The CMS ABA Toolkit’s 10, 20, and 30 Hour Example Has No Footnote. CASP’s Joy Pollard Says Other Thresholds Lack Evidence Too.

November 19, 2026

CASP’s Joy Pollard took the CMS ABA toolkit page by page at the Autism Law Summit, from hours by severity level to group therapy flags and telehealth limits.

Key Takeaways

  • Guidance with consequences: CMS says its August ABA toolkit sets no standard of care, but states are already weighing its recommendations as they respond to federal audits.
  • Thresholds without footnotes: Joy Pollard of CASP said the toolkit cites no research for its hours-by-severity example, its 10 percent group therapy flag, or its call for breaks between consecutive units.
  • A disputed starting point: Pollard said the 2021 baseline behind the toolkit’s 421 percent spending increase leaves out plans that had not yet adopted the ABA CPT codes.
  • Shared ground: The toolkit and CASP agree on Board Certified Behavior Analysts as reviewers, individualized treatment, and accreditation, and Pollard urged providers to engage state agencies early.

Joy Pollard’s slides at the Autism Law Summit carried page numbers from a federal document. The document was the CMS ABA toolkit, a guide of more than 170 pages that the Centers for Medicare & Medicaid Services released in August for state Medicaid and CHIP agencies. The citations, Pollard told the summit audience in late October, were deliberate.

“We are actually putting quotations and page numbers in here directly because we expect that your states are already responding to what has been occurring with OIG,” she said during the summit’s Legal Update session, referring to the HHS Office of Inspector General.

The toolkit comes with its own caveat. On page 7, CMS writes that it “does not include or represent clinical recommendations and should not be interpreted as establishing a standard of care or directing clinical judgment.” It also describes itself as “not a regulation.” Pollard’s argument, made provision by provision, was that several of its recommendations carry no citation, and that states considering them should ask what evidence stands behind each one.

Her critique comes from inside the guidelines CMS leans on. Pollard, who founded Behavior Change Institute, a California provider, and led it for 11 years, became the Council of Autism Service Providers’ first Vice President of Clinical Standards in August. She contributed to version 3.0 of CASP’s ABA practice guidelines as a Subject Matter Expert and was lead author of CASP’s telehealth practice parameters, which the toolkit cites.

What the CMS ABA Toolkit Says About Itself

CMS wrote the toolkit, it says on page 6, to support state Medicaid and CHIP agencies “in making decisions regarding ABA service provision.” By Pollard’s count, it cites the CASP practice guidelines 25 times, and she credited CMS with drawing on them accurately in many places. “There’s a lot of places where they got information right,” she said. The guidelines, released in April 2024, drew feedback from almost 100 Subject Matter Experts, she said.

Dan Unumb, President of the Autism Legal Resource Center, who presented alongside her, focused on what the toolkit declines to be. CMS, he said, had told readers its recommendations are not clinical standards of care, a distinction a CMS official also drew before Florida’s ABA task force in September. “Who are the clinical standards of care? Who decides that?” he asked.

Pollard’s answer was the field’s “best available scientific peer-reviewed evidence, along with consensus from our experts in our field.”

The Medicaid ABA Audits Behind the Toolkit

The toolkit follows a run of federal audits of state Medicaid ABA programs. The OIG found that Indiana made at least $56 million in improper fee-for-service Medicaid payments for ABA in 2019 and 2020, that Wisconsin made at least $18.5 million in 2021 and 2022, and, in a report issued in February, that Colorado made at least $77.8 million. The Colorado recommendations address session notes and signatures, what counts as billable ABA time, provider credentialing, and documentation supporting CPT code 97155.

Unumb did not wave the findings away. “Waste, fraud, and abuse is a serious issue,” he said, and of the audited providers, “Were there some bad actors in there? Yeah, probably.” Read closely, though, he said the reports show “it’s mostly documentation problems,” made worse by thin state guidance in the years after a new code set arrived in 2019. Auditors, he said, read the documentation rules for CPT code 97155 as requiring a modified treatment protocol in each billed session, a reading the industry’s coding guidance does not share.

“The answer to a documentation problem is not to cut rates, right?” he said. “The answer to a documentation problem is not to limit care.”

Pollard added that “an increase in utilization does not mean that services are clinically inappropriate, and it does not mean that fraud, waste, and abuse or improper services are occurring.” She listed rising autism prevalence and a growing workforce of technicians and analysts among the other drivers.

