OK, CMS Doesn't Actually Hate Specialists

It's just not entirely sure how to incentivize them

May 29, 2026

For me, writing is an intellectually stimulating outlet to express views on topics I find important and interesting. I enjoy tackling issues I think are misrepresented or misunderstood. Admittedly, I tend to skew “skeptical contrarian”, but I do my best to bring the evidence and explain my logic. Like most people, I have my own biases, but I try to be fair and measured. “Fair and measured” isn’t necessarily a winning strategy in the attention economy, and that’s OK.

A few weeks ago I published “Does CMS Hate Specialists? — an analysis of CMS/CMMI’s recent crop of value-based care models and their conspicuous lack of direct specialist accountability. I normally prefer clever titles to clickbait-y ones, but this one was perhaps a bit inflammatory. It was, of course, a rhetorical question — one that CMMI Director Abe Sutton reached out to directly to answer.

No, Sutton informed me, CMS does not, in fact, hate specialists…and he’d be happy to chat about it. How could I refuse?

Full credit; Abe was gracious and genuinely curious. He asked specifically what I thought about the Ambulatory Specialty Model and highlighted the potential of CARA’s Max Flex Option. He pointed out the challenges of incentivizing specialists, the limits of voluntary models, and the thought process behind CMMI’s current approach. Most importantly, he took the time to listen to what this Orthopedic Surgeon-turned-blogger had to say. That willingness alone says something.

The call confirmed that the people building these models aren’t indifferent to the specialist engagement problem. But they’re working within constraints that consistently produce middling outcomes. Sutton is acutely aware of past models’ failure, and he and his team are trying to learn from those failures. Based on our conversation, I think CMMI is open to specialist-led accountability.

Building on my previous article and my conversation with Abe, here’s what I think the current models miss and a well-designed model would look like.

The Flawed Thesis

I’ve often argued that CMS’ newest VBC programs — ACCESS, TEAM, ACO LEAD, and CJR-X — all send the same message: direct specialist participation is not a priority. Instead, these models are intentionally designed to shift accountability and control away from specialists. Despite evidence to the contrary, CMMI is betting that the path to better outcomes and lower costs runs around, not through, specialty care.

On the surface, this makes sense. Strip away program complexity, and the core argument is that primary care should serve as the filter, diverting patients away from high-cost specialty care and governing who gets expensive procedures. In theory, this leads to less unnecessary utilization, lower costs, and better stewardship of finite resources. In theory.

The problem is that the evidence to support this position is weaker than most think. Gatekeeping reduces utilization, but the literature fails to prove it reduces the right kind of utilization. Meanwhile, the knowledge gap that makes PCP-led management of complex conditions suboptimal has been documented extensively. Building care models around primary care accountability of specialty-focused conditions may paradoxically lead to low-value care.

This piece challenges that premise, then proposes something better. Care models built on flawed premises are destined to fail. The goal is to design an approach that produces the outcomes CMS wants; one that acknowledges current care gaps and builds bridges over them rather than gates in front of them.

Does Gatekeeping Work?

It’s widely accepted that routing all patients through primary care first reduces costs. A vestige of the Health Maintenance Organization concept, this tenet is a key underpinning of VBC. Substitute digital platforms for primary care and you get ACCESS. While it sounds good in theory, does it hold up to scrutiny?

The evidence that gatekeeping leads to higher-value care is sparse and murky. Studies supporting the idea that gatekeeping reduces specialist utilization are of limited quality. When Harvard Vanguard eliminated its 25-year gatekeeping system, specialist utilization barely changed. Meanwhile, gatekeeping lowers patient satisfaction and increases administrative burden. One study even found an association between gatekeeping and lower cancer survival rates.

CMS’ new VBC models emphasize co-management and care coordination between specialists and PCPs to address these issues. But this approach may actually be counterproductive. Specialist-only ambulatory care for osteoarthritis and low back pain is associated with improved functional status at slightly higher costs than non-specialty care. Meanwhile, co-care models cost substantially more and don’t improve function.

We simply lack robust evidence to support the claim that PCP gatekeeping of specialty care consistently drives high-value care. We also consistently underestimate the negative downstream effects of delaying definitive care — increased complexity, reduced treatment effectiveness, and paradoxically higher costs. Knowing when to refer is as important as knowing when not to.

The Knowledge Gap

When it comes to musculoskeletal conditions, the knowledge gap is well-documented. More than half of practicing physicians fail standardized MSK competency examinations. Many primary care physicians report low confidence in performing a musculoskeletal physical exam. Most NPs receive limited MSK education during their training. This is a reflection of systemic educational shortcomings, not an indictment of PCPs. Medical schools and residency programs don’t spend enough time teaching evaluation and management of MSK conditions, and efforts to address these shortcomings have produced mixed results.

I know what you’re thinking: artificial intelligence to the rescue. What better use case for AI than to upskill primary care and close the specialty-care knowledge gap? I agree in principle, but I’m not sure we’re there yet. Using LLMs to shift responsibility from specialists to PCPs doesn’t solve the liability issue. It probably makes it thornier.

