Half the money · five percent of people
A typical year is cheap. The premium is not priced off a typical year. These five spent $67,321. The cheaper half of the country spent $374.
The thin tan sliver on the right is the cheaper half of the country — 3% of the money.
One hundred people. The five orange squares spent half of everyone’s dollars.
In MEPS 2022 the top 5% of people accounted for 50% of spending. You crossed into that group at $30,206. This page ranks what that money actually bought — then what would move it.
100 people · Pay vs the 5% · What would move it · Evidence
Other rich countries have a 5% too. They pay less for it. These are the highest-spending large OECD systems — plus Japan and Singapore —2024 dollars per person (WHO 2023 for Singapore), then a modeled year for their expensive 5%.
Every peer on these tabs has a 5% that takes roughly half the money. The US does not have sicker 5% math — it has US prices. OECD 2024: $14,885 per person, 2.5× the OECD average. A 50% share on that stick is $148,850 for a top-5% year. MEPS 2023 says $72,918 because the household survey misses nursing homes and some bills. Click a peer: same concentration, cheaper unit prices.
Inpatient night, PPP, Wammes ~2013: US $3,180 vs Japan $432. Nights in the top 5%: US 6.6 vs Japan 97.7. Commercial US hospitals are often 2–3× Medicare — RAND on Evidence.
Same diseases. Follow a dollar through the United States versus the US commercial stack, then the twenty most expensive things in a top-5% year.
Yes, they make more — not enough to explain the whole gap. Specialists $316k PPP vs $98k (Sweden) to $202k (Australia). Generalists $218k vs $87k–$154k. Nurses $74k vs $42k in France. Debt service is ~$21–24k/year (Laugesen/Glied) and does not eat a $200k spread. Doctors are a slice. Hospitals and brands are the loaf.
No. Life expectancy 78.8 vs 80.7–83.9 in the other 10. Infant deaths 5.8 per 1,000 vs 3.6. Avoidable diabetes admissions 191 vs 125.6 per 100,000. The US is better on 30-day stroke mortality (4.2 vs 7.9) and has shorter specialist waits than Canada/UK/Sweden (6% wait 2+ months vs 39% in Canada; France 4%, Germany 3%). Acute rescue can be excellent. You are not buying a longer, healthier life with the extra $75k on a top-5% year.
Commercial stays are 2–3× Medicare (RAND). Peers pay a DRG or a budget.
Why US hospital prices are higher →Branded / specialty drugs$1.00US brands 4.22× other OECD (RAND/ASPE 2022). Unbranded generics are the one line the US often wins — not this pile.
Why US brand drugs are ~4× →The table, not the invoiceWho bargainsUS: hospital vs plan in secret. Medicare is a different bargain. Government does not set the commercial night. The must-have hospital extracts 2–3×, and the premium carries it.
How hospitals, insurers, and governments negotiate →In Job-plan sticker (Kaiser Family Foundation 2025 single $9,325 / family $26,993) plus Medicare payroll tax, Medicaid, and out of pocket. Private insurance is the primary card for 55% of people — unique among rich countries.
Out Each hospital and insurer writes a secret contract, often 2–3× Medicare (RAND). PBMs take list price on brands. 340B hospitals buy cheap and bill the job plan near ASP-plus. Claims, prior auth, and denial theater sit on top of the clinical bill.
In Job-plan sticker (Kaiser Family Foundation 2025 single $9,325 / family $26,993) plus Medicare payroll tax, Medicaid, and out of pocket. Private insurance is the primary card for 55% of people — unique among rich countries.
Out Each hospital and insurer writes a secret contract, often 2–3× Medicare (RAND). PBMs take list price on brands. 340B hospitals buy cheap and bill the job plan near ASP-plus. Claims, prior auth, and denial theater sit on top of the clinical bill.
Click a peer to split the extra US dollars. The recipe is 40% hospital prices, 28% brand drugs, 18% wages, 8% admin, 6% imaging / post-acute — a labeled mix, not a claims extract.
