Taking Control of MSK

Clarity on Process and Value as Prior Authorization Tightens and Costs Accelerate

By RaeAnn, Founder and Chief Executive Officer, HLTHWORKS

THE SHORT VERSION

Most health plans cannot say what they spend on musculoskeletal care, because the definition spans several code sets that no standard report assembles.
Trend is a price and mix story. Inpatient volume fell 51 percent while prices rose 36 percent, and demand keeps climbing. Volume is relocating, not falling.
Independent evidence puts virtual musculoskeletal savings at roughly 37 cents per member per month, and only where the program substitutes for in person care rather than adding to it.
Prior authorization, the lever plans have used to hold this category, is being narrowed and made public on a published federal schedule.
The replacement is measurement. A denominator the plan owns, an orchestration layer, a performance layer, 20 pathways, and steerage that providers earn in the data.

Every day a chief financial officer is looking for the same thing. Where the spend is, what is driving it, and what can be done about it. Today the question is musculoskeletal, and it comes in two parts. Are we reducing trend. And are we getting the return we were promised from our vendors. Most health plans cannot answer either half. The reason is not that the vendors are hiding the answer. The reason is that the plan never defined what it was measuring, so there is no denominator to measure anything against. That is a finance problem before it is a clinical one, which also means it is fixable. What MSK is, in code Musculoskeletal is not one code set. It is several, and they do not sit together in any standard report. A plan that wants to own this category has to assemble the definition itself.

WHERE	WHAT TO PULL	WHY IT BELONGS
ICD-10-CM Chapter 13	M00 to M99	Core musculoskeletal and connective tissue disease. Note that dorsalgia at M54 and joint pain at M25.5 already live here. They are not symptom codes.
ICD-10-CM Chapter 19	S00 to T88	Fractures, sprains, and dislocations. Also T84, complications of orthopedic implants, where revision volume actually lands.
ICD-10-CM Chapter 6	G00 to G99	Nerve root and plexus disorders and chronic pain, which drive much of the imaging and injection spend.
ICD-10-CM Chapter 18	R00 to R99	Gait and mobility abnormality, unspecified pain, and weakness.
ICD-10-CM Z codes	Aftercare and status	Joint replacement aftercare and presence of orthopedic implants.
CPT surgery	20000 to 29999	Procedures on the musculoskeletal system, including arthroscopy.
CPT radiology	70000 to 79999	Spine imaging at 72020 to 72120, whole spine at 72081 to 72084, extremity imaging in the 73000 series.
CPT rehabilitation	Therapy codes	Therapeutic exercise at 97110, manual therapy at 97140, gait training at 97116.
MS-DRG	MDC 08	Joint replacement at 469 and 470, hip replacement for fracture at 521 and 522, hip and femur at 480 through 482, cervical fusion at 471 through 473.
HCPCS Level II	L and E codes	Orthotics and prosthetics, plus braces, bone growth stimulators, walkers, and canes.
Pharmacy	Drug classes	Nonsteroidal anti-inflammatories, opioid analgesics, skeletal muscle relaxants, tumor necrosis factor inhibitors, antigout agents, and metabolic bone disease agents.
Facility claims	Revenue codes	Hospital based therapy, which professional CPT logic never sees.

Define musculoskeletal as Chapter 13 alone and four large things disappear. All trauma. Implant failure and periprosthetic infection. Post-acute aftercare and implant status. And the nerve root and chronic pain presentations that generate the imaging. The same definition also overcounts in one place almost nobody catches. Dentofacial anomalies and jaw disorders sit inside M00 to M99 and are almost always dental spend. They should be carved out explicitly. Two joins matter as much as the code ranges. Hospital based therapy bills on facility revenue codes, so professional claim logic alone will undercount rehabilitation. And the same biologic bills as a J code under the medical benefit and as a national drug code under pharmacy, so a view that does not join both will split that spend in half and understate each side. One warning for the lookback. Spinal fusion diagnosis related groups were restructured effective October 1, 2024. New groups were added and two were renumbered. A three year or four year lookback crosses that boundary, and if it is not adjusted the change will read as a shift in surgical volume that never happened.

