MHPAEA 2026: Health Actuaries Must Now Prove Parity Holds

From reviewing dozens of NQTL comparative analyses submitted to regulators, patterns emerge in where plans most frequently fail the parity test, particularly in prior authorization and network adequacy metrics. The Mental Health Parity and Addiction Equity Act (MHPAEA) has been federal law since 2008, but the September 2024 final rule fundamentally changes what compliance means for health actuaries. For the first time, plans must prove that their nonquantitative treatment limitations (NQTLs) are no more restrictive in practice, not merely in written policy language, using actual outcomes data.

For plan years beginning on or after January 1, 2026, the new provisions on discriminatory factors, evidentiary standards, relevant data evaluation, and the meaningful benefits standard took effect. Plans operating on calendar-year cycles are now in their first compliance year under the expanded framework. The actuarial work involved is substantial: building claims denial rate comparisons across six benefit classifications, quantifying prior authorization approval disparities between mental health/substance use disorder (MH/SUD) and medical/surgical (M/S) benefits, assembling network adequacy evidence, and packaging the results into comparative analyses that named fiduciaries must review and certify.

This article provides a technical walkthrough of what the 2026 provisions require, where the data collection and analysis work concentrates, and how to build a compliance infrastructure that will hold up under DOL scrutiny, given that every comparative analysis submitted to date has initially been found deficient.

100%

Initial Failure Rate on DOL-Reviewed Analyses

5.2x

BH Out-of-Network Rate vs. M/S (Inpatient)

$100/day

IRC §4980D Excise Tax Per Affected Individual

What Changed: The 2024 Final Rule’s Two-Phase Effective Dates

The Departments of Labor, Health and Human Services, and Treasury published the MHPAEA final rule on September 23, 2024 (89 FR 77586), implementing a two-phase effective date structure that health actuaries need to understand precisely.

Phase 1, plan years beginning January 1, 2025: General NQTL comparative analysis requirements, fiduciary certification, updated content requirements for comparative analyses, and 45-day response deadlines for providing comparative analyses upon request by participants or regulators.

Phase 2, plan years beginning January 1, 2026: The meaningful benefits standard, prohibition on discriminatory factors and evidentiary standards, relevant data evaluation requirements (outcomes data), and related provisions for comparative analyses. The final rule also extends MHPAEA requirements to individual health insurance coverage for the first time, significantly broadening CMS enforcement scope.

The practical implication: plans with calendar-year cycles are now operating under the full weight of both phases simultaneously. The Phase 2 provisions are where the actuarial workload concentrates, because they demand quantitative evidence of parity in outcomes, not just facially neutral policy language.

The NQTL Standard: “No More Restrictive” in Design and Operation

The core regulatory standard sounds simple: plans cannot impose NQTLs on MH/SUD benefits that are more restrictive, as written or in operation, than the predominant NQTLs applied to substantially all M/S benefits in the same classification. The "in operation" language is the 2024 addition that transforms the compliance exercise from a design review into a data-driven analysis.

NQTLs subject to this standard include prior authorization requirements, step therapy protocols, concurrent review for inpatient and outpatient services, network tier structures and provider admission standards, out-of-network reimbursement rate determination methodologies, and fail-first or treatment sequencing requirements. Plans must demonstrate parity across each of these NQTL categories within six benefit classifications:

Inpatient, in-network
Inpatient, out-of-network
Outpatient, in-network
Outpatient, out-of-network
Emergency care
Prescription drugs

From tracking NQTL compliance over recent years, the most common failure points concentrate in three areas. First, prior authorization: plans frequently require prior authorization for outpatient MH/SUD services at rates far exceeding comparable M/S services. Kennedy Forum data indicates MH services are 5.4 times more likely to require prior authorization than comparable medical services. Second, network tier structures: provider admission standards and credentialing processes for MH/SUD providers often impose stricter requirements than those for M/S providers, which contributes directly to the network adequacy disparity. Third, out-of-network reimbursement: plans commonly use different reimbursement methodologies for MH/SUD versus M/S providers, with MH/SUD rates benchmarking to lower percentiles of usual and customary databases.

The DOL’s 2024 Report to Congress identified six priority NQTL areas under active enforcement review: prior authorization for inpatient services (in-network and out-of-network), concurrent care review, provider network admission standards and reimbursement rates, out-of-network reimbursement rate determinations, MH/SUD provider network adequacy standards, and impermissible exclusions of key treatments including applied behavior analysis (ABA) therapy, medication-assisted treatment (MAT) for opioid use disorder, and nutritional counseling for eating disorders.