Ranked by what one US commercial event costs. Green is this country. Sourced when iFHP or Papanicolas posted a number; otherwise US price × that lever’s index, labeled modeled. How they pay is the money flow, not a second diagnosis.
Oncology, immunology, rare-disease vials that dominate the top 5% pharmacy line — not a generic blood-pressure pill.
Why US more US brand-name originator prices were 4.22× other OECD countries in 2022 (RAND/ASPE). France was among the cheapest. Generics are the exception: the US is often cheaper on unbranded generics, which is not what this pile is.
How United States pays PBM list plus 340B hospital-outpatient spread. No national bargain on the brand.
Open-heart year. Utilization is a bit higher in the US; the price gap is the story.
Why US more iFHP 2024 private-plan median: US $89,094 vs Australia $17,741. Same operation. US commercial hospitals do not take a national tariff.
How United States pays Each hospital and insurer writes a secret contract, often 2–3× Medicare (RAND). PBMs take list price on brands. 340B hospitals buy cheap and bill the job plan near ASP-plus. Claims, prior auth, and denial theater sit on top of the clinical bill.
Persistent top-5% — not a one-year crash. Three treatments a week, forever, until transplant.
Why US more USRDS-style Medicare spend per ESRD patient sits near $90k. Peers run dialysis as a budgeted service at lower unit cost. Not because they skip kidney failure.
How United States pays Each hospital and insurer writes a secret contract, often 2–3× Medicare (RAND). PBMs take list price on brands. 340B hospitals buy cheap and bill the job plan near ASP-plus. Claims, prior auth, and denial theater sit on top of the clinical bill.
HCUP #6 of hospital production cost. Shoppable for a lot of the cases — and still billed like an emergency in US commercial.
Why US more Hospital production cost about $31k (HCUP). Commercial modeled at 2.5× that. Peers pay a spine DRG, not a chargemaster plus implant markup.
How United States pays Each hospital and insurer writes a secret contract, often 2–3× Medicare (RAND). PBMs take list price on brands. 340B hospitals buy cheap and bill the job plan near ASP-plus. Claims, prior auth, and denial theater sit on top of the clinical bill.
US discharges for AMI match the 11-country mean (Papanicolas). They are not having more heart attacks in the billing file. They are paying more per stay.
Why US more Volume is similar. Price is not. Same ICU, same stent lab, US commercial multiple on the building.
How United States pays Each hospital and insurer writes a secret contract, often 2–3× Medicare (RAND). PBMs take list price on brands. 340B hospitals buy cheap and bill the job plan near ASP-plus. Claims, prior auth, and denial theater sit on top of the clinical bill.
Biggest US hospital cost line — $60B because of volume × ICU, not because it is the priciest DRG. Every peer has it.
Why US more HCUP production cost ~$24,790. Commercial modeled 2.5×. Japan’s top 5% spent 97.7 hospital nights at $432 a night vs US 6.6 nights at $3,180. More hospital, far less per night.
How United States pays Each hospital and insurer writes a secret contract, often 2–3× Medicare (RAND). PBMs take list price on brands. 340B hospitals buy cheap and bill the job plan near ASP-plus. Claims, prior auth, and denial theater sit on top of the clinical bill.
The crash on top of COPD, pneumonia, heart failure. Unshoppable. Price is the lever.
Why US more Cannot shop it. Peers still pay a posted DRG for the vent bed. US commercial does not.
How United States pays Each hospital and insurer writes a secret contract, often 2–3× Medicare (RAND). PBMs take list price on brands. 340B hospitals buy cheap and bill the job plan near ASP-plus. Claims, prior auth, and denial theater sit on top of the clinical bill.
Acute care is one place the US looks good: 30-day ischemic-stroke mortality 4.2 per 100 vs 7.9 mean in Papanicolas. The bill is still US-priced.
