When Inflation Reduction Raises the Cost: A Day-Two Reflection From the SNP Alliance Fall Forum

On day two of the SNP Alliance Annual Fall Forum, I sat in on "Impacts of IRA Implementation on Special Needs Plans," moderated by Mark Newsom, President and Founder of Health Evaluations, LLC. The speakers were Holly Brenier of Associated Care Ventures, Inc., and Ann Nguyen, Vice President of Medicare Part D at Health Care Service Corporation (HCSC).

The thought I keep coming back to was one Holly shared about what implementing the Inflation Reduction Act looks like at a small plan. In short: for a small plan, inflation reduction can mean cost increasing.

The Math of a Small Plan

Every change the law requires has to be built, whether a plan serves two million members or two thousand: benefit configuration, enrollment workflows, member notices, billing logic, call center scripts, pharmacy coordination, reporting, testing, staff training, and vendor system changes.

That cost is mostly fixed. What shrinks is the return: fewer members to use any change and to spread its cost across. When executing a change costs more than members engage with it, a policy built to lower costs raises them for the plan that carries it.

A small plan pays the same entry fee as a large one and spreads it across a fraction of the members.

What "Small" Looks Like

Associated Care Ventures is the parent organization of Simpra Advantage, an Alabama PPO offering an institutional SNP and a dual-eligible SNP. Its reported enrollment:

Source: CMS Special Needs Plan (SNP) Comprehensive Report, pulled from MimiLabs (dataset partcd.snpdata) on October 22, 2026. Reported counts; CMS suppresses one small I-SNP plan in each month shown, so true totals are higher.

Simpra's reported growth is almost entirely D-SNP growth, while the I-SNP base stays roughly flat.

As of October 2026, Simpra reported at least 3,554 SNP members, most of them in its I-SNP. Every SNP covers Part D, so the Part D redesign applies to all of them, just as it does for a plan with millions.

Small SNP organizations are common by count, even though enrollment is highly concentrated among a few large parents. Across the 130 parent organizations and one CMS summary group with reported SNP enrollment in October 2026:

Source: Author analysis of the CMS Special Needs Plan (SNP) Comprehensive Report and Monthly Enrollment by Contract/Plan/State/County, October 2026, joined by contract identifier; pulled from MimiLabs (datasets partcd.snpdata and partcd.cpsc_combined) on October 22, 2026. Enrollment figures are reported counts and exclude CMS-suppressed cells, so totals are lower bounds. The under-5,000 row includes CMS's non-attributable "Under-11" summary group (122 members); organizations whose enrollment was entirely suppressed are excluded.

Small SNP parent organizations are common by count; reported SNP enrollment is concentrated among a few very large organizations.

More than half of SNP organizations have fewer than 5,000 reported members, and together they serve 1.5% of reported SNP members. Four serve more than two-thirds. The largest has about 3.2 million reported SNP members, about 890 times Simpra's.

Now put Holly's point into numbers. Suppose a new requirement costs $500,000 to build and run in its first year. The figure is illustrative, but the math is not. That is about $140 per member for a 3,554-member plan, and about 16 cents for a 3.2-million-member SNP book. Same requirement. Same deadline. A nearly 900-fold difference per member.

The same fixed implementation cost creates a nearly 900-fold difference in per-member burden.

Why This Lands Hardest on SNPs

SNPs are small by design. Specializing in dual eligible, chronically ill, or institutionalized members is the point, and it makes every fixed cost heavier.

Some changes also reach SNP members least. Every Part D sponsor must offer the Medicare Prescription Payment Plan, which spreads out-of-pocket drug costs across the year, with its elections, notices, and billing. But full-benefit dual eligible members already pay little at the pharmacy because of the Low-Income Subsidy, and institutionalized members typically pay nothing. An I-SNP like Simpra's must build and run the program anyway, for very few likely users.

Lower drug costs are still worth pursuing. Holly's point is narrower and harder to argue with: the same requirement costs a different amount per member depending on who carries it.

Where AI Can Change the Math

If the problem is a fixed cost that does not scale down, the answer is to make it smaller, or to share it.

That is where AI earns its place in plan operations: not replacing the people who know this work, but speeding up what they already do. The judgment stays with the experts:

• Reading the rules. AI can read the final rule, HPMS memos, and revised layouts, compare them with last year, and draft the plan's to-do list for the expert to confirm.
• Building and checking the changes. AI can take the first pass at mapping a requirement to the files, fields, and workflows it touches and testing the output, so analysts spend their time on review and decisions.
• Watching the results. AI can check every member every month that CMS paid what the plan expected, and bring only the exceptions to the people who know what they mean.

A large plan staffs a team for each step. A small plan has a few experts doing all of them. AI multiplies their expertise, giving a 3,554-member plan's team more of the reach a 3-million-member plan buys with headcount.

Shared Infrastructure Is Economies of Scale, Shared

The problem does not stop at Part D. Every time CMS changes a file layout, a payment rule, a risk model, or an audit standard, a small plan faces the same work as a large one, with a smaller team.

That is the gap Falcon Health was built to close. On day one, I wrote that a plan's revenue cycle runs backward, through dozens of CMS files and months of retroactive change. Falcon processes all 70+ of those files (MMR, MOR, MAO-004, DTRR, Part D MOR) in one system, and reconciles what CMS paid, member by member, in both directions.

Because that system serves many plans, a change is built once and shared. When CMS revises a file layout or a payment rule, Falcon absorbs it for everyone, so a small plan gets the benefit of scale without the membership to pay for it. Inside Falcon, AI speeds up our experts and the plan's own team, while decisions about what a payment means stay with the people who know the work.

Some of that work is shared even more widely. My colleague and our CTO, Yubin Park, PhD, maintains the open-source hccinfhir package, which reads the 834 eligibility file and scores the 837 directly to a risk score. Any plan can use it, at any size.

The Question Worth Asking

When a change is measured, the usual question is "Did it lower costs?" Holly's point gives policymakers and plan leaders a better one: "Lower costs for whom, and who paid to build it?"

For a 3,554-member Alabama plan built mostly around an institutional SNP, the honest answer can be that the members gained a little, and the plan paid a lot. Policy written for scale can still work for small plans, if it counts the cost of execution, not just the benefit of the change, and if small plans do not have to carry that cost alone.

Every file. Every member. Every dollar. Reconciled.

See Falcon in Action

If your plan is small and the list of changes keeps growing, we'd like to show you what shared infrastructure looks like for your members, your files, and your payments.

Schedule a conversation with Falcon Health

This is my last reflection from the SNP Alliance Fall Forum. Next week, we'll be at the RISE Risk Adjustment Forum. If you'll be there, stop by our booth, #305. I'd love to say hi.