rexroad

Author here. This is Issue #6 of The American Healthcare Conundrum, a data-driven newsletter that analyzes CMS cost report data to find where U.S. healthcare dollars actually go.

For this issue I downloaded all 5,480 hospital cost reports from CMS HCRIS and extracted supply cost centers (medical supplies, implantable devices, drugs charged to patients). Key findings:

- Hospitals at the 90th percentile spend 3-7x more per discharge on supplies than the 10th percentile, even after adjusting for case mix - The gap persists within the same bed-size tier and teaching status - A 50-state ranking shows where the waste concentrates geographically - Conservative estimate: $28B/year in addressable waste

All analysis scripts and raw data pipelines are open source: https://github.com/rexrodeo/american-healthcare-conundrum/tr...

Previous issue on admin waste (Issue #5) hit the front page here 20 days ago. Happy to answer questions about methodology.

CMS now requires hospitals to publish machine-readable price files under the Hospital Price Transparency Rule (effective 2021, enforcement added 2022). Compliance rates are technically high. The problem is the files run tens of thousands of line items in inconsistent formats, negotiated rates vary by insurance plan, and chargemaster list prices bear no predictable relationship to what any patient actually pays.

RAND's Round 5.1 study analyzed actual paid claims rather than posted prices and found commercial insurers paid 254% of Medicare rates for identical procedures at the same hospitals. The transparency is on paper. The opacity is structural: a price list is only useful if a purchaser can act on it, and the moment of care is not when most patients can comparison shop. Which is why reference pricing works at the employer level (purchaser can act in advance on a population basis) but published hospital price lists have not moved the market.

Agreed on the diagnosis: the inpatient hospital market does not function as a competitive market. Patients arriving by ambulance do not shop on price. Insurance insulates consumers from marginal cost. Regulatory and capital barriers prevent new entrants. The HCRIS FY2023 data for 3,193 hospitals shows what this produces: a 2.6x median markup to actual cost, with the highest-markup hospitals rarely losing patients to lower-cost alternatives nearby.

Your examples of what works are instructive. Generic OTC, needles, blood panels all share the same conditions: price-visible, consumer-controlled, no third-party payment insulation. The RAND hospital data shows the identical procedure (hip replacement) costing $29K commercially in the US versus $14.7K in Germany and $8.7K in Spain. Same implant, same surgeon training requirements. The patient's insurance status varies; the procedure does not.

Montana's commercial reference pricing is not deregulation, but it creates the price-visibility function you are describing: a known floor price that makes it possible for a purchaser to act on price information. Employers have adopted it nationally with documented savings.

For-profit structure is part of the problem but not all of it. Issue 3 analyzed 3,193 hospitals using CMS HCRIS FY2023 cost reports. For-profit hospitals do have the highest cost-to-charge markups: 4.11x median. But nonprofits are 2.46x, and nonprofits hold 75.5% of total national hospital supply spend. The ownership form does not reliably determine pricing behavior when market conditions are the same: opaque prices, patients who cannot shop, no reference price to anchor negotiation.

Maryland has had all-payer hospital rate-setting since 1977 with a largely nonprofit hospital sector. It produces significantly better cost control than the national average. The structural fix for hospital pricing (commercial reference pricing at 200% of Medicare) works regardless of tax status because it creates a known floor price. That is what the RAND data and the Montana Medicaid experience both show. The mechanism matters more than the ownership form.

The direction is right. Total US healthcare spending is $4.87T for 335M people ($14,570/capita, CMS NHE 2023). Japan's per-capita is $5,790, with the highest life expectancy in the world and lowest infant mortality in the OECD. The annual national gap is approximately $3T.

The series is not arguing for a specific coverage structure. It is documenting where the excess goes, mechanism by mechanism. Four issues in, $128.6B is accounted for conservatively: drug pricing, hospital commercial markups, PBM extraction. Each mechanism has a defensible, operationally precedented fix that does not require redesigning the entire system. Montana Medicaid adopted commercial reference pricing at 200% of Medicare and measured no quality deterioration. The FTC documented PBM specialty drug markups and three states have already enacted clawback restrictions. None of that required universal healthcare.