If the definition is wrong, every return on investment calculation built on top of it inherits the error.

What the chief financial officer is asking

The question in the room is blunt. Are Hinge, Sword, Tailor Care, and the rest of the musculoskeletal vendors and providers going to deliver the return they sold us, and are outcomes actually better.

The market has started answering. In the Business Group on Health 2027 Employer Health Care Strategy Survey, fielded in June 2026 across 127 employers and 11 million covered lives, median trend is 8.5 percent for 2026 and 9.2 percent for 2027. Musculoskeletal is the second largest cost driver, cited by 68 percent of employers, behind cancer at 92 percent. Fifty eight percent of those employers plan to replace underperforming vendors. Eighty three percent are expanding performance guarantees. Seventy one percent are increasing outcome tied fees.

The Peterson Health Technology Institute surveyed 321 digital health purchasing decision makers in July 2026. Sixty eight percent of employers and 55 percent of health plans now use performance based contracts, and of those contracts, 85 percent put at least a quarter of fees at risk. Thirty eight percent of purchasers said they cannot independently verify vendor reported outcomes. Forty five percent said they cannot integrate vendor data with their own claims. Forty seven percent said fewer than a quarter of eligible members ever enroll.

Then read a filing. Hinge Health discloses in its most recent quarterly report that performance guarantees are included in most of its contracts, and that payouts on those performance guarantees have been immaterial to date.

Guarantees in most contracts. Payouts immaterial. And more than a third of buyers unable to verify the outcomes they are being billed against.

That is not a vendor problem. A performance guarantee only pays out when the buyer can prove the target was missed. If the plan cannot measure the outcome, there is nothing to prove, and the guarantee never pays. That is the whole explanation for payouts being immaterial, and the fix sits on the plan’s side of the table.

What the chief actuary sees

The actuary reconciles to paid claims, which makes the actuary the least sentimental reader of any vendor report.

Start with the best independent assessment available. The Peterson Health Technology Institute evaluated virtual musculoskeletal solutions in June 2024 and found that physical therapist guided programs improve pain and function comparably to in person physical therapy. The net budget impact was 4.4 million dollars in savings per one million commercially insured lives, which is roughly 37 cents per member per month. That result holds only if the price is 995 dollars per year and only if a quarter of in person physical therapy actually shifts to the virtual program. The economic model covers low back pain alone.

The mechanism matters more than the number. The savings come from substitution. A program layered on top of existing care substitutes for nothing and saves nothing. The assessment also declined to count reduced surgical intent as a proxy for reduced spending, which is the proxy several vendor savings models depend on. Of the 48 studies that met inclusion criteria, 44 were submitted by five of the companies being evaluated, and about three quarters were rated high risk of bias.

Now read the trend. The Health Care Cost Institute released its 2023 commercial claims report on October 13, 2026. Between 2019 and 2023, inpatient musculoskeletal admissions fell 51 percent, the largest decline of any inpatient subcategory, while the average price of those admissions rose 36 percent. Musculoskeletal procedures now account for 26 percent of all outpatient procedure spending. Musculoskeletal drugs are the single largest share of gross prescription spending at 17 percent, up 46 percent over the period, with point of sale price up 49 percent. Across all categories, price drove 90 percent of the increase in per person spending, while utilization grew 1.8 percent over five full years.

Set that against demand. Research published in the Journal of Arthroplasty in 2025 projects Medicare primary knee replacements rising from 1,013,533 in 2022 to 2,793,919 by 2040, and primary hip replacements from 594,981 to 2,174,382.

Volume is not falling. It is relocating. A plan still watching musculoskeletal through inpatient admissions believes this problem is getting smaller.

There is one more thing the actuary will notice, and it is worth stating plainly. No current independent source publishes a national musculoskeletal spending total. Federal expenditure data is organized by type of service rather than by condition, and the most recent independent estimates by condition run on data more than a decade old. The figure most often repeated in this market traces back to a report commissioned by a vendor and cannot be independently verified.

