Zimmet Healthcare offered one service when we started in 1993: Medicare Cost Reporting. No nurses, no analytics, no software. Just Cost Reports, and the work required to maintain the integrity of a finely tuned instrument that tracked expenses and calculated a facility-specific rate.

The process was nuanced, scrutinized, and consequential. Allowable costs were classified, allocated, audited, challenged, appealed, and eventually settled. Far from perfect, the Cost Report was traceable and demanded accountability. Medicare was not paying an abstract price. It was reimbursing allowable cost, subject to defined rules.

Then Came Change

May 12, 1998, changed the trajectory. That was the date CMS, then HCFA, released the Interim Final Rule establishing the initial SNF Prospective Payment System. “Rate shock” followed. Providers failed. Cost Reporting firms disappeared. We were spared much of the immediate fallout because New York was included in a demonstration that had already lured most providers into PPS a few years earlier. Had we been based elsewhere, I would not have lasted in long-term care. Either way, Medicare SNF reimbursement was never the same. I have seen precision and discipline descend into a pasty bolus of inconsistency and indifference. It’s wrong, it’s damaging, and to be honest, kind of gross.

The cost-based methodology had its flaws, exacerbated by publicly traded nursing home chains that pushed so hard they were pulled into bankruptcy. I am not advocating for a return to the old system. PDPM is exponentially more rational than the statistically deficient perversion of RUGs. That said, CMS and many state methodologies are built on a foundation that cannot bear the weight now being placed on it.

A Little Perspective

Historical context is telling. Medicare was not an easy sell to providers. Hospitals opposed federal price-setting; the program was blocked politically until the government promised reimbursement of reasonable cost, subject to rules, documentation, allocation, and audit. A reimbursement system identifies costs and makes the other party whole. A price system assigns a rate through formulas, weights, indexes, updates, assumptions, and the distorted or misreported experiences of others. The modern SNF system still uses the language of reimbursement, but the cost tether has frayed.

I am one of a small group of professionals still active on a national level who prepared SNF Medicare Cost Reports when they set facility-specific rates, pre-PPS, when Excel was a novelty and the medium carrying the numbers was accounting paper, sometimes taped together so one form could fit on a single page.

A Spoon-Bending RFI

That is why I cannot wrap my head around the 2027 PDPM Request for Information (RFI). CMS is trying to “bend the spoon.” It observed that case-mix scores increased under PDPM, a statement made as if discussing Schrödinger’s cat. Of course, scores increased; we calculated nearly the same increase within three months of launch.

PDPM replaced a therapy-dominated payment system with one that makes resident characteristics economically meaningful. The mystery is not whether the box contains higher case-mix. The mystery is why CMS believes the current reporting infrastructure can determine, with technical confidence, how much of that increase reflects acuity, model mispricing, or what the agency calls “case-mix creep.”

I take exception to the term because it collapses degrees of intent into one nefarious-sounding bucket. Improved capture of conditions that existed all along is not the same as aggressive coding, scheduling, or framing care around mispriced services. If providers are presented ARD options, choosing a day that captures a reimbursement-sensitive service instead of the next day that does not is obvious. CMS may consider that creep. I would call it the predictable behavior of rational operators inside the system CMS designed.

Rotting at the Core

The deeper problem is technical. CMS is asking whether payment reflects resource utilization, but the current system does not produce a valid resident-level resource-cost measure. The MDS classifies residents. Claims report covered days and billed items. Cost Reports summarize facility-level allowable expenses. None tie resource absorption or cost to a unique Medicare resident during a specific benefit period. That is not a data gap at the margin. It rots the equation from the center.

Before PPS, the system did not answer the same question perfectly, but it was organized around the right problem. Facilities took care to accurately report expenses, charges, utilization statistics, and other relationships because those numbers determined payment. Cost Reports were audited because they drove reimbursement. They told a clearer story because providers, auditors, and Fiscal Intermediaries had to settle relationships across populations, services, costs, and utilization variables.

Certified Distinct Parts were the accounting architecture that kept Medicare expenses from dilution into the building average. A nursing facility was not one economic unit simply because everyone lived under the same roof. Medicare admissions directly from the hospital had different needs than long-term custodial residents down the hall. More nursing time, more supplies, more laundry, more therapy coordination, more administration - not because the bed was certified, but because the resident in that bed required more.

Where Costs Belong

That distinction mattered because the Cost Report dictated where expenses belonged. If a facility operated 120 beds but only 30 were Medicare-certified, the Certified Distinct Part gave the system a place to focus the costs generated by the Medicare population. Without that structure, the math collapsed into Patient Days. A Medicare resident day and a long-term care resident day became equal units in an average, even when everyone in the building knew they were not consuming equal resources. The Certified Distinct Part did not create the cost. It prevented the cost from being averaged away.

