Specialty drugs: a price tier pretending to be a category
Half of American drug spending now flows through a category with no legal definition. Ask what makes a drug "specialty" and you'll hear: it's expensive, it's a biologic, it needs an infusion, it treats a rare disease, it has restricted distribution. These sound like the same drugs. They are not — and the fuzziness is profitable. The party that gets to call a drug specialty gets to route it: onto a coinsurance tier, into its own specialty pharmacy, out of your local pharmacy's hands entirely. So we did the obvious thing nobody does: we tested every public definition against every Part D drug and measured how much they agree.
First, the money
Drugs averaging $950+ per prescription — the CMS specialty-tier cost test — took 62% of Medicare Part D's $289B in 2024, while being just 3.8% of prescriptions. That share was 45% in 2020. Whatever specialty is, it is where the money went.
ⓘ How this was built
- CMS Medicare Part D Spending by Drug (annual, latest year)
For each year, the orange bar is the share of total Part D gross spending from drugs whose average cost per prescription (total spending ÷ total fills) is at or above CMS's ~$950 specialty-tier threshold; the green bar is those drugs' share of all prescriptions. Gross means before rebates.
Pipeline and calculations built by Claude (Fable 5) from the public datasets above. Educational only — verify against primary sources before acting.
The definitions don't agree
Part D plans currently place 514 drugs on their specialty tiers (median tier 5+ across ~5,500 plans). That's the revealed-preference definition — what plans actually do. Here is how many of those drugs satisfy each of the other public definitions:
ⓘ How this was built
- CMS Part D quarterly Prescription Drug Plan Formulary files
- FDA NDC Directory (openFDA)
- CMS Medicare Part D Spending by Drug (annual, latest year)
- FDA Orange Book exclusivity codes
- FDA REMS@FDA active program roster
For each NDC we take the median formulary tier across all ~5,500 Part D plans covering it in the latest quarterly file. A molecule counts as 'on a specialty tier' when its most-covered NDC has a median tier of 5 or higher. Each bar is the share of those specialty-tier drugs that also meet one other criterion: Average cost per fill is the molecule's total Part D spending divided by total prescription fills (claims-weighted, so one exotic presentation can't misprice a common generic); where the shown product is a named brand, its own brand row is used instead. Biologic = the product is BLA-licensed; injectable = injectable/infused/subcutaneous form or route in the NDC Directory; orphan = orphan-drug exclusivity (ODE code) on the FDA application; REMS = the molecule appears in an active REMS program.
Pipeline and calculations built by Claude (Fable 5) from the public datasets above. Educational only — verify against primary sources before acting.
Read that again: nearly every drug on a specialty tier passes the price test, but most fail every clinical one. 44% of specialty-tier drugs (228 of them) are not biologics, not injectables, not orphan drugs, and carry no REMS — nothing distinguishes them except what they cost. The category has one consistent meaning in practice, and it is a number on an invoice. That matters because the specialty label carries real consequences: percentage coinsurance instead of flat copays, mandatory specialty-pharmacy routing, and — for the pharmacist — whether you're allowed to dispense the drug at all.
The edges of the tier
On specialty tiers while averaging under $950/fill
CMS only permits specialty-tier placement above the cost threshold. A drug here either straddles the line across its presentations, moved recently, or its average is dragged down by partial fills — but each one shifts percentage coinsurance onto patients for a drug whose average cost doesn't clear the bar. Averages $100–$950/fill (below that, the molecule average usually describes a cheap sibling of the tiered product, not the product itself); sorted by prescription volume.
| Drug | Avg / fill | Median tier | Rx / yr |
|---|---|---|---|
| Opipza | $235.43 | 5 | 4,111,568 |
| Mycophenolate mofetil | $122.43 | 5 | 518,400 |
| Daptomycin | $790.29 | 5 | 152,673 |
| Darunavir | $853.34 | 5 | 63,546 |
| Mercaptopurine | $107.08 | 5 | 43,076 |
| Hydrocodone Bitartrate | $557.50 | 5 | 39,788 |
| Granix | $852.95 | 5 | 5,220 |
| Eraxis | $912.45 | 5 | 529 |
ⓘ How this was built
- CMS Part D quarterly Prescription Drug Plan Formulary files
- FDA NDC Directory (openFDA)
- CMS Medicare Part D Spending by Drug (annual, latest year)
For each NDC we take the median formulary tier across all ~5,500 Part D plans covering it in the latest quarterly file. A molecule counts as 'on a specialty tier' when its most-covered NDC has a median tier of 5 or higher. Average cost per fill is the molecule's total Part D spending divided by total prescription fills (claims-weighted, so one exotic presentation can't misprice a common generic); where the shown product is a named brand, its own brand row is used instead. This table lists specialty-tier drugs whose average lands between $100 and $950.