Data Requirements: Building the Comparative Analysis

The Phase 2 relevant data evaluation requirements, effective for 2026 plan years, specify the quantitative evidence plans must collect and analyze. This is where health actuaries are most directly engaged, because the work product is fundamentally actuarial: collecting claims data, constructing comparable metrics, and evaluating whether observed differences are statistically meaningful.

The required data elements include:

Data Category	Required Metrics	Comparison Basis
Claims Denials	Number and percentage of claims denied, broken out by classification	MH/SUD vs. M/S within each of the six classifications
Prior Authorization	Approval and denial rates, turnaround times	MH/SUD vs. M/S by classification and service type
Provider Reimbursement	Reimbursement rates for comparable services	Benchmarked to Medicare fee schedules or equivalent reference standard
Network Adequacy	Out-of-network utilization rates, in-network clinicians actively submitting claims, time and distance standards, provider-to-member ratios	MH/SUD provider network vs. M/S provider network
Appeals and Grievances	Rates of appeals and overturn outcomes	MH/SUD vs. M/S, including internal and external review

The network composition requirements are particularly detailed. For NQTLs related to network composition, the rule requires specific data elements including out-of-network utilization rates for MH/SUD versus M/S, the percentage of in-network clinicians who are actively submitting claims (a critical distinction from directory counts, since ghost networks remain pervasive), time and distance standards, and reimbursement rate comparisons. Milliman’s 2019 parity research documented the scale of the problem: behavioral health inpatient services were 5.2 times more likely to be rendered out-of-network compared to M/S inpatient, up from 2.8 times in 2013, an 85% increase in the disparity over five years. Substance use disorder inpatient was worse at 10.1 times, up from 4.7 times in 2013.

Remedial action trigger. If the data shows material differences in access to MH/SUD benefits compared to M/S benefits, plans must take "reasonable action" to address those differences, document the actions taken, and demonstrate that the actions produced results. This creates an ongoing monitoring obligation, not a one-time compliance check.

For health actuaries building this infrastructure, the data integration challenge is often the hardest part. Prior authorization data typically resides in utilization management systems separate from claims adjudication platforms. Network adequacy data requires provider directory verification against actual billing patterns. Reimbursement comparisons need a defensible benchmarking methodology, and Medicare fee schedules provide the most widely accepted reference point. Building automated data pipelines that can produce these comparisons on a quarterly or semi-annual cadence, rather than assembling them manually for each regulatory request, is the infrastructure investment that distinguishes sustainable compliance from ad hoc scrambling.

Prohibition on Discriminatory Factors

The 2026 effective date provision that may prove most challenging for actuaries is the prohibition on discriminatory factors. A factor or evidentiary standard is discriminatory if the information, evidence, sources, or standards on which it is based are biased or not objective in a manner that discriminates against MH/SUD benefits as compared to M/S benefits.

The proposed rule had included two specific exceptions: one for generally recognized independent professional medical or clinical standards, and another for reasonably designed fraud, waste, and abuse measures. The final rule did not finalize these exceptions. This is a significant tightening: plans cannot simply assert that clinical guidelines justify differential treatment of MH/SUD benefits without examining whether those guidelines themselves embed historical bias.

The practical consequence for actuarial work is that historical claims data, utilization patterns, or cost data that reflect pre-existing disparities in MH/SUD access may themselves be discriminatory factors under the rule. If a plan’s prior authorization criteria were historically more restrictive for MH/SUD services, the resulting utilization data reflects suppressed demand, not appropriate utilization levels. Building NQTL design on that suppressed-demand data would perpetuate the disparity and violate the rule.

This means actuaries must either source alternative benchmarks, such as published clinical guidelines from organizations without conflicts of interest, or adjust historical data to remove the effect of prior restrictions. The latter approach requires explicit assumptions and documentation about how the adjustment was performed, the rationale for the adjustment methodology, and the sensitivity of results to alternative adjustment approaches. ASOP No. 25 (Credibility Procedures) and general actuarial judgment about data appropriateness become important guardrails here.

The ERIC v. HHS lawsuit (filed January 17, 2025, in the D.C. District Court) specifically challenges this provision, arguing it exceeded the Departments’ statutory authority. The legal uncertainty does not eliminate the compliance obligation; it does mean that the precise boundaries of the discriminatory factors prohibition may shift depending on litigation outcomes.

Fiduciary Certification: What Named Fiduciaries Must Sign

For plan years beginning January 1, 2025, ERISA group health plans must have their named fiduciaries certify two things about the NQTL comparative analysis. This certification requirement applies to the Phase 1 provisions but carries forward through Phase 2 as plans update their analyses with outcomes data.

The two-prong certification requires:

Prudent process certification: The fiduciary engaged in a prudent process to select one or more qualified service providers to perform and document the NQTL comparative analysis.
Monitoring duty: The fiduciary has satisfied their duty to monitor those service providers as required under ERISA Part 4.

DOL guidance specifies minimum fiduciary actions: personally reviewing the comparative analysis, asking questions about the analysis and discussing it with service providers, understanding the findings and conclusions, and ensuring service providers provide assurance that the analysis complies with MHPAEA requirements.

The liability exposure here is often underappreciated. Plan sponsors remain liable for both the outcome of the analysis and the supervision and selection of third-party administrators (TPAs) and vendors who perform the analysis. From tracking the market, some TPAs do not certify that their comparative analyses comply with MHPAEA, creating a fiduciary gap where the plan sponsor bears full liability for work performed by a vendor that will not stand behind its own output.

For actuaries, this means the comparative analysis work product must be structured to support fiduciary review. That includes clear executive summaries, explicit statements of methodology, transparent presentation of results that a non-actuary fiduciary can understand, and documentation of any limitations or caveats. Actuaries performing this work should consider whether an actuarial opinion or certification letter, structured similarly to a Statement of Actuarial Opinion under statutory reporting, would be appropriate to accompany the comparative analysis and provide the "assurance" the regulation references.

Plans must also provide copies of comparative analyses to state authorities, participants who received adverse benefit determinations, and beneficiaries upon request, with a 45-day response deadline. This means the analysis must be in a state of perpetual readiness, not something assembled reactively when a request arrives.

The Meaningful Benefits Standard

Effective for 2026 plan years, the meaningful benefits standard requires that if a plan provides any benefit for a MH/SUD condition, it must provide meaningful MH/SUD benefits in every classification in which M/S benefits are provided. This is a cross-classification coverage completeness test.

The regulation’s illustrative example is instructive: a plan covers diagnosis and treatment for eating disorders but specifically excludes nutrition counseling for eating disorders in the outpatient in-network classification. Because nutrition counseling is a primary treatment for eating disorders, and the plan generally covers primary treatments for M/S conditions in that classification, the exclusion violates the meaningful benefits standard.

Regulators have flagged three treatment categories that frequently fail this test: applied behavior analysis (ABA) for autism spectrum disorder, medication-assisted treatment (MAT) for opioid use disorder (including buprenorphine and methadone), and nutritional counseling for eating disorders. Plans that carve out or exclude these treatments while covering analogous primary treatments for M/S conditions in the same classification face violation risk.

For health actuaries, the meaningful benefits standard intersects with benefit design and pricing work. Adding previously excluded MH/SUD treatments to a benefit schedule has direct cost implications. Actuaries pricing these additions should be cautious about using suppressed utilization data from periods when the treatment was excluded, since the exclusion itself depressed observed demand. External utilization benchmarks from plans that already cover these treatments, or published clinical literature on expected treatment prevalence, provide more appropriate pricing bases.

Enforcement Reality: The 100% Initial Failure Rate

The single most striking data point in MHPAEA enforcement is that every comparative analysis reviewed by DOL EBSA or CMS has been initially found insufficient or noncompliant. This has been consistent across multiple reporting periods.

The 2024 Report to Congress covering the period ending July 2023 documented:

39 initial letters requesting comparative analyses (combined EBSA and CMS)
55 insufficiency letters covering more than 40 NQTLs
32 initial determination letters finding MHPAEA violations
3 final noncompliance determinations (all CMS, all to a single health insurance issuer)
Over 120 total enforcement letters during the reporting period

EBSA devotes nearly 25% of its enforcement program to MHPAEA NQTL work, reflecting the regulatory priority. The agency has identified common deficiencies that health actuaries should treat as a checklist of what not to do:

No analysis prepared at all
Superficial or vague comparisons lacking meaningful evaluation
Failure to identify specific MH/SUD and M/S benefits affected by each NQTL
Focusing only on similarities rather than explaining differences
Conclusory assertions lacking specific supporting evidence
Missing benefit classifications (analyses that skip one or more of the six classifications)
Unexplained or poorly organized supporting documentation

CMS enforcement data from the same period shows 22 comparative analysis requests, 10 insufficiency letters, 19 initial determination letters finding violations, and 3 final noncompliance determinations. Cumulatively, EBSA reports that 7.6 million participants across 72,000 plans have directly benefited from corrections under CAA enforcement efforts since February 2021.

