US Healthcare Costs: Why Americans Are Delaying or Skipping Care

The Centers for Medicare & Medicaid Services (CMS) tracks national health spending through its National Health Expenditure Accounts (NHE) — the government’s authoritative count of every dollar spent on healthcare in the US, from hospital bills to insurance premiums to out-of-pocket copays. The most recent complete data, for 2024, put total national health expenditure at $5.3 trillion, or $15,474 per person, equal to 18.0% of US GDP — up from 17.6% in 2023. NHE data lags by roughly a year because it takes time to compile complete claims and spending records across the entire system; 2024 is the newest full-year figure as of this writing.

That single national number is not the same thing as what any individual household pays. National health expenditure includes hospital and physician spending, prescription drugs, nursing care, government public health activity, and the administrative cost of running insurance — most of it paid by employers, insurers and government programs, not directly out of a patient’s pocket. Household out-of-pocket spending — copays, coinsurance, deductibles and uninsured medical bills paid directly by patients — is a separate, smaller slice of that total, and it is the slice most directly tied to whether someone decides to seek care. Confusing a hospital’s list price, a national spending total, and a single patient’s final bill is one of the most common errors in healthcare reporting; this article keeps them separate throughout.

Health economists generally split “why healthcare costs so much” into three distinct questions, and conflating them is a common source of bad analysis: price (how much does each service cost), utilization (how much care do people actually use), and mix (what kind of care — routine versus specialist, generic versus brand drug, outpatient versus inpatient). Research comparing the US with peer nations, including work cited by the Commonwealth Fund and OECD, consistently finds that Americans do not use dramatically more healthcare than people in other wealthy countries — in some categories, like average length of hospital stay, the US is lower. What differs is price: US hospitals, physicians and drug manufacturers charge more per service than their counterparts elsewhere, and no single factor explains all of it.

Having health insurance caps financial risk — it protects against the worst-case, catastrophic bill. It does not make routine care free, and the gap between those two things is where cost-driven delay actually happens. Six terms do most of the work:

It would be inaccurate to say Americans are “refusing” doctors, and it would be inaccurate to say every skipped appointment is about money. Survey and health-services research point to a mix of overlapping reasons, several of which compound each other:

About one-third (36%) of adults say that in the past 12 months they have skipped or postponed getting health care they needed because of the cost — this exact figure and wording comes from KFF’s health-tracking polling, most recently reaffirmed in KFF’s April 2026 poll, “Health Care Costs, Expiring ACA Tax Credits, and the 2026 Midterms.” The question asks broadly about “needed” care in the prior 12 months — it is not limited to doctor visits, and it does not mean 36% of adults never see a doctor. The same polling found that 75% of uninsured adults under 65 went without needed care because of cost, that 18% of adults say their health got worse as a result of skipping or delaying care, and that 64% of the public is at least somewhat worried about affording healthcare costs, including insurance and out-of-pocket costs like office visits and prescriptions. The same poll found roughly 90% of respondents say healthcare costs will influence how they vote in the November 2026 midterms.

KFF is a nonpartisan health policy research and polling organization; its health-tracking poll is fielded periodically among a nationally representative sample of US adults. 36% should not be rewritten as “36% of Americans don’t see doctors” — that would misstate a broad, any-type-of-needed-care measure as a claim about routine doctor visits specifically, which the data does not support.

In 2025, 26% of American adults skipped some form of medical treatment because they could not afford it, according to the Federal Reserve’s Survey of Household Economics and Decisionmaking (SHED), published in the Fed’s Economic Well-Being of U.S. Households report and reported by USAFacts on May 18, 2026. This is down from 28% in 2024 and is the lowest share since 2022. The SHED measure asks specifically about “some medical treatment” skipped due to affordability in the survey year — a narrower, more specifically cost-tied question than KFF’s broader “needed care” measure.

The two numbers — KFF’s 36% and the Fed’s 26% — are not contradictory. They measure related but distinct things: KFF asks about any needed care skipped or postponed over 12 months for any reason tied to cost; the Fed asks specifically about medical treatment not obtained because a household could not afford it, in a single survey year. Different question wording, different reference periods and different survey populations reliably produce different percentages even when both are measuring real, overlapping phenomena. Neither number is wrong; they simply are not the same measurement.

The Fed data also shows a sharp income gradient: 38% of adults in households earning under $25,000 skipped medical treatment in 2025, compared with 13% of those earning over $100,000 — a nearly threefold gap that is one of the clearest single predictors in the dataset.

A deductible is why a technically-insured patient can still face a real financial barrier before coverage pays substantially. In 2025, the average general annual deductible for single coverage in employer plans was $1,886 — up 17% over five years and 43% over ten. Roughly 34% of covered workers are in a plan with a single-coverage deductible of $2,000 or more, and 88% of workers with single coverage have some deductible that must be met before most services are paid by the plan.

