If you're like most employers I talk to, you assume mental health parity is a carrier problem. You buy a plan that looks compliant: same deductibles, no separate visit limits, and you're done. The Department of Labor thinks you're off the hook? Not anymore.
Employers review the plan document and stop there. The real parity crisis sits in the automated workflows of your Benefits Administration system, your utilization management vendor, and even your EOB generator. These systems are quietly enforcing nonquantitative treatment limitations (NQTLs) that violate the Mental Health Parity and Addiction Equity Act, and most compliance audits never catch it.
Let me give you a concrete example. Your software treats a medical pre-certification request and a behavioral health pre-certification request as equal "transactions." But the logic governing them is wildly different. That difference is now a legal liability.
The Data Silo Parity Breach
Since the Consolidated Appropriations Act, 2021, any plan that applies an NQTL to mental health or substance use disorder benefits has to perform and document a comparative analysis against its medical and surgical NQTLs. That obligation sits with the plan sponsor, not the vendor, even when a carve-out administers the behavioral benefits. You can't just assert the processes are the same. You have to show the system applies them no more stringently than it applies to medical and surgical benefits. Here are three breakdowns I've seen in nearly every audit I've conducted.
1. The Single Vendor Fallacy (Mental Health Carve-Outs)
You might use one TPA for medical and a separate carve-out vendor like Lyra or Spring Health for behavioral. Each has its own pre-certification algorithm. On paper, both use medical necessity. But the systems don't talk to each other.
The medical system has no EAP history. The behavioral system has no PCP diagnosis. This fractured data environment creates a de facto higher barrier for mental health because the system has less data to approve a claim. Behavioral health claims get denied far more often than medical claims under the same "medical necessity" standard, close to twice as often in some analyses.
- Medical logic example: "Deny hip replacement if patient hasn't failed conservative care."
- Behavioral logic example: "Deny inpatient mental health stay if patient hasn't tried outpatient therapy first."
The fix: Demand side-by-side decision trees from your vendors. If the behavioral review uses fewer data inputs or a different credential level, you have a parity violation.
2. The Prior Authorization Binary Trap
Most BenAdmin systems, Workday and ADP among them, apply prior authorization rules as a simple binary: "Condition X requires auth" vs. "Condition Y does not." This hides the real problem.
The NQTL rules require comparing the frequency and intensity of reviews, not just whether they exist. For medical, you might review 2% of spine surgeries. For behavioral, you might review 90% of inpatient mental health stays.
Even worse, the system's routing logic often prioritizes medical over behavioral. A behavioral pre-auth might get routed to a less experienced nurse reviewer because the software assigns "low complexity" flags to mental health codes. Meanwhile, a medical pre-auth goes to a physician. The system enforces this disparity through automated routing rules.
- Audit the logic flow. Ask the vendor for the "decision tree" for both medical and behavioral pre-certs.
- Check if the trees are identical in depth and review level.
- Count the number of human overrides allowed for each path.
The fix: If the trees aren't identical, you have an NQTL gap.
3. The EOB Language Disparity
The Explanation of Benefits (EOB) is the user interface of your benefits system. It's the only proof the member sees.
When a medical claim is denied, the system typically generates a specific adjustment reason code: the procedure is excluded, the authorization is missing, or the diagnosis does not match the billing code. When a behavioral claim is denied, many older systems default to a generic "PR-1" (Deductible Amount) or a flat "not medically necessary" note.
Why? Many admin systems lack granular mental health denial codes. They simply don't have the logic to explain why therapy was denied, for example, "frequency exceeded plan maximum without documented clinical need."
This lack of granular data makes a legal comparative analysis impossible. You cannot prove parity if your system cannot generate the same data required to compare denial rationales.
The fix: Force your vendor to update EOB language so it uses identical denial wording across all claim types. Then run a 12-month comparative analysis of denial codes by category.
Where Enforcement Stands in 2026
The comparative analysis duty predates the 2024 rule. Congress added it in the Consolidated Appropriations Act, 2021, and the Departments have been requesting and reviewing NQTL analyses since. In September 2024 the agencies finalized rules that layer on content requirements, deadlines, and a duty to collect and evaluate data on material differences in access. Those new provisions are on hold. On May 15, 2025, the Departments announced they will not enforce the 2024 rule's new portions until a court resolves the ERISA Industry Committee's challenge, plus an additional 18 months.
That pause does not change the software problem. The comparative analyses the Departments still request under the statute look at the same mechanics: the decision trees, routing rules, and denial data behind every NQTL. An employer who waits for the litigation to end is carrying risk it could fix today. State regulators are not waiting either. Georgia's insurance commissioner reported logging thousands of parity violations across more than 20 insurers in 2025 and 2026.
The Bottom Line
The DOL is no longer just looking at your policy. They are looking at your data. If your system cannot produce a clean, granular, side-by-side comparison of how a denial was reached in the software, you are exposed.
The machine is the problem. Fix the machine, and you fix parity.
About the author: A 15-year veteran of health plan systems, benefits administration, and regulatory compliance. This analysis is based on actual audit findings from employer-sponsored plans.
This article is for general information only and is not legal, tax, or medical advice. Employers should consult their own advisors.