Is autism therapy becoming free childcare? Follow the money.

In our last piece on autism, we showed that most of the rise in diagnoses, from 1 in 150 children in 2000 to 1 in 31 in 2022, tracks changes in how autism is defined and how hard we look. One of the reasons we listed was simple: a diagnosis opens the door to help.

That door now has a price tag. This piece follows the money.

What ABA is, and what it costs

Applied behavior analysis (ABA) is the most widely funded therapy for autistic children. A trained technician works one-on-one with a child on skills like communication, play, toileting and following instructions, under a supervising analyst (a BCBA). It's billed in 15-minute units, mostly under a single code, 97153.

ABA is prescribed by the hour, and the hours can be enormous. "Comprehensive" programs run 20 to 40 hours a week, roughly a full-time job for a three-year-old. In 2025, state Medicaid programs paid between $42 and $91 an hour for the technician's time alone. At those rates, 30 hours a week for a year costs more than many people earn.

Every state now requires state-regulated private insurance plans to cover autism treatment; Tennessee was the last, in 2019. Medicaid followed: in 2014 the federal government told states that autism services for children must be covered, and by 2022, when Texas added it, every state's Medicaid program was reported to be paying for ABA.

The money

The federal Medicaid agency, CMS, published its own numbers in August 2026. Between 2021 and 2025:

Medicaid and CHIP spending on ABA rose from $1.94 billion to $10.1 billion, up 421%.
The number of children receiving ABA rose 189%.
The number of beneficiaries with an autism diagnosis rose 67%.
ABA billed for people without an autism diagnosis rose from $77.6 million to $1.47 billion.

The money grew faster than the children. Increase from 2021 to 2025 across US Medicaid and CHIP. Spending went from $1.94 billion to $10.1 billion; average therapy per child rose 22%, to 17.3 hours a week.Source: CMS, State Medicaid & CHIP ABA Toolkit, August 2026 (T-MSIS claims data).

Spending grew more than twice as fast as the number of children getting therapy. Part of the gap is more hours per child: the average rose 22%, to 17.3 hours a week.

The state numbers are steeper still.

What Medicaid pays for ABA, then and now. Millions of dollars a year. North Carolina figures are state fiscal years; Colorado’s are fee-for-service payments only, so the true total is higher.Source: Indiana FSSA, Nov 2025; NC DHHS to the Joint Legislative Oversight Committee on Medicaid, Mar 2026; HHS OIG A-09-24-02004 (Colorado).

Indiana spent about $17 million a year on ABA when coverage began in 2015–17. By 2023 it was $611 million, an increase its own Medicaid agency called "almost 3,000% in six years."
North Carolina went from $122 million to $544 million in three years, and projects more than $1.1 billion by 2027. Its health department told legislators the growth was "unlikely" to be "explained by increased access alone." 70% of the money went to children aged 1 to 6.
Colorado's fee-for-service payments nearly quadrupled between 2019 and 2024. In one case federal auditors examined, payments for a single child rose from $16,000 to $144,000 a year.

Private insurance shows the same shape. Among privately insured children aged 3 to 7 with autism, spending on outpatient behavioral therapy rose from $1,746 to $8,317 per child between 2011 and 2017.

Where the money has gone wrong

The federal Inspector General has audited Medicaid ABA payments in four states so far. In every one, all 100 sampled months of billing contained improper or potentially improper payments.

Indiana: at least $56 million improper.
Colorado: at least $77.8 million improper, much of it for therapy by technicians who weren't properly credentialed.
Maine: at least $45.6 million improper. Session notes counted lunch and naps as therapy time in 92 of the 100 samples.
Wisconsin: at least $18.5 million improper.

The money has also drawn investors. Private equity firms owned 574 autism centers in 42 states by 2024, most bought between 2018 and 2022. States with higher autism rates were more likely to have them.

A daycare by another name?

Action Behavior Centers (ABC) is a good example of the model, because it's the largest. Founded in Austin in 2016 and owned since 2022 by the private equity firm Charlesbank, it operates hundreds of centers in nine states.

By its own description, ABC's center-based program is "for children ages 18 months to 8 years." Most children receive "20–40 hours per week," in sessions of "several hours each," at centers open "Monday through Friday during regular business hours," year-round. The stated goal is school readiness. (ABC also runs an after-school program for 5- to 12-year-olds and, in Texas, home-based therapy up to age 13.)

Look at that schedule from a parent's side: a toddler is looked after by trained staff at a center, all day, every weekday, and insurance pays. Whatever the clinical content, that is the shape of a working day's childcare.

That's not only our observation. The federal Medicaid agency says it too. In its 2026 guidance to states, CMS wrote that "forty hours of ABA per week is not a best practice," and that

"these high-intensity weekly treatment schedules leave virtually no time for non-treatment activities like schooling and family time and can easily take on the characteristics of respite care and childcare."

