ABA Medicaid Audits Are Accelerating. Here Is What Reviewers Actually Look For.

The letter does not look like much. It arrives in a plain envelope or lands in a portal inbox, it references a date range and a list of member IDs, and it gives you a window measured in days to produce records for claims you submitted eighteen months ago. Nobody is accusing you of anything yet. Someone simply wants to see the paperwork behind services you already provided and were already paid for.

For a clinic with tight documentation, that letter is an afternoon of work. For a clinic where session notes were written from memory on Friday afternoons, where a supervising clinician's credential changed and nobody updated the file, or where the treatment plan on record predates the goals actually being run, the same letter is the beginning of a repayment demand. The difference between those two outcomes is decided long before the envelope arrives, which is why ABA Medicaid audit preparation belongs on your operations calendar rather than in your crisis plan.

Why ABA Medicaid audits are accelerating

The short answer is that spending grew faster than almost anyone in state government expected, and rapid growth in a public program reliably attracts examination.

North Carolina has become the most visible example. Data presented to the state's Joint Legislative Oversight Committee on Medicaid in March 2026 showed that between state fiscal years 2022 and 2025, the number of beneficiaries receiving ABA rose 249%, units of service rose 305%, and spending rose 347%. The state's own analysis noted that utilization growth far outpaced increases in autism diagnoses, and concluded that growth of that magnitude was unlikely to be explained by improved access alone. The state auditor has since opened a large-scale review examining providers, billing practices, and compliance with Medicaid rules, and figures cited publicly during that review have ranged widely depending on the baseline year used, which is worth keeping in mind when you see dramatic percentages in headlines.

North Carolina is not unusual, only louder. Nationally, Medicaid spending on ABA grew nearly 300% between 2019 and 2024, with wide variation by state, and the same committee materials cite increases of 607% in Indiana, 75% in Massachusetts, and roughly 2,000% in Nebraska. Growth curves like those do not distinguish between legitimate expansion of access and improper billing, which is precisely why auditors get sent to find out which is which. It is also the same pressure driving the rate reductions several states made in 2026, so most Medicaid-heavy practices are absorbing both at once.

What federal auditors have already found

The federal picture matters because it tells you what reviewers are trained to look for. The HHS Office of Inspector General has been conducting a state-by-state investigation into Medicaid autism therapy billing focused on improper payments and documentation failures, and audits completed in Maine, Wisconsin, Indiana, and Colorado have each turned up problems in the claim samples reviewed.

The findings cluster around a few themes rather than exotic schemes. Auditors identified payments made despite missing documentation, credentialing problems where the rendering provider did not hold or could not be shown to hold the required qualification, missing or expired evaluations and referrals, and instances of billing for treatment time during which children were reportedly napping or watching videos rather than receiving services.

That last category understandably dominates news coverage, and it is also the least relevant to most clinic owners reading this, because deliberate misbilling is not the risk profile of a typical practice. The first three categories are the real exposure. They describe paperwork failures rather than clinical ones, and they are entirely capable of producing a recoupment demand against a clinic that delivered every minute of care it billed.

What an ABA Medicaid audit actually examines

Reviewers work from the claim backward, and every link in that chain has to hold.

They start with whether a valid diagnosis and a current, signed treatment plan support the services billed, and whether a required referral or evaluation exists and falls within its effective window, which makes authorization and referral tracking an audit control as much as a billing one. They check that the rendering provider held the credential the code required on the date of service, and that supervision requirements were met and documented, which in states like Indiana now means demonstrating a specific ratio of supervision hours to technician-delivered hours, measured per member per month.

Then they read the session notes against the claim. The question is whether the note demonstrates that the billed service actually occurred, for the billed duration, targeting goals in the current treatment plan, delivered by the person identified on the claim. A note that records activities without connecting them to plan goals, or that shows a duration inconsistent with the units submitted, creates a finding regardless of what happened in the room.

The documentation gaps that catch honest clinics

Three patterns account for most of the trouble I would expect a well-run practice to encounter.

The first is drift between the treatment plan and the sessions. Programs evolve appropriately as a child progresses, and the authorized plan on file does not always evolve with them, so the notes describe good clinical work against goals that no longer match the approved document. The care was right and the record does not prove it.

