What the New CMS ABA Toolkit Gets Right About Caregiver Support, and What States Should Build Next

Yrenka Lolli-Sunderlin
Aug 13
5 min read

On August 4, 2026, the Centers for Medicare & Medicaid Services released the State Medicaid and Children's Health Insurance Program Applied Behavior Analysis Toolkit. The 173-page resource gives state Medicaid and CHIP agencies a detailed framework for reviewing how ABA is covered, authorized, delivered, documented, supervised, and audited.

The release immediately drew attention because of one number: 421%.

According to CMS, total Medicaid and CHIP payments for ABA increased from approximately $1.94 billion in 2021 to $10.1 billion in 2025. That is a 421% increase in four years.

The growth figures need to be explained carefully, however, because CMS uses two different beneficiary measures:

The number of beneficiaries with an autism diagnosis who received any Medicaid or CHIP service increased 67%, from 1.15 million to 1.92 million.
The number of children with an autism diagnosis who specifically received ABA increased 189%.

Both figures are accurate. They answer different questions. Using them interchangeably can exaggerate or obscure what is actually happening.

The spending increase is significant, but it does not prove that individual children are receiving unnecessary treatment. CMS itself identifies several contributors, including more children receiving autism-related services, earlier entry into treatment, greater demand for ABA, and increased use of ABA-specific billing codes. At the same time, the toolkit documents serious program-integrity concerns, including improper billing, insufficient documentation, kickbacks, and cases involving harm to children.

The appropriate response is neither denial nor panic. It is better program design.

What the Toolkit Actually Does

The toolkit is federal guidance for states, not a regulation and not a nationwide cut to ABA.

CMS states that the toolkit does not establish new federal requirements, reduce states' obligations under Early and Periodic Screening, Diagnostic, and Treatment (EPSDT), endorse a single treatment approach, or direct states to restrict access to medically necessary care.

Instead, CMS gives states suggested practices across areas that include:

Individualized treatment planning
Medical necessity and treatment intensity
Provider qualifications and credentialing
Supervision standards
Utilization management
Payment methodology
Documentation and auditing
Fraud, waste, and abuse prevention

The practical signal is still strong. States now have a common checklist they can use when rewriting Medicaid manuals, managed care contracts, prior authorization rules, provider enrollment standards, and audit procedures. ABA providers should expect more scrutiny. Families may also see changes in how treatment intensity, caregiver participation, progress, and reauthorization are documented.

The Overlooked Section: Caregiver Burden

One of the toolkit's most important discussions concerns parents and caregivers.

CMS recognizes that caregiver participation can help children use skills across people, routines, and settings. It also recognizes that families already devote substantial time to implementing recommendations, traveling to services, coordinating care, and trying to reach clinical teams.

That tension matters. Poorly designed caregiver requirements can become another barrier to care. If participation is treated as a rigid quota, families with multiple children, inflexible jobs, transportation problems, language barriers, or changing foster placements may be penalized for circumstances outside their control.

The toolkit therefore encourages flexibility. It discusses brief structured check-ins, conversations at drop-off or pick-up, telehealth coaching, and self-paced computer courses as options that may reduce the burden associated with caregiver involvement. Its state checklist also asks whether caregiver coaching is being treated as a comprehensive process instead of a replacement for provider-patient time.

That last distinction should guide the next phase of innovation.

Education Is Not a Substitute for Clinical Care

Families need understandable information, but they should not be turned into unpaid clinicians.

Clinical caregiver training is part of an individualized treatment plan. It should be directed by qualified professionals, tied to the child's goals, documented appropriately, and adjusted based on data and family circumstances.

Non-clinical caregiver education has a different role. It can help a parent:

Understand common behavioral concepts in plain language
Organize observations and questions for the clinical team
Prepare for a caregiver-training session
Practice how to communicate concerns or describe what happened
Locate appropriate professional or emergency resources when a situation exceeds an educational tool's scope
Access information in the family's preferred language and at a workable time

That educational layer should sit before, beside, and between clinical services. It should never diagnose, prescribe treatment, alter a treatment plan, or present generalized information as individualized clinical advice.

Where CareBridge Copilot Fits

CareBridge Copilot was built for this educational space.

It provides bilingual English and Spanish guidance designed by a BCBA, using plain language and safety boundaries that keep the licensed clinical team in charge. CareBridge is educational. It is not therapy, diagnosis, clinical supervision, crisis care, or an individualized treatment plan. It does not replace ABA, occupational therapy, speech therapy, medical care, school services, or the judgment of professionals who know the child.

Its potential value is not that it can replace expensive clinical hours. That would be the wrong promise and the wrong model.

The more responsible hypothesis is that accessible education may help families arrive better prepared, understand professional guidance more clearly, communicate more effectively, and feel less stranded during waitlists or between scheduled contacts. Whether that produces better engagement, improved access, or lower avoidable costs must be tested with real data. Those outcomes should not be assumed.

What States, MCOs, and Providers Should Test

The CMS toolkit creates an opportunity for carefully designed pilots. A caregiver-education pilot should measure more than logins or satisfaction. Useful questions include:

Did caregivers better understand the difference between education and treatment?
Did they feel more prepared to communicate with their child's team?
Was the resource equally usable in English and Spanish?
Did it reduce confusion without increasing pressure on caregivers?
Did the tool redirect clinical, safety, and crisis questions appropriately?
Did clinicians report better-prepared questions or more productive caregiver conversations?
Were there any unintended effects, including overreliance on the tool or delayed contact with a qualified professional?

Only after those questions are answered should organizations make claims about outcomes, efficiency, or cost savings.

The Real Choice Is Better Infrastructure

The debate is often framed as oversight versus access. That is too narrow.

States can protect children, strengthen program integrity, preserve EPSDT, and improve access at the same time. Doing so will require clear clinical standards, accountable providers, individualized care, and support systems that respect the realities of family life.

Caregiver education is one part of that infrastructure. Built responsibly, it can help families participate without asking them to carry the clinical system on their backs.

CareBridge Copilot is available for demonstrations and structured pilot discussions with state agencies, health plans, provider organizations, and community partners. To explore a pilot, contact us through the CareBridge Copilot website.

Disclosure: CareBridge Copilot is not endorsed by CMS. This article offers an independent interpretation of the August 2026 toolkit and describes a potential educational use case for future evaluation.

Sources: