> ## Content Index > Fetch the complete content index at: https://www.missionviewpoint.com/llms.txt > Use this file to discover other available public pages before exploring further. # ABA Never Built a Common Utilization Benchmark. The Wall Street Journal Did. - URL: https://www.missionviewpoint.com/aba-never-built-a-common-utilization-benchmark-the-wall-street-journal-did/ - Published: 2026-10-21T04:08:29.000Z - Updated: 2026-10-21T12:53:45.000Z - Author: Scott Dickson - Tags: Provider Operations, Provider: Centria, Provider: Cortica, Provider: Learn, Provider: Lighthouse Autism Center, Provider: Action Behavior Centers, Provider: HopeBridge **The Journal built a utilization benchmark from Medicaid claims. Now providers face a harder question:** > ***Can they produce the clinical data needed to explain the differences?*** Sixty-nine children in Arizona's Medicaid program switched autism-therapy providers between 2019 and 2024, all moving to the[ same large provider](https://www.missionviewpoint.com/provider-profile-centria-healthcare/). In the month before the switch, they **averaged about 13 hours of therapy a week**. After switching, they **averaged 24**. The children hadn't changed. **The organization delivering their care had.** That comparison comes from the Wall Street Journal's [latest analysis of autism-therapy billing](https://www.wsj.com/health/healthcare/autism-therapy-treatment-hours-private-equity-5781f2ca?mod=hp%5Flead%5Fpos7&ref=missionviewpoint.com), built from federal Medicaid claims the reporters accessed under a **research agreement**. Most of the attention will go to who owns the organizations billing the most hours. **The more consequential development for operators is quieter.** > ***This market has never had a common utilization benchmark, one that providers, payors, and platforms recognize and measure against.*** Payors run their own internal comparisons, and some large providers do too, but none of it is shared or accepted across the market. **The Journal didn't set out to fill that gap.** It built a claims-based benchmark to explain its findings, and in doing so produced a public reference point the industry **will now have to answer to**. [In June I argued](https://www.missionviewpoint.com/how-did-a-792-hourly-rate-for-autism-services-make-it-through/) that autism care never developed mature oversight partly because there was **no defensible definition of typical**. You can't flag an outlier without agreeing on what **normal looks like**. That argument turns out to be only **half right**. **The Journal didn't need clinical consensus to identify meaningful utilization differences.** It needed weekly hours per patient, a comparison group, and the switch events already sitting in the claims stream. Comparing the same children before and after a provider change **controls for some of the differences** in patient populations that make broader comparisons difficult. It doesn't eliminate every explanation for the increase, **but it makes the difference considerably harder to dismiss**. ## What the Benchmark Can and Cannot Tell Us The benchmark has real limits, and they're worth naming precisely, because the providers' objections focus on them. It's built on **Medicaid claims only,** so commercially insured patients aren't in it. It **measures billed services**, not independently verified care, and **says nothing about quality**. Five states were left out because their data was too thin to calculate weekly hours reliably. **Most importantly, claims data carries almost nothing about the child's clinical needs.** A claim records a billing code, a date, a provider, and units of time. It doesn't record a VB-MAPP score, a Vineland result, the goals in the treatment plan, or whether those goals were met. **Severity and progress are largely invisible to it.** Within those limits, the method is **more conservative** than the headlines suggest. The Journal weighted its comparison groups by each provider's state mix, which **it says produced smaller gaps** than raw averages would have. It filtered out duplicate claims and implausibly high weeks, and limited the switcher analysis to children with **at least ten weeks of service on both sides** of the change. Outside health-services researchers reviewed the approach and called it sound. This **isn't** a clinically validated standard for how many hours a child should receive. It **is a defensible utilization comparison** built entirely from administrative data that already existed. **That distinction matters for what comes next.** Every serious objection to the analysis points toward the same gap: the claims data doesn't establish **how impaired these children were** or **whether they improved**. The objection is **fair**. But much of the missing clinical information isn't missing from the world. **It's sitting in provider systems.** ## The Missing Link Between Utilization and Clinical Need The providers' principal defense was that their patients **had greater clinical needs**, making higher treatment intensity appropriate. That may well be true for some of them. But the **publicly reported provider responses** didn't establish a consistent, independently **comparable relationship between assessed need and delivered hours**. One provider said three-quarters of its patients fell into higher severity tiers, by its own measure. Others argued that the comparisons weren't apples to apples. [Action Behavior Centers](https://www.missionviewpoint.com/provider-profile-action-behavior-centers/) supplied four alternative utilization analyses, including one that excluded short-hour weeks early in treatment but counted them later. An outside researcher questioned **whether that approach was consistent with accepted methods** for measuring treatment dose. What remains missing from the public debate is **a way to connect utilization differences to comparable clinical evidence**. The frustrating part is that much of the evidence **probably exists**. Children entering these programs undergo **clinical assessments**. Treatment plans **establish goals**, and sessions **generate data on progress** toward them. Reauthorization requests to payors include **updated clinical information** and **justification** for continued treatment. Taken together, those records contain **much of the context** the claims data lacks. **The problem is the form it's in.