong learning for our radiologists, and provide an overarching goal and purpose for our team,” says Friedland, who previously served as chair of the quality and safety committee at Vassar Brothers Medical Center and the Health Quest network. “Quality became central to our identity, because as radiologists, we train for a long time to get this skillset, and we want to put that to its best use for our patients — but we can’t do that without continuous quality improvement.” 

Driven by this commitment to QI, HVR team members were eager to join the Learning Network, from radiologists to operational and clinical leaders and executives. Since ACR suggests the roles needed to participate in the Learning Network — including a patient care coordinator, radiologist, radiology manager, and scheduler — selecting the right team members for the program was a straightforward process. “The ACR does a very nice job framing the roles and responsibilities,” Wadsworth says. “We wouldn’t have known everything we needed to know to put a team together, but the descriptions they gave us made it clear.”

The HVR team met weekly with the rest of the cohort, in addition to regular internal meetings every week to apply what they were learning to their own practice. Several larger training sessions and homework assignments were required throughout the program, demanding an investment of time and energy from the entire team. 

“The commitment has to come from everyone because quality is a team sport,” Friedland says. “You need people with all sorts of skillsets to elevate quality. It’s an investment in your team and your practice that takes a lot of time and mental energy, but that investment ends up paying off in the long run.”

Besides addressing the specific follow-up issue, the broader benefit of the Learning Network was the built-in team-building component. “In many ways, it made our team even stronger because it increased our communication,” Friedland says. “It demonstrated how we all problem-solve together, and seeing that laid out in front of us was very powerful. We grew in admiration for each other and felt more connected as the project went along, and that’s the sine qua non of a successful project.”

Rethinking Structured Reports

Before joining the Improvement Collaborative, the HVR team thought they were doing a decent job of identifying incidental nodules and noting their recommendations for follow-up imaging. However, once they started digging into their data through the ImPower problem-solving process, they realized they weren’t consistent with their reports or their recommendations. Specifically, they noticed that radiologists weren’t uniformly following the Fleischner Society’s Guidelines for Management of Incidental Pulmonary Nodules Detected on CT Images, which recommend follow-up CT imaging at certain intervals, depending on the size of an incidental nodule detected in patients who are at least 35 years old. “One radiologist might say, ‘A nodule was noted,’ and it just opens up a million questions: How big is it? Where is it? Is it going to affect the patient? Should the patient do something about it?” says Wayne Wadsworth, chief data officer at HVR. “Just a small difference in how you interpret and dictate what you’re seeing can make a big impact.”

Unfortunately, these details weren’t always included in the templated reports that HVR radiologists used to dictate incidental findings. They didn’t necessarily indicate whether nodules were 6 or 8 millimeters in size, which would trigger different follow-up imaging recommendations according to the Fleischner guidelines. Before HVR could effectively follow up with patients whose findings fit these specific criteria, they first had to refine their standardized reports.

To that end, HVR convened a committee to determine which parameters to include in their revised reporting template, using the Fleischner guidelines as a compass. Through the process, they identified seven criteria they wanted to capture when noting incidental nodules — such as size, laterality, consistency and, of course, any recommendations for follow-up imaging.

Importantly, Laurie Wadsworth notes, this structured reporting template was “created by the docs, for the docs,” as HVR’s radiologists spearheaded these changes. Radiologists like Benjamin M. Hentel, MD, HVR’s vice president and director of quality, who served as the team’s physician leader in the Collaborative, and Donald Lien, MD, who also participated in the Learning Network, tested the templates themselves. After an iterative feedback process with the rest of the team, the docs then introduced the new reporting template to their radiology colleagues, aided by PowerPoint presentations and video instructions. 

“Executing these changes was simple because everybody was involved,” Laurie Wadsworth says. “The radiologists came up with it themselves, and that’s the best motivator you could have.”

Introducing the Shoulder Tapper

While the committee was coming up with the criteria to include in the nodule reporting template, Wayne Wadsworth was developing the software engine to capture this data consistently. 