The second is supervision documentation. Supervision typically happens, and it frequently goes unrecorded in a form that survives review, particularly when it occurs informally or when the supervising clinician assumes the session note covers it. Where a state sets an explicit ratio requirement, unrecorded supervision is functionally the same as absent supervision.

The third is timing. Notes written days after the fact tend to lose the specificity that makes them defensible, and in a specialty where a single technician may run several sessions a day across multiple clients, reconstructed detail is both harder to produce and easier for a reviewer to question.

How to prepare before a records request arrives

Audit readiness is mostly a habit rather than a project. The most effective version of it is a small internal sample review run on a regular schedule, where you pull ten or fifteen random claims from the prior quarter and walk them backward exactly as an auditor would, checking diagnosis, referral, treatment plan currency, authorization, credential, supervision record, and session note against the claim as submitted.

What that exercise produces is a list of gaps while they are still fixable and while the underlying pattern can still be corrected. Doing it quarterly means that when a real request arrives, you already know what your files look like, and you are assembling a response rather than discovering a problem.

Two supporting habits make the sample review more useful. Keep a current, written summary of each major payer's documentation and supervision requirements, since these changed materially in several states during 2026 and are the sort of thing that gets updated by bulletin rather than by phone call. And know your own production timeline, because many states impose short windows for producing records, with Indiana for example requiring referral documentation, signed treatment plans, and session notes within 72 hours of a request.

What a recoupment actually costs

The financial mechanics of an audit finding are worse than they first appear, because a review rarely stops at the sampled claims. When a sample produces an error rate, that rate can be extrapolated across a larger population of claims, which turns a finding on fifteen records into a repayment demand covering a year of billing.

That demand also arrives against money you have already spent. The wages were paid, the rent was paid, and the margin from those sessions was absorbed into operations quarters ago. Repaying it comes out of current cash rather than out of the revenue that generated it, which is why documentation exposure belongs in the same conversation as your cash reserves and runway rather than being filed under compliance.

Back to the letter

The clinics that handle a records request well are almost never the ones with the best lawyers. They are the ones whose files already told a complete story before anyone asked, because somebody made checking that story a routine rather than a reaction.

Scrutiny of ABA is not going to ease in the near term, and the growth figures driving it are a matter of public record in state after state. The practices that come through this period intact will be the ones that treated documentation as a financial control rather than as clinical housekeeping.

Frequently asked questions about ABA Medicaid audits

What triggers an ABA Medicaid audit?

Rapid growth in utilization and spending is the main driver at the program level rather than suspicion of an individual clinic. Several states have seen ABA spending rise sharply enough to prompt legislative review and auditor involvement, and reviewers are then sent to determine how much of that growth reflects expanded access versus improper billing.

What documentation do Medicaid auditors request for ABA?

Reviewers work backward from the claim, so expect requests for the diagnosis, the referral or evaluation and its effective dates, the current signed treatment plan, the authorization, proof the rendering provider held the required credential on the date of service, the supervision record, and the session notes for the dates billed.

What do auditors look for in ABA session notes?

Whether the note demonstrates that the billed service occurred, for the billed duration, targeting goals in the current treatment plan, delivered by the person named on the claim. Notes that record activities without tying them to plan goals, or whose duration does not match the units submitted, create findings regardless of what actually happened in the session.

What is Medicaid recoupment and how is it calculated?

Recoupment is repayment of money already received. Reviews rarely stop at the sampled claims: when a sample produces an error rate, that rate can be extrapolated across a larger population, turning a finding on fifteen records into a demand covering a year of billing, payable out of current cash rather than out of the revenue that generated it.

How long do you have to respond to a records request?

It varies by state and payer, and the windows are short. Indiana, for example, requires referral documentation, signed treatment plans, and session notes within 72 hours of a request. Knowing your own production timeline in advance is part of being ready, since there is no time to reconstruct a file once the request arrives.

If you want a clear view of where your documentation and billing exposure actually sits, and what it would mean for your cash position if a recoupment landed, that is exactly the kind of analysis our financial assessment provides.