** Assessments may be stored as PDFs or scored documents attached to a client record, rather than structured fields that can be queried across a population. Different clinicians use different instruments. Goal mastery lives in session-level data collection that doesn't necessarily roll up into a consistent patient-level trajectory. And the clinical record often doesn't join cleanly to billing data, so connecting what a child needed to what was billed becomes a **manual, case-by-case exercise**. For providers built through acquisition, **add multiple legacy systems** with **different data structures** to the list. [It's the same pattern I described in March](https://www.missionviewpoint.com/the-medicaid-autism-racket-a-systems-process-problem-hiding-inside-the-headlines/), when audit findings that looked like misconduct sometimes reflected documentation and compliance processes that **weren't designed to validate requirements** before claims were submitted. So when a reporter or payor asks a population-level question, such as whether children with similar assessed needs **receive similar hours** across clinics, or whether **hours decrease as goals are met**, a provider may be able to answer for one child by **manually reviewing the record**. **Answering the same question consistently for ten thousand children is a different proposition.** The clinical information may exist, but that doesn't mean the organization can turn it into evidence **at scale**. The benchmark is weak where provider data is weak. That's not a coincidence. It's the same gap **seen from the outside**. ## The Provider Who Can't Produce Its Own Benchmark Gets Judged by Someone Else's The Journal had to negotiate a research agreement to obtain its data. Payors didn't. Every Medicaid plan and commercial insurer already holds substantial claims data for its own members. They can **compare utilization** across participating providers, **track treatment intensity** over time, and **examine** what happens when members change providers within their **observable claims history**. Some payors already conduct these analyses. Earlier this year, UnitedHealthcare removed Action Behavior Centers from its Arizona Medicaid network, **citing annual per-patient costs three times** those of smaller competitors. Action disputed the comparison, arguing that its patients had greater clinical needs. UnitedHealthcare confirmed to the Journal that its comparison hadn't adjusted for severity. **That illustrates the problem.** A payor can identify an organization with **substantially higher utilization or cost**. The provider may have a legitimate clinical explanation, but demonstrating that explanation across a large patient population **requires a different set of data**. When one side has a **readily available utilization comparison** and the other cannot produce an **equally accessible clinical explanation**, the conversation starts with the numbers the **payor can see**. Cost per patient becomes the **metric**. Authorizations may **tighten**. Continued treatment intensity receives **more scrutiny**. Network decisions **can be influenced** by comparisons that don't fully account for clinical differences. > Federal Medicaid guidance questioning the relationship between treatment intensity and outcomes adds to that pressure. High-acuity providers delivering appropriate care could face the same scrutiny as organizations whose utilization is more difficult to justify, **because claims data alone cannot reliably distinguish them**. **The provider who can't produce its own benchmark gets judged by someone else's.** And the same gap shows up inside the business, where it's harder to see. ## When Lower Utilization Might Mean Better Care Picture a regional operator looking at a clinic whose delivered hours fell 12% last quarter. On most operational dashboards, that's a performance measure, and the natural response is to understand how to restore those hours. But the decline could mean **technician vacancies** and **late cancellations**, which represent an **access problem worth fixing**. It could also mean a group of children **met** treatment goals, **reduced** their recommended hours, or **transitioned** into after-school schedules. **That could represent treatment working as intended.** The two situations have very different implications, yet **both appear as lower utilization.** A typical practice management report can tell leadership what changed. It may not explain **why**. The reason a child's hours changed often lives in a **clinical note, treatment plan revision,** or **communication** with the family. It may never become a **structured data element** that can be analyzed alongside the schedule and billing record. An organization that can't distinguish between lost hours caused by operational problems and reductions resulting from clinical progress **risks managing both situations the same way**. Over time, without anyone explicitly deciding it should, the operation can lean toward preserving hours. That isn't necessarily a **failure of clinical judgment**. It's a consequence of managing an organization with **incomplete information**. And it is particularly consequential in an industry **where the primary measure of production is also a clinical treatment decision.