This was not theoretical. Medicare recognized Direct Care and other costs specific to the CDP. Post-surgical Medicare residents needed their sheets changed more often than stable long-term residents, so Pounds of Laundry became an allocation statistic. Nursing Administration and Social Services were allocated by Hours Worked because short-stay Medicare admissions consumed far more staff time. The statistics were not perfect. The point was that the system forced someone to ask whether the allocation reflected use. If Patient Days were wrong, you needed a better statistic. If the statistic was unsupported, an auditor could challenge it. That is Reimbursement.

What About Medicaid?

The same principle distorts Medicaid Rate Construction today. When states average unlike providers, populations, units, payer mixes, occupancy levels, and cost structures into broad rate classes, they create the appearance of fairness while preserving the distortions the old reimbursement machinery was designed to expose. A statewide average can look objective while burying the very differences that determine whether a provider is properly paid. It’s why I cringe when a profitable facility claims to “lose” a fixed amount per Medicaid day, when MedPAC treats Medicare as though it exists in a vacuum, and especially when CMS pulls stunts like this. It reveals institutional erosion on one side and a lost generation of Cost Reporting expertise on the other.

Ancillary charges were similarly disciplined for services that could be traced directly by resident and payer. Therapy, pharmacy, laboratory, medical supplies, and other resident-specific, medically necessary expenses were connected to charges, Cost Centers, and utilization patterns. It’s tricky, but meaningful Cost-to-Charge Ratios informed resource use. That logic still applies; the discipline has faded. It is tedious work and no longer impacts provider-specific rates. Another casualty of a rusted factory haunted by legacy reimbursement theory.

Cost Reports are still filed, but they are not consistently prepared, rarely even reviewed, and contain errors that distort the SNF Economy. As a pillar of Rate Construction, they cannot bear the weight of the RFI.

That is why Cost Report compendiums are so problematic. They collect fields, standardize labels, and create the appearance of comparability. But if the underlying reports are inconsistent, unaudited, and shaped by different state rules, preparer habits, allocation methods, and charge practices, the compendiums simply organize noise. They make bad data easier to quote. They turn self-reported artifacts into benchmarks. They give analysts the confidence of scale without the discipline of Reimbursement.

More Flaws

Nursing exposes the defect most clearly. Nursing is Routine, not Ancillary. It is the central service provided by a Skilled Nursing Facility, but resident-specific nursing cost is not directly identified through ordinary administrative data the way an Ancillary charge may be. The MDS can place a resident into a nursing category. A claim can show the covered stay. A Cost Report can show Routine cost at the facility level. But the system does not produce resident-level nursing cost.

That is the flaw in “case-mix creep.” The phrase assumes CMS can compare reported acuity to observed resource utilization in a technically coherent, comparable, and consistent manner. It cannot. Reported acuity comes from resident classification. Resource utilization, especially Routine nursing resource utilization, is not measured at the same level. CMS is comparing a resident-level classification output to a cost system that no longer preserves the resident-level relationship needed to validate it.

Budget neutrality has the same flaw. CMS intended PDPM to be budget neutral relative to RUG-IV, but RUG-IV was dominated by therapy utilization and had limited sensitivity to nursing-driven clinical complexity. Budget neutrality against RUG-IV is not resource neutrality. It is neutrality against the prior pricing model. If RUG-IV under-recognized certain resident characteristics, PDPM may look expensive because it recognizes what the prior model muted. That does not prove PDPM is right. It also does not prove creep. It proves the baseline is not a resource-cost standard.

Here Comes MDS

The MDS is part of the illusion. It matters, but it’s not the system. It can show that a resident grouped into a higher nursing category, generated NTA points, qualified for a clinical category, or carried reimbursement-sensitive conditions. It cannot prove whether the facility incurred the associated cost, whether that cost was Routine or Ancillary, whether charges were meaningful, or whether the Cost Report allocates cost properly. The MDS is a classification instrument, not a cost accounting system. In a budget-neutral environment, every additional dollar generated through case-mix management is not newly generated money - it’s taken from the rate pool and redistributed.

CMS is therefore asking a reimbursement question from inside a price system. The agency wants to know whether PDPM payment reflects resource utilization after weakening the machinery that once tied payment to resource utilization. The old system was imperfect, but it forced the right questions. Which population generated the cost? Was the service Routine or Ancillary? Could the cost be traced to charges, or did it require allocation? Did the allocation method reflect actual use? Could an auditor follow the logic?

Those questions did not disappear. The system simply stopped organizing payment around them.

Data is not measurement. Classification is not cost. Budget neutrality is not resource neutrality. A compendium is not an audit. Price is not reimbursement.

Maybe if the industry sends 2027 copies of The Matrix to DC, they’ll stop trying to bend the spoon. It is impossible. Why? Because there is no spoon, and providers already know that it is not the spoon that CMS bends. It is the facilities themselves.

Marc Zimmet is CEO of Zimmet Healthcare Services Group.