Why US more Better 30-day stroke survival does not require a 2–3× hospital contract. Other countries run stroke units on a tariff.
How United States pays Each hospital and insurer writes a secret contract, often 2–3× Medicare (RAND). PBMs take list price on brands. 340B hospitals buy cheap and bill the job plan near ASP-plus. Claims, prior auth, and denial theater sit on top of the clinical bill.
Same vial as the office. The job plan pays hospital outpatient rates so the tower keeps the spread. This is a US invention.
Why US more Peers infuse in a hospital too. They do not let the site of care multiply the brand price. No 340B-to-commercial spread.
How United States pays PBM list plus 340B hospital-outpatient spread. No national bargain on the brand.
HCUP #3. Repeat customer in the persistent third of the 5%.
Why US more Prevention can cut readmits. It does not explain a 2× night price. German and French heart-failure DRGs are the same disease, smaller invoice.
How United States pays Each hospital and insurer writes a secret contract, often 2–3× Medicare (RAND). PBMs take list price on brands. 340B hospitals buy cheap and bill the job plan near ASP-plus. Claims, prior auth, and denial theater sit on top of the clinical bill.
US does a bit more of these. iFHP private median still dwarfs peers.
Why US more Peterson-Kaiser Family Foundation: US private PCI can run 3–10× other countries’ private plans. Stent plus night plus lab — priced as a luxury good.
How United States pays Each hospital and insurer writes a secret contract, often 2–3× Medicare (RAND). PBMs take list price on brands. 340B hospitals buy cheap and bill the job plan near ASP-plus. Claims, prior auth, and denial theater sit on top of the clinical bill.
US has more obesity and more diabetes. After you account for prevalence, Papanicolas finds diabetes hospitalizations look less outlier. Asthma still does. The stay is still expensive.
Why US more Sicker metabolic mix is real. It is not a 2× hospital price. The extra dollars are mostly the invoice, not the A1c.
How United States pays Each hospital and insurer writes a secret contract, often 2–3× Medicare (RAND). PBMs take list price on brands. 340B hospitals buy cheap and bill the job plan near ASP-plus. Claims, prior auth, and denial theater sit on top of the clinical bill.
US actually does fewer inpatient hips than peers now (shift to outpatient). Cost per surgery is still the highest.
Why US more iFHP: US $29,006 vs Australia $14,986. Peterson-Kaiser Family Foundation: peers do ~3× as many inpatient hips and still pay less each. Volume is not why the US bill is bigger.
How United States pays Each hospital and insurer writes a secret contract, often 2–3× Medicare (RAND). PBMs take list price on brands. 340B hospitals buy cheap and bill the job plan near ASP-plus. Claims, prior auth, and denial theater sit on top of the clinical bill.
Papanicolas: US did more knees. Peterson-Kaiser Family Foundation 2022: US now among the fewest inpatient knees — they moved to outpatient, where commercial prices stay high.
Why US more iFHP: US $26,340 vs Australia $13,609. Implant plus OR plus night. Peers buy the same cobalt chrome on a schedule.
How United States pays Each hospital and insurer writes a secret contract, often 2–3× Medicare (RAND). PBMs take list price on brands. 340B hospitals buy cheap and bill the job plan near ASP-plus. Claims, prior auth, and denial theater sit on top of the clinical bill.
AFib ablation, ICD, pacemaker. Device list prices in the US are a known outlier (Papanicolas flagged devices as under-measured and expensive).
Why US more The generator costs the hospital a few thousand on GPO and bills as a five-figure line. Peers negotiate devices nationally or in the DRG.
How United States pays Each hospital and insurer writes a secret contract, often 2–3× Medicare (RAND). PBMs take list price on brands. 340B hospitals buy cheap and bill the job plan near ASP-plus. Claims, prior auth, and denial theater sit on top of the clinical bill.
HCUP #2 on headcount. Typical delivery is cheap; the NICU tail is a top-5% year. US C-section rate is higher (33 vs mean 25 per 100 births).
