Self-Inflicted Wounds
CMMI's Complexity Problem
CMS proposed CJR-X two weeks ago essentially ensuring that, by October 2027, virtually every joint replacement performed in an acute care hospital will be part of an accountable care program. For the first time in Medicare history, we’ll have mandatory, nationwide episode-based payments. Like it or not, VBC will soon be unavoidable.
Is this progress or an expensive administrative solution to a problem largely of CMS’s own making?
Is There Still Meat on the Bone?
Original CJR ran for eight years and generated decent savings — roughly $112 million over two years in its later performance years. As anyone who has studied or participated in these models knows, the majority of those savings came from reductions in post-acute care. Sending patients home instead of to skilled nursing facilities generates the biggest financial wins. Better surgical outcomes and lower complication rates are well and good, but discharge destination is the value champion.
Studies show that VBC and FFS payment models deliver more or less equivalent quality. For example, a Health Affairs analysis found that “changes in quality of care, measured by complication rates, emergency department visits, and mortality, were not associated with bundled payment participation.” Programs like CJR reduce costs, but don’t necessarily improve outcomes. As I’ve already pointed out, those reduced costs come from reduced post-acute care spending. Once that’s been exhausted, outcomes driven cost savings may not pick up the slack.
CJR-X is a tweaked version of a similar model, applied nationwide. The post-acute care reduction ship has already sailed in many places. Fewer patients are being discharged to rehabs and SNFs. There’s another factor at play too, site-of-service shift. CMS’ own data shows that, by the end of the original CJR model, 70% of elective joint replacements were being performed in the outpatient setting. By definition, those patients go home. (Note: CJR-X expands to total joints performed in HOPDs. Impact TBD.)
If most savings in the original model came from avoiding SNFs and rehabs, where will the savings come from in the new model? Are we reaching the point of diminishing returns? Original CJR saved $1,142 per episode. CMS projects CJR-X will save $725M over a 5-year period. With around 750,000 Medicare joint replacements performed annually, that’s less than $200 savings per episode — a significant reduction. Notably, cost reduction isn’t even an explicit goal of CJR-X.
The tradeoffs are scale and competition. Everyone who’s not part of TEAM will be forced into CJR-X. Benchmark comparisons are regional, not internal. Program goals may be shifting away from costs savings towards care coordination and quality improvements. These are noble goals, but to what end? Is all the added complexity justified, or are there better ways to achieve the same results?
The Wound CMS Inflicted on Itself
Between 2000 and 2024, inflation-adjusted Medicare reimbursement for primary hip and knee replacement fell 55%. CMS currently pays approximately $1,160 for a joint replacement, including 90 days of postoperative care. In real terms, that’s less than it paid twenty years ago for procedures whose outcomes and efficiencies have improved meaningfully in the setting of an aging, less healthy population.
Reducing unit price in the face of inflation-driven overhead increases pushes rational actors to increase volume. Annual Medicare joint replacement surgeries doubled from 2000 to 2019. Over that same time period, reimbursements experienced negative CAGRs. If the contention is that FFS incentivizes volume, no one should be surprised when lower payments lead to more procedures.
Of course, that assumes that “volume over value” is a valid argument. The easy conclusion is that lower reimbursement produces unnecessary procedures — surgeons stretch indications to compensate for smaller margins. The evidence to support that assumption isn’t as strong as we think.
Joint replacements have pretty well-defined criteria with fewer gray areas than, say, back surgeries. There will always be quick-to-cut surgeons, but overutilization of joint replacements remains an open question. A controversial 2015 study deemed 34% of knee replacements “inappropriate.” That study had significant methodological flaws and was based on outdated appropriateness criteria. In some populations, joint replacements are arguably underutilized.
Volume growth is real, but shifting demographics, not surgical opportunism, is the main driver. What we have is an aging population that’s being diagnosed with hip and knee arthritis at increasing rates. There is absolutely no need to do unnecessary joint replacements — surgeons will need to double their caseload by 2050 just to meet expected demand. Tinkering with payment models won’t change this dynamic.
No one has a great answer for the demand side problem. Mandatory bundles don’t reduce the number of people who need a new hip or knee. Episode-based accountability doesn’t slow the demographic and metabolic trends driving volume. The supply of fellowship-trained arthroplasty surgeons is not infinitely elastic. Current reimbursement trends may make the subspecialty less attractive to trainees and force more surgeons to opt out of Medicare. An access problem is bubbling underneath the cost control conversation.