CMS also notes that more hours don't reliably mean better results: "high levels do not equate to better outcomes compared to lower levels." It suggests about 10, 20 and 30 hours a week for children at the three levels of autism severity, and tells states to watch for treatment plans that never change, a sign ABA may be "functioning primarily as respite or childcare."

The Wall Street Journal's March 2026 investigation of the industry reported that ABC billed Colorado's Medicaid program for some of the most hours per patient of any provider, about 33 hours a week per child. It also reported that employees could earn bonuses of up to $15,000 a year tied to hours of treatment delivered, and that some former employees said they felt pressure to recommend high-hour plans. Those are the Journal's findings and former employees' accounts, not findings by any regulator. We could not find a public response from ABC.

None of this means the children in these centers aren't autistic, or that the staff aren't helping them. It means the payment system rewards hours, and the cheapest way to deliver many hours is a full-day center for young children.

So are parents chasing the label?

This is the question at the heart of the claim, and here the evidence gets more complicated than the money.

What supports it:

Coverage does increase diagnoses. After states required private insurers to cover autism treatment, the number of children diagnosed rose about 13%, and about 18% by the third year. That's a real response to money.
The pattern of who gets diagnosed flipped. Through the 2000s, autism was diagnosed most often in children from wealthier neighbourhoods. By 2020, the CDC found the opposite: rates were slightly higher in lower-income neighbourhoods, and white children had the lowest rate of any group. That's the period when state Medicaid programs, which cover lower-income families, were adding ABA. The CDC's 2022 report notes that "another reason for differences in prevalence could be whether children have insurance coverage."
Clinicians elsewhere describe it. In Australia, where disability funding has been tied to an autism diagnosis, the Royal Australian College of General Practitioners told a 2025 parliamentary inquiry it had seen a "steady increase in higher levels of autism diagnoses" since the scheme began, and that tying funding to diagnosis encourages families to seek them.
Paying parents has been tried. In Minnesota, federal prosecutors have charged a provider with paying parents kickbacks of hundreds of dollars a month to enroll their children in Medicaid autism services, some of whom allegedly had no autism diagnosis.

What cuts against it:

No study has found parents seeking a diagnosis for childcare. Research on why parents pursue a diagnosis finds worry, wanting answers and wanting help. Nobody has measured a childcare motive, and nobody has yet tested whether Medicaid paying for ABA caused the income flip. CMS says it's still studying diagnosis trends itself.
The timing is messy. The income pattern was already shifting by 2018, while many states only started paying for ABA between 2016 and 2022. Better outreach to under-served families, which the CDC documents, also explains it.
Most diagnosed children get no ABA at all. In 2025, only 14% of Medicaid beneficiaries with an autism diagnosis received any ABA. If a diagnosis were a ticket to free childcare, far more would be using it. Waitlists for therapy commonly run months.
It isn't free for families. ABA programs expect parents to take part, and Virginia, for one, requires direct family involvement at least weekly and refuses to pay for sessions used for "respite or child care." Mothers of autistic children earn on average 56% less than mothers of children without health conditions. These families tend to lose working hours, not gain them.
The fraud found so far is on the provider side. Every audit and prosecution above targets companies billing for hours, unqualified staff or non-autistic children. Where parents were involved, in Minnesota, providers allegedly recruited and paid them.

Does all this therapy work?

Honestly, less well than the spending suggests. A 2018 Cochrane review found only "weak evidence" that early intensive behavioral intervention helps some children, based on just five studies. A 2023 review of 252 studies in the BMJ found that most apparent benefits disappeared when outcomes were measured by people who didn't know which children had been treated.

The US military's insurer, TRICARE, has tracked thousands of children in ABA. In 2019 it told Congress that 76% showed little or no change in symptoms after 12 months, and 9% got worse. Its 2026 report found average improvements of about 5% and said it had "no information to know if these changes are clinically significant." (In 2025 the National Academies disagreed with TRICARE's reading and concluded ABA does meet its evidence standard. The debate is live.)

ABA may well help many individual children. But "40 hours a week" is not what the evidence supports, and the government paying for it now says so.

The bottom line

The money is real. Medicaid's ABA bill grew fivefold in four years, faster than the number of children, and some of the biggest providers run full weekday centers for toddlers that, by the federal government's own description, can work like childcare. A payment system that rewards hours will produce hours.

But the claim that parents are driving the rise in autism diagnoses to get free childcare isn't supported by evidence yet. Coverage nudges diagnosis rates up, and that deserves study. Yet most diagnosed children get no ABA, the families involved generally lose income, and the abuse found so far has come from the companies doing the billing.

That's why our verdict is Unproven. The questions to ask are about the system: who diagnoses, who prescribes the hours, and who profits from both.

If you're the parent of an autistic child weighing ABA, ask the provider how many hours they recommend and why, how they'll measure progress, and when the plan will be reviewed. A good program will have clear answers.

Sources

Share this fact-check

← All articles