25.27 — The US Healthcare System

A patient breaks an ankle and goes to an emergency department. The hospital bills $18,000. The insurer's contracted rate for that care is $6,200, so $11,800 is written off and never paid by anyone. The patient owes $2,400 because they have not met their deductible for the year. The orthopaedic surgeon who set the ankle is not employed by the hospital and bills separately. The imaging is billed separately again. Four documents arrive over eight weeks, three of which look like bills and one of which says in large letters that it is not a bill.

Nothing here is a mistake. This is the system operating exactly as designed, and every one of those steps is a transaction in software that somebody built.

If your clients are American healthcare organisations, this chapter is the map. Even for a pharmaceutical client it matters, because the payer decides whether their product is used at all.

The one sentence that explains everything else

The United States does not have a healthcare system. It has several, sorted by who you are, and they interact.

Most working-age people get insurance through an employer. Most people over 65 get Medicare from the federal government. Lower-income people may get Medicaid, run jointly by federal and state governments. Veterans, active service members and Native Americans have their own systems. Some people buy individual coverage on a marketplace. And some have nothing.

Because coverage is tied to employment, changing job, losing a job, turning 65, or having your income cross a threshold can all change your insurance, your doctor, your drug coverage and your costs. This churn is a permanent, structural feature and it drives an enormous amount of administrative work — including the re-verification problem in Chapter 25.25 and the eligibility redeterminations in Chapter 25.30.

The size of it

In 2024 the United States spent $5.3 trillion on health care — 18.0 percent of the national economy, and $15,474 per person. Medicare accounted for around 21 percent of that spending, at about $1.1 trillion, and Medicaid for about 18 percent, at roughly $932 billion.

Two facts about that number are worth carrying, because they end most casual arguments.

The largest categories are hospital care and physician services. Prescription drugs are around a tenth of total spending. Which means drug prices, however visible and however contentious, are not the arithmetic centre of American health spending.

And the United States spends far more per person than any comparable country without better population health outcomes on most measures. The best-supported explanation is that prices are higher — for the same procedure, the same device, the same drug — rather than that Americans consume more care. Administrative complexity adds a substantial further layer, and that layer is largely what this Part's remaining chapters describe.

The players, and what each one wants

Follow the incentives and behaviour that looks irrational becomes predictable.

Payers — insurance companies, employers who fund their own plans, and government programmes. They collect premiums or tax money and pay claims. They want lower cost per member, and they manage that by negotiating prices, deciding what is covered, and requiring approval before expensive things happen.

Providers — hospitals, physician groups, laboratories, imaging centres, pharmacies. They deliver care and want to be paid adequately and promptly. Under traditional payment, more services means more revenue, which is the incentive that value-based payment in Chapter 25.29 was invented to change.

Pharmacy benefit managers sit between drug manufacturers and payers, negotiating rebates, running the covered-drug list and processing pharmacy claims. They earn from the spread between what they negotiate and what they charge, plus fees, and they are the reason a drug's list price and its net price differ so widely.

Wholesalers and distributors move the physical product. Three of them handle the large majority of American drug distribution.

Manufacturers want their product on the covered list at a good position with as few restrictions as possible.

Employers pay a large share of premiums for their workers and want the total cost to stop rising.

And patients, who in this arrangement are usually not the ones choosing or paying most of the price, which is the fundamental reason ordinary market pressure works so weakly here.

Medicare, part by part

Medicare covers people aged 65 and over, plus younger people with certain disabilities or end-stage kidney disease. Its four parts confuse everyone at first and are simple once laid out.

Part	Covers	Note
A	Inpatient hospital, skilled nursing, hospice	Usually no premium
B	Doctor visits, outpatient, equipment	Monthly premium
C	Medicare Advantage	Private plan replacing A and B
D	Outpatient prescription drugs	Private plans

Original Medicare means Parts A and B, where the government pays providers directly and beneficiaries often buy supplemental coverage for what is left over.

Medicare Advantage — Part C — is the alternative: a private plan paid a fixed amount per member per month to provide everything Parts A and B cover, usually with extra benefits and a restricted network. It has grown to cover around half of all Medicare beneficiaries, which is why plan quality ratings and risk adjustment (Chapter 25.29) have become such large industries.

Part D covers outpatient drugs through private plans with their own formularies. Recent legislation changed its economics significantly, including a cap on what beneficiaries pay out of pocket in a year and the introduction of negotiation of prices for selected high-spend drugs — the first time the federal programme has directly negotiated drug prices, and a substantial change to manufacturers' long-term planning.

Medicaid, and why it is thirty-eight programmes

Medicaid covers low-income people and is jointly funded by federal and state governments but administered by the states within federal rules.