The 421 Percent ABA Spending Figure and Its Baseline

The number that has traveled furthest sits on page 6. Between 2021 and 2025, according to Medicaid data cited in the toolkit, the number of children with an autism diagnosis receiving ABA rose 189 percent, and spending on ABA for those children rose 421 percent. Total Medicaid and CHIP ABA payments, the toolkit says on page 15, grew from about $1.94 billion in 2021 to $10.1 billion in 2025.

Pollard questioned the starting point. The baseline, she said, “is not representative of all of the healthcare plans because when those initial analyses were conducted, not all of the healthcare plans had adopted the new AMA CPT codes.” Some still bill under HCPCS codes, she said, which also leaves them out of the current use data. “So we can’t necessarily take that baseline data and look at the percentage increase that they’re reporting and conclude that the driver is fraud, waste, and abuse,” she said.

She made a similar point about hours. The toolkit reports that Medicaid and CHIP beneficiaries averaged 17.33 hours of weekly ABA in 2025, a 22 percent increase since 2021. A data note two pages later shows that the figure counts every ABA code from 97151 through 97158, so assessment, caregiver training, and group sessions sit alongside direct treatment. The note also says 2023 was excluded as an outlier.

ABA Hours by Severity Level and Medicaid Soft Limits

On page 37, the toolkit describes best practice as aligning the hours in a child’s treatment plan with autism severity levels, giving as an example “10, 20, and 30 hours for children in levels 1, 2, and 3, respectively.” The severity levels are cited to the DSM-5-TR. The hours example has no footnote.

“The toolkit first and foremost does not cite any evidence to support that particular recommendation,” Pollard said, “and the DSM severity levels that describe the level of support needs were not validated specifically for this use.” A child with a level two diagnosis, she said, “is going to present very differently than another child with the same level of severity.” She pointed to page 13, where the toolkit itself says the intensity and duration of ABA “vary according to the child’s needs and response to intervention.”

Under Medicaid’s children’s benefit, the toolkit says on pages 54 and 55, states may not impose hard limits but may use soft limits that can be exceeded through prior authorization. States’ existing weekly ABA hour limits run from 16 to 40 hours. Pollard asked that any threshold come with its evidence and clear criteria for exceeding it. Without those, she said, “it could be very easy for this to turn more into a hard limit.”

Group Therapy, Telehealth, and Consecutive ABA Units

A figure on page 34 lists documentation flags for potentially erroneous or improper billing. Among them, beside identical documentation across children and a lack of caregiver engagement, is “participation in group therapy that accounts for more than 10 percent of treatment time.” The line carries no source.

Pollard was careful about what the flag means. “CMS is not stating that if this is exceeded, that it’s clinically inappropriate or that it’s improper. They’re just saying it warrants additional review,” she said. Her objection was to the number. “We know that there’s not clinical evidence to support this 10 percent threshold specifically for many children,” she said, adding that a larger share of group services “would be very appropriate” for some, depending on their goals.

On telehealth, a summary list near the end of the toolkit urges states to “reduce over-reliance on telehealth for supervision and service delivery,” and the body text says diagnostic and treatment authorization visits “should occur in person,” a standard Virginia’s proposed Medicaid ABA rules would apply to autism evaluations. The toolkit does cite CASP’s telehealth practice parameters, the document Pollard led. Her complaint was with the rest of the section. “There is a lot of references about best practice with no citations,” she said. Whether telehealth fits, she said, depends on the child, the program, and the family, and many states also have telemedicine parity laws.

The toolkit’s page 36 recommends that states “set a policy that prohibits more than a specified number of consecutive units to be delivered without a break,” and elsewhere notes that activities of daily living such as toileting and feeding need room in the schedule. It does not give a number. “They don’t specify a specific number of units consecutively because there isn’t one that is related to best practice,” Pollard said. “There isn’t any research that I am aware of.”

Where CASP and the CMS ABA Toolkit Agree

Pollard closed with the provisions she welcomed. The toolkit says states and managed care plans “should consider hiring experienced BCBAs” to conduct utilization reviews, and she encouraged providers whose plans have not done so to raise it. It repeatedly ties treatment intensity to individual need, and it calls accreditation a best practice, which MassHealth is phasing in as a requirement for ABA providers. The Autism Commission on Quality, the accreditor the toolkit names, was founded by CASP, as the toolkit notes.

Her advice for the months ahead was procedural. Providers who belong to a state association, she said, should join its public policy committee, learn which toolkit recommendations their Medicaid agency is weighing, and start by “understanding what is the problem that they are trying to solve.”

She offered an example from Unumb’s work with a payer that had raised concerns about clinic-based services. After many conversations, she said, it became clear that “the concern wasn’t the setting; it was generalization.” Once that was on the table, she said, “they were able to open the door and start talking about those concerns.”