Same Protein, Two Prices

Snapshot

The problem: a price made of payments

Since 2018, Medicare has paid most clinical laboratory tests at the volume-weighted median of the rates private payers paid, under the Protecting Access to Medicare Act (PAMA). After a run of congressional delays, the Consolidated Appropriations Act, 2026 fixed the current cycle. Laboratories reported between May 1 and July 31, 2026 the payments they had received between January 1 and June 30, 2025. After exclusions, data from 6,304 laboratories produced medians for 1,528 of the 1,947 codes CMS collects. Those medians become the 2027 rates, with any cut limited to 15% a year through 2029. CMS estimates that the reset will save about $1 billion a year and puts the average change at about 16% lower. Final rates are due in November.

Three definitions decide what enters a median, and each turns on what happens after adjudication. Applicable information is each private payor rate "for which final payment has been made during the data collection period." The private payor rate "does not include information about denied payments." And the rate is the payer's full allowed amount, including the patient's share: CMS's guidance tells a laboratory paid $130 by the payer, with $20 still owed by the patient, to report $150. Private payors include Medicare Advantage plans and Medicaid managed-care organizations. The median, in short, is built only from claims that were paid, at the amounts finally allowed, inside a fixed window.

For a high-volume chemistry code, more than a thousand reporting entities contribute and no single book of claims moves the result; the median for hemoglobin A1c draws on 1,633 TINs. A new code is different. Its median is set by a few dozen reporting entities, 29 to 101 TINs for the five codes discussed here, in the first months of the code's life, while payers are still deciding what to pay. Five Category I codes for neurodegenerative disease, created for 2025, are a case in point: they had no national Medicare price during the six months in which their private-payer data was collected.

Recovr works after adjudication, on the remittances and claims a billing system already produces. This study asks what the 2027 neuro prices reveal about how private payments become public prices, and what laboratories can do with that.

The evidence

Five sources stand behind the figures that follow.

The first is CMS's preliminary 2027 rate file, published September 21, 2026, which lists for each of 1,947 codes the 2026 rate, the private-payer weighted median, the number of reporting taxpayer identification numbers (TINs) and the 2027 rate after the 15% cap.

The second is the raw private-payer file CMS released with it: 6.55 million records, each a payment rate and the number of tests paid at that rate, covering 399 million tests paid from January to June 2025. The file carries no payer or laboratory identifiers, and it suppresses codes reported by fewer than ten TINs. For the five neuro Category I codes we recomputed CMS's weighted medians from it and matched them to the cent.

The third is CMS's payment determinations for the codes new in 2025: the preliminary and final determinations of September and November 2024, and the gapfill amounts that Medicare's contractors proposed in April 2025 and that were finalized in September and November 2025.

The fourth is the rulebook: 42 CFR 414.502, CMS's PAMA reporting FAQs (updated July 14, 2026), the CPT instructions for proprietary laboratory analyses (PLA) codes, and the coverage policies of the payers named below, read in October 2026.

The fifth is Recovr's own records: denial outcomes across our customers, laboratories and physician groups, which record for each denial code the share of denied lines later paid, and reconsideration records at one national payer. Our customers' books hold too few claims for the neuro codes themselves to analyze, so Recovr data appear only where noted.

Rising against the tide

CMS counted 186 codes whose median exceeds the 2026 rate, 169 unchanged and 1,171 lower, and it flags 775 codes that will take the full 15% cut in 2027. The generic neurodegeneration codes sit among the 186. Only reductions are capped, so their 2027 rates equal their medians.

2027 rates are CMS's preliminary figures. PLA: proprietary laboratory analysis code. CSF: cerebrospinal fluid. † Capped at a 15% cut; the medians are $432.79 for 0035U and $11.86 for 83520.

Every code in the table that rose by more than 5% was new in 2025 and gapfilled, so it had no national price that year. The other proprietary codes behave differently. The cerebrospinal-fluid amyloid ratio, 0358U, is essentially unchanged, and four Alzheimer's proprietary tests, 0206U, 0207U, 0412U and 0503U, have medians at 98.0% of their 2026 rates; Finding 5 suggests why that number recurs. Several others, including the cerebrospinal-fluid assays 0445U and 0459U, drew no private-payer data at all; CMS took them up at a public meeting on September 15 and 16 and will price them by crosswalk or gapfill. Among the neuro tests, the prion test 0035U falls furthest: its median is 20% below its current rate, so the cap holds its 2027 rate at $459.84. The generic immunoassay code 83520, which CMS once proposed as the price for the new codes, falls 31% on its median and 15% in 2027.

A price set in a vacuum

The new codes reached the fee schedule by an unusual route. In September 2024 CMS proposed to crosswalk the amyloid, tau and neurofilament codes to 83520, a generic immunoassay paid $17.27. Its advisory panel had favored half the rate of 0358U, about $130. In November 2024 CMS changed course. It wrote that, "given widespread public input as to the broader public health implication of this test," it "understands why the crosswalk to 83520 is also not appropriate," and it assigned the codes to gapfill, leaving Medicare's contractors to price them for 2025.

So during the six months whose payments now set the 2027 rates, Medicare had no national price for these tests. Its contractors proposed local amounts in April 2025, midway through the window: $93.17 for the amyloid codes, $88.37 for the tau codes and $70.48 for neurofilament light chain in 43 of 57 payment localities, and $130.25 in the other 14. The final gapfill amounts, published in September and November 2025, set the 2026 rates at $128.92 and $116.23. Private payers, over the same six months, paid medians of $175.00 for amyloid and $210.00 for phosphorylated tau and neurofilament light chain, more than any contractor proposed. The window also straddled the first FDA clearance of a blood test to aid in diagnosing Alzheimer's disease, Fujirebio's Lumipulse G pTau217/β-Amyloid 1‑42 Plasma Ratio, on May 16, 2025.

Same protein, two prices

One code took the other road. 0479U, a proprietary pTau217 assay, was proposed for the same $17.27 crosswalk in September 2024, and in November CMS kept it there: "Finalize as proposed; CMS continues to agree this is the most appropriate crosswalk." Its private-payer median, from fewer than ten TINs, is $18.13, so that is its 2027 rate. The generic phosphorylated-tau code, whose descriptor names pTau217 as an example, is set at $210.00. The CPT rule for proprietary codes leaves little room: when a PLA code is available for a test, it "takes precedence," and the test should not be reported with another code. Which code a test carries, not what it measures, decides whether Medicare pays $18 or $210.

The crosswalk also leaked into the generic codes. In the first half of 2025, private payers paid 345 phosphorylated-tau tests, 6.0% of the volume, at exactly $17.27, the rate for 83520 and the figure CMS had proposed. They paid 10 total-tau tests at $17.27 and 81 neurofilament tests at $13.82, which is 80% of $17.27. In all, 9.1% of phosphorylated-tau tests and 12.0% of neurofilament tests were paid $20 or less. The public file cannot say which payers these were, but the pattern is consistent with fee schedules that priced the new codes from the September 2024 proposal, or from 83520, and were not revisited once CMS chose gapfill.

One code, a sixteen-fold spread

Quinn observed that the distribution of reported prices is diffuse rather than bell-shaped, and that a median ignores the outliers at either end. The raw file shows how diffuse. For phosphorylated tau, a tenth of tests were paid $25.91 or less and a sixth $420 or more; the 90th percentile is 16 times the 10th.

Private-payer payment per test, January to June 2025, at percentiles of tests paid

The prices are not random. Six of the ten most common payments for phosphorylated tau, $420, $315, $210, $176.40, $132.30 and $88.20, are exactly 100%, 75%, 50%, 42%, 31.5% and 21% of $420, and together they account for 42% of the tests. For beta-amyloid 1‑40, the eight most common prices that are round percentages of $185.90, from 40% to 200%, account for half the tests; the most common, $260.26, is 140% of $185.90 and covers 26% of the volume on its own. That is the pattern one would expect if contracts paid a percentage of some reference price rather than a negotiated rate for the test, in a year when the reference Medicare normally supplies did not exist. The file does not say whose reference prices $420 and $185.90 were.

Only paid claims vote

The rules turn post-adjudication work into pricing. A denied claim contributes nothing to the median: the regulation excludes "information about denied payments," and CMS's FAQs add that a claim paid before the window but denied during it is not reported, because its final payment is $0. A claim paid after the window closes does not count either: "Only final paid claims received during the data collection period are applicable information." Payments that cannot be tied to a specific HCPCS code, such as bundled or claim-level payments, are not reported, and neither is anything paid under a miscellaneous code. What is reported is the allowed amount on each paid test.

Three consequences follow for an applicable laboratory. An overturned denial that is paid inside the window enters the median at its allowed amount. An underpayment corrected inside the window is reported at the corrected amount; one never disputed is reported at the amount the payer allowed. And a laboratory's weight in the median is the number of tests it was paid for: the phosphorylated-tau median rests on 5,765 tests reported by 101 TINs, the beta-amyloid 1‑40 median on 4,840 tests from 57. The same logic covers claims that never produced a remittance at all, the subject of our study The Denial That Never Arrives: a claim with no adjudication has no vote.

