New CMS Medicaid Rate Guide Takes Effect July 1: What MLTSS and ILOS Documentation Shifts Mean for Actuaries Certifying Now

The CMS 2026-2027 Medicaid Managed Care Rate Development Guide, published February 19, 2026, applies to all rating periods beginning between July 1, 2026 and June 30, 2027. It requires separate service category documentation for MLTSS programs and a new standalone ILOS cost certification, converting discretionary data-gathering into formal actuarial opinion deliverables. The certifications it governs are being written against a base period drawn from a population that no longer exists.

Key Takeaways

• Appendix B makes ILOS documentation a signed deliverable, with the ILOS Cost Percentage defined as total ILOS expenditures over total actual capitation payments for the rating period.
• Blended MLTSS rate cells are no longer sufficient. Four service categories now need separate documentation, each with its own data source, trend, and credibility assessment.
• A new ILOS enters its first certification cycle with zero state-specific experience, so the actuary is opining on analog data and clinical assumptions rather than the state's own record.
• CMS will issue formal comment letters where ILOS rates lack state-specific utilization data, suspending approval until a data collection plan is filed.
• National enrollment fell from roughly 94 million in early 2023 to about 67 million by mid-2026, with state disenrollment rates from 12% in North Carolina to 57% in Montana.

What the Guide Adds

Twenty-five states operate MLTSS programs as of 2025, up from 12 in 2011, delivering long-term services and supports through capitation to elderly adults, people with physical disabilities, and people with developmental disabilities. Earlier guides let states present blended rate cells aggregating costs across service types. The 2026-2027 guide requires each service category documented and justified on its own.

The four are personal care, home health, residential services including assisted living and adult family care homes, and facility-based services including nursing facility care. Their cost drivers do not move together. Personal care is largely a labor cost sensitive to minimum wage and home care workforce turnover. Home health mixes clinical staffing with aide services and often tracks post-acute discharge patterns rather than custodial trend. Residential placement follows bed availability and waiver capacity. Facility-based care follows clinical acuity, length of stay, and state staffing mandates that have moved materially since 2020.

A blended cell carries all four assumptions without exposing any of them, which is why attribution after the fact is nearly impossible. When a blended MLTSS rate proves inadequate, nothing in the certification distinguishes a personal care utilization surge from a residential placement shift or a nursing facility cost spike. Each category section now needs its data source and disaggregation method, category-specific trend, a credibility assessment, and a reconciliation into the blended total the contract actually uses.

The second change is Appendix B. Every certification must carry a summary of actual plan ILOS costs from claims and encounter data, the ILOS Cost Percentage computed as total ILOS expenditures excluding short-term IMD stays over total actual capitation payments, and an actuary's certification that the data is accurate and consistent with CMS guidance.

Certifying a Cost With No State Experience

Appendix B is the more consequential of the two because it converts a data-gathering question into a signed opinion. Prior guides treated ILOS costs as a rate component and left the documentation standard to the actuary. The new structure specifies the elements, requires the signature, and attaches a review consequence.

The problem is sharpest exactly where states are most active. When a state launches a hospital-at-home program, a specialized housing placement, or expanded respite, the first certification happens with no state-specific utilization or cost data for that item. The opinion has to rest on analog data from comparable programs elsewhere, plan projections, and clinical assumptions, and ASOP No. 49 sets a credibility bar that first-cycle data is unlikely to clear without qualification.

Three paths follow, each with a different cost. Certify on the best available data with a disclosure naming the credibility limitation, the analog basis used, and the uncertainty range. Issue a qualified opinion constraining what the certification covers for the ILOS component. Or decline to certify that element, which leaves the service in fee-for-service until state-specific data accumulates and removes it from the managed care benefit the enrollee actually uses.

The March 11, 2026 Managed Care Monitoring and Oversight CIB made the consequence explicit: rates for ILOS not supported by state-specific utilization data draw formal comment letters requiring a data collection plan before approval. That is a shift from completeness review, where CMS checked that components were present and reasoning documented, to a mechanism that suspends approval.

The exposure runs on the calendar. A state submitting in late spring for a July 1 effective date and receiving comment letters in May or June cannot finalize rates until the data collection plan is filed, while the MCO has already committed network capacity and staffing against expected capitation. Amendment timelines documented across the 2023-2024 unwinding ran three to nine months depending on contract provisions and CMS review queues, which is the length of the gap the plan finances.

The Base Period Describes a Population That Has Gone

Every requirement above assumes the underlying experience data can support it, and that is the part the guide cannot legislate.

National Medicaid enrollment peaked near 94 million in early 2023 when continuous enrollment ended, then fell to roughly 67 million by mid-2026, with about 27 million disenrolled and state rates spanning 12% in North Carolina to 57% in Montana. Claims forming 2026 base periods therefore describe neither the pre-COVID mix, which excluded millions retained under continuous enrollment, nor a post-OBBBA equilibrium that does not yet exist.

The shift has not finished. OBBBA work requirements and a six-month redetermination cycle beginning in late 2026 will keep composition moving through the rating period the certification covers, so the documentation standard rises at the point the data supporting it is least stable.

The rate range constraint then closes the remaining flexibility. Federal regulations at 42 CFR 438.4(c)(2)(iii) cap the spread between the upper and lower bounds of a certified range at 5% of the lower bound, with up to 1% of in-period plan adjustment inside it. That cap exists to stop states certifying ranges so wide they carry no information, and under current conditions it can be narrower than honest uncertainty warrants.

The consequences are asymmetric in a specific way. If actual enrollment lands 8% below certified assumptions rather than the 5% the range accommodates, a fresh certification is required before adjusted capitation can begin, on the same amendment clock that ran three to nine months during the unwinding. States managed that period with MLR corridors, risk corridors, and mid-year amendments, and the same instruments are what the 2026 certification has to anticipate at signature rather than discover at reconciliation.