Quality Measures Could Be Linked to Outcomes With Dynamic Oncology Pathways

Patient-reported outcomes (PROs) and artificial intelligence (AI) could help community oncologists better assess how patients are actually doing on treatment than relying on oncology clinical pathways designed around trial data

and guidelines. he shift from static to dynamic pathways built around patient feedback was the focus of a panel moderated by David Jackman, MD, medical director of clinical pathways at Dana-Farber Cancer Institute, at The American Journal of Managed Care’s 2026 Patient-Centered Oncology Care conference in Nashville, Tennessee.

Edward “Ted” Arrowsmith, MD, MPH, medical director of pathways at OneOncology, described the discussion as “more aspirational” than data-driven, with no formal process specifically in place at OneOncology to include the patient voice in its pathways.

“I know at the Patient-Centered Oncology [Care] conference that’s kind of a terrible thing to say, but it’s a true thing to say,” he said. However, Arrowsmith added, “We do have a lot of informal ways that the patient voice is included.”

The Challenge for Current Pathways to Capture Patient Feedback

The core problem with capturing patient feedback within pathways is a design issue, according to Erin Crum, MPH, senior director of quality, strategy and innovation at McKesson. Pathways center on clinical evidence, guideline recommendations, and utilization management, so they are not built to continuously take in patient feedback or weigh what she called “softer variables.” Patient data also sit in separate places, including survey tools, third-party systems, and modules embedded in the electronic health record, and little of this is processed and connected at the point of decision-making.

“Are we expecting our clinicians to be flipping through screens and different product or vendor tools for symptom management while they’re trying to think through what’s the next best step with our patient?” Crum asked. “We need to figure out a way that we can be smarter about that and really bring that extra layer of synthesis and translation at the point of care.”

Patient perspective does reach OneOncology’s pathways informally, Arrowsmith said. Physicians who vote on pathways spend their days treating those diseases, practices run patient surveys, and the network looks for gaps between trial data and real-world outcomes, such as treatments that worked in trial populations but are not reaching patients in rural clinics.

Stephen “Fred” Divers, MD, chief medical officer of the American Oncology Network, said the gap has become harder to ignore. At a recent Community Oncology Alliance board meeting, leaders discussed how existing studies on cost of care, site of service, and outcomes lack patient experience data. He pointed to growing government support for PROs as well as interest from life sciences companies.

“It’s probably time that this surfaced and became the third leg of that stool of patient-centered oncology care,” he said, adding, “How do we incorporate their voice and their patient experience on their journey in such a way that it helps us make appropriate clinical decisions?”

How PROs Could Reshape Pathway Decisions

Arrowsmith said PROs could narrow the gap between clinical trial and real-world performance by capturing adverse events quickly and documenting how patients feel on treatment. If a regimen produces a poor patient experience across a network, he said, it may belong off the pathway or be limited to a narrower subgroup, particularly if predictors of poor tolerance can be identified.

OneOncology currently revisits pathways at each disease group meeting, prompted by FDA label changes, National Comprehensive Cancer Network category shifts, or new publications. A PRO- or toxicity-based trigger would be a welcome addition, Arrowsmith said.

Divers argued that review needs to happen in real time rather than retrospectively, and that it should be built into agreements with payers and networks. If costs reach parity across sites of care, documenting pathway adherence, lower toxicity, and patient experience could help practices differentiate themselves, he said.

McKesson’s Practice Insights reporting platform has been designated as a Qualified Clinical Data Registry by CMS for more than 10 years, a designation that allows the company to build custom quality measures, including PRO-based measures. For instance, these measures track outcomes such as whether patients with elevated pain scores see improvement within 30 days or whether distress improves or resolves within a set period. Crum said the aim is to move beyond asking patients about pain or distress to measuring whether practices act on the answers.

“I think if we can start to think about actually monitoring and measuring our success with these different interventions, based on the feedback that we’re hearing from our patients, that’s going to be much more impactful to their care and their potential outcomes,” she said.

Why Static Pathways Still Matter in Community Oncology

Even as the panelists called for more dynamic tools, Arrowsmith said formal, committee-driven pathways retain value. When there could be 3 or 4 FDA approvals across disease states in the same week, but many clinics only have 1 physician and a few advanced practice providers treating a wide range of cancers, there is little time to compare new options. He added that disease group meetings allow physicians to share best practices for logistically complex therapies, including radiopharmaceuticals, and that the trust built there has helped OneOncology adopt innovative treatments.

Jackman noted that pathways committees are also an important “human in the loop,” as many AI models draw on decisions made in those meetings.

“When something first gets approved by the FDA, we don’t have any real-world experience,” Jackman said. “We have the clinical trial experience, but we don’t have a mountain of data on which to base [decisions] yet, so it’s coming a lot from those review meetings to set us in the right direction.”

How AI and Integrated Data Could Change Pathways

The ability to pull scattered data together at the point of care is the evolution in decision-making tools that most excited all the panelists. Arrowsmith highlighted the ability to use AI to integrate multiple data streams in the exam room to inform both treatment and bigger-picture strategic decisions. The number of patients living with incurable cancer for years, sometimes decades, has been a defining change during his career.

“It’s really a different type of medicine we’re practicing now in oncology,” he said.

Crum said the answer comes down to technology used responsibly, so clinicians are not toggling between different AI tools on the side of the screen. She pointed to the “intelligent medical record” concept from McKesson’s chief product officer. Today’s electronic health record is largely a static record of entered data, she said, and the next step is to pull disparate data sources together and solve interoperability issues so the best decision can be made in the moment.

Divers said he was excited because the tools make him “a much better doctor than I could have ever been before.” He said intelligence has been democratized and the playing field has been leveled, although key opinion leaders are still needed for nuances that AI cannot supply. He expects these tools to become part of patient interactions, clinical decision support, revenue cycle support, and clinical trial enrollment.

“Once we make that leap from static to dynamic, how cool of a universe is that when you know everything that there is to know and you can deploy it at a moment’s notice?” Divers asked. “That’s close to nirvana.”