Part B spent $60.2B. Its fastest-growing line is not a drug

Part B grew 20.3% in one year, more than four times Part D's rate. Split the increase by code and the top of the list is not Keytruda — it is a group billed by the square centimetre, up 21× since 2020, used by 77,104 people at about $101,000 each.

Published 2026-10-23Data FY2024 Part B annual file · 799 codesRead 7 min

The short version

Medicare Part B spent $60.2 billion on drugs in FY2024, up 20.3% from $50.0 billion; over the same year Part D grew 4.6%. 41% of the $10 billion increase came from codes billed by the square centimetre — skin substitutes. 68 codes went from $0.37 billion in 2020 to $7.76 billion in 2024, 21× in four years, for 77,104 patients: roughly $101,000 each. Strip that category out and the rest of Part B grew 12.9%.

The total first

Part B covers drugs given in a clinician's office — infusions, injections, oncology. Part D is the pharmacy benefit. Two separate ledgers.

YearSpendingCodes with spend
2023$50.0B639
2024$60.2B709

$10 billion in a year, 20.3%. Part D, five times larger at $285 billion, grew 4.6% over the same period. The problem with Part B is not its size, it is its slope.

Five of the top ten increases are not drugs

Ranked by absolute dollars added in 2024, five of the top ten are described "per square centimetre":

RankCode / name2024 spendAddedYoY
1Q4271 Complete ft, per sq cm$1.17B+$1.16B+18,769%
2Q4205 Membrane graft, per sq cm$1.44B+$1.10B+325%
3Q4191 Restorigin, per sq cm$0.73B+$0.73B+30,150%
5Q4164 Helicoll, per sq cm$0.64B+$0.59B+1,143%
4Vabysmo (J2777)$1.92B+$0.62B+47.9%
6Keytruda (J9271)$5.99B+$0.55B+10.2%

Those percentages cannot be read. Going from $60,000 to $11.7 million is +18,769%; going from zero to $73 million is +30,150%. A huge percentage only says the base was tiny.

So this article ranks by dollars throughout. Switching the sort did not change the finding — which is exactly why this category is worth writing about.

The category on its own

The 68 codes billed per square centimetre:

YearSpendingShare of Part B
2020$0.37B1.4%
2022$1.19B3.0%
2023$3.57B7.1%
2024$7.76B12.9%

Four years, 1.4% to 12.9%. In 2024 the category was 12.9% of Part B and 41.2% of its growth.

The patient count is nothing like proportional: 77,104 beneficiaries in 2024, up from 59,925. About $101,000 per beneficiary — the median spend per beneficiary across all Part B drugs is $2,691.

The code count is growing too: 52 codes with spend in 2023, 68 in 2024. Sixteen new codes; the category is still expanding.

What the actual drugs look like

Remove that category and 641 codes remain: $52.4 billion, up 12.9%. The top ten:

RankDrug2024 spendYoYPatients
1Keytruda$5.99B+10.2%75k
2Darzalex Faspro$2.53B+21.9%27k
3Prolia$2.43B+11.0%701k
4Eylea$2.37B−24.9%293k
5Opdivo$1.93B+1.2%27k
6Vabysmo$1.92B+47.9%173k
7Orencia$0.93B+3.1%33k
8Imfinzi$0.84B+14.1%15k
9Entyvio$0.77B+5.7%20k
10Gammagard Liquid$0.77B+3.4%26k

Top ten = $21.6 billion, 35.6% of the file. More concentrated than Part D's 29%, but growing slower than the code category above.

Inside the eye-injection group there is a swap: Vabysmo +47.9%, Eylea −24.9%. One drug added $620 million, the other lost $790 million. Whether that is patients switching or prescribing expanding is not in this file — it records no prescription flows, and we do not guess.

We re-ran the ranking

Our first pass sorted by year-on-year percentage. The leader was Q4271 at +18,769% — a great headline and a useless analysis, because it started from a tiny base.

So we recomputed by absolute dollars. Five of the top ten were still the same category. Two independent orderings pointing at one thing is what made this publishable.

The mirror-image lesson: any "fastest-growing" list published without its base and its threshold can turn this category into "one drug exploded".

What this file cannot answer

  • Why it grew. Coverage policy, coding practice, or genuine utilisation? The annual file carries spend, dosage units, claims and beneficiaries. It carries no payment rules.
  • Who sells it. There is no manufacturer field in the Part B file — only HCPCS code, description and generic name. Manufacturer analysis needs an NDC crosswalk we did not use here.
  • Whether it is abuse. $101,000 per patient is a fact, not a verdict. This data has no case-level records, and we make no such claim.

FAQ

Can Part B and Part D be added together?
Not simply. Both files count by drug or code, but they cover different settings — office-administered versus pharmacy-dispensed — and the same product can appear in both. Adding them double-counts.
Why treat "per square centimetre" as its own category?
Because the unit of measure differs. These are priced by area, one code per product, with no dose-strength concept. Mixed into a general ranking, they permanently occupy the top of any "fastest-growing" list.
Are all 68 codes skin substitutes?
Our filter is "description contains per square centimeter" or "HCPCS starts with Q4". That is a description-level rule, not an official taxonomy. Code definitions belong to the CMS HCPCS file; we do not redefine them here.
Data notes
  • Source: CMS Medicare Part B Spending by Drug, data year 2024, annual file with 799 HCPCS codes; the series uses the 2020–2024 columns in the same file.
  • Method: "skin substitute category" = description contains per square centimeter or HCPCS starts with Q4, 68 codes; growth checked under both absolute-dollar and year-on-year rankings.
  • Excluded: rows CMS flagged as outliers in 2023 or 2024 (90 rows), leaving 709 codes in the totals (639 with 2023 spend).
  • Limits: no manufacturer field; no payment or coverage policy; spend per beneficiary is total spend ÷ beneficiaries, not out-of-pocket cost.
LIVE · API / MCP
Slice it by code, by year, by patient yourself

Signup comes with a key and an MCP config: query spend, dosage units, claims and beneficiaries by HCPCS code, brand or year, as structured records rather than a file — your agent can keep asking follow-ups.

2,000 free reads included with signup; same method as this article, results recomputable.