The Disappearing Specialty, Part I

When the Floor Disappears

"In our mind, the only way this is going to be addressed and solved is in a virtual first, AI-driven manner. In the coming years, the biggest cancer centers will be virtual first." That's Color Health CEO Othman Laraki, quoted last week in STAT, on the occasion of his company becoming the first virtual oncology practice to earn ASCO Certified status. The framing is bold and explicitly aspirational. The biggest cancer centers — by which I take Laraki to mean the names you would put on a list — are going to be replaced by something that doesn't quite exist yet, in a model that is very modern.

What gets less attention in that sentence is the word "biggest." The biggest cancer centers don't get to be biggest because they decided to be. They are accredited. Commission on Cancer accreditation. NCI Comprehensive Cancer Center designation in many cases. State licensure. Each of those regimes carries floor requirements. Some of those floors are clinical. Some of those floors include specialty palliative care. When a virtual-first model claims parity with the gold standard while clearing a different — and lower — bar, the floor isn't simply lowered. It disappears for the patients on the other side of it.

The Frame, and What It Implies

Color Health is a serious effort. They've raised real money, partnered with the American Cancer Society, MSK Direct, Carrum, Maven, OpenAI, and UCSF. Their CEO is articulate about a real problem: cancer is scaling faster than oncology can. New science, new drugs, new genomic stratifications, new survivorship populations — the cognitive load on a single oncologist is rising every year, and the workforce isn't keeping up. Laraki isn't wrong about the diagnosis.

The proposed solution is virtual-first, AI-driven cancer care, oncologist-led, with a multidisciplinary tumor board, an OpenAI-built Cancer Copilot, and a 50-state medical group. Color isn't alone in this space. Hinge Health is running the same play for musculoskeletal care. Maven is running it for fertility and women's health. Carrum is running it for surgical bundles via Centers of Excellence. Vynca is running a parallel play for serious illness management with virtual specialty palliative care at the center. The employer-sponsored benefits market is being divided into specialty verticals, each promising better outcomes at lower cost than the fragmented system most employees encounter.

The rhetorical move that does the work in Laraki's quote is the comparison to "the biggest cancer centers." That's the comparison the new model is being sold against. That's the comparison the new certification is meant to substantiate. And that's the comparison that requires actually being clear about what the biggest cancer centers do, and why.

What Comprehensive Cancer Centers Are Required to Have

Three regimes define what counts as a serious cancer center in the United States, and they don't fully overlap.

NCI Comprehensive Cancer Center designation is the most prestigious. There are 57 NCI CCCs as of 2025. The designation is primarily about research — added depth and breadth of basic, clinical, and population science, with substantial transdisciplinary integration. It's renewed every five years through peer review, and it carries clinical care expectations as embedded scaffolding for the research mission. Most NCI CCCs are also accredited by the Commission on Cancer.

Commission on Cancer accreditation is the floor for serious cancer programs. Over 1,500 programs are CoC-accredited, and they care for roughly 70% of newly diagnosed cancer patients in the country. The 2020 CoC Optimal Resources for Cancer Care Standards include Standard 4.5: Palliative Care Services. The standard requires palliative care services to be available, that patients with serious illness be screened for palliative care needs, and that there be a defined pathway to those services. Programs are surveyed every three years.

NCCN guidelines, which most large cancer centers use as their treatment frame, recommend specialist palliative care consultation for patients with refractory symptoms, complex decision-making conflicts, severe distress, or advanced disease. The picture is uneven. Mo and colleagues reviewed all 53 NCCN treatment guidelines in 2021 and found palliative care mentioned in 57%. Specialist palliative care referral was specifically named in only 17% of the guidelines that mentioned PC at all. Even with the variability, NCCN sets a more demanding bar for PC integration than what's appearing in virtual-first models that claim NCCN concordance.

ASCO's own clinical practice guideline, first published by Ferrell and colleagues in 2017 and reaffirmed since, recommends early integration of dedicated interdisciplinary palliative care for patients with advanced cancer or high symptom burden, ideally within eight weeks of diagnosis of advanced disease. The evidence base for that recommendation includes Temel's 2010 NEJM trial, the Bakitas trials, Zimmermann 2014, and most recently Temel's 2024 JAMA paper — the STEP trial — which showed that telehealth-delivered specialty palliative care was non-inferior to in-person specialty palliative care for patients with advanced lung cancer. The 2024 paper matters here because it removes the obvious objection: virtual specialty palliative care is now evidence-supported. Whether to include it in a virtual-first model is no longer a question of feasibility. It's a question of choice.