Which is the argument for everything that follows. If the number does not exist outside the plan, the plan has to build it inside.

What the chief compliance officer is asking

The compliance officer has a different problem, and it arrives on a published deadline.

Under the CMS Interoperability and Prior Authorization final rule, impacted payers, which include Medicare Advantage organizations, Medicaid and CHIP fee for service and managed care, and qualified health plan issuers on the federal exchanges, must return expedited prior authorization decisions within 72 hours and standard decisions within seven calendar days. Every denial has to carry a specific reason. Payers must post prior authorization metrics publicly on their own websites, with the first set due March 31, 2026. The prior authorization programming interface follows beginning January 1, 2027. Add roughly two dozen state gold carding laws on top of that.

The direction is unmistakable. Prior authorization is being made faster, narrower, more automated, and public. And it matters more in this category than in any other, because musculoskeletal is where prior authorization volume concentrates.

The lever most plans have used to hold musculoskeletal spend is being regulated away, on a published schedule. Nothing replaces it unless the plan builds the replacement.

So the three executives are asking one question in three dialects. The chief financial officer wants to know whether the vendors will deliver what they sold. The chief actuary wants to know why the trend has not moved. The chief compliance officer wants to know how the plan absorbs a faster and more transparent prior authorization process without losing control of the category.

The answer is the same for all three. Replace friction with measurement, and build it before the deadline rather than after it.

What actually moves the number

Four interventions have real evidence behind them. Each one also has a limit, and a plan that knows the limit builds a better pathway than one that does not.

The timing of physical therapy. In a study of 753,450 adults with new low back pain in the Military Health System, patients who began physical therapy within 14 days, compared with those who began between 15 and 90 days, had roughly half the odds of advanced imaging and spinal injections, and lower odds of surgery and opioid use. Two year low back pain costs were 1,828 dollars against 3,031 dollars. Only 16.3 percent received physical therapy within 90 days at all. The limit worth knowing is that this compares early therapy to late therapy among patients who receive it. It is not a finding that therapy beats watchful waiting.

Restraint on imaging. The American College of Radiology’s 2021 criteria describe uncomplicated acute low back pain as a benign, self-limited condition that does not warrant imaging. It is imaged anyway about a quarter of the time in primary care. In a Veterans Affairs cohort of 405,965 patients with nonspecific low back pain and no red flags, those who received a lumbar MRI within six weeks went on to surgery at 1.48 percent against 0.12 percent, used opioids at 35.1 percent against 28.6 percent, and generated acute costs of 8,082 dollars against 5,560 dollars. A HEDIS measure for this already exists, and the national rate says about one in four patients is imaged within 28 days.

Prehabilitation. A 2023 meta-analysis of 48 randomized trials found moderate certainty evidence that prehabilitation improves function before knee replacement and for roughly six weeks after it. Length of stay falls by about half a day in knee replacement and not significantly in hip. The post-acute case rests largely on one Medicare analysis of 4,733 joint replacements, where any post-acute care was 54.2 percent with preoperative therapy against 79.7 percent without, and skilled nursing was 31.1 percent against 43.7 percent. The limit is that the data is from 2008 and 2009, before joint replacement moved outpatient, so that baseline no longer holds. Prehabilitation is worth building. It is not worth promising a number on.

Fall prevention. About one in four adults 65 and older falls each year, more than 14 million people, and the age adjusted fall death rate rose 21 percent between 2018 and 2024. The protocol is screen, assess, intervene. The supporting evidence is observational: patients identified as at risk who actually received a fall plan of care had lower odds of fall related hospitalization. A randomized trial published in 2026 was null, though it was underpowered. Build it, measure it, and claim no more than that.

Notice what these four have in common. Every one of them is a sequencing decision, not a product.

The plan

Five phases. Each one has an owner, a timeline, and a gate, and no phase starts before the gate behind it closes. The infrastructure build is four months. Everything after that runs on quarters.

PHASE 1. DATA AGGREGATION AND ASSESSMENT

Aggregate three to four years of paid claims across every code set above. Trend each component separately: inpatient surgery, outpatient surgery and site of service, physical therapy, imaging, pharmacy, and devices. Split price from volume. Carve out dental. Adjust for the spinal fusion grouper change.

