meilynx

Reference · Healthcare

HHS Section 1557 and Patient Care Decision Support Tools, Explained

45 CFR 92.210 — Nondiscrimination in the use of patient care decision support tools

Last reviewed September 5, 2026

Section 1557 of the Affordable Care Act prohibits discrimination in health programs and activities receiving federal financial assistance. The 2024 final rule added 45 CFR 92.210, which applies that prohibition to patient care decision support tools, automated or not, and creates two duties for covered entities: identify tools that use protected-class inputs, and mitigate the risk of discrimination from them. Both have applied since 1 May 2025.

Who it applies to

Covered entities under Section 1557: health programs and activities receiving federal financial assistance from HHS, including hospitals, physician groups, health plans, and state health agencies. The rule reaches the tools they use in care, whether built in-house, bought from an EHR vendor, or embedded in a payer's utilization-review workflow.

The three parts of 92.210

  • 92.210(a): no discrimination on the basis of race, color, national origin, sex, age, or disability through the use of patient care decision support tools.
  • 92.210(b): an ongoing duty to make reasonable efforts to identify tools that employ input variables measuring a protected class.
  • 92.210(c): for each identified tool, reasonable efforts to mitigate the risk of discrimination from its use.
The rule does not require a specific test or a specific outcome. It requires that the entity knows which tools use protected-class inputs and can show what it did about the risk.

How it connects to ONC HTI-1

ONC HTI-1's decision support interventions criterion requires certified health IT developers to make source attributes available, including whether race, ethnicity, sex, date of birth, or health status are used. Those attributes are the developer-side input to the covered entity's 92.210(b) determination. The determination itself, and the mitigation, remain the covered entity's.

What reasonable mitigation looks like

  • An inventory of decision support tools with the protected-class input determination recorded per tool and revisited when a tool changes.
  • Bias or outcome testing performed by the organization or its vendor, with metrics and a review cadence.
  • Clinician review of tool output before it informs care, and the ability to override.
  • A nondiscrimination policy, the 92.8 notice, and a grievance procedure that reach AI-assisted care.

Control mapping

What a reviewer expects to be able to see.

ObligationWhat the system must doEvidence a reviewer expects
Identification (92.210(b))Know which decision support tools are in use and which use protected-class inputsA tool inventory with a per-tool determination and its date
Ongoing identificationRevisit the determination when a tool changesConfiguration change history and re-determination records
Mitigation (92.210(c))Make reasonable efforts to mitigate discrimination risk per toolTesting by whom, metrics, review cadence, clinician override
Nondiscrimination and notice (92.210(a), 92.8)Cover AI-assisted care in the policy, notice, and grievance procedureThe policy, the posted notice, and the procedure

Key dates

  • 5 July 2024Section 1557 final rule effective date for most provisions.
  • 1 May 202545 CFR 92.210 patient care decision support duties apply.

Primary sources

Common gaps

Where covered entities most often fall short of 92.210.

  • No inventory of decision support tools. The identification duty presumes a list. Entities that cannot say which tools run cannot say which use protected-class inputs.
  • The determination made once. The duty is ongoing. A tool whose model or prompt changed since the determination has to be looked at again.
  • Mitigation with no record. Clinician override exists in practice and nowhere on paper. OCR asks what the entity did; the answer needs a date and an owner.
  • Notices that stop at the front desk. The 92.8 notice and grievance procedure predate AI-assisted care and never mention it.

Last reviewed September 5, 2026. This reference summarises publicly available regulatory guidance and is provided for general information. It is not legal advice. Obligations depend on an institution's charter, registration status, size, and activities. Verify against the primary sources cited above and consult counsel before relying on any summary here.

Regulatory updates

When a regulator changes what an AI examination asks for, hear about it first.

Short notes on SR 26-2, NYDFS 500, FINRA, the NAIC bulletin, the EU AI Act, and the employment-AI statutes, plus what we ship. A few emails a month.