A Single Personal Care Services Code Dwarfs All of Medicaid's Clinical Spending
New York State alone billed $71.7 billion in personal care services under a single Medicaid procedure code from 2018 through 2024. That's 61.31% of the entire national total for that code, from a state with roughly 6% of the U.S. population.
The code is T1019, which covers hourly personal care aide services. And the numbers around it reframe what Medicaid actually pays for.
One Code Outspends All of Clinical Medicine
From 2018 through 2024, T1019 generated $122.7 billion in Medicaid spending across 1.1 billion claims. That figure exceeds the combined total of every office visit (codes 99202 through 99215), every surgical procedure (codes 10000 through 69999), and every J-code drug injection billed to Medicaid over the same period, by more than $18.7 billion.
To put that in concrete terms: all Medicaid office visits combined totaled $78.5 billion. All surgical CPT codes added $14.7 billion. All J-code drug injections added $10.8 billion. Together, those three categories represent the core of what most people think of as medical care. T1019, a code for aide hours, outspent all three combined.
For taxpayers and program administrators, this is a structural fact about Medicaid that rarely surfaces in public debate. The program's largest single line item isn't a blockbuster drug or a complex surgery. It's hourly aide time, billed at scale, concentrated in a small number of states.
The New York Effect
The geographic concentration behind T1019 is as striking as the total. New York's $71.7 billion in T1019 spending from 2018 through 2024 represents 61.31% of the national total. The second-highest state, Massachusetts, accounted for 6.98%. The top five states (New York, Massachusetts, Missouri, New Jersey, and Virginia) collectively covered 80.27% of all national T1019 spending.
New York's per-enrollee spending of $9,578.32 was more than 7 times the national average of $1,299.60. The gap between states is even wider when you look at the full distribution. Texas, with 5.7 million Medicaid enrollees, spent $331.77 per enrollee on T1019. The District of Columbia, with 262,771 enrollees, spent $5,130.42 per enrollee. That's a 15-fold difference between two jurisdictions operating under the same federal program rules.
| State | T1019 Spending | Per-Enrollee Spend | Share of National Total |
|---|---|---|---|
| New York | $71.7B | $9,578 | 61.31% |
| Massachusetts | $8.2B | $4,201 | 6.98% |
| Missouri | $5.9B | $4,637 | 5.05% |
| New Jersey | $4.1B | $1,896 | 3.50% |
| Virginia | $4.0B | $2,109 | 3.42% |
These disparities matter because Medicaid is a federal-state partnership with shared financing. When one state's per-enrollee spending on a single code is more than 28 times another's, the federal government is effectively subsidizing very different program designs, not just different populations.
States That Weren't in the Picture Are Now
The concentration in a few large states doesn't mean the rest of the country is static. Alabama's T1019 spending grew from $36.8 million in 2018 to $136.8 million in 2024, nearly a fourfold increase. Arkansas grew from $3.7 million in 2018 to $84.2 million in 2023, a 23-fold increase in five years.
Both states started from a low base, so their share of national spending remains small. But the trajectory is clear: states that historically billed little for personal care services are now building programs at scale. If that pattern continues across more states, the national T1019 total, already larger than all Medicaid clinical spending combined, will grow substantially.
The question that the data raises but doesn't answer is whether New York's per-enrollee T1019 spending of $9,578.32, more than 28 times Texas's $331.77, reflects differences in eligibility criteria, reimbursement rates, program design, or the underlying health needs of enrollees, and whether the states spending more are actually delivering better outcomes for the people receiving those aide hours.
Ask a follow-up
Run one of these against the same data. The question is filled in for you.
Explore the data yourself
Run your own queries against 240M+ rows of federal health data using natural language — powered by AI.
Start analyzing