OWNER Finance and actuarial, with clinical informatics.

TIMING 30 days.

GATE One musculoskeletal denominator that finance, actuarial, and clinical services all sign. No vendor reporting is accepted against any other number.

PHASE 2. NETWORK INTELLIGENCE

Bring provider data and vendor data into one place. The orchestration layer routes the member and sequences the care. The performance layer measures provider, setting, vendor, and pathway against the denominator from Phase 1. Neither layer is a product you buy. Both are owned by the plan.

OWNER Data and technology, with network management.

TIMING 90 days.

GATE The performance layer reproduces the actuary’s paid trend within an agreed tolerance. Until it does, it is a dashboard, not a measurement system.

PHASE 3. PROTOCOLS AND JOURNEY MAP

Rank the top 20 by spend and build each one. Put the fall prevention protocol in the 20 regardless of where it ranks, because it guards against the most expensive events in the category. Every pathway documents the same six things: the physical therapy timing rule, the imaging rule, the surgical indication, the prehabilitation step, the post-acute plan, and the medication set. Build the first five completely before starting the rest, because the second pathway is where you learn what the template was missing.

OWNER Clinical services, with medical policy.

TIMING First five pathways in one quarter. The remaining fifteen across the next two quarters.

GATE Every pathway has a named owner, one primary measure, and a documented exception path.

PHASE 4. MONITOR, THEN DISPLACE

Overlay the musculoskeletal team on both layers. Watch prehabilitation, surgery, and post-acute rehabilitation as one sequence rather than three separate events. Measure every vendor contribution inside the actuary’s trend rather than inside the vendor’s report. As the trend moves, displace what is not producing, and use the performance guarantee language to do it.

OWNER The musculoskeletal team, with finance.

TIMING Continuous monitoring. Vendor displacement decisions by the third quarter.

GATE Each vendor has a documented contribution to paid trend, or a termination date.

PHASE 5. STEER, AND THEN REWARD

Member and provider report cards inside every pathway. Identify the best physicians and the best settings on quality, outcome, satisfaction, and total cost of the episode. Then reduce friction for the providers who earn it. Centene announced a national performance based gold card program on October 13, 2026, launching in November for more than 160,000 eligible providers across select outpatient services, with qualification based on approval history, documentation quality, and compliance with medical policy. Musculoskeletal is not named in it.

OWNER Network management and medical management.

TIMING Follows the first displacement decisions, from the fourth quarter.

GATE Every gold card decision traces to a measure in the performance layer, and utilization is monitored after the card is issued, not only before.

That gate is there for a reason. The National Association of Insurance Commissioners warned in December 2025 that gold carding may increase service utilization and put upward pressure on cost. The one published payer evaluation, an imaging program in Vermont, found that utilization among participating providers rose above both its own baseline and that of non-participating providers. Removing friction without measuring what follows is not a strategy.

Be careful with designation itself. Two evaluations of a national center of excellence program for joint replacement and spine surgery found that designated hospitals did not reliably outperform others. For spine, no difference reached significance. Where steerage has demonstrably worked, the mechanism was a mandatory expert second opinion paired with benefit design, not the designation. One large employer reported finding spine surgery medically unnecessary in about half the cases it reviewed.

So the sequence matters. Measure first, steer second, and reduce friction last, for the providers your own data says have earned it. Musculoskeletal is the obvious place to point the mechanism, because prior authorization volume, provider variation, rising incidence, and spend all converge in this one category. Whoever applies it here first, with the measurement underneath it, sets the standard everyone else answers to.

Inside one pathway

Take knee osteoarthritis moving toward replacement, which sits near the top of almost every plan’s 20. Here is what the six elements look like when they are actually written down.

