{
  "study": {
    "slug": "medicare-part-d-spending-concentration-2023",
    "title": "The few drugs that drive most of Medicare's Part D bill, 2023",
    "standfirst": "Medicare Part D paid $275.9 billion for 3,598 drugs in 2023. The 100 costliest — just 2.8% of the list — account for $176.2 billion, 63.8% of the bill. Ten drugs alone are a quarter of it, and 61.5% of priced drugs cost more per dose than a year earlier.",
    "desk": "financial-distress",
    "article_type": "Original Research",
    "published": "2026-06-16",
    "issue": 77,
    "doi": "10.5072/fonteum/medicare-part-d-spending-concentration-2023",
    "url": "https://fonteum.com/research/medicare-part-d-spending-concentration-2023",
    "methodology_version": "cms-part-d-spending/v1"
  },
  "data_as_of": "2026-06-16",
  "datasets": [
    {
      "slug": "cms-provider-data-catalog",
      "name": "CMS Provider Data Catalog",
      "publisher": "CMS — Provider Data Catalog",
      "upstream_url": null
    }
  ],
  "key_findings": [
    {
      "number": "63.8%",
      "finding": "of Medicare's $275.9B Part D drug bill in 2023 — $176.2B — went to the 100 costliest drugs, which are just 2.8% of the 3,598 drugs on the file. The 10 costliest alone are 25.8%, and the top 50 cross half the bill",
      "dataset": "cms-provider-data-catalog"
    },
    {
      "number": "$18.27B",
      "finding": "Medicare Part D spending on Eliquis (apixaban), the single costliest drug, 6.6% of all Part D drug dollars on its own. Seven of the ten costliest are brand-name diabetes, anticoagulant, and respiratory treatments reaching millions of beneficiaries",
      "dataset": "cms-provider-data-catalog"
    },
    {
      "number": "$1,250,918",
      "finding": "average Part D spend per beneficiary for Strensiq, a rare-disease drug reaching 207 people — the other road to a large bill, where price not volume drives the total. Revlimid runs $104,412 per beneficiary and Humira $71,898",
      "dataset": "cms-provider-data-catalog"
    },
    {
      "number": "61.5%",
      "finding": "of the 3,390 drugs with a year-over-year price on file cost more per dose in 2023 than in 2022; over five years 65.0% of priced drugs had a rising price. Among material drugs the steepest 2023 climbs were Zejula (+43.1%) and Erleada (+32.1%)",
      "dataset": "cms-provider-data-catalog"
    },
    {
      "number": "$275.9B",
      "finding": "total Medicare Part D drug spending across 3,598 drugs in the 2023 file, source released 2025-05-19. Brand-name drugs hold 90.6% of the dollars. Every figure is a count or ratio over published drug-level records — no prescriber or beneficiary is named",
      "dataset": "cms-provider-data-catalog"
    }
  ],
  "faqs": [
    {
      "q": "Which drug costs Medicare Part D the most?",
      "a": "Eliquis (apixaban), a blood thinner. Medicare Part D spent $18.27 billion on it in 2023 — 6.6% of the program's entire $275.9 billion drug bill, on a single drug, reaching 3.93 million beneficiaries. Ozempic ($9.19B) and Jardiance ($8.84B) are next."
    },
    {
      "q": "How concentrated is Medicare Part D drug spending?",
      "a": "Heavily. The 100 costliest drugs are only 2.8% of the 3,598 drugs on CMS's 2023 spending file, but they account for $176.2 billion — 63.8% of the total bill. The 10 costliest alone are 25.8% of spending, and the top 50 drugs cross half of it."
    },
    {
      "q": "Why do so few drugs account for so much spending?",
      "a": "Two reasons sit underneath the concentration. Most of the costliest drugs are brand-name treatments for common chronic conditions — diabetes, blood clots, COPD — taken by millions of beneficiaries; brand-name drugs hold 90.6% of all Part D dollars while generics carry the volume. A second, smaller group of specialty and rare-disease drugs reaches only hundreds of patients but costs hundreds of thousands of dollars each."
    },
    {
      "q": "Are Medicare Part D drug prices rising?",
      "a": "For most drugs on the file, yes. Of the 3,390 drugs with a year-over-year price recorded, 61.5% cost more per dose in 2023 than in 2022. Over a five-year horizon, 65.0% of priced drugs had a rising average price per dosage unit. The figures describe price per dose, separate from how much total volume changed."
