The Six New 2027 Adaptive Behavior CPT Codes & Impact on your Tech Stack

Last November, before the new adaptive behavior CPT code descriptors were public, I published an article exploring what I thought the six new codes might mean for technology platforms and providers.

Much of that article reflected my own hypotheses about where the code set was heading. I expected broader changes around evaluation workflows, care coordination, technology-assisted assessment, and structured re-evaluation.

Now we have our first public look.

On July 14, CMS released the proposed 2027 Medicare Physician Fee Schedule (MPFS), which includes the six new adaptive behavior CPT codes approved by the AMA CPT Editorial Panel at its September 2025 meeting and reviewed by the RUC HCPAC in January 2026.

Before getting into the details, it's worth pausing on the work itself. Getting a code family revised—deleting temporary codes, revising eight existing descriptors, and creating six new ones through the CPT Editorial Panel and RUC process—represents years of effort by the clinicians, coding experts, and advocates who shepherded it through. Whatever the industry ultimately makes of the new codes, that work deserves recognition.

Two important caveats before going further.

First, the codes are not official until the AMA publishes the 2027 CPT® Professional Edition later this year.

Second, this is a proposed rule, subject to a 60-day public comment period; nothing is final until CMS issues the final rule in November. The ABA Coding Coalition has appropriately cautioned providers and payors against making changes to their EMR systems or operations until the official publication. I agree with that guidance.

With that said, the proposed rule gives providers, technology companies, and payors their first real opportunity to understand the direction of the new code set.

The published descriptors are more focused than I expected.

Rather than introducing entirely new categories of adaptive behavior services, the six new codes primarily recognize work that providers have long performed but that has been difficult to represent within the existing code set.

Ironically, I think that makes the technology story even more compelling.

The Six New Codes at a Glance

Although there are six new CPT codes, they really introduce three new capabilities into the adaptive behavior code set.

Harmful Behavior Services (97X1X, 97X2X, 97X4X, 97X5X)

Codes supporting assessment and treatment of patients exhibiting harmful behavior, delivered by two technicians in an environment customized to the patient's behavior, with add-on codes (97X2X and 97X5X) for each additional technician present.

Notably, these codes accompany the deletion of Category III codes 0362T and 0373T, which described similar multi-technician services for patients exhibiting destructive behavior. In other words, this work was already represented in the code set—just in temporary codes.

The new structure moves it into permanent Category I codes, updates the terminology from "destructive" to "harmful" behavior, and restructures staffing as a two-technician base with per-technician add-on codes.

Non-Face-to-Face Professional Services (97X3X)

A new code recognizing physician/QHP work performed away from the patient, including reviewing and analyzing treatment data and session notes, modifying treatment targets or protocols, determining the need for additional assessment, developing discharge or transition plans, and reviewing revised protocols with technicians.

Direct Physician/QHP Adaptive Behavior Treatment (97X6X)

A new code recognizing adaptive behavior treatment with analysis delivered directly by the physician or other qualified healthcare professional, face-to-face with one patient.

CMS also proposes contractor pricing for the six new codes, consistent with the existing adaptive behavior code family, which remains contractor priced. Existing codes 97151–97158 would remain on Medicare's permanent telehealth list, while telehealth status for the six new codes was not addressed in the proposed rule; the ABA Coding Coalition has said it plans to request their addition during the comment period.

Most discussion around these changes will naturally focus on reimbursement.

I think the longer-term impact may be on software architecture.

One Code Immediately Stood Out

Of the six new codes, 97X3X immediately caught my attention.

Not because it creates a new clinical service.

Because it recognizes work that has traditionally happened behind the scenes.

• Reviewing treatment data.
• Analyzing session notes.
• Changing treatment targets.
• Modifying protocols.
• Planning transitions.
• Reviewing revised protocols with technicians.

Experienced BCBAs have always performed these activities.

What's changed is that they are now explicitly recognized as a distinct adaptive behavior service.

That's significant.

These aren't simply documentation activities. They're clinical decision activities.

The Technology Challenge

The 2019 adaptive behavior code set fundamentally changed how ABA services were categorized and billed.

The 2027 changes are different.

Rather than redefining care delivery, they better recognize work that has historically existed around assessment, treatment, and clinical oversight.

That may ultimately prove just as important for technology platforms.

Most ABA practice management platforms were designed around appointments, documentation, authorizations, scheduling, and billing.

The new codes ask platforms to represent more of the work that has always existed.

Take 97X3X.

Can today's platforms:

• Assemble treatment history, session notes, and outcome data into a meaningful clinician review workspace?
• Version treatment protocols as they evolve, so protocol revisions become tracked events rather than overwritten documents?
• Maintain traceability between assessments, treatment decisions, protocol revisions, and ongoing services?
• Capture structured QHP review time separately from general administrative work?
• Support technician guidance and protocol review as structured workflows rather than simply additional notes?

