The Big Cycle Comes for Physical Therapists

Is physical therapy heading toward its own financial reckoning?

In Episode 32 of our podcast, Alex and I ran musculoskeletal care through the same systems logic Ray Dalio uses to explain how countries go broke. I want to walk you through what came out of it, because the reason it all lands right now is simple: the pressure we've discussed in the abstract for years is finally showing up in the 2026 numbers. For most of my career, "the system will have to change" was a thing we said at conferences and then went back to our schedules. This year it's in the rule.

How physical therapy goes broke

Dalio's new book treats national insolvency as a system that fails on a schedule, a short cycle of five to eight years riding inside a long cycle of fifty to a hundred. Run MSK care through that lens and the picture gets uncomfortable fast. Large systems are taking write-offs. Payers and financiers are backing away from operations that stopped returning what they promised. I read this as the first tremor of a two-to-three year reorganization, and I'm calling it now.

Here's why it's no longer theoretical. The 2026 Medicare Physician Fee Schedule finalized a permanent efficiency adjustment that reduces the work value of our evaluation and re-evaluation codes, the exact diagnostic reasoning that only a physical therapist does. The headline conversion factor went up, yes, but most of that bump is one-time congressional money that expires, which means another cut is already waiting for us in 2027. Read the "increase" for what it is and plan accordingly.

Your evaluation pays $90. Your treatment pays $30.

The clearest illustration lives inside our own fee schedule. Our highest-paying code is the initial evaluation, somewhere around $90 to $100. The next one down, therapeutic exercise and neuromuscular re-education, pays roughly $30 to $35. Three times the value for the diagnostic work over the delivery work! The system has already told us where it thinks our expertise lives.

Most of us keep over-delivering the low-value half anyway. So design your day around the thing the system pays a premium for. Be readily available for the formal evaluation, own the rehab diagnosis and the prognosis and the plan, and let your assistant partners deliver what you set. That one shift moves you toward your top of license and off the high-volume treadmill that quietly caps your access and your revenue at the same time.

Your patients are already running your plan through ChatGPT

Picture a time-constrained patient leaving your clinic. On the walk to the car, they open a chatbot, type in their symptoms and the plan you just gave them, and ask whether it holds up. Thirty seconds, and now they have a second opinion.

I'm not raising that to scare anyone. I'm raising it because of what just changed underneath it. A June 2026 study in Nature Medicine found that general-purpose frontier models outperformed purpose-built clinical AI tools across medical knowledge, clinician alignment, and real physician queries, and the clinicians preferred them. That “second opinion” is genuinely competent now, and we underestimate how fast patients, seniors included, are getting comfortable using it. So write every plan as if it will be reviewed, because it will be.

The money is moving to primary care, and we're in the specialty bucket

While we were watching the cuts, CMS built a new front door and forgot to invite us in. Advanced Primary Care Management codes went live in January 2025, a monthly, non-time-based payment for the wrap-around, team-based work of primary care. They picked up roughly a ten percent raise for 2026 and gained new behavioral health add-ons. Underneath that, the innovation models keep testing what happens when you pay an advanced primary care team a capitated monthly amount and hold them accountable for cost and quality, and LEAD, the ten-year ACO model arriving in 2027, extends that logic with prospective, population-based payments.

Medicare Part B is a closed system. When CMS funds the primary care and behavioral health teams more richly, that money comes from somewhere, and right now we sit in the specialty bucket that pays for it. The definition of the "advanced primary care team" is being drawn as we speak, and our name is not on it. We have a window to insert ourselves as a funded, first-line service for the movement, function, and behavior-change work we already do better than anyone. That window will not stay open forever.

Your discharges might be creating your own return visits

Here's a clinical reason this matters beyond economics. Tissue remodels over months and years. Collagen, tendon, bone, and muscle each recover on their own timeline. Our episodes of care run in days and weeks. So we treat people until symptoms quiet down, call it a successful “discharge,” and then watch a good number of them bounce back into a new episode later. We've been rewarded for volume and habituation instead of durable change when we should be rewarded for maintaining a longitudinal relationship with our patients.

A system that finally pays for outcomes over visits rewards the care we actually trained to provide. And the evidence is landing right on time. Duke researchers, including Trevor Lentz and Chad Cook, published work in late 2025 showing that removing financial barriers to early physical therapy led to more evidence-based care and cut downstream utilization. That is the argument we've made for twenty years, now with the receipts. When interoperability lets payers and referring providers pull our outcomes straight from the EMR, and it will, I want us holding data like that.

AI is safer than your intuition wants to believe

Iatrogenesis is one of the largest preventable costs in medicine, which is worth remembering before we reflexively distrust a machine. Think about the Waymo comparison. Across more than 170 million driverless miles, the Waymo Driver was involved in 92 percent fewer serious-injury crashes than human drivers. It also ran over a cat once, and got stuck making a U-turn, and those are the stories we tell. The safety profile is dramatically better.

There's a version of this we control. AI has a training phase and a tuning phase, and the tuning happens through real exposure in real settings. The clinicians on site are the tuners. The more we bring these tools into practice, the more we shape them to work for us instead of around us. Sitting it out hands the tuning to someone else.

The choice in front of us

If you want proof the front door is real, it opens this month. The ACCESS Model, a ten-year test of technology-supported chronic care, begins its first cohort in July 2026. It covers chronic musculoskeletal pain, and it publishes risk-adjusted outcomes so patients and referring clinicians can choose where to go based on who actually gets results. Care that starts with a technology-enabled intake, outcomes made transparent at the clinic level, patients steered toward the practices with the best three, six, and nine-month numbers. That's the digital front door I described on the episode, arriving as federal policy rather than as a prediction.

(Note: make sure you read up on the ACCESS co-manager/referring clinician role in the model!)

Every system under pressure faces the same fork: evolve the thing you have, or let it break and get rebuilt around you. The models stopped being models to me. This is a revolution in how care gets paid for, and it's underway. If we go with it, prove our value beyond what payers currently reimburse, and plant ourselves in the parts of the system gaining dollars instead of losing them, this profession gets more rewarding, more individualized, and finally legible to the rest of medicine.

Or we can keep our paper charts and our goniometers and wait to find out what the restructuring decides for us. I know which one I'm choosing.

I never liked those goniometers anyway.

Sources & further reading

The systems frame the whole argument borrows from. Ray Dalio, How Countries Go Broke: The Big Cycle (Avid Reader Press, 2025). Publisher page

Where the 2026 cuts to our evaluation codes actually live. APTA's breakdown of the proposed fee schedule, including the efficiency adjustment and the primary-care-over-specialty tilt. APTA

The CMS landing page for the Advanced Primary Care Management Codes, accessed 7.5.26.

The code the advanced primary care team is being built around. CMS, Advanced Primary Care Management Services. CMS

The digital front door, as federal policy. CMS ACCESS Model, a ten-year test of technology-supported chronic care covering chronic MSK pain, launching July 2026. CMS

The consumer-AI threshold we're now on the wrong side of if our plans are sloppy. NYU Langone study of frontier models versus specialized clinical tools. Nature Medicine

The evidence that early PT access cuts downstream cost. Duke Orthopaedic Surgery on the Lentz and Cook no-copay study. Duke

Check out my recent article in OT Potential: “Therapy and Value-Based Care Models”