Physical therapy timing. Measure the interval from first diagnosis to first therapy visit, and report it by medical group. The rule is expressed in days, not in months.
Imaging. Weight-bearing plain films where indicated. Advanced imaging reserved for findings that would change the surgical plan. Track the share of knee episodes carrying an MRI before any therapy at all.
Surgical indication. Document what has to be true before authorization: conservative care attempted over a defined period, functional limitation captured with a standard instrument, and imaging consistent with the complaint. The second opinion belongs here, because that is where the documented savings in steerage programs actually came from.
Prehabilitation. A defined program before the surgical date, with completion tracked rather than assumed. Expect better function going in and roughly half a day of length of stay. Do not build a business case on more than that.
Post-acute plan. The discharge destination is decided before surgery, not at discharge. Home with outpatient therapy is the default. Skilled nursing becomes an exception that someone documents.
Medications. The set is defined up front, including what is used for pain control after discharge and for how long.

Then one owner and one primary measure. For this pathway the measure is the share of members who complete therapy before surgery is authorized. Everything else on the list is diagnostic.

One caution carries across all 20. The strongest evidence in this field, on therapy timing and on imaging restraint, comes from low back pain. It does not transfer automatically to knee, hip, or shoulder. Build each pathway on the evidence for that condition, and where the evidence is thin, say so inside the pathway document rather than borrowing confidence from a different body part.

The close

Prevention, quality, satisfaction, outcome. All of it measured inside the plan, against a denominator the plan owns.

A health plan cannot outsource accountability for a category it cannot measure. It can only outsource the explanation.

WHERE HLTHWORKS WORKS

HLTHWORKS builds Phase 1 with health plans, the denominator, and the performance layer that every phase after it depends on. If you cannot say what your musculoskeletal spend actually is before the next vendor renewal, that is the place to start.

ALSO LISTEN

Redesigning Healthcare: Unfiltered, Episode 8. Insourcing musculoskeletal care for health plans and health systems.

Sources

Business Group on Health, 2027 Employer Health Care Strategy Survey, fielded June 2026.
Peterson Health Technology Institute, Virtual Musculoskeletal Solutions Health Technology Assessment, June 2024.
Peterson Health Technology Institute, 2026 State of Digital Health Purchasing, fielded July 2026.
Health Care Cost Institute, 2023 Health Care Cost and Utilization Report, released October 13, 2026.
Jones and colleagues, Trends in Medicare Arthroplasty Procedure Volume: Projecting From 2025 to 2040, The Journal of Arthroplasty, 2025.
Hinge Health, quarterly report on Form 10-Q, August 2026.
Centers for Medicare and Medicaid Services, National Correct Coding Initiative Policy Manual 2026, and FY2025 inpatient prospective payment system changes.
Centene Corporation, National Performance-Based Gold Card Program announcement, October 13, 2026.
Centers for Medicare and Medicaid Services, Interoperability and Prior Authorization Final Rule, CMS-0057-F, January 2024.
Childs and colleagues, Implications of early and guideline adherent physical therapy for low back pain on utilization and costs, BMC Health Services Research, 2015.
Jacobs and colleagues, Observational Study of the Downstream Consequences of Inappropriate MRI of the Lumbar Spine, Journal of General Internal Medicine, 2020.
American College of Radiology, Appropriateness Criteria, Low Back Pain, 2021 update.
Punnoose and colleagues, Prehabilitation for Patients Undergoing Orthopedic Surgery, JAMA Network Open, 2023.
Snow and colleagues, Associations Between Preoperative Physical Therapy and Post-Acute Care Utilization Patterns and Cost in Total Joint Replacement, Journal of Bone and Joint Surgery, 2014.
Centers for Disease Control and Prevention, Older Adult Falls Data, 2026, and the STEADI initiative.
National Association of Insurance Commissioners, Prior Authorization White Paper, December 2025.
Mehrotra and colleagues, Evaluation of Centers of Excellence Program for Knee and Hip Replacement, and Evaluation of a Center of Excellence Program for Spine Surgery, Medical Care, 2013.

HLTHWORKS builds artificial intelligence governance, scalability, sustainability, oversight, audit readiness, and a mature enterprise solution for medical groups, health systems, health plans, payers, and the vendors, plus investors serving them.

Taking Control of MSK