Those aren't billing capabilities.

They're workflow and data architecture capabilities.

The harmful behavior codes raise a parallel set of questions.

Can the platform:

• Schedule and document multiple technicians delivering a single service?
• Capture overlapping technician time accurately?
• Represent staffing composition for authorizations, reporting, and payor documentation?

Supporting these services isn't simply adding new billing codes.

Representing the operational work behind those codes is considerably more challenging than adding the codes themselves.

Preparing Without Jumping Ahead

The ABA Coding Coalition is right to caution providers against reconfiguring systems before the final rule and official CPT publication.

But there's an important difference between changing systems and understanding your operations.

Providers can begin asking questions today that will be valuable regardless of how the final rule evolves.

How is non-face-to-face clinical work captured today?

Many organizations have no structured record of the time BCBAs spend reviewing data, revising protocols, preparing transition plans, or guiding technicians.

How are treatment protocol changes tracked?

Does your system know what changed, when it changed, who changed it, and why?

Or does each revision simply overwrite the previous version?

How clean is the data supporting clinical review?

A code recognizing review and analysis of treatment data implicitly raises expectations that the underlying data is structured, complete, and connected to treatment targets.

How are multi-technician services documented today?

Organizations already delivering these services can begin evaluating whether existing documentation supports the component requirements reflected in the proposed descriptors.

None of these activities require changing software today.

They're exercises in understanding operational maturity.

Questions to Ask Your Platform Vendor

It's too early—and unfair—to judge individual vendors.

Every major ABA platform now has time to evaluate these changes before implementation in 2027.

But providers should begin asking different questions during platform evaluations.

Not simply:

"Will your platform support the new CPT codes?"

Instead:

• How will non-face-to-face professional work be represented and time-tracked?
• How will treatment protocol revisions be versioned over time?
• How will multiple-technician encounters be scheduled, documented, and reported?
• Will these activities become structured workflows or simply additional documentation templates?
• How will they connect to authorizations, treatment planning, reporting, and billing?

Those answers may prove far more important than whether six new billing codes appear in a future software release.

What About Innovation at the Edges?

A fair question is whether these codes leave room for the more innovative models emerging across autism care—virtual delivery, AI-assisted clinical work, and developmental or relationship-based models like DRBI.

The honest answer: the codes don't appear to create new pathways for any of them. But they're structurally agnostic in ways worth noting.

Virtual care is already inside the guidelines. Codes 97151–97158 sit on Medicare's permanent telehealth list, and CMS proposes keeping them there for 2027. Whether the six new codes receive telehealth status remains open—the ABA Coding Coalition plans to request their addition during the comment period.

And 97X3X is non-face-to-face by definition. For distributed care models that depend on remote clinical oversight, that's not a stretch of the code. It's the code.

AI-assisted work fits differently. These are time-based codes billing human professional time, and 97X3X specifies work personally performed by the physician or QHP. AI can inform a clinician's review and analysis—nothing in the descriptors prohibits decision support. But the billable service remains the professional's judgment, not the tool.

That's a meaningful distinction. The codes recognize clinical analysis without prescribing how clinicians arrive at it.

For DRBI and other developmental or relationship-based models, the picture is unchanged: these codes describe adaptive behavior services delivered by QHPs and technicians, full stop.

What the new codes recognize—structured review, protocol evolution, professional decision-making—happens to be the same infrastructure any evidence-based care model needs. But that's an observation about infrastructure, not a reimbursement pathway.

Looking Back

When I wrote about these forthcoming codes last year, I expected the biggest changes to come from entirely new categories of adaptive behavior services.

The published descriptors point in a different direction.

Rather than expanding the taxonomy of adaptive behavior services, they better represent work providers are already doing every day.

• Reviewing information.
• Analyzing progress.
• Refining treatment.
• Guiding technicians.
• Making clinical decisions.

Those activities have always existed.

The new code set simply recognizes them more explicitly.

Bottom Line

The published descriptors took a different path than I expected.

But they reinforced something I've come to believe over the past year.

The future of autism technology isn't simply about documenting services.

It's about representing the work that surrounds those services—clinical review, protocol evolution, staffing complexity, and professional decision-making—in ways that are structured, connected, and actionable.

For investors, that shift extends well beyond ABA: value is moving toward systems that support clinical decisions, not just clinical documentation.

These six new adaptive behavior CPT codes don't create that future.

They acknowledge that it's already here.

Note: The six new adaptive behavior CPT codes remain unofficial until publication of the 2027 CPT® Professional Edition by the AMA. The CMS Physician Fee Schedule discussed here is a proposed rule and remains subject to public comment and revision before the final rule is issued. Descriptor language discussed in this article is based on the CMS proposed rule.