Why US more A bit more C-sections. A lot more per delivery and per NICU day. France and Sweden run births on a tariff; they are not skipping the NICU.
How United States pays Each hospital and insurer writes a secret contract, often 2–3× Medicare (RAND). PBMs take list price on brands. 340B hospitals buy cheap and bill the job plan near ASP-plus. Claims, prior auth, and denial theater sit on top of the clinical bill.
Surgeon, cardiologist, intensivist, oncologist — billed on top of the facility. US specialists $316,000 PPP vs $98k (Sweden) to $202k (Australia).
Why US more Doctors do make more. Papanicolas: US specialists ~2× UK/Germany; generalists $218k vs $87k–$154k. Training debt does not eat a $200k gap. This is part of the pile — smaller than hospital prices and brand drugs.
How United States pays Specialists $316k PPP, generalists $218k. Highest ratio to national average wage in the 11-country set.
Planning, regulating, managing, and fighting. Papanicolas: 8% of US health spending vs 1–3% in the other 10. NHEA private insurance keep ~10%.
Why US more Every other country on these tabs has one price list or a handful of funds. The US runs a claims industry. 54% of US doctors called insurance paperwork a major problem. That labor is in the premium.
How United States pays 8% of spending on governance/admin (Papanicolas) plus ~10% insurer keep on private premiums.
US scans a lot of machines (2nd in MRI units). Medicare MRI prices sit near peer public fees. Commercial is ~3× Medicare.
Why US more Papanicolas 2013: US $1,145 vs Australia $350, Netherlands $461. Peterson-Kaiser Family Foundation 2022: volume of MRIs is similar to peers; private price is the outlier. Quality of the magnet is not 3×.
How United States pays Medicare MRI is near peer public fees. Commercial is ~3×. The magnet is not 3×.
7.4% of the top 5% MEPS bill, ~0% of the cheap half. Nurses and aides, not wellness. US nurse pay $74,160 PPP vs $42,492 in France.
Why US more Wammes: rehab/post-acute is a bigger slice in the US and Canada. Unit wages are higher. Peers use cheaper long stays (Japan) or budgeted community nursing.
How United States pays Each hospital and insurer writes a secret contract, often 2–3× Medicare (RAND). PBMs take list price on brands. 340B hospitals buy cheap and bill the job plan near ASP-plus. Claims, prior auth, and denial theater sit on top of the clinical bill.
The middle person spent $1,361. The top 1% averaged $147,071. Premiums spread the tail across everyone. Raising deductibles on the healthy half does not shrink this pile. See invested premiums for pay-in vs the median person. The same 5% does not take 50% of physician or hospital-CEO pay — that cut is on pay vs the 5%.
$72,918 does not look like half the health-care system until you add. The top 5% average 10× this room’s mean ($7,292). Five people × 10× = 50% of the money. That is the identity, not a coincidence. Each bar is one person. Green is the cheap half. Orange is the expensive five. The last bar is the top 1%.
| Who | People | Each (avg) | Their pile | Share of spending | Share of premiums |
|---|---|---|---|---|---|
| Bottom 50% (14 of them $0) | 50 | $433 | $21,650 | 3.0% | 2.3% |
| Next 45 | 45 | $7,621 | $342,940 | 47.0% | 36.8% |
| Top 5% except the last person | 4 | $53,531 | $214,123 | 29.4% | 23.0% |
| Top 1% | 1 | $150,467 | $150,467 | 20.6% | 16.1% |
| Top 5% together | 5 | $72,918 | $364,590 | 50% | 39% |
| All 100 | 100 | $7,292 | $729,180 | 100% | 78% of stickers |
They should take 50% of spending. They take about 39% of premiums, because the Kaiser Family Foundation single sticker ($9,325 × 100 = $932,500) is bigger than this room’s bills. The leftover $203,320 is insurance load in this model — admin, profit, and the fact that MEPS spending is not an ESI claims file. If the pool instead spent the NHEA employer-plan mean ($8,000 × 100), the same five bills would be 46% of what the plans paid. One top-5% year is 7.8× one single premium, or 2.7× the family sticker. Band averages are Peterson-Kaiser Family Foundation MEPS 2023. Overall mean is implied so the top 5% are exactly half. Shape inside each band is modeled. Premium is Kaiser Family Foundation 2025 single, a different year — the ratio is the point, not a 2023 premium file. Where that sticker actually goes is the pie below.
