Recently, ABA providers across Massachusetts received recoupment letters demanding hundreds of thousands of dollars back within 30 days. No warning. No prior engagement. It won't be the last service to be audited.
MassHealth had audited 2024 claims and concluded providers had billed outside a supervision ratio many in the field say was never clearly communicated as a condition of payment. The estimated overpayment: $16.8 million.*
CMS has identified the services most vulnerable to fraud, waste, and abuse as HCBS, ABA, home health, personal care, and non-emergency medical transport. ABA is the documented case. It is unlikely to be the only one.
Whether the methodology holds up legally is still being contested. What isn't contested is the financial reality: providers on thin margins, with limited cash reserves, facing six-figure demands on a 30-day clock. Some have stopped accepting MassHealth members. Some have made cuts.
Most of these providers were operating in good faith. That hasn't mattered.
Being ready for this environment isn't about hoping the audit skips you. It's about what happens the day it doesn't.
At the billing level: do the systems in place ensure that what was billed accurately reflects what was delivered? Attendance, authorization status, level of support, supervision ratios where applicable. Billing accuracy isn't just good practice anymore. It's the first line of defense, and the systems need to be right.
At the organizational level: if MassHealth or a managed care organization comes asking, can the organization respond in days, with clean queryable records that surface what actually happened for any given claim, participant, or time period. The difference between a defensible organization and a struggling one is rarely the underlying practice. It's whether the documentation is accessible when it counts.
That's not a compliance posture. It's an infrastructure problem wearing a compliance costume. Most good-faith organizations already have the data. It's in 125 bankers boxes, not a queryable database that can answer a question across fifty cases in a day instead of six weeks. Having that kind of readiness is what builds organizational credibility. That credibility opens the door to push the system to better meet the needs of those we support: to say, make an exception here, to better serve this person, this population. It's not a task. It's the mission.
* Behavioral Health Business, "'Can I Survive Here?': Massachusetts Autism Therapy Providers Rattled by Contentious Medicaid Clawback Effort" (May 1, 2026). Acuity News, "Massachusetts ABA Audit Crisis: Providers Fight $19M MassHealth Recoupment" (April 20, 2026).
We don't need more documentation. We need smarter systems.
June 17, 2026
When a group home's funding model stops fitting the people living in it, the instinct is almost always the same: we need more data.
It's the wrong diagnosis. In most agencies I've worked with, the data exists. It's in shift notes, incident reports, daily logs, and behavior tracking sheets. The problem isn't volume. It's that nobody designed those systems to capture what a funding conversation actually requires, and even when the right information is in there, it's buried in narrative, scattered across platforms that don't talk to each other, and would take hundreds of staff-hours to compile into anything a DDS Area Director could act on.
We don't have a data problem. We have a systems problem.
Here's what that looks like in practice.
Picture a group home that has been running well for years. Four men, all in their late forties or early fifties. Staff know them. The routines are established. The funding model reflects the support needs that were documented when the home opened, or the last time anyone sat down and really looked.
That was a while ago.
What's happening now is harder to see than a crisis, which is exactly what makes it dangerous. One of the men has Down syndrome and has started showing early signs of cognitive decline. These things can be easy to miss or explain away. He's a little slower in the morning. He needs a second prompt to start his shower where one used to be enough. He forgot his medication on Tuesday, which hadn't happened in years. He's been quieter at meals.
Individually, each of these is easy to attribute to a bad night's sleep or a busy week. Collectively, they are a pattern. But only if someone is capturing them.
The picture across the rest of the home is similar in kind, if different in detail. Another resident's mobility has changed enough that he now uses a walker in the house. Morning routines that took 45 minutes take over an hour. There's a real fall risk that didn't exist two years ago. A third man has new cardiac medication that requires timing coordination with meals and monitoring for side effects that staff weren't trained for when they were hired. A fourth has hearing aids: expensive, easy to lose, and essential for the communication and engagement that make every other part of his day work. When they're not in, or not working, he withdraws. Staff spend time they don't have troubleshooting a device nobody was trained to maintain.
None of this shows up as an incident. None of it triggers a critical event report. It's just the slow, real accumulation of what it means to grow older, and it's something the people this sector supports are doing for the first time, in large numbers, right now.
The funding model for this home was set based on who these men were five years ago. It doesn't describe who they are today.
This is where the systems conversation gets serious.
If your documentation system prompts staff to note whether routines were completed but not how many prompts it took, you are not capturing the drift. If behavioral observations live in shift notes and medical complexity lives in a separate health record that program staff rarely access, you have two partial pictures that nobody is connecting. If the walker and the hearing aids and the medication timing protocol are each documented somewhere but never aggregated into a coherent support picture, you cannot make the case that this home's needs have fundamentally changed.
You have a compliance record. You don't have a funding argument.
This is also where AI tools are starting to show real promise, not replacing clinical judgment, but surfacing patterns in narrative data that would otherwise require hundreds of staff-hours to compile. The gradual shift from one prompt to three, documented across six months of shift notes, is invisible to the human eye at scale. It doesn't have to be.
The fix isn't telling staff to document more. It's building systems that prompt the right information at the point of care, connect the data across platforms, and surface it in a form that managers and executives can actually use.
That last part matters more than people think.
When the data is structured correctly, it doesn't just support a conversation with your area director. It changes what that conversation looks like.
Instead of "we think this home needs a higher funding level," you're saying: "Over the last eight months, we've documented a consistent increase in prompting across morning and evening routines for three of the four residents. We've tracked 14 medication administration flags, a measurable increase in time-on-task for personal care, and two near-miss fall incidents. We've also begun early-stage dementia monitoring for one individual in consultation with his physician. Here's the trend. Here's what we're projecting. Here's what we're asking for."
That's a different meeting.
And when the area director engages, when DDS adjusts the model, when the home gets the staffing and training it actually needs, the people who made that possible are the ones who wrote those shift notes. Not the executive who sat in the meeting. Not the management team who helped frame the argument.
The staff did it. And they need to know that.
Closing that loop, telling your direct support team that their documentation drove a funding change that changed the lives of the people they support, isn't just good management. It's accurate. And it's the difference between staff who document because they have to and staff who document because they understand why it matters.
Building systems that do all of this (systems that prompt well, connect across platforms, surface data in usable form, and close the loop back to the people doing the work) is not a one-time project. It's an ongoing discipline. Systems fail not because people stop caring, but because the system itself was never fully documented in the first place. Staff turn over and take institutional knowledge with them. Workarounds become the de facto process, unwritten and untrained. New learning from a difficult incident, a regulatory change, or a better approach observed somewhere else never makes it back into the system. Bit by bit, what's actually happening drifts from what was built.
The agencies that get this right treat their documentation systems not as artifacts to be filed, but as living infrastructure that needs regular attention, honest assessment, and the willingness to rebuild when something isn't working.
Most don't get there on their own. Not because they lack the commitment, but because the work requires a specific kind of capacity that most agencies can't justify maintaining full-time: someone who can stress-test existing systems, close documentation gaps, develop training materials, assess whether the culture actually reinforces what the manual says, and help management staff build the habits that keep it all from drifting again.
That capacity doesn't need to live inside the organization. It needs to show up with enough skill and frequency to keep the system honest.