Adam Omary and Jeffrey A. Singer
How Overdiagnosis, Perverse Incentives, and Absent Oversight Created Medicaid’s Fastest-Growing Spending Crisis
As we recently argued, the American healthcare system’s structure of open-ended reimbursement, subjective diagnostic criteria, and fee-for-service billing creates powerful incentives for psychiatric overdiagnosis across every condition. Nowhere have those incentives produced more dramatic consequences than in Medicaid-funded autism therapy.
Autism and Overdiagnosis
Autism spectrum disorder is a neurodevelopmental condition characterized by difficulties with social interaction and communication, restricted interests, and repetitive behaviors. Historically, autism was rare. But in the last two decades, autism diagnoses have risen across every wealthy nation that tracks them. In the United States, its prevalence was fewer than 1 in 150 people prior to the year 2000, according to the Centers for Disease Control and Prevention (CDC). The CDC’s most recent report on autism, published in 2025, places its prevalence at 1 in 31.
But autism is up to 91 percent heritable, meaning almost all of the variation in whether a person develops the condition is attributable to genetic factors. It is one of the most genetically determined psychiatric conditions known to exist. Such a heritable trait should have broadly stable prevalence in a population across generations. Changes in autism prevalence are therefore more likely to reflect changes in diagnostic sensitivity than genuine changes in the underlying condition.
The natural response is to ask how we can be sure the increase is not simply better screening continuing to identify previously missed cases. The answer lies in the screening itself. The CDC’s prevalence estimates use imprecise surveys such as the Social Responsiveness Scale (SRS), which asks parents to rate statements about their child, like “Would rather be alone than with others,” “Has difficulty making friends,” and “Is regarded by other children as odd or weird.” These items describe behavioral traits common to social anxiety, introversion, and ordinary shyness and cannot reliably distinguish autism. Yet high scores on such instruments are routinely used as proxies for clinical diagnosis in prevalence studies.
None of this diminishes the reality of autism for individuals with significant functional impairment. But we are almost certainly overdiagnosing autism on a massive scale. That conclusion would be troubling enough as a matter of public health, as every false or marginal diagnosis dilutes clinical attention and resources away from children with genuine, functionally impairing conditions. However, the overdiagnosis problem does not stop at the clinic door. In a system where diagnosis triggers entitlement to publicly funded services, diagnostic inflation becomes the first link in a chain that leads to wasteful spending and outright fraud.
The Spending Explosion
The primary service attached to an autism diagnosis is applied behavior analysis (ABA) — therapy for autism designed to teach communication and social skills. In the last decade, Medicaid spending on ABA therapy has defied any reasonable expectation anchored to the actual prevalence of disabling autism.
Across the eight states with publicly available ABA data included in this report—Indiana, Minnesota, Colorado, Massachusetts, Nebraska, North Carolina, Wisconsin, and Maine— Medicaid spending on autism therapy grew from a combined $347 million to over $2.2 billion in recent years, a 561 percent increase.
Individual states present even more striking trajectories. The figures and tables below summarize ABA spending growth documented by state Medicaid reports and federal audits across the eight states examined for this briefing.
Medicaid ABA Spending by State, 2017–2024
Minnesota’s spending growth is the most extreme in percentage terms. Reimbursements for Early Intensive Developmental and Behavioral Intervention (EIDBI) for autism surged from roughly $671,000 in 2018 to $342.8 million in 2024—more than a 50,000 percent increase. The number of Minnesota EIDBI Medicaid recipients grew from 400 to 5,705 between 2018 and 2024, a 1,326 percent increase. More striking is that the average spending increased from only $1,678 per recipient in 2018 to over $60,000 per recipient in 2024, a 3,481 percent increase far outpacing the increased number of diagnoses and claimants.
Indiana’s Medicaid ABA spending grew from $21 million in 2017 to $611 million in 2023, a 2,810 percent increase, while ABA’s share of the total Medicaid budget grew from 0.2 percent to 3.4 percent. Indiana’s historical policy of reimbursing providers at 40 percent of whatever they billed, rather than using standardized rates, was a significant driver of this growth.
A Wall Street Journal analysis of Indiana Medicaid billing records illustrates how the old system was exploited. One provider, Piece by Piece Autism Centers, received $29 million from Indiana Medicaid in 2023 to treat 84 patients, approximately $340,000 per child, a figure that surpasses what the state typically spends on treating a newly diagnosed lung cancer patient or covering a year of nursing home care. The provider raised its listed therapy prices from $200 per hour to $1,600 per hour under the 40 percent reimbursement system, yielding Medicaid payments of $640 per hour for routine therapy that could be administered by Registered Behavior Technicians with little more than a high school diploma. When Indiana implemented a uniform reimbursement rate in 2024, spending fell to $445 million, or by 27 percent, without any reduction in the number of children served.