Pipeline and calculations built by Claude (Fable 5) from the public datasets above. Educational only — verify against primary sources before acting.
Averaging over $950/fill without specialty-tier placement
The mirror image: drugs that pass the cost test that plans keep on ordinary tiers — often protected classes or brands with rebate deals that make ordinary-tier placement worthwhile. The tier, in other words, is a choice, not a property of the drug.
| Drug | Avg / fill | Median tier | Rx / yr |
|---|---|---|---|
| Mayzent | $11,072.57 | 3 | 6,850 |
| Omvoh | $10,413.69 | 2 | 511 |
| Fragmin | $4,316.47 | 4 | — |
| Vivjoa | $2,837.47 | 4 | 114 |
| Clemastine Fumarate | $2,756.27 | 2 | 7,234 |
| Trelstar | $2,463.95 | 4 | 6,223 |
| Zenpep | $2,305.26 | 4 | — |
| Annovera | $2,250.63 | 4 | — |
ⓘ How this was built
- CMS Part D quarterly Prescription Drug Plan Formulary files
- FDA NDC Directory (openFDA)
- CMS Medicare Part D Spending by Drug (annual, latest year)
For each NDC we take the median formulary tier across all ~5,500 Part D plans covering it in the latest quarterly file. A molecule counts as 'on a specialty tier' when its most-covered NDC has a median tier of 5 or higher. Average cost per fill is the molecule's total Part D spending divided by total prescription fills (claims-weighted, so one exotic presentation can't misprice a common generic); where the shown product is a named brand, its own brand row is used instead. This table lists drugs above the threshold whose most-covered NDC sits on an ordinary tier (1-4).
Pipeline and calculations built by Claude (Fable 5) from the public datasets above. Educational only — verify against primary sources before acting.
Every specialty-tier drug, scored
The full list — all 514 drugs Part D plans currently place on specialty tiers, each held against every criterion. Sort by any column; the ✗ columns are the point.
| Revcovi elapegademase-lvlr | 5 | $407,851.20 | ✓ | ✓ | ✓ | ✗ | ✗ | 98 |
| Cablivi caplacizumab | 5 | $217,354.87 | ✓ | ✓ | — | ✗ | ✗ | — |
| Livmarli maralixibat chloride | 5 | $123,450.53 | ✓ | ✗ | ✗ | ✓ | ✗ | 64 |
| Bylvay odevixibat | 5 | $123,160.93 | ✓ | ✗ | ✗ | ✓ | ✗ | 52 |
| Sohonos palovarotene | 5 | $106,046.68 | ✓ | ✗ | ✗ | ✓ | ✗ | 31 |
| Ruconest c1 esterase inhibitor recombinant | 6 | $104,916.91 | ✓ | ✓ | ✓ | ✗ | ✗ | — |
| Miplyffa arimoclomol citrate | 5 | $103,916.52 | ✓ | ✗ | ✗ | ✓ | ✗ | 18 |
| Cholbam cholic acid | 5 | $76,135.04 | ✓ | ✗ | ✗ | ✗ | ✗ | 78 |
| Mavenclad cladribine | 5 | $73,968.61 | ✓ | ✗ | ✗ | ✗ | ✗ | 2,420 |
| Actimmune interferon gamma-1b | 5 | $63,834.62 | ✓ | ✓ | ✓ | ✗ | ✗ | — |
| Firazyr icatibant acetate | 5 | $60,798.09 | ✓ | ✗ | ✓ | ✗ | ✗ | 3,316 |
| Juxtapid lomitapide mesylate | 5 | $59,762.07 | ✓ | ✗ | ✗ | ✗ | ✓ | 117 |
| Daybue trofinetide | 5 | $59,393.36 | ✓ | ✗ | ✗ | ✓ | ✗ | 667 |
| VYVGART Hytrulo efgartigimod alfa and hyaluronidase (human recombinant) | 5 | $52,674.07 | ✓ | ✓ | ✓ | ✗ | ✗ | — |
| Gattex teduglutide | 5 | $50,477.66 | ✓ | ✗ | — | ✓ | ✗ | 7,085 |
ⓘ How this was built
- CMS Part D quarterly Prescription Drug Plan Formulary files
- FDA NDC Directory (openFDA)
- CMS Medicare Part D Spending by Drug (annual, latest year)
- FDA Orange Book exclusivity codes
- FDA REMS@FDA active program roster
For each NDC we take the median formulary tier across all ~5,500 Part D plans covering it in the latest quarterly file. A molecule counts as 'on a specialty tier' when its most-covered NDC has a median tier of 5 or higher. Average cost per fill is the molecule's total Part D spending divided by total prescription fills (claims-weighted, so one exotic presentation can't misprice a common generic); where the shown product is a named brand, its own brand row is used instead. Criteria columns as defined above; one row per molecule, showing its most plan-covered product.
Pipeline and calculations built by Claude (Fable 5) from the public datasets above. Educational only — verify against primary sources before acting.
Drugs become "specialty" every quarter
Because the category is a tier, membership changes with every formulary file. The latest quarter moved no drugs across the boundary — but the boundary moves. When it does, a flat copay can become 25–33% coinsurance overnight, and a pharmacy can lose the prescription to a PBM-owned specialty pharmacy.
- Jan 20262 onto · 13 off
- Apr 20260 onto · 0 off
ⓘ How this was built
- CMS Part D quarterly Prescription Drug Plan Formulary files
- FDA NDC Directory (openFDA)
Each consecutive pair of quarterly formulary files is diffed per NDC; a crossing is a median tier moving from below 5 to 5+ (onto) or the reverse (off). The same diff posts tier-change events to the change feed and email digests.
Pipeline and calculations built by Claude (Fable 5) from the public datasets above. Educational only — verify against primary sources before acting.
Examples below are from the last quarter with movement (Jan 2026).
| Drug | Moved | Median tier |
|---|---|---|
| Fingolimod Hydrochloride | onto specialty | 4 → 5 |
| Furoscix | onto specialty | 1 → 5 |
| Ravicti | off specialty | 5 → 1 |
| Wegovy | off specialty | 5 → 4 |
| Zepbound | off specialty | 5 → 4 |
| Ustekinumab-Ttwe | off specialty | 5 → 3 |
These moves now post to the change feed each quarter — put a drug on your Drug List and the email tells you when it goes specialty.
What we can't show you — and what we estimate anyway
The most consequential specialty designation is the one with no public dataset at all: PBM limited-distribution lists — which drugs may only be dispensed by the PBM's own specialty pharmacy. The three largest PBMs each own one of the three largest specialty pharmacies, so classifying a drug as specialty routes its revenue in-house. Those lists are contractual and confidential, like the rebates on the big picture page. But confidential doesn't mean unestimable — it means the estimate must be labeled. (What has escaped the contracts — subpoenas, audits, lawsuits — is collected in our plain-language guide.) So, labeled:
The 514 drugs on Part D specialty tiers carried roughly $84B in gross Part D spending in the latest spend year. Industry market-share analyses put the PBM-owned share of specialty dispensing at roughly 65–75%. If that share holds across this set, then somewhere around $55–$63B a year of Part D specialty revenue is dispensed by pharmacies owned by the same companies that decide which drugs are "specialty". That circularity — classifier and beneficiary being the same firm — is our best explanation for why the category keeps growing faster than any clinical definition of it.
What we think is happening, stated as hypotheses the public data is consistent with, not as findings: drugs drift onto specialty tiers because percentage coinsurance shifts cost to patients and tier placement steers volume to affiliated pharmacies; expensive brands stay offspecialty tiers when rebate deals make preferred placement worth more than the routing; and the FTC's interim PBM report (2024) documented the mechanism we can't measure — affiliated pharmacies being reimbursed above independent ones for the same specialty drugs.
ⓘ How this was built
- CMS Part D Spending by Drug × our specialty-tier set (above)
- Drug Channels Institute specialty-pharmacy market-share analyses (PBM-owned share)
- FTC interim staff report on PBMs (July 2024) — qualitative only
ESTIMATE, not an audit: gross Part D spending summed over the drugs on specialty tiers this quarter, multiplied by an industry-estimated 65-75% PBM-owned dispensing share. The true routing depends on confidential contracts and varies by drug; spend is gross of rebates; the spend year and formulary quarter differ by months. Treat the range as order-of-magnitude.
Pipeline and calculations built by Claude (Fable 5) from the public datasets above. Educational only — verify against primary sources before acting.
What we can show — everything above — is the public shadow of that machine: a category defined by price, applied by tier, changing quarterly.
Method notes: "on a specialty tier" = median tier ≥ 5 across Part D plans in the latest quarterly formulary file, one representative NDC per molecule. The cost test uses average Part D spend per prescription (the closest public proxy for a 30-day-equivalent fill) against CMS's ~$950 threshold. Biologic = BLA-licensed; orphan = FDA orphan-drug exclusivity on the application; REMS = active program on the REMS@FDA roster. Full details on the methodology page. New to the topic? Start with the plain-language specialty drugs guide.