clinicians.build interactive · July 31, 2026

The 26%

Nebraska's Medicaid director is proudest of a number that isn't a termination count. It's 74% — the share of enrollees his state confirms compliant without asking them for anything. The other 26% get a letter and thirty days, and that cohort is the whole story. Here are 1,000 renewals, drawn to the real CMS ledger, and what happened on each side of that line.

38.0M
renewals cleared ex parte — never asked
31.0M
the automation could not clear
27.4%
of that cohort lost coverage on paperwork
0%
of the cleared cohort did — by definition

1,000 renewals, one state, twelve months

Every dot is one-thousandth of the renewals that came due in that state between May 2024 and April 2025. Press play and the automation runs first.

cleared ex parte renewed on a returned form terminated — procedural found ineligible still pending
 
renewals due (12 mo)
cleared without an ask
procedural loss inside the un-cleared cohort
paperwork losses per ineligibility finding
80/20 lens — and the trap in it

A renewal the state closes from data it already holds cannot end in a procedural termination. A renewal that requires the enrollee to act ends in one about 27% of the time nationally — and in Nevada, Maryland, Washington, Minnesota and Missouri, above 50%. Every point of ex parte reach is thousands of people who never have to prove anything to keep their insurance.

Now the part that should make you suspicious of your own metric. Washington has the highest ex parte rate in the country (80.6%) and the third-highest procedural loss rate inside its un-cleared cohort (54.7%). Pennsylvania has the second-lowest ex parte rate (13.9%) and the lowest un-cleared loss rate (7.1%). Automate harder and what's left is the hard cases — a selection effect, not a failure. Judge a state on the bar chart alone and you will rank them backwards.

Across the whole population the ordering flips again: Pennsylvania loses 6.2% of all renewals due to paperwork, Washington 10.6%, Nevada 14.0%. Which is the honest read: ex parte reach explains roughly a quarter of the variance in who loses coverage, and notice design, response windows and follow-up explain the rest.

Inside the un-cleared cohort, by state

Share of renewals the automation did not clear that ended in a procedural termination — not an ineligibility finding, a paperwork one. Click any state to load it into the field above.

Bars are on a common 0–60% scale. Hover for the underlying counts. Read this ranking with the selection effect in mind: a state near the top may simply have automated away every easy case.

Now add a second gate

Everything above is the renewal machine, which has been running for years and is measured monthly. Starting today, a second gate sits in front of it in 44 states: the community engagement requirement. Nebraska went live in May; its first terminations — about 200 people — land tomorrow. Twenty million more follow on a January 1, 2027 deadline.

The off-ramp is the medically frail exemption, and it is implemented as a diagnosis code list, written state by state. Nebraska's layers severity on top of the codes — not just a cancer diagnosis, but whether it produced an inpatient stay, whether there's a wheelchair, whether the condition is currently doing something to you. Other states' lists, per Gonshorowski, include wrist splints. Or every diabetes code, flat.

why a clinical informaticist should care

Two states, one statute, one phrase — and two completely different populations through the door. That is a value-set design decision doing the work of a policy decision. There is no steward, no version number, no change log, and no published sensitivity. The renewal data on this page exists because CMS built a reporting requirement for it. Nothing equivalent exists for the exemption list.

The build: pull your state's medically-frail code list out of the state plan amendment appendix or the eligibility manual, run it against a Synthea cohort or a de-identified count from your own registry, and answer one question — what fraction of your Medicaid panel clears it automatically? If that number is low, your ED is the appeals process, and you should be able to say so with a figure attached.

“Automating exemptions just automates the denial.” It genuinely can. But the counterfactual isn't a careful human — it's a form nobody receives, or receives twice, or can't read. The chart above is what the manual path actually produces. What it can't show you is the person who assumed she'd lost coverage and stopped going to appointments while still exempt; she never files an appeal, so she appears in no metric on this page. A zero appeals count is at least as consistent with people giving up as with determinations being right, and nothing in the current instrumentation separates the two.

Three caveats that matter

1. “Un-cleared” is not exactly “got a letter.” CMS reports ex parte renewals, so the residual (renewals due minus ex parte renewals) includes a small group the state disenrolled on data alone without a form. It is a close proxy for the letter cohort, not an exact one.

2. Nebraska's 74% is a different denominator than the 42.8% shown here. Gonshorowski's figure is work-requirement compliance clearance using income data; this is the twelve-month share of renewals due that closed ex parte. Same machinery, different question.

3. States amend these numbers. Outcomes are reported roughly three months after the renewal month; this uses the latest submission per state-month. Several states carry footnotes for held terminations or partial reporting, which are not modeled here.

Read the primary source: Tradeoffs on Nebraska →