The insurer does not keep the hospital money. Of a private-insurance dollar, about 10% stays with the plan (admin, taxes, profit). Hospitals get the biggest slice — and in the national accounts that slice is the whole building, inpatient plus hospital outpatient. Doctors and independent clinics are the office-visit / outpatient pile. Retail drugs are the pharmacy counter, not the infusion chair in the tower.
Single sticker $9,325 (Kaiser Family Foundation 2025). Family sticker $26,993. Shares are CMS NHEA 2023 private health insurance — the whole PHI dollar, including Marketplace, not only job coverage. In the 100-person room, hospitals would get about $354,350 of the $932,500 in stickers; the plans would keep about $93,250. Large-group fully insured simple loss ratios sat near 91% in 2025 (Kaiser Family Foundation) — the keep is in the same neighborhood as this 10%. About two-thirds of workers are in self-funded plans, where the “insurer” is a TPA and the keep is an ASO fee, usually thinner than 10%; the hospital / doctor / drug split of the claims is the same idea. Sourced mix; dollars on 2025 stickers are that mix applied, not a 2023 premium file.
High blood pressure is the most common label in this group. It is not why the year costs $67,321. The dollars are hospital stays, specialty drugs, and stacked complications. AHRQ says this out loud.
Inpatient care is 27.9% of the top 5%’s dollars — about $18,783 of a $67,321 year — and 0.1% of the cheap half’s dollars. One admission usually clears the $30,206 cut by itself. Office visits are most of what the cheap half buys.
Drugs are 27.8% of top-5% dollars, about $18,715 in that average year, vs 11.2% of the cheap half. IQVIA: specialty medicines are 54% of US medicine spending. New 2023 launches had a median annual price above $150,000. Oncology and rare-disease launches sit near $300,000 per patient. Blood-pressure generics do not do that.
Source: AHRQ MEPS Statistical Brief 560 (share of the top 5% bill); IQVIA Institute, The Use of Medicines in the U.S. 2024 / outlook to 2028 (specialty share and launch prices).
75.1% of adults in the top 5% have two or more AHRQ “priority” conditions (heart disease, diabetes, cancer, arthritis, stroke, and the rest of that list). In the cheap half, 49.9% have none. The expensive event is often the crash: sepsis, heart failure, a diabetes complication — not the diagnosis on the problem list.
Sepsis is $60 billion because 2.4 million people landed in a hospital with it in one year — about 1 in 14 stays — and each of those stays cost the hospital about $24,790 to produce. It is not the most expensive stay on the list. It is the one that is both common and an ICU-level bill.
| Condition | Stays | Hospital cost each | Total |
|---|---|---|---|
| Sepsis | 2.42 million | $24,790 | $60.0 billion |
| Newborns | 3.49 million | $5,614 | $19.6 billion |
| Heart attack | 582,000 | $28,062 | $16.3 billion |
| Spine / spondylosis | 447,000 | $31,324 | $14.0 billion |
Newborns beat sepsis on headcount and still cost a third as much, because a typical delivery is not nine days in intensive care. A heart attack or a spine operation costs more per stay; there are not 2.4 million of them. Sepsis is also a headline code: once an infection (lungs, urine, a diabetic foot, a gut leak) goes body-wide, septicemia is the principal diagnosis and the pneumonia or the diabetes does not get the bar. The bill is ICU days, a ventilator, pressors — not a $12 antibiotic. You cannot shop that at 2 a.m. Spine and osteoarthritis ($5.6 billion) often you can. Hip fracture ($7.7 billion) sits in between. AHRQ saw sepsis hospitalizations up about 40% from 2016–2021 ($31.2 billion → $52.1 billion of hospital cost, broader definition). Some of that is COVID and an older population. Some is hospitals writing the code more often. Even so, septicemia is #1 for Medicare, Medicaid, private insurance, and the uninsured.