    },
    {
      "q": "What is the most expensive drug per patient in Medicare Part D?",
      "a": "Among drugs with at least $100 million in total spending, Strensiq (asfotase alfa), a treatment for a rare metabolic bone disease, had the highest per-beneficiary cost: an average of $1,250,918 across just 207 beneficiaries in 2023. Soliris ($483,709) and Takhzyro ($415,315) follow."
    },
    {
      "q": "Does high spending on a drug mean it is overpriced or overused?",
      "a": "No. This study counts where Medicare's drug dollars go; it does not assess whether any drug was worth its price, appropriately prescribed, or fairly priced. A large bill can reflect a common condition, a genuinely costly therapy, or a high unit price — the spending figures alone do not distinguish among them, and this study draws no such inference."
    },
    {
      "q": "Can I reproduce these figures?",
      "a": "Yes. Every number is a direct count or ratio over the public cms_part_d_spending_by_drug table — CMS's Medicare Part D Spending by Drug file for data year 2023, source released 2025-05-19 — restricted to the per-drug 'Overall' rows so manufacturer breakouts are not double-counted. The exact SQL for the concentration curve, the top-drug ranking, the per-beneficiary extremes, and the price-change split is published in the reproducibility block below."
    }
  ],
  "citation": {
    "apa": "Fonteum Research. (2026, June 16). The few drugs that drive most of Medicare's Part D bill, 2023. Fonteum Research, Issue 77. https://doi.org/10.5072/fonteum/medicare-part-d-spending-concentration-2023",
    "url": "https://fonteum.com/research/medicare-part-d-spending-concentration-2023"
  },
  "reproducible_sql": "-- How concentrated is Medicare Part D drug spending — and are the prices of the\n-- drugs that dominate the bill still rising? Fully reproducible query.\n--\n-- Question: across the drugs Medicare Part D pays for, how much of the program's\n-- total drug bill is carried by a handful of products, who are they, and what is\n-- happening to their per-unit prices? The lead figure: the 100 costliest drugs —\n-- 2.8% of the 3,598 drugs on CMS's published spending file — account for 63.8%\n-- of the $275.9B Part D drug bill ($176.2B). These are spending facts over a\n-- published drug-level file, NOT a quality, fraud, or pricing-fairness judgment.\n--\n-- Source:\n--   public.cms_part_d_spending_by_drug — CMS \"Medicare Part D Spending by Drug\"\n--     public-use file, published annually via data.cms.gov (Summary Statistics\n--     on Use and Payments / Medicare-Medicaid Spending by Drug). Data year 2023;\n--     source release 2025-05-19. Public, read-only. License: US-Government-Works\n--     (17 U.S.C. Sec. 105). methodology_version = 'cms-part-d-spending/v1'.\n--\n-- Universe + double-count guard: the file ships TWO kinds of rows. An \"Overall\"\n--   row (manufacturer_name = 'Overall') gives the all-manufacturer total for each\n--   drug; additional rows break that total out by manufacturer. Summing every row\n--   double-counts. This study reads the 3,598 \"Overall\" rows ONLY — one per drug —\n--   which is the published Part D total. The other 10,711 of the 14,309 rows are\n--   manufacturer breakouts and are excluded from every aggregate below.\n--\n-- Grain: each row is a drug (brand x generic), not a provider. No NPI, no\n--   beneficiary, no prescriber is present or named anywhere in this study.\n\n-- ============================================================================\n-- (1) Universe reconciliation — the published file at a glance, Overall rows only.\n-- ============================================================================\nSELECT\n  count(*)                                                           AS drugs,\n  count(*) FILTER (WHERE manufacturer_name <> 'Overall')             AS mfr_breakout_rows_excluded,\n  (SELECT count(*) FROM public.cms_part_d_spending_by_drug)          AS rows_in_file,\n  sum(total_spending)::numeric::bigint                               AS total_spend,\n  max(data_year)                                                     AS data_year,\n  max(source_release_date)                                           AS source_release\nFROM public.cms_part_d_spending_by_drug\nWHERE manufacturer_name = 'Overall';\n--  drugs 3,598 · mfr_breakout_rows_excluded 10,711 · rows_in_file 14,309\n--  total_spend 275,924,520,551 ($275.9B) · data_year 2023 · source_release 2025-05-19\n\n-- ============================================================================\n-- (2) HEADLINE: spend concentration. Rank drugs by total spend and read the\n--     cumulative share. 10 drugs = a quarter of the bill; 50 = half; 100 = 63.8%.\n-- ============================================================================\nWITH ranked AS (\n  SELECT total_spending,\n    row_number() OVER (ORDER BY total_spending DESC)                 AS rn,\n    sum(total_spending) OVER ()                                      AS grand\n  FROM public.cms_part_d_spending_by_drug\n  WHERE manufacturer_name = 'Overall'\n)\nSELECT\n  round(100.0 * sum(total_spending) FILTER (WHERE rn <= 10)  / max(grand), 1) AS top10_pct,\n  round(100.0 * sum(total_spending) FILTER (WHERE rn <= 25)  / max(grand), 1) AS top25_pct,\n  round(100.0 * sum(total_spending) FILTER (WHERE rn <= 50)  / max(grand), 1) AS top50_pct,\n  round(100.0 * sum(total_spending) FILTER (WHERE rn <= 100) / max(grand), 1) AS top100_pct,\n  round(100.0 * sum(total_spending) FILTER (WHERE rn <= 250) / max(grand), 1) AS top250_pct,\n  (sum(total_spending) FILTER (WHERE rn <= 100))::numeric::bigint            AS top100_spend\nFROM ranked;\n--  top10 25.8% · top25 38.9% · top50 50.7% · top100 63.8% · top250 80.5%\n--  top100_spend 176,157,336,695 ($176.2B). 100 of 3,598 drugs = 2.8% of the list.\n\n-- ============================================================================\n-- (3) The top of the list — the 10 costliest drugs and their share of the bill.\n--     Seven are brand-name diabetes / anticoagulant / respiratory treatments;\n--     two (Humira, Revlimid) are specialty drugs reaching far fewer beneficiaries.\n-- ============================================================================\nWITH o AS (\n  SELECT *, sum(total_spending) OVER () AS grand\n  FROM public.cms_part_d_spending_by_drug WHERE manufacturer_name = 'Overall'\n)\nSELECT\n  brand_name, generic_name,\n  total_spending::numeric::bigint                                    AS spend,\n  round(100.0 * total_spending / grand, 1)                          AS pct_of_total,\n  total_beneficiaries                                               AS beneficiaries,\n  round(avg_spend_per_beneficiary)::bigint                          AS spend_per_bene\nFROM o\nORDER BY total_spending DESC\nLIMIT 10;\n--  Eliquis    Apixaban       $18.27B 6.6% 3,927,848 benes $4,652\n--  Ozempic    Semaglutide    $9.19B  3.3% 1,464,468 benes $6,278\n--  Jardiance  Empagliflozin  $8.84B  3.2% 1,882,768 benes $4,695\n--  Trulicity  Dulaglutide    $7.36B  2.7%   938,731 benes $7,844\n--  Xarelto    Rivaroxaban    $6.31B  2.3% 1,324,165 benes $4,765\n--  Trelegy Ellipta           $4.46B  1.6% 1,050,583 benes $4,241\n--  Humira(Cf) Pen Adalimumab $4.42B  1.6%    61,474 benes $71,898  <- few patients, huge bill\n--  Farxiga    Dapagliflozin  $4.34B  1.6%   993,909 benes $4,369\n--  Januvia    Sitagliptin    $4.09B  1.5%   843,391 benes $4,850\n--  Revlimid   Lenalidomide   $3.86B  1.4%    36,967 benes $104,412 <- specialty cancer drug\n\n-- ============================================================================\n-- (4) The second road to a big bill — price per beneficiary, not volume. Among\n--     drugs with >= $100M total spend, the highest per-beneficiary costs are\n--     rare-disease therapies reaching only a few hundred patients each.