When it comes to AI, high-quality outputs depend on high-quality inputs — effective prompting requires an appropriate knowledge base. Real-world clinical scenarios differ from carefully constructed case vignettes. Spotting hallucinations and dubious recommendations is difficult when you don’t know what you don’t know. Comfort managing specialty conditions comes from direct experience and learned expertise, two things AI cannot give you. Faced with uncertainty, most physicians err on the side of caution — i.e., referral — AI or not.

MSK complaints represent the second most common reason patients visit a primary care physician. The evidence suggests the clinicians managing most of those patients lack the tools to do it well. Incentivizing them to take on more of this care in the name of cost savings is the other side of the low-value coin. If overutilization (unnecessary procedures) is heads, underutilization (delayed care) is tails.

Gatekeeping and the knowledge gap are at odds with each other and with the idea of value-based care. Treatment decisions should be driven by the people most qualified to make them.

Solutions

The “PCP-as-Gatekeeper” approach has its flaws, but there are ways to address some of the shortcomings while preserving the spirit of what CMS is trying to achieve. The solutions presented below are applicable to any value-based approach to specialty care looking to strike a balance between resource stewardship and timely, appropriate intervention.

1. Close the Knowledge Gap

Evidence suggests that PCPs lack the confidence and knowledge to optimally manage MSK conditions or make timely referral decisions. The most direct fix is accessing specialist expertise at or near the decision point — not downstream from it.

The tools to do this already exist. Several studies suggest that structured, non-face-to-face specialist input (e.g., eConsults and Expert Medical Opinions) is clinically appropriate in many common MSK scenarios. A meaningful portion of the referral pipeline can be handled with a structured specialist touchpoint without a formal visit. One study showed eConsults prevented unnecessary in-person orthopaedic consults 61 percent of the time. PCPs found the consults valuable 94 percent of the time, and 97 percent of them contained actionable advice.

Specialist input early in the care process and upstream of the referral decision works best to avoid low-value care. Improper imaging, three months of fruitless conservative treatment, and misplaced patient expectations are a setup for failure in the form of wasted time, money, and resources. Early structured specialist input reduces referrals, improves quality when referrals do happen, and achieves higher patient satisfaction than gatekeeping models that simply create bottlenecks.

This workflow already exists through various digital platforms, but suboptimal implementation has produced underutilization. Medicare’s existing interprofessional consultation codes pay between $18 and $74 for specialist time, depending on duration. A 20-minute written review of a complex MSK case pays $37. The same specialist seeing that patient in the office bills at three to four times that rate. Worse, the codes can’t be billed at all if the eConsult leads to a face-to-face visit within 14 days. A specialist who reviews a case, identifies a patient who needs to be seen, and schedules an appointment gets paid nothing for the triage work that most would consider a more efficient use of resources. In such cases, the 14-day rule acts as a penalty rather than an incentive.

If upstream specialist input is genuinely valuable, payment has to reflect that. Aligning eConsult reimbursement with the complexity of the clinical work would increase specialist participation and still result in net cost reduction. Eliminating the 14-day rule would help. Thoughtful value-based care model design could also “monetize” eConsults through shared savings in a capitated or condition-specific model — which brings us to an interesting wrinkle in ACO LEAD, the CARA Max Flex Option.

2. Use CARA’s Max Flex Option

Built on the principles of BPCI-A, CARA is the mechanism by which ACOs can contract with specialists for care episodes (with help from CMS). CARA offers two participation structures: the Default Approach and the Max Flexion Options. The former uses CMS-constructed episode parameters with limited customization; the latter allows ACOs to select an existing episode and customize it by specifying trigger codes, episode length, quality measures, and performance adjustment parameters. (Max Flex arrangements are subject to CMS review for clinical appropriateness and operational feasibility).

The Max Flex Option creates an interesting opportunity to design condition-specific MSK bundles tailored to certain patient populations (global subcapitation) with comprehensive care pathways and embedded surgical episodes. Such arrangements may be better suited to achieve the goal of true co-management between PCPs and specialists. Think Integrated Practice Units and a multi-disciplinary approach based on building two-way streets, not throwing up gates.

The Default Approach will undoubtedly be easier to establish and will be the preferred path for most ACOs. Implemented intelligently, Max Flex represents an interesting opportunity for ACOs to treat specialists as collaborators, not subordinates.

Global subcapitation and other forms of well-structured risk-sharing are not solely the purview of CMMI models. Such arrangements are possible at the commercial insurance, employer-focused, and direct care levels. The main barriers are will, patience, sophistication, and incentive alignment.

3. Embed Specialist Expertise

The third solution is simpler and more immediately actionable — put a specialist Medical Director inside the ACO’s (or other entity’s) operations.

Not as a consultant or “Preferred Provider” contractor but as an operational resource. Someone with the requisite knowledge to interpret the specialty care utilization data, help navigate complex referral decisions, identify variation in pre- and post-acute utilization, and translate claims signals into clinical action. Someone who knows if a 20 percent SNF discharge rate after joint replacement is good or bad and can tell if you have a case mix problem, a surgical quality problem, a post-acute contracting problem, or all of the above.