The consequence is that eligibility rules, covered benefits, provider payment rates and administration differ by state. A vendor building for Medicaid is not building one product; they are building something that must accommodate genuinely different programmes with different data and different rules. Chapter 25.30 covers this in detail because it is a large, distinctive area of service work.

The Affordable Care Act of 2010 allowed states to expand eligibility to adults up to a defined income level with heavy federal funding, and states chose differently — which is why coverage for a person with the same income can differ entirely depending on the state line they live on.

Commercial insurance and managed care

Managed care means the insurer actively manages what care is delivered rather than simply paying bills. The plan types differ in how tightly they do that.

HMO — you must use the plan's network, and typically need a referral from a primary care doctor to see a specialist. Cheapest, most restrictive.

PPO — a network with better rates, but out-of-network care is still partly covered, and no referral is needed. More expensive, more freedom.

EPO — a network with no out-of-network coverage but no referral requirement.

POS — a hybrid with a gatekeeper and some out-of-network coverage.

And the tools the plan uses to manage cost are the ones your systems will interact with constantly.

Networks and contracted rates. The listed price of care is nearly meaningless; what matters is the negotiated rate between that insurer and that provider.

Prior authorisation. Approval required before certain services or drugs, described operationally in Chapter 25.28.

Step therapy. You must try the cheaper option first and show it failed.

Formularies with tiers. Drugs are placed on tiers with different patient costs, and placement is negotiated with manufacturers.

Concurrent and retrospective review — checking, during or after care, that it was necessary and appropriate.

The words on the patient's bill

These terms appear in every claims system and every patient-facing product, and they are worth being exact about.

Premium — what is paid monthly for coverage, whether or not any care is used.

Deductible — what the patient pays before the plan begins paying.

Copay — a fixed amount per visit or prescription.

Coinsurance — a percentage of the cost, so the patient's share rises with the price.

Out-of-pocket maximum — the annual ceiling on the patient's share, after which the plan pays fully. This is the number that determines whether an illness is survivable financially.

Explanation of benefits — the document the plan sends showing what was billed, what was allowed, what the plan paid and what the patient owes. It is not a bill, it says so, and patients pay it anyway constantly.

And balance billing — where an out-of-network provider bills the patient for the difference between their charge and what the plan paid. The No Surprises Act now protects patients from this in emergencies and in situations where they had no realistic choice of provider, such as an out-of-network anaesthetist at an in-network hospital, and it created a dispute resolution process between the provider and the plan instead.

Why the costs are what they are

Four explanations do most of the work, and a client's strategy conversations assume you know them.

Prices. Negotiated separately by thousands of payers and providers, with wide variation for identical care in the same city. Nobody with power in the transaction has both the information and the incentive to shop.

Fee-for-service incentives. Paying per service rewards volume rather than health. This is the diagnosis behind every value-based payment reform.

Administrative complexity. Every provider bills many payers, each with different rules, forms, coverage policies and appeal processes. Every payer runs enormous machinery to process and adjudicate. A meaningful share of national health spending is consumed by this activity, and it is the direct cause of the revenue cycle industry in Chapter 25.28.

And fragmentation. A patient's care is spread across organisations that do not share records, so tests are repeated, drugs interact unnoticed and follow-up is missed. That is the problem the interoperability work in Chapters 25.31 and 25.32 exists to solve.

For perspective, briefly

One paragraph of comparison prevents the American arrangement from feeling like the natural order of things.

Most wealthy countries achieve universal coverage in one of three ways: a tax-funded national service employing providers directly, as in the United Kingdom; compulsory social insurance through regulated non-profit funds, as in Germany and Japan; or a single public insurer paying private providers, as in Canada. India runs a mixed system with large private out-of-pocket spending alongside expanding public insurance schemes. All of them ration in some way — by waiting time, by covered benefit, by capacity — and the American system rations too, mainly by cost and coverage status. The honest framing is not that one system rations and another does not, but that each chose a different rationing mechanism and a different administrative burden.

What this means for the work

Three practical consequences for anyone building software here.

Eligibility is a moving target. Coverage changes constantly, so any system holding a patient's insurance details needs a re-verification path, and any analytics using coverage data must handle people appearing and disappearing.

The same clinical event produces several different records. A hospital's record, the physician's record, the claim, the remittance, the pharmacy fill, and the patient's own bill. They will not agree with each other, and reconciling them is a large fraction of healthcare data engineering.

And the payer, not the doctor, is often the decision point. A treatment is available when it is covered. That single fact is why so much of the industry's software effort goes into coverage, authorisation and claims rather than into clinical care itself.

Next: Chapter 25.28, how the money actually moves — revenue cycle management, the coding systems, claims, denials and prior authorisation, in operational detail.