He built the earliest version of this engine back in 2017 to capture follow-up recommendations from imaging reports. He updated the engine a few months before the QI collaborative began, because he knew they would have to pull data about incidental lung nodules as part of the Learning Network project. He launched the latest version in June 2022 — creating an automated reporting system that prods radiologists to dictate specific details and recommendations for each nodule noted. 

Wayne Wadsworth incorporated AI and natural language processing (NLP) into the engine, creating a data “macro” that automates pieces of the report. For example, depending on the size of the nodule noted, the macro automatically prompts the radiologist to make a follow-up recommendation based on the Fleischner criteria. If radiologists skip over any of the required criteria, the engine virtually “taps them on the shoulder” with a reminder to fill in the missing details. 

“We affectionately call this tool the Shoulder Tapper,” Friedland says. “The engine reads all of our reports and if there are fallouts, it’ll send a message saying, ‘You left out this piece of the Fleischner criteria,’ or ‘You left out the recommendation,’ even if the recommendation is ‘No further follow-up.’ That direct feedback to the radiologist altered the way we practice and moved us toward best practices in a very gentle way.”

This automation saved HVR from having to do extensive training or correcting to get radiologists on-board with the new reporting template. Instead of having to remind them which details to include or requesting an addendum later, the Shoulder Tapper prompts them automatically. As a result, radiologists quickly adopted the new template within weeks of rolling out the macro — drastically improving the data collected by the practice.

“Once we started capturing those details automatically, it became a game changer,” Wayne Wadsworth says. “Now you can start seeing trends over time,” which helps track each nodule through follow-up imaging. “We never could have done that before had we not looked closely at what we were really stating in our reports.” 

Communicating Recommendations 

Consistently reporting incidental nodules and making appropriate follow-up recommendations only addressed part of the problem, though. The other piece of the puzzle was actively communicating these recommendations to patients.

Prior to the Learning Network, HVR took a “wait and see” approach to bringing patients back for imaging. After recommending a follow-up exam in a report, they would wait a couple of months before sending a follow-up letter. They assumed they were being proactive by sending any reminder at all, but the shared learning environment of the improvement collaborative revealed opportunities to improve.

“We noticed that patients were late in coming back,” says Tony Abbate, director of clinical operations. “If they needed to get a follow-up exam in three months, it wasn’t happening for four or five months or even at all. So, we started sending a letter sooner versus later.”

To draw attention to their recommendations, which often went unnoticed in the stack of paperwork patients received, HVR began sending notices to patients within three days after the initial imaging report. Then, they scheduled a regular cadence of touchpoints urging patients to return for timely follow-up. To manage this process, Wayne Wadsworth developed a tool that tracks follow-up recommendations, prompting multiple reminders to monitor whether patients return on time.  

This tool pulls follow-up recommendations from all of HVR’s imaging reports, compiling them into a dashboard that Abbate and his team use to track patients over time. After the initial notice, “We give patients about 45 days to get the follow-up done on their own,” Abbate explains. “We want to give them time to reach out to their doctors and schedule the exam.”

If, after 45 days, a patient has not completed a recommended follow-up exam, the system prompts the team to send out a reminder letter. “We can also send SMS (text message) or email,” Abbate says, “but we find that in our practice, the letter seems to work best.” 

The letter moves the patients off the dashboard for another 30 days, giving them time to act. If they still haven’t responded after that time, they pop back into the dashboard and a second letter goes out. As a final attempt, Abbate’s team will call patients directly. 

“The key is not just to send one letter and stop,” Abbate says. “We try to go that extra step, especially for emergency room patients because those are the patients who are usually lost in follow-up,” he says, partly because they may not have a referring physician to remind them. “We can’t order exams for patients,” Laurie Wadsworth adds. “We can only communicate the findings and ensure that the patients know they need to follow up.”

While some patients schedule exams on their own after receiving a reminder, others need more direction. “The letters are causing patients to call us and say, ‘What follow-up are you talking about?’ Sometimes they had no idea there was even a finding,” Abbate says, “but because of the letter, they’ll call us, we’ll explain it, and they’ll end up scheduling the exam.”