Other states saw growth in Medicaid-funded ABA therapy for autism grow from between 35 percent to 1,700 percent in just 4 to 7 years, to the tune of tens to hundreds of millions of dollars. In many cases, only fee-for-service payments were reported, which constitute a fraction of total Medicaid reimbursements. Across all observed states and payment models, autism therapy spending grew far faster than the underlying rate of autism diagnoses, which itself grew far faster than any plausible change in the actual prevalence of the condition. The full scale of Medicaid autism therapy spending across all fifty states and all delivery systems remains unknown, but the available data suggest it is substantially larger.
Fraudulent Payments
The U.S. Department of Health and Human Services Office of Inspector General has now completed audits of Medicaid autism therapy payments in four states: Indiana, Colorado, Maine, and Wisconsin. In every state, 100 percent of sampled enrollee-months contained at least one improper or potentially improper claim. Confirmed improper payments totaled $198 million, or 31 percent of the $638 million audited.
The categories of noncompliance were strikingly uniform across all four audits. Session notes routinely failed to document what therapy was actually provided, for how long, or by whom. Providers billed for children napping, eating, playing on tablets, watching movies, or going down water slides. In Colorado, one provider billed 151 hours of therapy in a single month. Staff without credentials delivered therapy across multiple states. Facilities submitted session notes that staff had photocopied, cloned between patients, or signed before sessions concluded.
Many children had never even received a valid diagnostic evaluation. In one Indiana case, a child accumulated $677,448 in Medicaid payments from age 2 to 11 based on a single 2014 referral with no subsequent reassessment. None of the four states had ever conducted a postpayment review of ABA claims before the federal audits. The programs operated, in effect, on an honor system without verification.
The Cost-Benefit Question
Even setting fraud aside entirely—granting, for the sake of argument, that every dollar of Medicaid ABA spending was disbursed to a legitimate provider for a service actually rendered—a fundamental question remains: do the services themselves justify the expenditure?
This is a question that public policy routinely asks of other interventions. Physical therapy for children with motor impairments can be evaluated against long-term mobility and labor force participation outcomes. Special accommodations for children with dyslexia can be evaluated against literacy and educational attainment. In these cases, there is a plausible argument that early investment in targeted intervention produces lasting human capital gains that exceed the cost of the intervention.
The evidence base for ABA is more complicated. For children with moderate-to-severe autism—the population for which ABA was originally developed—the research literature provides evidence of meaningful gains in adaptive behavior, communication, and social skills, particularly when therapy is initiated early and delivered at high intensity. The Council of Autism Service Providers’ practice guidelines emphasize that therapy should be individualized, with the scope and intensity calibrated to each child’s assessed needs, and that “comprehensive” high-intensity therapy should be reserved for children with significant deficits across multiple developmental domains.
But the population now receiving Medicaid-funded ABA has shifted dramatically. The diagnostic data show that the growth in autism identification is concentrated among children with mild symptoms and no significant functional impairment, the very population for which the evidence of lasting benefit from intensive behavioral therapy is thinnest. The question is not whether ABA has any benefit for any child, but whether the marginal dollar spent on 30 to 40 hours per week of clinic-based ABA therapy for a child with no significant functional impairment produces returns comparable to alternative uses of the same public funds.
The data suggest it does not. Indiana’s ABA Working Group found that, in 2024, 50 percent of children on Medicaid-funded ABA received more than 25 hours per week, and 34 percent received more than 30 hours per week, despite the fact that full-time therapy is medically necessary in only the rarest circumstances. North Carolina found that the generalized increase in service intensity was not explained by any change in the severity of children’s conditions, any revision to clinical guidelines, or any alteration of coverage policy. Multiple states found that treatment plans were not individualized; the same number of hours was recommended for nearly every child regardless of assessed need.
What Needs to Change
What needs to change is not simply more oversight, more audits, or more rules layered onto an already complex system. Those responses treat the symptoms, not the cause.
As we argued in our earlier analysis of psychiatric overdiagnosis, when diagnosis is subjective and payment depends on diagnosis, the system will reward expanding the definition of illness. Autism therapy is where that dynamic has become most visible and most costly.
Real reform involves realigning incentives: tightening diagnostic standards, linking the level of care to documented functional impairment, and restoring financial accountability to both providers and payers, so that neither those delivering care nor those receiving it are completely shielded from its costs.
Until then, we should not be surprised by what we are seeing. The system is doing exactly what it was designed to do.
And the cost is not just measured in dollars. It is measured in diverted resources, diluted care, and the children with serious, functionally impairing autism who must now compete for attention in a system stretched far beyond its clinical purpose.
Cato intern Alex Williams contributed to this research.