That person also fills another, less appreciated function that has equal importance: clinical education and embedded expertise. Payment model design doesn’t fix the MSK knowledge gap in primary care, easy access to specialist-level reasoning at the point of decision does. That’s exactly what happens when you put a credible specialist in the building, make them an integral part of the care team, and leverage them to build specialty infrastructure that makes sense.

4. Flip the Script

The most ambitious solution is one CMS and others haven’t yet entertained: stop treating specialists as downstream vendors in someone else’s accountability model. Let specialist groups organize as the primary accountable entity with full cost of care risk for a defined population’s MSK spend, the data access to manage it, and the clinical authority to act on it. Partner with primary care for longitudinal chronic disease management. Own the specialist utilization decisions that drive the majority of the cost.

Evidence from BPCI-A supports this approach at the episode level. In that model, preoperative optimization served as a form of chronic disease intervention and uncovered some important lessons about the interplay between primary and specialty care. For instance, patients who stop smoking before surgery are more likely to remain smoke-free and perioperative glycemic control drives lasting behavior change for patients with diabetes. Orthopedic groups could build the proper infrastructure, aggregating around value-based capabilities, positioning as CARA Preferred Providers, and developing the analytics to take on episode risk directly. They just need a model that makes it worthwhile.

CMS could create such a vehicle explicitly, a specialist-led population health model with the same structural features LEAD provides for primary care ACOs. The evidence base and precedent exist. As yet, the model does not. Until it does, one option is to use Max Flex, embed expertise inside the organization, and make sure specialists are deeply involved. Accountability is coming one way or another; the main question is whether specialists are co-designers or bystanders, inheriting whatever primary care and payers build without them.

A Few Thoughts on The Ambulatory Specialty Model

What about ASM? Hasn’t CMMI already created a mandatory specialist accountability model? Isn’t that exactly what I’m asking for?

It’s a fair question, and ASM deserves both credit and criticism.

ASM is absolutely one answer to the question “Does CMS hate specialists?” The model explicitly acknowledges that specialists need direct clinical and financial accountability. It’s mandatory, carries two-sided risk, requires collaborative care arrangements with PCPs, and provides episode-level performance data. Participants are exempt from MIPS reporting requirements for the duration of the model, meaningful admin burden relief that any successor model should replicate. All of these things make sense. Holding primary care accountable for downstream specialty costs and specialists accountable for upstream care decisions might not produce the intended results. ASM tests the alternative.

That said, ASM as designed falls short of what genuine specialist-led accountability requires and leaves some opportunities on the table.

Specialists could rightly view ASM as another hidden payment cut with CMS taking money off the top. Some practices may run the numbers and ultimately decide that absorbing any potential penalties is better than dealing with the hassle. Others may view ASM as another reason to consider opting out of Medicare altogether.

ASM initially covers only low back pain (LBP) and heart failure. Notably, lower extremity joint replacement isn’t in the model — a significant missed opportunity. While hip and knee arthritis care lends itself to episodic, standardized care pathways, LBP is a more complex condition to manage. ASM chose the conditions with the highest Medicare spend, not necessarily the conditions where specialist accountability is most likely to work.

ASM requires participating specialists to build formal Collaborative Care Agreements with PCPs, including bi-directional data sharing, co-management protocols, and closed-loop referral communication. However, the requirements are minimal by design, and the scoring structure rewards avoiding a penalty rather than collaborating well. Both parties will take the path of least resistance to compliance, chasing metrics without necessarily improving care.

The peer-relative benchmarking structure of ASM also creates an asymmetric information problem. CMS and ACOs can use episode performance scores to construct narrow networks, but specialists have no reciprocal visibility into who is steering patients away from them or why.

Closing Thoughts

To be fair, CMS doesn’t actually hate specialists. The people building these programs are smart, well-intentioned, and have the difficult task of balancing incentives and political realities. ASM is evidence they understand that specialist accountability needs to be part of the value-based care equation rather than an afterthought. CMS is tying program design to an agenda that emphasizes chronic disease prevention, whole-person care, and improving patient health without expensive interventions. That framework is applicable to specialist-led models too.

The PCP-as-gatekeeper narrative underpinning some of CMMI’s models is less certain than it appears. Routing care through primary care only reduces costs if it prevents inappropriate procedures, and the evidence that gatekeeping selectively filters unnecessary utilization is thin. The knowledge gap that makes PCP-led MSK management suboptimal is well documented and underappreciated.

Specialists — the people CMS (and others) keep trying to route around — may be the most underutilized lever for achieving exactly what everyone says they want: better chronic disease outcomes, lower total cost of care, and patients who enter and exit the system healthier than they were before.

ASM is one answer, but perhaps not the best one.

At the end of the call, Abe suggested I might want to join “the team.” I think he was just humoring me, but I’m always open to opportunities to work on hard problems. Being a blogger/critic is the easy part — anyone can do that. Putting thoughts into action is harder... and far more fruitful. As history has proven, that’s a difficult task in healthcare reform.

I’ve already got some ideas and a blueprint — what I call “BPCI-X.”

More to come.

Discussion about this post

Ready for more?