Speed decides which side of the line a recovery lands on. In Recovr's reconsideration records for a physician book at one national payer, the payer decided reconsiderations a median of five days after filing, and confirmed payments followed a median of 16 days after filing. At that pace, a denial worked in the month its remittance arrives can be paid inside the window. One that waits in a queue until summer cannot.

The definition of a private payor adds a twist. Medicare Advantage plans count, and a plan that pays the Medicare rate net of the 2% sequestration reduction pays 98% of the Medicare rate. Fifty-five of the 1,526 codes have medians at 98.0% of their 2026 rate, to one decimal place, among them the Alzheimer's tests 0206U, 0207U, 0412U and 0503U. For 0503U, a blood test that combines amyloid and tau measurements, 429 of the 1,239 tests, 35%, were reported at exactly $735.00, which is 98% of $750. The pattern is consistent with laboratories reporting Medicare Advantage payments net of sequestration, although CMS's FAQs describe the reportable rate as 100% of the payer's allowed amount. If such payments keep setting a median, the test loses about 2% with each reporting cycle.

Coverage decides which claims are eligible

For these tests, coverage policy decides which claims can reach the median at all. Cigna's coverage policy 0580, effective July 15, 2025, calls beta-amyloid and phosphorylated-tau testing in plasma or cerebrospinal fluid experimental, investigational or unproven, and lists 82233, 82234, 84393, 84394, 0479U and 0503U among its codes. BlueCross BlueShield of Tennessee, in a policy revised September 4, 2025, and Harvard Pilgrim draw a finer line. Cerebrospinal-fluid amyloid, and the ratios of total tau and p-tau181 to amyloid beta 1‑42, meet their criteria, as does the plasma ratio of p-tau217 to amyloid beta 1‑42 "using an FDA-approved test." Plasma or serum biomarkers measured as individual markers, including p-tau217 and amyloid beta 1‑40, do not, and neither do multianalyte or algorithmic tests. Under those policies a pTau217 result can be covered as part of a qualifying ratio and excluded as a stand-alone plasma marker.

Coverage denials are not all equally final, and the denial code usually says which kind a laboratory has. Across our customers and all procedure codes, lines denied as non-covered (CARC 96) were later paid 50% of the time, and those denied for medical necessity (CARC 50) 37%. Lines denied as not a benefit of the patient's plan (CARC 204) were paid 7% of the time. If a payer's system has not loaded a new code, the denial is often administrative, and administrative denials recover best: lines denied for missing or invalid information (CARC 16) were later paid 82% of the time, and 95% when the remark was M51, a missing or invalid procedure code. Modifier denials (CARC 4) were later paid 85% of the time.

The new price has to reach the remittance

On January 1, Medicare's payment for most laboratory codes will fall. CMS flags 775 codes that take the full 15% cut in 2027, and for all but eight of them the cap means a further cut in 2028. The neuro codes move the other way, which creates an underpayment risk of its own. Where a contract pays a percentage of the current Medicare fee schedule, the payer should apply the 2027 rate to 2027 dates of service. A payer that keeps paying the 2026 rate underpays every test.

* If a payer keeps paying the 2026 rate at 100% of Medicare. 2027 rates are preliminary.

A remittance that still shows $128.92 for 84393 in February 2027 is not a rounding difference; it is 39% short. For codes whose rates fall, the check runs the other way: a contract with fixed rates should not shrink because Medicare's did.

Results scoreboard

What we would tell a peer

Treat the next window as a deadline. Under CMS's general rule, payments finalized between January 1 and June 30, 2028 will set Medicare's rates from 2030, so work denials and underpayments as they arrive in that half-year: a claim denied in May and paid in August is cash, not a data point. From January 1, 2027, reconcile every neuro-code payment against the 2027 rate and the contract percentage, and dispute those that still carry 2026 amounts. Search remittances for 84393, 84394 and 83884 lines paid at $17.27, $13.82 or similar amounts, and test them against the contract while the payer's dispute window is still open. Bill each test under the code that describes it, remembering that a PLA code takes precedence when one exists. Read each coverage denial against the policy behind it: at two of the three plans we reviewed, cerebrospinal-fluid ratios and an FDA-cleared plasma pTau217/amyloid ratio meet criteria where stand-alone plasma markers do not. Keep payment data at the HCPCS level and at the allowed amount, the form in which it is reported, and by payer, so that each contract can be checked. And read the 2% line on Medicare Advantage remittances twice: against the contract, and against CMS's description of the reportable rate as 100% of the allowed amount.