On a typical morning in the clinic, Lawrence Shulman, MD, does what oncologists across the country do: he logs into his electronic health record (EHR)and starts searching. Not for a diagnosis — he already knows what his patients have. He’s searching for the key data relevant to that patient’s cancer care, buried somewhere in a sprawling digital chart that might contain years of lab results, imaging reports, genomic panels, and notes from a dozen specialists.

“One of my time sinks, but also one of my concerns from a safety and quality point of view, is not finding data that might be critical for a patient’s care,” said Shulman, a medical oncologist and professor of medicine at the Perelman School of Medicine, University of Pennsylvania, Philadelphia. This information is often “scattered in all different places in the chart and not generally well cataloged.”

“I think that’s a big worry,” he said.

It is a worry shared across oncology. Cancer care is uniquely data-heavy — patients accumulate years of treatment records, lab results, genomic data, and imaging, often across multiple institutions. EHRs were supposed to streamline clinical practice and improve safety, freeing physicians to devote more time to patient-facing care. Instead, research and clinician experience suggest the technology has become a relentless documentation machine that swallows their days, follows them home at night and wedges itself between them and the patients sitting across the exam room.

Now, with cancer incidence rising and the oncology workforce struggling to keep pace, the pressure to find solutions is intensifying. Emerging AI tools and calls for payment reform may offer support, but whether the system can adapt fast enough to relieve a specialty already stretched thin remains an open question.

The Inbox That Never Closes

The numbers tell a stark story.

A 2025 study published in Journal of the National Cancer Institute found that oncologists’ EHR time increased dramatically between 2019 and 2022. The study, which tracked more than 15,600 oncology specialists across over 43 million ambulatory visits, found that overall EHR time for oncologists rose 16% in that time. Inbox message volume rose 19% and patient-initiated messages grew by 34%.

Among oncology specialists, medical oncologists and hematologists had the heaviest EHR load. By 2022, medical oncologists and hematologists were spending about 9.6 h/wk in the EHR followed by gynecologic oncologists at roughly 5.4 h/wk.

The increasing time spent managing the EHR inbox signals a fundamental shift in how patients and physicians interact, one born out of the COVID pandemic, explained lead study author A. Jay Holmgren, PhD , assistant professor of medicine and a researcher in the Center for Clinical Informatics and Improvement Research at the University of California San Francisco.

“It was really the onset of COVID where the awareness of the portal account, because [patients] needed to use it for telemedicine, increased dramatically,” Holmgren said. “A big part of that EHR time rise comes from an increased demand for asynchronous portal-based messaging.”

The pandemic, in other words, taught millions of patients that they could reach their doctor with a few keystrokes. And they haven’t stopped.

This reliance on portal-based messaging now “just reflects the new normal of what patients are expecting from their physicians,’ he noted.

Oncologists often find themselves fielding messages from frightened patients who may be navigating complex regimens and grappling with tough questions about their symptoms, treatment, or prognosis. Once the messaging begins, “suddenly you’re down this rabbit hole having messaged 20 times back and forth,” with the resolution often being, “‘oh, geez, we need to have you come in for a visit,’” Holmgren said.

What’s Lost When Screens Take Over

Portal messages are only one layer of a growing EHR workload. Underneath them sits a documentation system that several experts described as fundamentally broken.

Arturo Loaiza-Bonilla, MD , explained that before even greeting a patient, an oncologist may need to track down lab results, check pathway requirements for clinical trials, review toxicities, coordinate care across specialties, and wade through secure chat messages.

All this adds time, said Loaiza-Bonilla, systemwide chief of Hematology and Oncology at St. Luke’s University Health Network, Allentown, Pennsylvania, and co-founder and chief medical officer at Massive Bio, a company that connects patients to clinical trials using AI.

What’s worse: Sifting through this glut of information can make it surprisingly hard to read oncology chart notes and identify the most immediate and relevant information, Shulman said.

“I think documentation is so far off the rails that people should really be embarrassed about what we’re doing,” said Shulman, who has spent decades working on EHR implementation and oncology informatics.

Research supports their concern. A phenomenon known as “note bloat” — in which clinical notes balloon with unnecessary information — has been well documented in the medical literature. One longitudinal analysis found that EHR note length grew by roughly 60% between 2009 and 2018, and a separate study reported that US clinical notes can be up to four times longer than those from physicians using the same EHR systems in other countries.

Billing and liability incentives are among the key drivers of unnecessary documentation in clinical practice: physicians may document more than is clinically useful to justify higher-level billing codes and reduce legal risks. In 2021, CMS revised its coding guidelines to help reduce these documentation pressures, but early studies suggest the changes had limited impact on note length or time spent on the EHR.

The problem is that “we’re incentivized to put a lot of stuff in our notes,” said Shulman. “They should pay us for documenting what’s important for patient care and quality and research, not to just fill the dots with what everybody would admit is not meaningful information.”

The Personal Toll

The documentation burden is not just a time problem — it’s a physician well-being problem.

In Medscape’s 2024 Physician Burnout & Depression Report, more than 60% of physicians surveyed said that too many bureaucratic tasks, including charting and paperwork, contributed most to their feelings of burnout, and nearly 24% pointed specifically to EHRs.

