The LEAD RFA is Here

It’s an exciting new model, but our earlier caution still applies.

In January, we wrote about why you shouldn't trust early underwriting on LEAD, organized around five structural unknowns that would determine whether existing REACH or MSSP models could rapidly produce credible LEAD forecasts. The RFA, published March 31, resolves most of those unknowns. 

Some answers are familiar, while others are genuinely novel. The net effect is a benchmark methodology sufficiently different from either REACH or MSSP that responsible underwriting will take real time to get right.

What's New or Noteworthy

Alignment is more frequent. Hybrid alignment allows ACOs to add net new participant TINs twice a year (compared to annually in REACH or MSSP) and allows for monthly additions of voluntarily-aligned beneficiaries (compared to quarterly in REACH). This means a more continuous enrollment and management cycle for ACOs.

Groups will have restrictions on switching options in LEAD. In ACO REACH, some entities (like Pearl) owned both Global and Professional REACH ACOs, allowing practices to “graduate” from Professional (lower risk) to Global (higher risk) as they built savings momentum. LEAD will make that more challenging by preventing TINs from switching between ACOs owned by the same “convener” in a given three-year period.

High Needs is a distinct beneficiary category with its own financial architecture. As suspected, High Needs beneficiaries are carved into a separate benchmark with a concurrent HCC risk model (V2) and a risk score growth cap between 3% and 8% (the exact figure still TBD). The eligibility criteria are expansive, as in REACH, and include mobility impairment, frailty indices, high risk scores, prior unplanned admissions, or prior utilization of extensive SNF stays. Once qualified, a beneficiary remains High Needs permanently. 

The baseline is static, permanently. LEAD locks in a three-year historical baseline (CY2024–2026 for PY 2027 entrants) for the entire performance period. Historical expenditures are recalculated each year based on the current TIN list, but the base years never change. This eliminates the ratchet effect of rebasing.

The Accountable Care Prospective Trend (ACPT) introduces a third trending dimension. Trending is a three-way blend: two-thirds from a realized spending trend and one-third from a prospectively set administrative growth rate, subject to guardrails that widen annually—from +0.3/−0.2 percentage points in PY 2027 to +1.5/−1.0 by PY 2031. This mechanism sustains a "savings wedge" between benchmarks and realized spending.

CARA brings specialty accountability into the ACO framework. CMS Administered Risk Arrangements allow Global Risk ACOs to establish episode-based risk arrangements with specialists. ACOs may either adopt any of a set of pre-determined CMS-constructed episodes, or create novel constructs in negotiation with Preferred Providers. Nothing like this exists in current ACO programs.

Voluntary alignment benchmarks are rebuilt from scratch. REACH used a regional-rate-only benchmark for newly voluntarily aligned beneficiaries. LEAD eliminates that, instead basing voluntary alignment benchmarks on historical spending of currently voluntarily aligned beneficiaries, capped at 10% deviation from claims-aligned benchmarks. 

AI-inferred risk adjustment is no longer theoretical. In January, we noted that the CMMI Director had discussed inferred risk but that the LEAD announcement hadn't signaled this direction. The RFA changes that. CMS will shadow-test an AI-inferred risk model for the Aged & Disabled population in 2028, blend it at one-third weight in PY 2029, two-thirds in PY 2030, and fully replace the traditional HCC model by PY 2031. For a ten-year model, this is a consequential mid-stream transition with no actuarial precedent in Medicare ACO programs. While the details of the inference approach are not known, removing the administrative overhead of risk adjustment marks an important evolution in value-based care.

Why You Should Still Wait on Enrollment Decisions

Our January framework identified five categories that would determine underwriting reliability and speed. The RFA confirms that LEAD borrows selectively from REACH and MSSP while introducing elements present in neither, meaning responsible reconstruction will take more time.

• The alignment methodology is recognizable but adds mid-year TIN additions and restructured voluntary alignment benchmarks requiring new modeling. 
• Benchmark construction uses mechanisms familiar from MSSP’s historical-first approach, with regional or prior savings adjustments layered on, as appropriate. Underlying modifications and eligibility criteria do not map cleanly onto either predecessor.
• The trending approach is an important evolution: no existing underwriting model captures the three-way ACPT blend with annually widening guardrails.
• Risk adjustment starts on familiar V28 ground but layers in a separate V2 of the High Needs concurrent model, variable risk score growth caps, and phased AI-inferred risk—making multi-year forecasting a fundamentally different exercise. 
• And quality measurement adds two phased eCQMs, and a redesigned graduated CI/SEP methodology, and a return of REACH’s High Performers Pool funded by unearned withholds.

The LEAD model’s design is ambitious and an important step forward for Medicare value-based care. But the number of novel, interacting components means credible projections require building new tools for actuarial pricing, not adjusting old ones.

If someone is offering you confident LEAD underwriting today, ask how they're modeling the ACPT guardrail interaction with regional efficiency adjustments. Ask how they're establishing a benchmark for High Needs beneficiaries with a risk model that hasn't been published yet. Ask what assumptions they're making about a risk score growth cap CMS itself hasn't finalized.

The RFA gave us answers. It also gave us new questions. Both deserve rigorous investigation before you commit.