Not every deductible is unaffordable, and high-deductible plans paired with a Health Savings Account (HSA) can be a reasonable trade-off for a healthy person with cash reserves — lower premiums in exchange for more first-dollar exposure. The affordability problem is concentrated among households without savings to cover that gap, and it is why the timing of care — deciding whether to go now, early in a plan year with an unmet deductible, or wait — can become an explicitly financial decision rather than a purely medical one. ACA marketplace plans and employer plans both use deductible structures, though exact averages differ by plan type and metal tier.

Prescription drug affordability shows up in patient behavior more directly than almost any other cost category, because a prescription is a recurring, visible price rather than a one-time bill. KFF’s tracking data found 27% of adults did not fill a prescription, 19% cut pills in half or skipped doses, and 31% took an over-the-counter drug instead of filling a prescription, all due to cost in the past 12 months. In total, 43% of adults took at least one cost-related medication shortcut — up from about a third in the prior year’s tracking. Lower- and middle-income households report these behaviors at meaningfully higher rates than higher-income households.

Brand-name drugs, complex formulary rules, pharmacy-network restrictions and drug-specific coinsurance (rather than a flat copay) all compound the price a patient sees at the counter. This article does not offer guidance on any individual medication decision — patients concerned about medication cost should talk to a prescriber or pharmacist about generic alternatives, manufacturer assistance programs or formulary options, not stop or ration a prescription without medical guidance.

The most recent nationally representative federal estimate — Survey of Income and Program Participation data from December 2021, analyzed by KFF-Peterson Health System Tracker and published February 2024 — found that people in the US owe at least $220 billion in medical debt, held by roughly 20 million people (nearly 1 in 12 adults). About 14 million people (6% of adults) owe more than $1,000, and about 3 million (1%) owe more than $10,000. Separately, KFF’s Health Care Debt Survey found roughly four in ten adults report some form of healthcare-related debt under a broader definition that includes credit cards used for medical bills and informal borrowing from family.

The Consumer Financial Protection Bureau (CFPB) separately estimated $88 billion in medical bills sitting on consumer credit reports, based on June 2021 data published in March 2022, affecting about 15 million Americans. POLICY UPDATE: the three major credit bureaus voluntarily agreed in 2022 to remove paid medical collections regardless of balance, unpaid medical collections under $500, and to apply a 12-month grace period before any medical debt appears on a report — those voluntary changes remain in effect. A subsequent CFPB rule that would have banned medical debt from credit reports entirely was finalized but then vacated by a federal court in the Eastern District of Texas on July 11, 2025, at the CFPB’s own request after new agency leadership paused its implementation earlier that year. As of 2026, there is no federal ban on medical debt appearing on credit reports; roughly 15 states have their own restrictions, and the CFPB issued an interpretive rule in October 2025 asserting that federal law preempts those state protections — a legal question still being contested.

Delaying care can, in some cases, allow a condition to progress to a point where treatment is more complex, more expensive, and requires a longer recovery — a routine issue caught early may need a simple intervention, while the same issue caught late may require hospitalization. But this is not a universal law. Many conditions do not worsen meaningfully over weeks or months of delay, and some care can safely be postponed without consequence. The honest claim, and the one supported by the evidence in this article, is narrower than “delaying care always costs more.”

Some patients who delay routine or primary care eventually use an emergency department once a condition becomes acute — emergency rooms are legally required to treat emergency conditions regardless of ability to pay (under EMTALA), which makes them a backstop, not a substitute, for primary care. It would be inaccurate to characterize all emergency-department use as inappropriate: emergency care is essential, and many ER visits reflect genuine emergencies unrelated to any prior delay. Where it does connect to affordability is cost and continuity: ER care is typically billed at a higher rate than an equivalent primary-care visit, and network status and insurance rules can add further billing complexity, especially if the ER or its treating physicians are out-of-network.

A “surprise bill” is a bill a patient did not expect and could not reasonably have avoided — most commonly, care from an out-of-network provider at an in-network facility (an anesthesiologist or radiologist a patient never chose, for instance), or emergency care at an out-of-network hospital. The federal No Surprises Act, effective since January 2022, bans balance billing patients directly in these situations: patients pay only their normal in-network cost-sharing, and the insurer and provider settle the remaining payment amount between themselves, either by negotiation or through an independent dispute resolution (arbitration) process.

On August 12, 2026, the full, en banc 5th US Circuit Court of Appeals (based in New Orleans) ruled that part of the formula regulators use to calculate the Qualifying Payment Amount (QPA) — the benchmark insurers use as their starting offer in No Surprises Act billing disputes — was unlawfully calculated. The court found federal agencies improperly let insurers include “ghost rates” (contracted rates for services a provider never actually delivered) in QPA calculations, and improperly excluded bonus and incentive payments from the calculation — both features that tended to lower the benchmark price insurers offered providers. The ruling sided with healthcare providers, including the Texas Medical Association and air-ambulance operators, who argued the formula systematically underpaid them. Insurers won one narrower point: they can still exclude one-off, single-case agreements — such as certain air-ambulance contracts — from the contracted rates used to calculate the QPA.

