Ten million Americans will lose Medicaid coverage under OBBBA work requirements. A meaningful share of them qualify for a medical-frailty or caregiving-role exemption — but only if someone files the physician attestation. We built that workflow, at scale, with a real physician of record.
Three inputs. Every number below is a model with its source named; replace the defaults with your own and the arithmetic updates. When you want the real cohort, the ROI report below uses your de-identified panel.
The model assumes every at-risk member who is not documented loses coverage, which overstates loss (some will comply or self-attest) and ignores members who lose coverage for non-hour reasons, which understates it. It is for sizing a conversation, not a budget line. Your ROI report replaces these assumptions with your panel.
OBBBA adds work requirements for able-bodied adult Medicaid enrollees. States must verify compliance. Enrollees who fail verification lose coverage. For health systems, disenrollment is revenue lost and uncompensated care incurred. The exemption process is the relief valve — if you file.
CMS's interim final rule took effect July 31, 2026. States are building auditable condition lists and wiring the prior twelve months of claims and encounter data in as a verification source. What is in your structured record is now an input to the state's determination.
The window to identify at-risk patients, flag exemption candidates, and pre-file attestations before first redetermination.
States begin applying work-verification. First redetermination cycles start rolling. Disenrollment begins.
Without intervention: 10M disenrolled nationally (CBO), ~$18M margin loss per 10K affected patients at a typical health system (our model; Guthrie’s own on-record figure is $38M — see below).
Guthrie — a nonprofit system across rural Pennsylvania and New York — told Spotlight PA it stands to lose as much as $38 million in net revenue from the federal Medicaid changes. About 70% of its patients carry government-funded insurance; 16% are on Medicaid. So it did something most systems have not done yet: it started treating enrollment retention as an operating function instead of a policy problem.
Not a national projection. One mid-size rural nonprofit's own figure, given on the record by its chief operating officer. Run the same arithmetic against your Medicaid mix before your budget cycle closes.
Read the Spotlight PA report →Out of 2.8 million Pennsylvanians on Medicaid — 22% of the state. State officials also anticipate a $20 billion federal funding loss over ten years beginning 2028. Every state has a version of this number.
PA DHS data reports →Guthrie expanded its volunteer program so patients can meet the monthly requirement on campus — greeting visitors, the gift shop, events, deliveries — and is calling Medicaid recipients directly to walk them through the new rules. Pennsylvania’s human services secretary called the system a trusted resource for helping people meet the requirement and stay covered.
See their program →This is the right instinct and the right economics — a covered patient is a reimbursed patient, so retention belongs in revenue cycle. But it only closes one of the two doors. Community engagement works for the patient whose barrier is a scheduling gap. It does nothing for the patient whose barrier is clinical: the COPD exacerbation, the chemotherapy calendar, the psychiatric admission, the dementia caregiving load at home. Asking that patient to volunteer 80 hours is asking them to do the thing their condition prevents.
Peer-reviewed data published in JAMA Health Forum in August 2026 sizes that group: across all 40 expansion states, 19.8% of eligible adults are at risk of missing the hours — and the study's lead author flags the specific population caught in between, those with “substantial health concerns that may not rise to the level of a documentable disability.” Too sick to hold predictable hours, not disabled enough for an automatic pass. For a health system, that is a named, findable cohort sitting inside your own EHR.
Those patients need the other door: a physician-signed frailty attestation filed against the federal “significantly impairs” standard. Same retention goal, same revenue logic, different instrument — and it is the one your organization already has the clinical evidence to produce. A complete Medicaid retention program runs both. Guthrie has built one. We build the other.
Size the exemption-eligible cohort in your panel →Three jobs, one platform. We identify who qualifies, Mira drafts the attestation, our medical director reviews and signs, or sends it back with the gap named. Review window agreed at contract. Full audit trail.
Upload an anonymized patient list. Our eligibility engine cross-references ICD-10, medications, chart notes, and caregiving flags against OBBBA's six exemption categories. Output: a ranked list of candidates with likely-qualifying criteria pre-cited.