Specific penalties for noncompliance include the IRC Section 4980D excise tax of $100 per day per affected individual. For an unintentional failure with reasonable cause, the annual cap is the lesser of 10% of the aggregate amount paid or incurred by the employer for group health plans, or $500,000. For violations identified on IRS examination, the minimum tax is $2,500, enhanced to $15,000 for more-than-de-minimis violations. State-level enforcement has also produced significant penalties: Washington State fined Regence Blue Shield $550,000 for alleged MHPAEA violations, and United Behavioral Health agreed to a $15 million settlement plus corrective actions in DOL’s first federal parity litigation.

The ERIC Lawsuit and Current Enforcement Status

On January 17, 2025, the ERISA Industry Committee (ERIC) filed suit in the D.C. District Court (Case No. 1:25-cv-00136) challenging several provisions of the 2024 final rule, arguing they exceeded the Departments’ statutory authority. On May 12, 2025, the court granted abeyance, and the Departments announced they will not enforce the new 2024 final rule provisions pending resolution of the lawsuit plus an additional 18 months.

This enforcement pause applies only to the provisions new to the 2024 final rule. The following remain fully enforceable: the 2013 MHPAEA regulations, CAA 2021 NQTL comparative analysis requirements, the 45-day response deadline for providing analyses upon request, participant notification requirements, and all statutory MHPAEA parity obligations. Private litigation under ERISA Section 502 (29 U.S.C. Section 1132) also remains available, meaning participants and beneficiaries can bring claims directly.

For health actuaries, the enforcement pause does not eliminate the compliance obligation; it creates a window to build compliant infrastructure without immediate regulatory pressure on the new provisions. Plans that use this window productively will be in a stronger position when enforcement resumes. Plans that treat the pause as permission to delay will face a steep implementation curve when the abeyance lifts, likely with the full weight of the 2026 outcomes data requirements taking effect simultaneously.

Reimbursement Disparities: The Data That Drives the Rule

The regulatory urgency behind the 2024 final rule reflects documented and worsening disparities in MH/SUD access relative to M/S benefits. Understanding the underlying data helps actuaries calibrate the analytical framework and anticipate where their plan’s data is most likely to show parity gaps.

Milliman’s parity research, commissioned by the Kennedy Forum and widely cited by regulators, documented that behavioral health office visits were 5.4 times more likely to be out-of-network compared to primary care office visits. For children, that ratio reached 10.1 times. Substance use disorder services showed even larger gaps: SUD inpatient services were 10.1 times more likely to be out-of-network than M/S inpatient, a 113% increase in the disparity between 2013 and 2017.

Reimbursement rate analysis reveals a structural driver behind network adequacy failures. Primary care reimbursements were 23.8% higher than behavioral health reimbursement rates relative to Medicare benchmarks in 2017, up from 20.8% in 2015. In 11 states, primary care reimbursements exceeded behavioral health office visit rates by more than 50%. A Congressional Budget Office analysis found that commercial and Medicare Advantage plans paid 13 to 14% less than Medicare fee-for-service for mental health care while paying 12% more for other specialties. When MH/SUD providers are systematically paid less than M/S providers for comparable cognitive evaluation and management services, fewer MH/SUD providers participate in networks, driving the out-of-network utilization gap that the parity analysis must then document.

Provider participation data reinforces the pattern: 42% of all Medicare physician opt-outs are psychiatrists, despite psychiatrists comprising a small fraction of all physicians (KFF). Only 35% of psychiatrists accepted Medicaid in 2014-2015, compared to 73% of primary care physicians. Over 50% of providers listed in Medicaid mental health directories had not seen patients in the previous year, a ghost network problem that the rule’s requirement to track in-network clinicians "actively submitting claims" is specifically designed to address.

Mental health and substance use treatment spending has remained a small fraction of total healthcare spending at 2.2 to 2.4% for MH and 0.7 to 1.0% for SUD, never exceeding 1%. During the same period (2013-2017), suicides increased 14.6% to 47,173 and substance-related deaths surged 45.5% to 109,813. This combination of flat spending and rising morbidity provides the policy context for why regulators are demanding outcomes-based evidence.

Compliance Roadmap for Health Actuaries

Building a sustainable MHPAEA compliance infrastructure requires coordinated effort across data engineering, actuarial analysis, legal review, and fiduciary governance. From reviewing the common deficiency patterns in DOL enforcement letters, here is the sequencing that produces the most defensible work product.

Phase 1: Data collection infrastructure (months 1-3). Map data sources for each required metric: claims denial rates from the adjudication system, prior authorization data from utilization management platforms, provider reimbursement from fee schedule and claims payment files, network adequacy from provider directory and billing data, and appeals data from grievance tracking systems. Build automated extraction pipelines that can produce standardized output on a quarterly cadence. The single biggest source of deficiency findings is missing data, so completeness across all six benefit classifications is the first priority.