Research has also linked EHR workload and emotional exhaustion. A 2020 study found, for instance, that primary care clinicians in the top quartile of after-hours EHR time had more than 12 times the odds of high exhaustion compared with those in the bottom quartile, whereas those with the highest message volumes had more than six times the odds.

Oncologists can face similar strains on their well-being, especially when EHR time extends beyond clinic hours and follows them home. Holmgren’s JNCI study, for instance, found that medical oncologists logged more than 3.5 h/wk of EHR work outside of scheduled clinic time, including evenings, days off, and weekends — after hours work that’s “concerning given the role of EHR burden in physician burnout,” he and colleagues explained.

Loaiza-Bonilla referred to this extra work as “pajama time” — the hours spent after clinic, at home in pajamas, finishing documentation. That time is taken from families, from rest, and from the mental reserves that oncologists need to deliver empathetic, high-stakes care.

This added time burden may pull oncologists away from the bedside and has raised concerns about safety. A 2022 analysis of 28 studies across medical fields suggested that higher EHR workload was associated with increased error rates and potential risks to patient safety.

A Policy Problem

Experts interviewed for this piece returned to the same fundamental issue with EHRs: the payment system. Technology can generate efficiencies, but if the incentive structure rewards documentation volume over quality, and if portal messaging remains largely uncompensated, increasing reliance on the technology can also do the opposite: burden oncologists and waste time.

The core problem is that US healthcare payment is built around discrete, in-person visits. When a physician spends 20 minutes responding to a complex thread of portal messages, that work doesn’t fit neatly into the fee-for-service model.

While billing codes for patient messages do exist — CMS introduced them in 2020 — they’re cumbersome in practice. Physicians must determine whether the messages meet billing criteria, attach a diagnosis code, document the time, and navigate restrictions like not being able to bill for a message sent within 3 days of an in-person visit or 7 days after.

Research indicates that organizations that have enabled physicians to bill for inbox messages have found that less than 1% of message threads are billed — and that the reimbursement is often insufficient to support the care teams necessary to sustain this kind of work.

Holmgren suggested that the US experiment with hybrid payment models — combining elements of fee-for-service with capitated or per-member-per-month structures that give physicians a baseline payment for patient interactions, including portal-based communication.

But he cautioned that any flat-rate approach has its own problems. He noted that an upcoming study he’s working on shows that 1% of patients generate 20% of messages, so a physician with these high-volume message patients would be undercompensated under a flat-rate model while a colleague without those patients would be overcompensated.

Still, without a new reimbursement model, Holmgren sees physicians getting increasingly overwhelmed and eventually refusing to use the portal. And “that’s going to really be worse for patients, especially in visit wait times,” he said.

Shulman echoed the frustration about the current approach to documentation. “Somebody’s got to stand up and say, this is a failed experiment, and how are we going to change this to incentivize documentation that in fact accomplishes what we wanted to accomplish,” he said.

The workforce numbers only sharpen the urgency as concerns over an oncology shortage persist amid growing patient demand.

“The math is pretty simple,” Shulman said. “We’re not going to have a lot more oncologists in the country, and we do have a lot more cancer patients.”

A Way Back to the Bedside

There are glimmers of progress. New ambient AI tools are showing promise at reducing the EHR workload by listening to physician-patient conversations in real time and automatically generating clinical notes, summarizing longitudinal data and patient histories, reconciling outside records, and preparing clinicians for visits.

Some academic medical centers have built triage teams that route portal messages through administrators and nurses before they reach physicians, filtering scheduling requests and straightforward clinical questions. AI scribes are giving oncologists some time back in their day.

Shulman described a project in development that would use large language models to extract and synthesize data across the EHR and present it in an organized, clinician-friendly format.

“That would be both time saving and is also safer because you would be much less likely to miss critical data or to misinterpret data,” said Shulman.

Organizations like Massive Bio, which has partnered with the American Cancer Society, are using AI to surface clinical trial options at the point of care — work that Loaiza-Bonilla said can improve equity and outcomes while reducing administrative burdens.

Loaiza-Bonilla also noted that AI tools have given him back some time at the bedside, allowing him more freedom to maintain eye contact with patients, take in the silence that can come with a difficult prognosis, and respond to empathy cues, instead of being buried behind a screen.

While AI can help provide clinical infrastructure, Loaiza-Bonilla cautioned that, as the technology takes over data-heavy tasks, clinicians may rely on it too early in training, which can lead to gaps in foundational reasoning. AI works best, he said, as a support layer — not as a replacement for clinical judgement.

“AI can process information, but it really cannot carry the responsibility that comes with being the physician,” said Loaiza-Bonilla, who works at the intersection of AI and oncology and writes a column for Medscape Medical News.

Bradley Malin , PhD, a professor of biomedical informatics at Vanderbilt University Medical Center in Nashville, Tennessee, believes that more patient portal messaging, despite the added workload, is a good sign — it reflects a shift toward a more engaged, two-way relationship between patients and clinicians that can lead to better care.

But the success of this shift depends on building systems that support the physician-patients relationship rather than suffocates it.

The question now is whether the healthcare system can change fast enough to allow oncologists to focus on the heart of their work: to look patients in the eye and guide them through their care.

The experts interviewed for this piece had no relevant disclosures.