ASCO Certified Is Not a Substitute

Color earned ASCO Certified status in May 2026. ASCO Certified is built on the Oncology Medical Home framework, a joint initiative of ASCO and the Community Oncology Alliance first published in 2021 and updated in 2025. The OMH standards cover seven domains: patient engagement; availability and access; evidence-based medicine; team-based care; quality improvement; goals of care, palliative, and end-of-life care discussions; and antineoplastic safety.

The fact that one of the seven domains has "palliative" in its name is doing an enormous amount of work in how the certification is being received. It is not doing the work people assume.

The OMH "Goals of Care, Palliative, and End-of-Life Care Discussions" domain is a process standard. It addresses whether the practice has documented workflows for goals-of-care conversations, whether it has advance care planning processes, whether symptoms are screened and addressed, and whether end-of-life care is discussed appropriately. It does not require an HPM-board-certified clinician on staff. It does not require a referral pathway to specialty palliative care. It does not require any of the elements that distinguish primary palliative care — the work every oncologist should be able to do — from specialty palliative care, the work that requires a trained interdisciplinary team.

A virtual practice can pass an OMH site survey by demonstrating that its oncologists discuss goals of care with patients, document advance care plans, and address symptoms in their notes. None of that is wrong. None of that is sufficient.

Let me be careful. ASCO Certified is a real signal. The OMH framework is genuinely more rigorous than what most community oncology practices were doing five years ago. Practices that earn it are, on average, doing better on the things the framework measures than practices that don't. The problem isn't that the framework is bad. The problem is that the framework is being used as evidence for claims it doesn't actually support — specifically, the claim that a certified virtual practice is delivering care equivalent to a comprehensive cancer center.

When the field was small and palliative care was new, the gap between primary and specialty PC didn't matter much because both were rare. Now that primary palliative care competencies are increasingly embedded in oncology training and certification, the gap matters more. A practice can pass primary palliative care assessments while having no actual access to specialty palliative care for the patients who need it. That gap is where the floor disappears.

Where the Floor Goes When the Floor Goes

The accreditation floor doesn't disappear evenly. It disappears most for the patients who need it most.

Consider who the employer-sponsored cancer benefits market is built to serve. Color reports a 2.8:1 ROI in year one. That ROI is calculated against a population of employed people with employer-sponsored insurance — most of whom are working-age, most of whom either don't have cancer or, statistically, have early-stage cancer that can be cured with timely intervention (and yes, all the outlier young people with terrible cancer who I see crowd in on my mind as I write this and try to make me soften this point with anec-data). The model's value proposition is screening earlier, diagnosing faster, getting people back to work. Each of those propositions is reasonable. None of them describes the patient with metastatic disease whose oncologist has run out of effective lines of therapy, whose symptoms are escalating, whose family is fragmenting under the cumulative load, whose decision-making is no longer about which chemotherapy regimen but about whether further treatment is consistent with what they actually want.

That patient exists. Demographically, that patient is more likely to be older, more likely to be Black or Hispanic, more likely to be on disability, more likely to have aged into Medicare. Employer benefits programs are not structured around that patient because that patient is no longer the employer's patient — by the time the disease has progressed that far, the patient has often left employment (or never had a job that included health insurance to begin with). The financial incentives of the model align with serving the patient before progression, not after. The exclusion of specialty palliative care from the model isn't an oversight. It's consistent with who the model is paid to serve.

Comprehensive cancer centers exist inside this same market. They sell services to insurers, including employer-sponsored insurance. But they don't get to filter their advanced-disease patients out the way a virtual-first model effectively does, because comprehensive cancer center accreditation requires them to accept and serve patients across the full disease trajectory. CoC accreditation requires palliative care services available to all patients, not just to the ones whose insurance status makes them attractive to keep. NCI CCC designation requires research and clinical infrastructure that includes serious illness care. The accreditation regime isn't just a stamp. It's the mechanism that prevents cancer programs from cherry-picking the patients they serve.