Compressed reimbursement also makes independent practice less economically viable. Consolidation of independent groups is demonstrably inflationary. Once surgeons are employed, market power increases, facility fees tacked on, and the balance of power shifts. As a price setter, Medicare is somewhat protected from consolidation forces. But vertical integration and site-of-service payment differentials undoubtedly increase costs for the government, too. Mandatory VBC may be the proposed solution, but at least one study found that physician-led bundles save more money than hospital-led bundles. CMS appears to be pivoting away from surgeon-led VBC models.
Fee compression drives increased volume and consolidation. Volume and consolidation are inflationary, and inflation justifies cost controls. VBC models generate modest net savings but come with significant administrative cost. Meanwhile, the underlying structural forces compound and shifting demographics drive utilization. CMS has been treating a condition it created and a demand problem it can’t control with a remedy that could make the condition worse.
A Wrinkle in TEAM
The FY2027 IPPS proposed rule containing CJR-X includes an interesting wrinkle: an RFI asking whether grandfathered physician-owned hospitals should be allowed to participate in TEAM. Unfortunately, it’s not a full reopening of the ACA’s Section 6001 moratorium on POHs — but it’s something.
What the RFI would do is generate controlled, comparative data on the two arguments the AHA has used for years to justify the ban. First, that physician-owned hospitals cherry-pick healthier patients. Second, that their apparent cost efficiency is an artifact of that selection rather than genuine care delivery improvement.
Recent analyses across roughly 740,000 beneficiaries found no evidence of cherry-picking. Commercial prices were 17% lower with potential annual Medicare savings estimated at $1.1 billion. As expected, the AHA pushed back on those findings which were commissioned by physician advocacy groups. TEAM participation by POHs could settle the argument once and for all. If both sides truly believe in their stance, they should welcome a narrowly scoped opportunity to prove themselves right. So far, that’s not what’s happening.
BCBSA has already submitted a letter to the DOJ asking for limited Section 6001 exceptions. They explicitly want surgical-specialty facilities, including orthopedic, spine, and cardiac hospitals, carved out. These are the exact facilities that would generate the most meaningful comparison data. That a commercial payer commented on a proposal 11 months before it was announced is telling and, frankly, more of the same game most are tired of playing.
The AHA and Federation of American Hospitals are expected to file on the same side before the June 9 comment deadline. If these efforts succeed, CMS will have passed up the one provision in its own rule that pointed toward a structural answer rather than an administrative one. It will have once again capitulated to self-protectionism at the expense of meaningful progress.
A Different Test
One way or another, I keep coming back to same conclusion: CMS is making things unnecessarily complicated. The ROI on these models just doesn’t seem to justify their complexity, especially if more effective alternatives are possible. I’ve written before about what a genuinely different model might conceptually look like — The Simple Model. The core argument is straightforward: we keep testing variations of the same complicated hypothesis when the simpler one hasn’t been tested.
A targeted, voluntary test of reimbursement restoration for independent orthopedic practices — with episode cost and quality tracked through existing registry infrastructure and claims data — would test the second-order case directly. CMS has the data, does it really need convoluted, administratively burdensome reporting mechanisms?
It’s a simpler model than CJR-X without the need for 230-pages of proposed rules or administrative infrastructure implementation at thousands of hospitals. The net effect could be slowing or reversal of the consolidation trend and volume migration to lower-cost settings. The net Medicare math is likely favorable. If not, the cost of the experiment is bounded and you’ve learned something.
CMS knows how to adjust conversion factors and construct complicated reimbursement models. Based on the results so far, it may be going about things the wrong way. The same proposed rule that would mandate episode accountability at every hospital in America contains a question about whether POHs should be allowed to compete. It’s a small but extremely important provision. The salve for CMS’ self-inflicted wounds has been visible the whole time.
Primum non nocere.
Ben Schwartz, MD, MBA
Editor-in-Chief
Great piece Ben. Do you think collection of patient reported outcomes and analysis of the results will shed more light on the question of unindicated surgery? In other words, an insufficient clinically important difference between pre-and postoperative scores could be revealing. Or low preoperative scores could indicate an overly aggressive approach to surgery.