\n-- ============================================================================\nSELECT\n  brand_name, generic_name,\n  total_spending::numeric::bigint                                    AS spend,\n  total_beneficiaries                                               AS beneficiaries,\n  round(avg_spend_per_beneficiary)::bigint                          AS spend_per_bene\nFROM public.cms_part_d_spending_by_drug\nWHERE manufacturer_name = 'Overall' AND total_spending >= 100000000\nORDER BY avg_spend_per_beneficiary DESC\nLIMIT 5;\n--  Strensiq  Asfotase Alfa   $258.9M    207 benes $1,250,918\n--  Soliris   Eculizumab      $223.5M    462 benes   $483,709\n--  Takhzyro  Lanadelumab     $194.4M    468 benes   $415,315\n--  Firdapse  Amifampridine   $146.3M    356 benes   $411,048\n--  Acthar    Corticotropin   $377.8M    942 benes   $401,104\n\n-- ============================================================================\n-- (5) Are prices rising? The file carries each drug's 1-year and 5-year change\n--     in average spend per dosage unit. A majority of priced drugs rose on both\n--     horizons. (Percentages use the non-null base; 208 drugs lack a 1-year\n--     change and 175 lack a 5-year CAGR.)\n-- ============================================================================\nSELECT\n  count(*) FILTER (WHERE change_avg_spend_per_dosage_unit IS NOT NULL)         AS yoy_base,\n  count(*) FILTER (WHERE change_avg_spend_per_dosage_unit > 0)                 AS rose_yoy,\n  round(100.0 * count(*) FILTER (WHERE change_avg_spend_per_dosage_unit > 0)\n        / count(*) FILTER (WHERE change_avg_spend_per_dosage_unit IS NOT NULL), 1) AS rose_yoy_pct,\n  count(*) FILTER (WHERE cagr_avg_spend_per_dosage_unit IS NOT NULL)           AS cagr_base,\n  count(*) FILTER (WHERE cagr_avg_spend_per_dosage_unit > 0)                   AS cagr_pos,\n  round(100.0 * count(*) FILTER (WHERE cagr_avg_spend_per_dosage_unit > 0)\n        / count(*) FILTER (WHERE cagr_avg_spend_per_dosage_unit IS NOT NULL), 1) AS cagr_pos_pct\nFROM public.cms_part_d_spending_by_drug\nWHERE manufacturer_name = 'Overall';\n--  yoy_base 3,390 · rose 2,084 · 61.5% rose 2022-2023\n--  cagr_base 3,423 · positive 2,225 · 65.0% positive 5-yr CAGR (2019-2023)\n\n-- ============================================================================\n-- (6) The steepest 1-year per-unit price climbs among material drugs (>= $50M\n--     spend), so the list is not dominated by tiny products with noisy ratios.\n-- ============================================================================\nSELECT\n  brand_name, generic_name,\n  total_spending::numeric::bigint                                    AS spend,\n  round(100.0 * change_avg_spend_per_dosage_unit, 1)                AS yoy_unit_price_pct,\n  change_period\nFROM public.cms_part_d_spending_by_drug\nWHERE manufacturer_name = 'Overall'\n  AND total_spending >= 50000000\n  AND change_avg_spend_per_dosage_unit IS NOT NULL\nORDER BY change_avg_spend_per_dosage_unit DESC\nLIMIT 5;\n--  Zejula   Niraparib   $161.5M  +43.1%  2022-2023\n--  Entyvio  Vedolizumab $100.0M  +36.5%  2022-2023\n--  Lumakras Sotorasib   $140.2M  +34.9%  2022-2023\n--  Erleada  Apalutamide $1.06B   +32.1%  2022-2023\n--  Metformin ER Gastric  $191.1M +27.3%  2022-2023\n\n-- ============================================================================\n-- (7) Why so concentrated — brand vs generic. The partition is\n--     lower(brand_name) = lower(generic_name): a generic is published under its\n--     own ingredient name. Generics are 601 of 3,598 drugs but only 9.4% of the\n--     dollars; brand-name drugs carry 90.6% of the bill.\n-- ============================================================================\nSELECT\n  count(*) FILTER (WHERE lower(brand_name) = lower(generic_name))             AS generic_drugs,\n  (sum(total_spending) FILTER (WHERE lower(brand_name) = lower(generic_name)))::numeric::bigint AS generic_spend,\n  round(100.0 * sum(total_spending) FILTER (WHERE lower(brand_name) = lower(generic_name))\n        / sum(total_spending), 1)                                            AS generic_spend_pct\nFROM public.cms_part_d_spending_by_drug\nWHERE manufacturer_name = 'Overall';\n--  generic_drugs 601 · generic_spend $25.83B · generic_spend_pct 9.4% (brand = 90.6%)",
  "license": "U.S. Government Works (federal sources; 17 U.S.C. §105)",
  "generated_by": "Fonteum — https://fonteum.com",
  "notes": "Aggregate, source-traced figures frozen to the snapshot above. Reproduce by running reproducible_sql against the cited federal dataset; no per-entity records are included."
}