News|Articles|September 25, 2026

Community Oncologists Ready to Innovate, but Barriers Remain

Listen

0:00 / 0:00

Key Takeaways

Prior authorization remains the dominant friction point, with nearly universal reporting of delays that widen the gap between innovation development and patient access in community oncology.
AI uptake is rising rapidly, but predictive model use is uncommon due to workflow misalignment, lack of EHR-integrated platforms, and concern that AI may introduce new operational or clinical risks.
Clinical trial enrollment is limited by scarce local options, restrictive eligibility criteria misaligned with pharmacology, and administrative burdens including lengthy consent and delayed translations for non-English speakers.
Implementation of bispecific T-cell engagers and CAR T is desired but hindered by inadequate infrastructure, staffing, training, and reimbursement dynamics, particularly outside 340B-enabled environments.
Community-generated real-world evidence can materially change regulatory expectations, exemplified by reduced post-infusion observation requirements for tarlatamab that improve access for small cell lung cancer.

Community oncologists cited prior authorization as a care barrier, even as AI adoption doubled in 2026, according to the Oncology Care Index.

Nearly all community oncologists say prior authorization and payer requirements are stalling delivery of innovative cancer care, even as artificial intelligence (AI) adoption in community practices has more than doubled year over year, according to findings from the 2026 Oncology Care Index (OCI) discussed during a panel at a recent industry conference.

The survey, conducted by Johnson & Johnson (J&J) and drawing on responses from 109 community oncologists and 100 practice administrators, found that 78% of clinicians believe they are uniquely positioned to accelerate the reach of innovation into the communities where patients live, and 92% said they incorporate new therapies into practice even before those therapies appear in formal guidelines. Yet 79% of respondents agreed there remains a persistent gap between where innovation is developed and where it actually reaches patients, and 97% cited prior authorization as a major source of friction.

“There is not true value delivered to the system if the innovation doesn't reach the patient,” said Mahadi Baig, MD, MHCM, vice president and head of solid tumors, US Medical Affairs, at Johnson & Johnson, who moderated the panel. Baig noted that roughly 85% of patients with cancer receive care in the community setting, making that gap a central concern for the field.

AI Adoption Is Accelerating, but Trust in Predictive Tools Lags

The data show AI use among community oncologists jumped from 19% in 2025 to 49% in 2026, with nearly 40% of practice administrators reporting similar uptake. Confidence in the technology is high, with 79% of oncologists saying they're comfortable using AI to guide treatment decisions, but that confidence hasn't translated into deeper use. Only 19% said they actually use predictive AI models, and 45% said they worry AI could introduce as many problems as it solves.

Stephen “Fred” Divers, MD, chief medical officer of the American Oncology Network, said the disconnect reflects where the technology currently sits rather than a lack of enthusiasm.

“I'm certainly not the AI domain expert. I mean, I'm not capable of writing code, et cetera,” Divers said, adding that adoption “really depends on the use case.” He described practices being “inundated with multiple single-point solutions day in, day out” rather than a single platform integrated into the electronic health record—something he said would ease the burden considerably.

Divers said AI for tasks including clinical decision support and revenue cycle management but acknowledged that most tools still require additional training data before they function as true plug-and-play solutions.

Lucio Gordan, MD, president and managing physician of Florida Cancer Specialists & Research Institute, who will step into a chief of clinical innovation officer role at McKesson in January 2027, said his practice is moving more aggressively into AI heading into 2027, including ambient documentation tools, chart summarization, and AI-assisted clinical trial matching that he said has already improved trial accrual. Looking ahead, Gordan described a near-term goal of having AI automatically generate orders for follow-up labs and imaging based on physician conversations with patients.

“It's not to replace us; it's to give us more time to spend with who is the most important one in the picture,” he said of the technology's intended role.

Clinical Trial Design Still Excludes the Patients Most Likely to Need Access

Panelists were candid about structural barriers keeping community patients out of trials. Baig cited OCI data showing 60% of oncologists and 50% of administrators said they have limited trial options to offer patients, even though 97% of survey respondents said a more pragmatic trial design that reflects the patients they actually see would meaningfully boost enrollment.

Siddhartha Devarakonda, MD, of the Swedish Cancer Institute, argued that eligibility criteria often fail that pragmatism test.

“The best trial for a patient is a trial that they can enroll to, and the number one exclusion criteria for a patient is, ‘Can I get to the cancer center or not?’” he said.

He pointed to a phase 1 trial in which a drug that is 97% metabolized by the liver nonetheless excluded patients with a creatinine clearance below 60—a criterion he called disconnected from the drug's actual risk profile. Devarakonda also described piloting the use of AI to condense lengthy informed consent documents, saying a 50-page consent form summarized into 2 pages showed no meaningful loss of information while improving patient comprehension. He pointed to a related barrier for non-English-speaking patients, noting that the wait for a translated 40-page consent form can leave a patient with metastatic disease sitting in limbo, a delay he said AI-assisted translation could resolve in seconds.

Gordan agreed the barriers are largely systemic rather than a matter of will.

“The willingness to have clinical trials, enroll patients, and advance science and advance new drug development is there 100%,” he said, pointing instead to regulatory and paperwork burdens that make standard-of-care treatment far simpler to deliver than trial enrollment.

Divers offered a concrete example: a front-line CDK inhibitor trial for first-line metastatic breast cancer that, despite representing an unambiguous standard-of-care fit, remained difficult to fill.

Implementation Gaps Persist for Complex Novel Therapies

The OCI data also showed more than half of practices want to implement newer, complex therapies, such as bispecific T-cell engagers and chimeric antigen receptor T-cell therapies, within the next year, but only 35% feel adequately equipped to do so, citing staffing, training, and financial barriers.

Divers described building an outpatient step-up dosing protocol for bispecific therapies at American Oncology Network after cost barriers, including the absence of a 340B-eligible site in his region and limited access. The program has since supported more than 5000 doses.

Devarakonda noted that trials for complex therapies are often designed around large academic centers with extensive specialty support, which can leave community sites unprepared once a drug reaches approval. He pointed to Amgen's tarlatamab (Imdelltra) as an example of a drugmaker responding to that gap, saying the company did a "pretty good job in trying to modify their protocol subsequently with reduced monitoring time to make the effort to show that it's doable.” He also credited J&J's amivantamab (Rybrevant) program for similar responsiveness.

Divers pointed to tarlatamab as a case study in how real-world data from community practices can change label requirements. Just days before the panel, the drug's label was updated to reduce mandatory post-infusion observation from an overnight hospital stay to 2 to 4 hours.³ He attributed the shift to outcomes data submitted by community oncology programs treating small cell lung cancer patients who couldn't easily reach a tertiary center.

Closing the panel, Baig said progress will depend on continued alignment across payers, regulators, and drug developers.

“That will take a little more alignment across different stakeholders, reducing the friction of payer authorization, investing heavily in workflow-aligned AI solutions and tools, really developing pragmatic designs in partnership with both academic and community centers, as well as heavily investing in education,” he said.

References