Phase 2: Comparative analysis methodology (months 2-4). Design the analytical framework for each NQTL category. For prior authorization, the comparison must examine both the written criteria (which services require authorization for MH/SUD versus M/S) and operational outcomes (approval rates, denial rates, turnaround times). For network adequacy, develop provider-to-member ratio and time-and-distance metrics that use actively billing providers rather than directory counts. For reimbursement, establish the benchmarking methodology against Medicare fee schedules or another defensible reference standard. Document the methodology in sufficient detail that it could be replicated by an independent reviewer.

Phase 3: Initial analysis and gap identification (months 4-6). Run the comparative analysis across all NQTL categories and classifications. Identify where material disparities exist. For each disparity, determine whether it reflects a design issue (the NQTL criteria themselves are more restrictive for MH/SUD) or an operational issue (the criteria are facially neutral but produce disparate outcomes). This distinction matters because the remedial action differs: design issues require policy changes, while operational issues may require changes to claims processing, utilization management vendor practices, or provider network recruitment.

Phase 4: Remedial action and documentation (months 6-9). For identified disparities, develop and implement corrective actions. Document the actions taken and establish monitoring metrics to evaluate whether the actions produce improvement. The regulation requires not just taking action but demonstrating results, so build the monitoring cadence into the ongoing compliance calendar.

Phase 5: Fiduciary review and certification (months 9-12). Package the comparative analysis, methodology documentation, results, remedial actions, and monitoring plan for fiduciary review. Prepare an executive summary that enables non-actuarial fiduciaries to understand the findings and conclusions. Support the fiduciary certification process by being available to answer questions and provide additional context.

Ongoing: Quarterly monitoring and annual refresh. The outcomes data evaluation requirement is not a one-time exercise. Establish a quarterly monitoring cadence for key metrics (denial rates, authorization rates, out-of-network utilization) and a full annual refresh of the comprehensive comparative analysis. Maintain the analysis in a perpetual state of readiness for the 45-day response obligation.

No Actuarial Standard of Practice Exists for Parity Analysis

One notable gap in the professional infrastructure: no formal Actuarial Standard of Practice (ASOP) has been issued specifically for MHPAEA parity analysis. The American Academy of Actuaries submitted comments to HHS on the interim final rules regarding the use of the terms "qualified and licensed actuary" and "member in good standing," but no ASOP comparable to ASOP No. 6 (Measuring Retiree Group Benefits Obligations) or ASOP No. 45 (The Use of Health Status Based Risk Adjustment Methodologies) provides specific guidance for parity comparative analysis methodology.

Actuaries performing this work must rely on general ASOPs, particularly ASOP No. 23 (Data Quality), ASOP No. 25 (Credibility Procedures), and ASOP No. 41 (Actuarial Communications), plus the regulatory guidance itself. The absence of a dedicated ASOP means there is no consensus standard for how to measure "material" disparity, what statistical tests are appropriate for comparing denial rates across classifications, or how to adjust for differences in case mix between MH/SUD and M/S populations that might explain observed outcome differences.

Milliman’s February 2025 white paper on navigating the latest mental health parity rules, authored by Katie Matthews, Travis Gray, and Stoddard Davenport, represents the most detailed published actuarial guidance currently available. Until the Actuarial Standards Board addresses this gap, the Milliman methodology and the DOL’s own enforcement guidance function as the de facto practice standard.

Why This Matters for Health Actuaries

MHPAEA compliance is no longer a legal department exercise with occasional actuarial input. The 2024 final rule’s outcomes data requirements place quantitative analysis at the center of the compliance framework. Health actuaries are the professionals best equipped to build the data infrastructure, design the comparative analysis methodology, interpret the results, and communicate findings to fiduciaries in terms that support informed certification.

The enforcement pause from the ERIC lawsuit provides a practical window, but it also creates a risk that plans will defer the infrastructure investment until enforcement resumes. Plans that build the data pipelines and analytical frameworks now, during the window, will have the advantage of testing and refining their methodology before it is subject to regulatory scrutiny. Plans that wait will face a compressed timeline when enforcement resumes, potentially with multiple years of outcomes data to retroactively analyze.

The underlying disparities documented in the Milliman/Kennedy Forum data and the DOL’s own enforcement experience make clear that most plans will find parity gaps when they conduct honest comparative analyses. The question is not whether gaps exist but how plans identify, quantify, and remediate them. That is fundamentally actuarial work, and health actuaries who develop expertise in MHPAEA compliance analysis are positioning themselves in a practice area with sustained demand and growing regulatory attention.