Health and Care: what the evidence shows

The evidence on Health and Care in the United States is strongest on what is spent and who is covered, and weakest on why outcomes differ and which reforms would change them. Federal data show $5.3 trillion in health spending in 2024, 92.0% of people insured for some or all of the year, and life expectancy of 79.0 years, the highest on record. Studies of what insurance does for health disagree in ways that matter for policy, and much of what people most want to know, such as which reforms would lower costs without harming care, remains unsettled.

In plain words

Some facts about health care are well measured. We know roughly how much the country spends, how many people have insurance and how long people live.

Other questions are much harder. Does giving people insurance make them healthier? The best studies give different answers depending on what they measure and for how long. Why does the United States spend more than other rich countries? Researchers point to prices, to how care is organized, to the health of the population and to many other factors, and they weigh them differently.

This page sets out what is known, where good researchers disagree, and what nobody knows yet. It is a companion to the main Health and Care page and the Who decides page.

Defined terms

National health expenditure (NHE). CMS’s estimate of total U.S. spending on health care and related activity.

Uninsured rate. The share of people without health insurance. The Census Bureau reports coverage for “some or all of the year,” which differs from coverage at a single point in time.

Life expectancy at birth. The average number of years a newborn would live if current age-specific death rates persisted. It is a summary of current mortality, not a forecast for any person.

Age-adjusted death rate. A death rate standardized to a fixed population age structure, so that years with different age mixes can be compared.

Randomized controlled trial. A study in which chance decides who receives an intervention, so differences in outcomes can be attributed to the intervention.

Quasi-experimental study. A study that uses a policy change or natural variation (for example, some states expanding Medicaid and others not) to estimate effects when randomization is not possible.

Evidence on health and care spending

U.S. health spending data come mainly from the National Health Expenditure Accounts maintained by CMS. For 2024, CMS reports total spending of $5.3 trillion, growth of 7.2% from the prior year, $15,474 per person, and 18.0% of gross domestic product.⁠¹

By payer, CMS reports for 2024: private health insurance $1,644.6 billion (31% of the total), Medicare $1,118.0 billion (21%), Medicaid $931.7 billion (18%), and out-of-pocket spending $556.6 billion (11%).⁠² By service, hospital care was $1,634.7 billion, physician and clinical services $1,109.7 billion, and prescription drugs $467.0 billion.⁠² CMS projects health spending to reach 20.6% of GDP in 2034.⁠²

International comparison requires care because definitions differ. Using OECD data, the Peterson-KFF Health System Tracker reports that the United States spent $14,775 per person on health in 2024, compared with an average of $7,860 among comparable high-income countries, and 17.2% of GDP compared with a peer average of 11.2%.⁠³ The 17.2% figure differs from CMS’s 18.0% because the OECD’s health accounting framework counts spending differently. Both measures agree on the main point: U.S. spending is far higher than its peers’.

Interpretation. Why U.S. spending is higher is contested. Some researchers emphasize higher prices for the same services; others emphasize administrative complexity, the mix of services, technology use, or underlying population health. These explanations are not mutually exclusive, and the weight given to each shapes which reforms seem promising.

Health insurance coverage: what the data show

The Census Bureau’s annual report on health insurance, based on the Current Population Survey, found that in 2024, 92.0% of people, or 310 million, had health insurance for some or all of the year.⁠⁴ Private coverage, at 66.1%, remained more common than public coverage, at 35.5% (people can hold both).⁠⁴ Between 2023 and 2024, private coverage rose 0.7 percentage points, driven by direct-purchase coverage, while public coverage fell 0.8 points, driven by a 1.3-point drop in Medicaid coverage.⁠⁴

Measurement caution. Coverage estimates vary across federal surveys because they ask about different time frames and use different methods. Being insured also does not guarantee access: a person can have coverage with high deductibles or limited provider networks. Coverage is a measure of financial protection, not of care received.

Health outcomes: what the data show

The National Center for Health Statistics reports that U.S. life expectancy at birth was 79.0 years in 2024, up 0.6 year from 78.4 in 2023.⁠⁵ The age-adjusted death rate fell 3.8%, from 750.5 to 722.1 deaths per 100,000 standard population.⁠⁵ The leading causes of death in 2024 were heart disease, cancer, unintentional injuries, stroke, chronic lower respiratory diseases, Alzheimer disease, diabetes, kidney disease, chronic liver disease and cirrhosis, and suicide, which replaced COVID-19 as the tenth leading cause.⁠⁵ The infant mortality rate was 552.5 deaths per 100,000 live births in 2024, not significantly different from 560.2 in 2023.⁠⁵

Interpretation. Mortality reflects much more than medical care. Several leading causes, such as unintentional injuries, suicide and chronic liver disease, are shaped heavily by behavior, economic conditions and public health policy. Comparing U.S. outcomes with those of other countries and attributing the gap to the health care system alone would overstate what the data show.

Mental health and substance use: what the data show

The National Survey on Drug Use and Health, conducted for the Substance Abuse and Mental Health Services Administration (SAMHSA), estimated that in 2024, 23.4% of adults, or 61.5 million people, had any mental illness in the past year, and 5.6%, or 14.6 million, had a serious mental illness.⁠⁶ Among people 12 and older, 16.8%, or 48.4 million, had a substance use disorder in the past year.⁠⁶ Among adults with any mental illness, 52.1% received mental health treatment in the past year; among adults with serious mental illness, 70.8% did.⁠⁶

Measurement caution. These are survey-based estimates from a structured interview, not clinical diagnoses. Definitions of “any mental illness” are broad, and treatment counts do not show whether the treatment was adequate or effective.