When a model can claim parity with the gold standard while operating outside the accreditation regime, the protection that accreditation provides for vulnerable patients evaporates. The patients who get filtered out don't disappear. They go somewhere — usually to a worse place, with less integrated care, more often in crisis, more often in the ED, more often dying in the hospital instead of at home. The first-order effect is invisible to the new model's quality metrics because those patients aren't in the denominator.

This is the part of the argument that doesn't get made because it doesn't show up in the dashboards. Equity in cancer care doesn't live in the screening numerator. It lives in who is and isn't in the denominator at the end of the disease trajectory. The accreditation floor is one of the few mechanisms we have for keeping all the patients in the denominator. When that mechanism is routed around, the dashboards keep looking great because the people who aren't in them are by definition not being measured.

Where I Might Be Wrong

This argument has at least four points where the case could be weaker than I think.

First, Color and the broader virtual oncology cohort may genuinely be coordinating with local specialty palliative care teams behind the marketing, even though the marketing doesn't say so. If that's true, the criticism here lands more as a marketing critique than a clinical one. I'd love one of them to show me their published referral pathway and a shared metric for the proportion of advanced-disease patients receiving specialty palliative care consultation within eight weeks of advanced diagnosis. None of the public materials I've reviewed include either. I'd update if Color or a peer publishes both.

Second, the OMH framework may be exactly the right standard for a virtual care layer that is meant to complement, not replace, comprehensive cancer center care. If the explicit positioning were "we're a layer on top of your existing oncology, not a substitute for it," the accreditation argument would shift. But that's not the positioning. Laraki's quote is specifically about replacement.

Third, the evidence that PC integration improves outcomes in the population virtual-first models actually serve — working-age, employer-insured, often early-stage — is thinner than the evidence in advanced disease. It's possible that for the population the new model is built around, primary palliative care via the oncologist is sufficient and specialty PC is not where the value lies. I think this is partially true and structurally inadequate as a defense, because the model markets itself for "any cancer type at any stage."

Fourth, the historical accreditation regime has its own failures. Many CoC-accredited programs technically meet Standard 4.5 by having a palliative care service that is poorly resourced, hard to access, or used only at end of life. The floor is real but it is also low. A reasonable counterargument is that the new model could clear a higher functional bar than many accredited programs do, even without the formal requirement. I'd accept that as a hypothesis. I haven't seen the data that proves it.

What Comes Next

Three things the field can be louder about, starting now.

The first is what we mean when we say a model has palliative care. The OMH framework and primary palliative care competencies aren't bad. They are necessary and insufficient. Specialty palliative care is a different thing, and we shouldn't let that distinction collapse. When a press release says "goals of care, palliative, and end-of-life care discussions," the right next question is whether there is an HPM-boarded clinician anywhere in the model.

The second is what accreditation actually protects. CoC Standard 4.5 isn't a check box. It's the mechanism that prevents cancer programs from filtering out the patients whose care is most expensive and most complex. When a virtual-first model claims comprehensive cancer center parity without it, the right response isn't to celebrate the certification — it's to ask which patients the model is built to serve, and which it isn't.

The third — and this is where the next two parts go — is to look harder at the structural and economic conditions that make the disappearance possible. The accreditation gap is the policy infrastructure layer. The next layer is the field marketing layer, where palliative care has spent twenty years trying on different euphemistic names in an effort to drive referrals, and is now watching that vocabulary get harvested by people who didn't sign up for the work. Part II will look at that. The layer underneath both of those is the data and AI layer, where the algorithm that drives the new cancer care economy has no signal for the work specialty palliative care actually does. Part III will tackle that.

Each layer disappears specialty palliative care in a different way. Together they describe an emerging cancer care economy in which the work we do doesn't have a place to sit.

The Disappearing Specialty is a three-part R&R series on how specialty palliative care is being structurally disappeared from the next generation of cancer care.

Part I — When the Floor Disappears (you are here)
Part II — The Vocabulary Heist (next Tuesday)
Part III — No Signal for the Work (two weeks)

I am a palliative care physician, educator, and professional strategery expert. Known for turning rounds into rants and rants into teaching points. Rounds & Rants represents my views — not those of any institution or professional membership organization where I hold a role. I don't write on their behalf and they don't vet what I publish.