Mira is our Medicaid Intelligence Review Agent. She drafts the §1902(e)(4) exemption attestation in plain language, cites the supporting diagnoses and encounters, and formats for your state's required fields.
Our medical director reads every attestation Mira drafts and either signs it or sends it back with the specific gap. The review window is agreed at contract, not assumed. Your compliance team gets the audit trail: draft, edits, signature, date.
Mira is the AI agent. Our medical director is the physician. medfrail is the platform. Every exemption your health system files carries physician attestation. This is what physician-governed AI actually looks like — not a chatbot, not a form-mill.
Most AI in healthcare is anonymous. An output arrives, you have no idea what model produced it, what training data shaped it, or whether a physician ever looked at it. That's the surveillance-era default.
Mira is the opposite. She has a name, a scope, a version number, and a supervising physician. Every draft is logged with its edits and signature. She never submits an attestation our medical director hasn't read. When an auditor asks "who made this decision?" — the answer is two humans and an agent, in that order.
That's the attestation era. It's the only form of AI that will survive the next ten years of healthcare audits.
Send an anonymized panel summary (counts by ICD-10 group, age band, and caregiving flag — no PHI). Within two business days we return a confidential report with:
No commitment. No PHI required (de-identified counts only). Reviewed by our medical director.
We follow up within one business day to confirm file format and security requirements.
The economics of this product only work if your revenue protection exceeds our cost, meaningfully. Our pricing is structured so that's the only way anyone signs anything.
Prices below are indicative. Final per-attestation pricing is set with the reviewing physician’s fee, which is flat and identical whether the attestation is signed or declined — the fee never depends on the outcome.
Yes. The statute's exemption categories (§1902(e)(4) and associated state-plan amendments) include medical-frailty, caregiving-role, disability, and related criteria. Each requires documentation — most commonly a physician's written attestation accompanied by supporting chart evidence. The question isn't whether attestation works; it's whether you have the capacity to file them at scale before your patients hit redetermination. That's what medfrail solves.
Our physician of record is a Board-Certified Internal Medicine physician licensed in the states medfrail serves. They serve as physician of record for medfrail attestations across the Managed and Self-Serve tiers. Having one physician sign every attestation means a single, consistent standard of review — and a single named physician an auditor can interview. Enterprise deployments use a multi-physician pool; our medical director remains the lead reviewer.
Plainly: we do not accept PHI today. The ROI report runs on de-identified panel counts. Before any PHI is exchanged, a BAA is signed and the intake path is one the reviewing physician controls. We do not currently operate a HIPAA-audited cloud environment, a FHIR ingestion service, or a SOC 2 program; those are scoped for the Managed tier and will be described as built, not before. If your security team needs a data-flow diagram of what exists now, we will draw the honest one.
Not yet. Today the input is a CSV/XLSX of a de-identified panel. EHR integration (FHIR R4 against Epic, Cerner, Athena, Meditech) is scoped for the Managed tier and is not built. We will not describe it as available until it is running against a live system.
Mira sends the draft back to the queue with a human-readable flag explaining the gap. A dedicated review specialist (Managed and Enterprise tiers) works with your team to either provide additional documentation or remove that patient from the cohort. No attestation is ever submitted without a physician signature, period.
The ROI report: two business days from a panel summary. A first batch of attestations: we scope it with you after the BAA, and the binding constraint is the physician review window, which we agree in writing rather than promise. We would rather tell you a true date than a fast one.
Same physician of record, same attestation infrastructure, different buyer. The consumer product at medfrail.com serves individual families directly at $99/attestation. This enterprise product serves health systems and Medicaid plans at scale. Both share the same clinical quality standard — every attestation is physician-signed.
Your Medicaid panel's first redetermination cycle is closer than your budget process thinks. The systems that identify, file, and protect ahead of the curve will keep their patients. The systems that don't will write uncompensated care into next year's variance report.
Get your Exemption ROI Report →