Thanks for sharing the OSU piece, good read. A few things it surfaces that complicate the "discount cards help patients" narrative: discount card companies still contract with PBMs to set pricing, they don't bypass them entirely. The savings often come out of the pharmacy's margin rather than the PBM's: the article shows a pharmacy receiving $5 on a drug it acquired for $15, with the PBM still collecting a transaction fee either way. GoodRx was also fined by the FTC for selling patient health data to advertisers without authorization, so there's a second extraction happening on the data side.

The deeper point the card market reveals: a profitable arbitrage layer can consistently undercut the insurer's "negotiated" rate, which tells you the negotiated rate isn't really a discount. Generic apixaban costs £1.16 per 30-day supply in the UK. Medicare's gross cost for Eliquis (same molecule) is $862.

Your example captures two distinct extraction mechanisms in one transaction. The $25 to $125 gap is spread pricing: the PBM pockets the difference between what they pay the pharmacy and what they bill the plan. The deductible non-application is a separate mechanism: by routing through their own channel, the PBM ensures that cost doesn't reduce your out-of-pocket maximum, extending your exposure for the year.

The FTC's 2024 Interim Report documented $7.3B in specialty drug markups from the Big 3 PBMs in a single year. The Ohio Auditor found PBM spread pricing extracted $224.8M from one state's $2.5B Medicaid drug budget annually.

Author here. Issue #4 is now live — pharmacy benefit managers.

Three companies process 80% of US prescriptions. The FTC spent two years investigating them and documented $7.3B in specialty drug markups at PBM-owned pharmacies alone. Ohio's state auditor found $224.8M in spread pricing extracted from one state's Medicaid program in a single year.

Six mechanisms, $30B/year booked conservatively. Running total across four issues: $128.6B.

github.com/rexrodeo/american-healthcare-conundrum

The NHS prices in the Issue #2 analysis are not subsidies. They are the generic reimbursement rates from the UK Drug Tariff after patent expiry on each molecule. Apixaban (Eliquis) costs £1.16 per 30-day supply on the Drug Tariff. That is what the open market charges for the active compound once patent protection expires. The NHS does not manufacture it or subsidize the price: that is the market rate for the molecule itself.

The $862 Medicare gross cost for the same molecule is not explained by active R&D recovery, either. The IRA's first ten negotiated prices (effective January 2026) cut Medicare gross costs 40-70% per drug, which does not happen if those gross costs were development-cost-justified.

The data supports this. The AMA's 2024 Prior Authorization survey found 93% of physicians report PA requirements delay medically necessary care. Twenty-nine percent reported a PA delay causing a serious adverse event for a patient. Seven percent reported PA contributed to a patient death.

The requirement that patients fight for care isn't just a frustration. It's a documented cost driver: Health Affairs (2025) puts the total system-wide cost of prior authorization at $93.3B/year, including $35.8B borne directly by patients navigating the process. The persistence required to appeal a denial is unevenly distributed across income, education, and time availability. That is a structural equity problem as well as a cost problem. Issue #5 of this series covers the full mechanism.

The RAND Round 5.1 study (2023) puts US commercial insurer payments at 254% of Medicare rates for identical procedures. That's the mechanism behind the international gaps — it's not complexity or quality, it's that commercial insurers negotiate against chargemaster list prices rather than against cost. The HCRIS cost-to-charge analysis (3,193 hospitals, FY2023) puts median markup at 2.6x actual costs.

Thank you for the feedback and support! I've added some charts to the repo and have charts in the substack.

Personal experience with specific interventions reflects something real: US cancer survival rates, cardiac procedure outcomes, and access to cutting-edge treatments are genuinely strong for people with good coverage. That's not disputed.

The cost-outcome tradeoff shows up at the population level. US life expectancy: 77.5 years. Spain: 83.6. UK: 81.6. Infant mortality: US 5.4 per 1,000 vs. Spain 3.4, UK 3.7 (OECD 2023). The US spends $14,570 per capita. Spain spends $3,300, UK $4,100. If the premium were buying 10 extra years of life expectancy and half the infant mortality, it might be worth the argument. The data shows the opposite at population scale.

The newsletter's framing isn't that US clinical quality is poor. It's that the US is paying $3T more per year than Japan (same life expectancy, lowest infant mortality in OECD) for aggregate outcomes that are worse.

Fair point about KPI gaming, and it's a real problem in value-based care. But the fix in Issue #3 (commercial reference pricing at 200% of Medicare) is a price cap, not a quality incentive structure, so it doesn't directly create the risk-avoidance problem you're describing.

Montana Medicaid has used 200% Medicare reference pricing since 2015. Published evaluations haven't shown measurable quality deterioration or patient-selection effects in that program. The RAND Round 5.1 study underlying the savings estimate controlled for case mix, so it's comparing equivalent procedures at equivalent acuity. Risk adjustment is still genuinely hard at the individual level, and the concern is well-founded for P4P schemes. It's a separate question from whether commercial payers should pay 254% of Medicare rates for the same surgery at the same hospital.

Your example captures two distinct extraction mechanisms in one transaction. The $25 to $125 gap is spread pricing: the PBM pockets the difference between what they pay the pharmacy and what they bill the plan. The deductible non-application is a separate mechanism: by routing through their own mail-order or in-network channel, the PBM ensures that cost doesn't reduce your out-of-pocket maximum, extending your exposure for the year.

The FTC's 2024 Interim Report documented $7.3B in specialty drug markups from the Big 3 PBMs (CVS Caremark, Express Scripts, OptumRx) in a single year. The Ohio Auditor found PBM spread pricing extracted $224.8M from one state's $2.5B Medicaid drug budget annually. This is the subject of the next issue being released this Sunday.

Those figures are in the right range, and the full picture is larger. The CMS NHE 2023 data puts total US healthcare administration at roughly $1.1-1.7T annually (depending on methodology), building on Woolhandler and Himmelstein's 2020 Annals paper ($812B in 2017 dollars). The per-capita comparison against 10 OECD peers: US $4,983 vs. a peer average of $884. That 5.6x gap is the number Issue #5 of this series will examine in detail, covering three separate computation methodologies and why the estimates range so widely. All source code will be in the repo: https://github.com/rexrodeo/american-healthcare-conundrum

The policy lever that addresses this is billing standardization, not just insurer reform. Countries like Germany and Switzerland run much lower admin under private insurance through standardized claims formats and all-payer rate setting.

The Dutch model is a useful counterexample to the argument that you need a single-payer structure to contain costs. Netherlands uses regulated private insurers with community rating and risk equalization, yet achieves per-capita spending well below the US (roughly $7,200 vs. $14,570 in 2023 OECD data). The direct insurer-hospital negotiation you describe is also how Germany's sickness funds operate.

The US equivalent would be all-payer rate setting. Maryland has run a statewide all-payer hospital rate system since 1977 with documented cost containment. Issue #3 of this series focuses on a lighter-weight near-term version: capping commercial hospital payments at 200% of Medicare (already used by Montana Medicaid and thousands of self-insured employers). The Dutch model shows a stronger structural fix is feasible. The question is political path, not technical feasibility.

The obesity adjustment is worth quantifying. US adult obesity: 42% (CDC). UK: 28%, Australia: 31%, Germany: 22%. Those gaps are real, but they don't explain a 2.5x per-capita spending differential. The Commonwealth Fund's 2021 analysis controlled for age, income, and chronic condition burden; the US still spent roughly $5,000 more per capita than the next-highest spender (Switzerland).

Obesity also matters less than assumed in hospital pricing: a hip replacement costs $29,000 commercially in the US regardless of patient BMI, vs. $15,000 in Germany and $9,000 in Spain (iFHP 2024). The cost structure is in the pricing system. Johns Hopkins researchers estimated eliminating US obesity would reduce healthcare spending by about 12%, real but not 2.5x. Repo with methodology: https://github.com/rexrodeo/american-healthcare-conundrum

Correction on ownership breakdown: A CMS cost report expert flagged that my CTRL_TYPE mapping in the HCRIS processing script was wrong — I had for-profit and nonprofit hospital categories swapped. The corrected figures: for-profit hospitals have a 4.11x median markup (highest), nonprofits 2.46x, government 2.22x. The 254% commercial-to-Medicare finding ($73B savings estimate) is unaffected — that's from RAND, not my HCRIS analysis. Corrected code and a full audit report are on GitHub. This is exactly why the code is open-source.

The wage adjustment is worth testing with data. Japan's GDP per capita on a PPP basis is roughly $47,000 versus the US at $80,000, a 1.7x income gap. The per-capita healthcare spending gap is $14,570 vs $5,790, a 2.5x ratio. Healthcare costs outpace the income gap by a meaningful margin even on PPP terms.

The outcome data is what makes the adjustment argument hard to sustain. Japan has the highest life expectancy in the OECD (84 years) and the lowest infant mortality (1.7 per 1,000). If higher spending were buying proportionally better outcomes, the wage argument would carry more weight. The US spends 2.5x more and gets worse population health statistics. PPP narrows the gap, it doesn't close it.

Those figures are consistent with what Issue #5 (still a couple weeks out) of this series computes from CMS NHE 2023 data and OECD health statistics. The 10-peer OECD average lands at $884 per capita, putting the US at 5.6x. Scaled to 335M people, that's $1.37T in excess admin annually.

The Woolhandler/Himmelstein 2020 figure ($812B) updates to $1.13-1.66T in 2023 dollars when adjusted for healthcare inflation. The CMS narrow admin estimate ($410B) plus CAP's billing complexity analysis ($496B) gives a $906B floor. Those three methodologies agree on the floor, disagree on the ceiling. Issue #5 covers all three and explains why the range is so wide. Coming soon.

The rebate pass-through rule (effective 2028) is a real step, and worth tracking. But rebate retention is one of six extraction mechanisms the Big 3 PBMs use. The FTC's Interim Reports I and II (2024-2025) documented $7.3B in specialty drug markups alone, separate from rebate games. The Ohio Auditor found PBM spread pricing extracted $224.8M from a single state's $2.5B Medicaid drug budget in one year.

The rebate rule doesn't touch spread pricing, formulary manipulation, or self-preferencing to vertically integrated pharmacies. Issue #4 (scheduled for releases 3/22) of this series covers the full mechanism stack and what each proposed reform actually targets. Repo: https://github.com/rexrodeo/american-healthcare-conundrum

Thanks for the meta-analysis reference. The 141-259% range tracks with what I see in the HCRIS data. The variance across hospitals is enormous — even within the same bed-size category, the P75/P25 ratio for cost-to-charge is 2.5-3.4x. Hospitals in the same peer group are charging wildly different amounts for equivalent services. All the scripts are in the repo if you want to dig into the hospital-level data: github.com/rexrodeo/american-healthcare-conundrum

The MLR incentive question is one I'm digging into for a future issue. The short version: the ACA's 80/85% MLR floor was supposed to constrain overhead, but vertical integration changed the math. When UnitedHealth's Optum division provides services to UnitedHealthcare's members, those internal payments count as "medical expenses" for MLR purposes. The money stays in-house but reports as care delivery. On the denial rate point: 15-17% initial denial rate, 80%+ overturned on appeal, but less than 1% of patients actually appeal. That gap between the overturn rate and the appeal rate is where the profit lives. If you deny 100 claims and only 1 patient appeals, you've effectively reduced payouts on 99 claims at the cost of processing 1 appeal. I'll have the numbers on this in a later issue.

The cross-subsidy argument is one hospitals use to justify high commercial rates: "Medicare underpays, so we have to make it up on commercial." The HCRIS data lets you test this. If cross-subsidization were the full story, you'd expect cost-to-charge ratios to be tight — hospitals would charge commercial just enough to cover the Medicare shortfall. Instead, the median markup is 2.6x across all hospitals, and 3.96x for nonprofits. That's not cross-subsidy. That's pricing power in a concentrated market.

You're right that there's no single bad actor, and that's exactly the framing of this series. Each issue isolates one mechanism with one savings estimate. The 254% figure is RAND's. What I added is the HCRIS cost-to-charge analysis across 3,193 hospitals showing the variance by ownership type.

The surprise was nonprofit hospitals: median markup of 3.96x actual operating costs, versus 2.39x for for-profit and 1.87x for government hospitals. That's hard to square with the narrative that nonprofits deserve their tax exemptions ($28-37B/year) because they serve charitable purposes.

On the self-funded employer point — you're correct that self-funded plans have more negotiating latitude, and thousands of them already use reference pricing (capping hospital payments at a percentage of Medicare). That's actually the policy fix this analysis proposes. Montana Medicaid implemented it and saved $47.8M. The question is why it isn't the default.

Author here. The 254% figure comes from RAND Round 5.1. I built a Python pipeline on CMS HCRIS cost reports (FY2023, 3,193 hospitals) to compute cost-to-charge ratios by ownership type. The surprising finding: nonprofit hospitals have a median markup of 3.96x actual costs. All scripts are in the repo. Happy to discuss methodology.