People 65 and older are 18.1% of the population and 40.5% of the top 5%. Children are 21.6% of people and 4.9% of the expensive group. Medicare pays 30.8% of top-5% dollars; private insurance pays 44.4%; the patient pays 8.0% at the counter. The job-plan sticker on the journey is partly pre-paying this tail — including other people’s parents.
Other rich countries have sepsis, joints, and cancer. They do not pay US commercial hospital rates (often 2–3× Medicare — see RAND on Evidence) or US specialty-drug list prices, and they do not park infusions in a 340B hospital outpatient department so the spread can be billed to a job plan. The country tabs at the top of this page put a number on that: same 5% concentration, OECD prices. Office, outpatient, and ER are still 32.5% of the top 5% bill — about $21,879 — and a lot of that is hospital-owned clinics, not a cheap office.
MEPS also misses nursing homes. The real tail, including people who left the community for a facility, is sicker and more expensive than this survey.
Only about one in three people in the top 5% are still there next year. The other two-thirds had a bad year — surgery, injury, a cancer treatment year — and then drop down. Wellness pamphlets do not prevent a crash; cash menus can change the price of a planned one.
Left graph: how often the label shows up in the expensive group. Right: what the whole US system spends on a condition (Dieleman, 2019). High blood pressure is everywhere in the 5% and is not a top dollar line. Diabetes and muscle/joint disease are both common and expensive.
Ranked by whether it hits the dollars above — not by what sounds caring on a benefits slide.
Inpatient is the line the cheap half never uses. Planned joints, much spine, and a lot of imaging and outpatient surgery can move to a posted-price ASC. That is this site’s Surgeries floor and the hospital floors. Sepsis, heart attack, newborns, and respiratory failure cannot be shopped. Those need hospital prices that are not 2–3× Medicare — Maryland’s all-payer rates are the live US proof (one stay, one rate, whoever pays). Who sits in that room: hospitals, insurers, and governments.
Drugs are another 28% of the top 5% bill. Site of care is the markup: the same vial costs a job plan more in a 340B hospital outpatient department than in an office. Biosimilars and Medicare negotiation attack list price; moving the chair attacks the spread. See 340B on Evidence.
About 34% stay expensive next year — heart failure, advanced diabetes, cancer on specialty drugs, kidney failure. That is where chronic-care programs can pay for themselves, if they prevent the sepsis / heart-failure / “diabetes with complication” admission. They do not replace price reform. A prevented admission at US commercial rates is a bigger save than the same admission at Medicare rates.
Two-thirds of the 5% are a one-year shock. Catastrophic coverage and stop-loss are the right shape. High deductibles on the healthy 50% collect little (they averaged $374) and do not lower sepsis prices. The family premium on the journey is $26,993 because the tail is socialized across every paycheck — which is the point of insurance — at US prices, which is the problem.
The cheap half already spends almost nothing. Gym credits and annual physicals do not touch sepsis, a $150k specialty drug, or a commercial hospital day. Prevention of diabetes and heart disease can shrink the persistent group over decades. It is not a same-year claims lever, and it does not excuse the price of the year you already had.
Every graph on this page is tagged to a survey or a cost file on Evidence. MEPS undercounts the sickest (no nursing homes). HCUP hospital lines are costs, not what a job plan paid. Dieleman is the whole country, not the top 5% slice. The ranking is the argument: hospital + specialty drugs + stacked complications × US prices — not “people go to the doctor too much.” Country tabs use OECD 2024 per person × the 5% share; that is not MEPS.