Bringing Patients Back

Since refining the way its practice reports incidental lung nodules and communicates follow-up recommendations through the improvement collaborative, HVR has made enormous headway in bringing patients back. “Our initial data showed that only about 20% of our patients were coming back in a timely fashion because our communications were inconsistent,” Laurie Wadsworth says. “Our goal was 70%, but now we’re above 90%, and we’ve been maintaining 100% consistency with our radiologists’ communication of the findings. It all boils down to communicating better with patients because the majority of them don’t even know these nodules are there — and that’s the scariest part.” 

Timely follow-up has improved patient outcomes by providing clear results and timely service that impact care. Even if these follow-up scans lead to additional exams, biopsies or even cancer diagnoses, Abbate says, “the good news is that it got diagnosed today and not two or three years down the road, when the chances of beating it are lower.”

Abbate, who speaks with patients often as part of HVR’s follow-up process, says the feedback has been incredible. Even patients who receive follow-up imaging from another provider call HVR back in gratitude. “They call us and say, ‘I got this done elsewhere, but I know you had my back, so thank you for taking care of me,’” Abbate says. “There’s no better feeling than knowing you potentially saved a patient’s life.”

Referring physicians have been equally appreciative of HVR’s extra efforts to ensure timely follow-up. Initially, the QI collaborative team wondered if referrers would object to them sending letters directly to their patients, but even the skeptical physicians have recognized how these reminders help protect them and their patients. 

“Once they realize you’re trying to make sure the patient doesn’t get lost or fall through the cracks, they're grateful to have someone else looking out for them,” Abbate says. “The best part is that the referring physicians don’t have to do anything extra. It makes their life easier because instead of them following patients manually through a spreadsheet, we can do that for them electronically on our dashboard. The system is just there saying, ‘I’ve got your back.’”

Plus, the standardized imaging reports made findings clearer for patients and referring providers to understand, which led to more streamlined follow-up care. “We’re getting fewer phone calls from our clinicians asking, ‘What modality do you want to see? Do you want this study with or without contrast? What body part were you talking about?’” Friedland says. “Previously, all these protocoling issues got cleaned up on the back end but it really impacted the front-end efficiency of our practice.” 

Expanding the Circle

Lessons from the Learning Network continue to drive improvement throughout HVR’s practice. The same committee that participated in the improvement collaborative continues to meet monthly to review the group’s reporting template, recommendation engine and new quality measures — constantly tweaking the process to continue improving outcomes. 

Now that the team has had practice tackling follow-ups for incidental lung nodules, they’re applying the same problem-solving process to other types of findings. For example, several cardiothoracic surgeons have approached the group for help tracking patients with aneurysms who require follow-up imaging.

“How do we capture these patients? Because they’re coming from everywhere,” Friedland says. “They’re coming from emergency rooms, they’re coming from internal medicine, they’re coming from chest X-rays and CTs, so we must widen our circle.  How can we all collaborate on these cases so patients don’t fall through the cracks? It’s like throwing a pebble in the water. The circle keeps widening as you focus on your patients, and it just ripples across your whole organization.”

By leading this charge toward improved quality measures and enhanced patient outcomes, HVR is establishing a more proactive role for radiology in the big picture of comprehensive care. 

“We’re no longer just passively receiving patients,” Friedland says. “We’re performing patient care coordination functions, which is not traditional for radiologists. We’re taking a non-traditional role by widening the scope of what radiologists can do and changing the way we’re perceived by both the medical centers and the patients. Now we’re seen as the experts in not only initial imaging but also follow-up, and there’s no better position for a radiologist than to be the subject matter expert.” 

Benefits of Participation in the ACR Learning Network

• Rigorous training in proven quality improvement (QI) strategies and processes through the ImPower Program and quality coach training that sites can use for future QI work.
• Guidance from national leaders to get the most out of participation and achieve sustained diagnostic excellence.
• The opportunity to become a regional and/or national leader in one of the collaboratives’ focus areas and continue participation in the Learning Network community.
• Continuing education credits for ImPower program participation and completing a QI initiative.

Learn how to apply at the ACR Learning Network web page.