U.S. health spending grew fastest for the highest-income group

The discovery

The researchers asked whether per-person U.S. health-care spending changed at the same rate across income groups from 2005 through 2023, and whether overall national trends concealed a growing difference between the highest and lowest income fifths.

The research question and why it matters

A 2016 Health Affairs analysis reported that spending differences among high-, middle- and low-income Americans narrowed from 1963 to 1999 and then widened through 2012. The new study uses the same federal survey family to extend that question through 2023 and separates the years around Affordable Care Act expansion and the 2018 survey redesign.

What researchers found

Across 2005–2023, adjusted real spending grew 1.8% per year in the highest-income quintile and 0.0% in the lowest-income quintile, a statistically significant difference. The direction briefly narrowed after the Affordable Care Act's major coverage expansions: from 2013–2017, growth was 4.4% a year in the lowest quintile and 0.7% in the highest. It diverged again from 2018–2023, when spending rose 2.5% a year at the top and fell 2.5% at the bottom. The adjusted per-person gap exceeded $4,700 in 2023.

Results at a glance

Key results from the tested systems

1.8% vs 0.0%

annual adjusted growth, 2005–2023

Spending grew in the highest income fifth and showed no growth in the lowest fifth after adjustment for age and health measures.

>$4,700

adjusted 2023 gap

The estimated per-person difference between the highest and lowest income quintiles exceeded $4,700 in 2023.

+2.5% vs −2.5%

annual growth, 2018–2023

The highest-income group's spending increased while the lowest-income group's spending declined during the most recent modeled period.

How the research worked

The team analyzed annual public-use data from the Medical Expenditure Panel Survey Household Component, which represents the civilian, noninstitutionalized U.S. population. People were assigned to income quintiles using family income relative to the federal poverty level. Generalized linear models estimated inflation-adjusted total, outpatient, inpatient, emergency-department and prescription-drug spending. The models accounted for survey design and adjusted for age, self-reported health, chronic conditions and documented survey changes, with separate growth periods for 2005–2013, 2013–2017 and 2018–2023.

Subjects or systemHuman

Research designRepeated cross-sectional analysis of a nationally representative federal household survey

Evidence basePublic-use person records from the 2005–2023 Medical Expenditure Panel Survey Household Component. AHRQ's annual sample tables total 580,031 person-year records before any study-specific exclusions; the final analytic record count was not stated in the accessible article text.

How to interpret this design

The design determines what kind of conclusion the evidence can support. Direct measurement strengthens the reported observation, while generalization beyond the tested subjects, material, place or conditions requires additional evidence.

Because the research involved people, it speaks directly to the participants and outcomes measured. It may still apply differently to people outside the eligibility criteria, age range, clinical setting, geography or follow-up period.

What strengthens or limits the finding?

Nineteen years of nationally representative survey data, survey weights, health and age adjustment, sensitivity analyses and reported confidence tests support the descriptive trend. The analysis cannot determine whether the gap reflects prices, use, unmet need or another cause, and survey changes complicate comparisons across the full period.

The result is meaningfully informative, but identifiable limitations could alter the size, reach or causal interpretation of the finding.

Funding and disclosure context

The launch record does not yet reproduce a complete funding statement; readers should consult the paper's declaration. The complete conflict-of-interest declaration should be checked in the original publication rather than inferred. Funding or a disclosed relationship does not by itself invalidate a result, but it is relevant when judging design choices, analysis and the need for independent replication.

What it means

The national spending slowdown was not shared evenly. Because the largest differences appeared in outpatient care and prescription drugs rather than emergency or inpatient care, the pattern is consistent with income-related differences in routine access, service use, prices or all three. Spending is not the same as health or quality, however, so the analysis signals a disparity that needs explanation rather than proving that lower spending caused worse outcomes.

Beyond the abstract

Deeper analysis

The middle years are an important counterexample

The full-period averages hide a temporary reversal. From 2013 through 2017, after major Affordable Care Act coverage expansions, spending grew more quickly in the lowest-income group and the gap narrowed. That does not isolate the policy's causal effect, but it shows that the long-term divergence was not an unbroken or inevitable trend.

Where the dollars diverged changes the interpretation

The strongest differences appeared in outpatient services and prescriptions. Inpatient and emergency spending did not follow the same consistent income gradient. That pattern makes access, insurance payment and discretionary use plausible explanations, but the dataset cannot distinguish a useful visit from a low-value one or identify care that never occurred.

Spending is a system measure, not a health score

More spending can represent better access to beneficial treatment, higher prices for the same treatment, greater illness, unnecessary care or a mixture of these. The study's value is that it reveals how differently the system's dollars moved across income groups. Determining whether those dollars translate into fair access and better outcomes requires linked clinical and utilization research.

Keep the claim in proportion

What it does NOT prove

It does not prove that lower-income people received less necessary care. The survey measures payments and service use, not whether each service was medically needed or beneficial.
It does not show that income itself caused the spending divergence. Insurance type, reimbursement, geography, availability, affordability and other factors may contribute.
It does not separate higher prices from greater use within the reported growth rates, so a larger dollar total cannot be interpreted as a larger quantity of care.
It does not represent people living in institutions such as nursing homes, and it should not be used to make individual treatment decisions.

Important limitations

MEPS relies partly on self-reported and imputed information and does not capture uncompensated care, over-the-counter drugs or people in institutions.
The survey instrument changed in 2007 and was redesigned in 2018; the authors adjusted for those changes and did not estimate growth across 2017–2018, but measurement may still differ over time.
Pandemic-era survey operations had lower response rates, and uncommon events such as hospitalizations become imprecise when divided into income quintiles.
The models adjusted for age, self-reported health and chronic conditions, but residual differences in health need and other unmeasured factors remain possible.
The design is descriptive and cannot determine whether the post-2018 divergence came from affordability barriers, reimbursement differences, changing service mix or another mechanism.

How this fits with previous research

Questions still unanswered

How much of the recent divergence is caused by different prices versus different quantities or types of care?
Which services—preventive care, chronic-disease management, specialty visits or high-cost medicines—account for the largest income differences?
Does lower spending among lower-income groups reflect unmet need, and how does it affect health outcomes over time?
How much do insurance design, state policy, provider supply and local prices change the pattern?

Government verification and context

Relevant U.S. government resources

These resources serve different purposes. A registry can verify what researchers planned, a repository can locate government-funded work, and an agency page can supply authoritative background. None automatically proves that this paper's conclusion is correct.

Study verificationNational Library of Medicine / NIH

PubMed record search ↗

Federal biomedical-index search keyed to this paper's DOI or title. It can confirm indexing and expose linked identifiers when a record is available; the journal paper remains the primary source.

Authoritative contextNational Institutes of Health

Understanding clinical research ↗

NIH background on how clinical research is designed, reviewed and interpreted. This is contextual guidance, not independent confirmation of the study's result.

Reuse note: Facts and discoveries are summarized here in original language. We link to government material instead of copying it wholesale, and we do not reuse agency logos, photographs, charts or third-party material unless the specific reuse rights are verified.

Sources and provenance