Caregiving: what the data show

AARP and the National Alliance for Caregiving estimated in their 2025 study that 63 million Americans are family caregivers, nearly 50% more than in 2015.⁠⁷ Over 40% provide high-intensity care, only 22% have received training for the tasks they perform, half report a negative financial impact, and seven in ten are employed.⁠⁷

Measurement caution. Caregiver counts depend on how a survey defines caregiving, and changes over time can reflect both more caregiving and greater recognition of it. The study is produced by advocacy and research organizations; its methods should be read alongside its conclusions.

Where researchers disagree: does insurance improve health?

The question of whether health insurance improves health is central to many policy arguments, and the strongest studies do not give one simple answer.

The Oregon Health Insurance Experiment. In 2008 Oregon used a lottery to fill a limited number of Medicaid slots for low-income uninsured adults, which created a rare randomized comparison.⁠⁸ Researchers found that in the first one to two years, Medicaid coverage increased outpatient visits, hospitalizations, prescriptions and emergency department use; reduced financial strain and “virtually eliminated” catastrophic medical expenditures; and “substantially reduced the prevalence of depression”.⁠⁸ It “had no statistically significant effects on blood pressure, cholesterol, or cardiovascular risk”.⁠⁸

Medicaid expansion and mortality. Miller, Johnson and Wherry linked federal survey data to death records and compared states that expanded Medicaid under the ACA with states that did not. They report that “individuals in expansion states experienced a 0.132 percentage point decline in annual mortality,” a 9.4% reduction relative to the sample mean, among the near-elderly adults they studied, with the effect driven by fewer disease-related deaths and growing over time.⁠⁹

How to read both. Supporters of coverage expansion read the two bodies of work together: Oregon shows clear financial and mental health benefits, and the mortality study suggests physical health benefits that take longer to appear than Oregon’s follow-up allowed. Skeptics emphasize that the only large randomized trial found no significant change in key physical health markers, and that quasi-experimental comparisons between states can be affected by other differences between those states, although the mortality study reports that placebo tests and alternative specifications support its result.⁠⁹ Both groups agree that coverage reduces financial strain and increases use of care. They disagree about how much better health that care buys, and therefore about cost-effectiveness.

Where researchers disagree: supply regulation

Certificate-of-need laws illustrate a different kind of dispute. NCSL reports that 35 states and Washington, D.C., operate CON programs as of January 2025.⁠¹⁰ Supporters argue the programs control costs by preventing excess capacity and protect access in underserved areas; opponents argue they shield incumbent providers and that there is little evidence they control costs or improve quality.⁠¹⁰ Because CON programs differ widely in what they review, studies of one state’s program may not apply to another’s.

Where researchers disagree: financing the future

The 2025 Medicare Trustees Report projects that the Hospital Insurance trust fund will be depleted in 2033, after which revenue would cover 89% of costs, and notes that the fund has not met the trustees’ test of short-range financial adequacy since 2003.⁠¹¹ Projections like these depend on assumptions about prices, use of services and the economy, and they change from year to year: the 2025 depletion date was three years earlier than the year before.⁠¹¹ Analysts disagree about whether the remedy should come mostly from revenue, from payment reductions, from benefit changes or from broader reform.

What is unknown about health and care

Much that matters for policy is not yet known.

• The long-run health effects of coverage for working-age adults, beyond the time frames that randomized studies have followed.
• Which cost-control strategies work without harming access or quality. Price regulation, competition, payment reform and supply controls each have supporters and mixed evidence.
• How much of the U.S. outcome gap with peer countries is due to the health system rather than to injuries, overdoses, diet, inequality and other factors.
• The true scale and cost of unpaid caregiving, which falls outside most spending accounts.
• How public trust in health institutions affects outcomes, including vaccination and treatment adherence, after the COVID-19 pandemic.

Competing interpretations

Health and care evidence supports several reasonable interpretations.

A spending-efficiency reading holds that the main problem is high cost relative to results, pointing to spending far above peer countries,⁠³ and life expectancy that, though at a record⁠⁵ is shaped by causes outside medicine.

A coverage-protection reading holds that the main benefit of public programs is financial security and access, supported by Oregon’s findings on financial strain and depression,⁠⁸ whatever the effect on physical markers.

A limits-of-medicine reading holds that many leading causes of death respond more to public health, economic and social conditions than to more medical care, and that spending priorities should shift accordingly.

A market-failure versus government-failure dispute runs through all three: some see high prices as the result of too little regulation, others as the result of regulation that limits competition. The CON debate is one instance.⁠¹⁰

Application to this framework

• My Vote Forward presents evidence alongside priorities so that a participant ranking Health and Care can see what is known and what is disputed. See Priorities.
• Evidence does not settle values. Even agreed facts, such as Oregon’s results, support different conclusions depending on how one weighs financial security against physical health and cost. See Tradeoffs.
• Numbers on this page carry dates, denominators and sources so they can be checked and updated. See Measurement.
• When a specific proposal reaches the framework, its own evidence should be presented with the same distinction between data, research findings and interpretation. See Proposals.

Unresolved questions

• How should a public resource present studies that disagree, without implying false balance or false certainty?
• Which outcome measures (mortality, financial strain, self-reported health, access) should matter most when people compare health proposals, and who should choose?
• How should evidence pages handle annual revisions, such as changing trust fund projections, without confusing readers?
• Can local health data be presented at a level that helps communities without exposing small groups? See Collective priorities.