** ## What ABA Technology Platforms Need to Connect The way forward isn't simply a better rebuttal letter or another productivity dashboard. **It's connecting what the organization already knows**. That starts with **capturing assessment results as structured data** at intake and reassessment, while **preserving the differences** between clinical instruments and the limits of what their scores can tell us. It means recording **why recommended or delivered hours changed**, whether because of clinical progress, a school transition, a family request, a staffing shortfall, or another reason, as information that can be analyzed rather than a sentence buried in a progress note. And it means **linking those records** to authorized, scheduled, delivered, and billed hours. With that foundation, an operator could begin answering questions that are difficult to address today. Do children with similar assessed needs receive **comparable** treatment intensity across clinics? How do hours **change** as goals are met? What happens when a child **transfers** from another provider? Are reductions in delivered hours driven by **clinical recommendations, family preferences,** or **operational constraints**? These analyses wouldn't independently determine medical necessity. Clinical measures aren't perfectly standardized, and a benchmark **can't replace individual clinical judgment.** But they would make patterns visible and give clinicians and operators a better basis for reviewing them. Some of the answers may be **uncomfortable**. Better to discover those patterns through **internal analysis** than through an external investigation. The infrastructure that lets a provider defend its hours to a payor is the same infrastructure that tells its own leadership **what those hours mean**. That's where ABA technology platforms have **something to prove**. Practice management systems already know what was authorized, scheduled, delivered, and billed. Clinical systems hold assessments, treatment goals, and evidence of progress. Sometimes those capabilities live in the same platform. Often they don't. And even when they do, the **information rarely comes together** in a way that lets an operator explain treatment intensity across an entire organization. A platform that connects those records, makes changes in recommended hours **explainable**, and allows providers to **examine utilization alongside clinical need** would be delivering something more consequential than another productivity dashboard. This is also where the current enthusiasm for AI needs some perspective. AI may help extract information from clinical documents, identify changes in treatment plans, or surface utilization patterns that warrant review. But none of that eliminates the **need for consistent data definitions, reliable integrations,** and **clinical governance.** An AI model that can summarize a treatment plan isn't the same as an organization that can explain its treatment intensity **across thousands of clients**. The latter requires **organizational infrastructure**, not just another application feature. ## The Benchmark Is Already Here The three Wall Street Journal investigations I've written about this year trace a progression. The March reporting raised questions about whether services were **properly documented** and whether **claims met the requirements** governing reimbursement. The June investigation exposed **reimbursement structures** and **billing patterns** that existing controls had **failed to catch**. Now the October investigation **examines the amount of treatment itself**, using claims data to compare organizations, track patients moving between providers, and measure treatment intensity over time. Each story moves closer to a question that administrative controls alone cannot answer. Not just whether an hour was billed correctly, or whether it was reimbursed at an appropriate rate, **but whether the treatment intensity was supported by the child's clinical needs.** The Journal hasn't answered that final question. Its benchmark measures utilization, not clinical appropriateness. But it has demonstrated how much can be learned from claims data alone, and how **exposed providers can be** when they cannot connect those patterns to clinical evidence. The industry doesn't need to wait for a universally accepted standard for appropriate treatment hours before improving that visibility. Providers can begin by **connecting the clinical and operational information** they already collect. Platforms can make those connections easier, more consistent, and more useful across entire organizations. **If providers don't build that capability, payors and regulators will continue making decisions with the data they have, rather than the data they need.** --- *Source: The Wall Street Journal, “Autism-Therapy Workers Say Big Chains Pressured Them to Boost Treatment Hours,” October 20, 2026.* *Previously on MissionViewpoint:* [*The Medicaid Autism “Racket”? A Systems & Process Problem Hiding Inside the Headlines*](https://www.missionviewpoint.com/the-medicaid-autism-racket-a-systems-process-problem-hiding-inside-the-headlines/) *(March 2026) and* [*How Did a $792 Hourly Rate for Autism Services Make It Through?*](https://www.missionviewpoint.com/how-did-a-792-hourly-rate-for-autism-services-make-it-through/) *(June 2026).*