What the ruling does: it requires federal agencies to revise how the QPA benchmark is calculated for insurer-provider billing disputes. What it does not do: it does not repeal the No Surprises Act, and it does not change what a patient can be billed directly — patients remain protected from surprise balance bills regardless of how this business-to-business formula is eventually recalculated. The court’s own majority opinion said the change need not cause “all-out chaos,” since agencies can continue using the current formula temporarily while a replacement is developed. This is a dispute over how insurers and providers split a bill between themselves, not a change to patient protections — and it should not be read as legal advice for any specific billing dispute.

Federal price-transparency rules now require hospitals to publish negotiated rates and give patients “good-faith estimates” for scheduled care, and insurers must provide comparable cost-estimate tools. In practice, transparency helps most when a patient has a real choice of provider, time to shop before a non-emergency procedure, and a comparison tool that is actually usable — none of which apply in an emergency, and not all of which apply even for routine care in areas with few competing providers. Compliance with hospital price-posting rules has also been uneven since the requirement began. The fair assessment is that price transparency is a necessary but not sufficient tool: it has not solved affordability on its own, and it has not failed either — it works best as one piece of a larger set of changes, not a standalone fix.

KFF’s 2026 polling found health costs are now tied with gasoline and transportation as the top affordability worry for US households, ahead of many other categories. Survey evidence consistently shows a share of households — cited at roughly half in some 2026 polling — report cutting back on other essentials, including groceries, utilities or savings contributions, to cover healthcare costs in a given year. This article does not fabricate an individual household’s budget; the honest framing is the “ordinary decision” itself: do I make the appointment now, or wait until payday? — illustrative framing, not a reported quote from any specific person.

Healthcare spending moves through several layers before it reaches a provider: a patient (or their employer, on their behalf) pays premiums to an insurer; the insurer negotiates rates with and pays healthcare providers; government programs (Medicare, Medicaid, ACA subsidies) pay or subsidize premiums and care for eligible populations; and taxes fund the government side of that chain. At the point of care, the patient separately owes deductibles, copays and coinsurance directly. Pharmaceutical and medical-device manufacturers sit further upstream, setting list prices that insurers and pharmacy-benefit managers then negotiate down to varying degrees — which is part of why the same drug can have wildly different final prices depending on a patient’s specific insurance plan.

ACA subsidies: the enhanced premium tax credits first introduced during the pandemic and extended through 2025 were not renewed by Congress, and expired at the end of 2025. The “subsidy cliff” has returned: households with income above 400% of the federal poverty level lose eligibility for premium tax credits entirely once they cross that line. KFF estimates subsidized marketplace enrollees will pay 114% more on average for premiums in 2026 (roughly $888 to $1,904 per year), and projects average effectuated marketplace enrollment could fall from 22.3 million in 2025 to somewhere between 16.5 and 17.5 million in 2026 — a decline of roughly 5 million people, concentrated heavily among households just above the 400% FPL threshold, where sign-ups fell about 44%.

Uninsured rate: CDC’s National Health Interview Survey found the uninsured rate held steady at 8% in 2025 (11.6% among adults 18-64), though the number of uninsured people grew by about 800,000, including 300,000 children, even as the rate stayed flat — a sign of population growth outpacing coverage growth. KFF and other analysts expect the uninsured rate to rise in 2026 and beyond as the effects of expired ACA subsidies and separate Medicaid eligibility changes show up in survey data; the Census Bureau’s more comprehensive American Community Survey data for this period was not yet available as of this writing.

Hospital financial pressure: a rising uninsured population can increase uncompensated care costs for hospitals, which may in turn pressure hospital finances — this has been reported as a risk facing the sector in 2026 commentary, though it does not apply uniformly to every hospital, and this article does not claim it as a proven, system-wide outcome.

Surprise billing: see the dedicated section above on the August 12, 2026 5th Circuit ruling on the QPA formula.

Artificial intelligence is already being deployed against several of the cost drivers described above, mostly on the administrative and operational side rather than the clinical one. AI-assisted scheduling and patient navigation can reduce missed appointments and no-shows. Automated clinical documentation can cut the time clinicians spend on billing-related paperwork, one contributor to administrative overhead. Telehealth platforms, increasingly AI-assisted for triage and intake, can lower the cost and time barrier for some primary-care and mental-health visits. AI-assisted diagnostic tools are being piloted in imaging and pathology to help clinicians work faster, and AI-driven billing-automation tools are being used by both providers and insurers to reduce claims-processing costs.