{"id":57434,"date":"2024-05-08T17:21:00","date_gmt":"2024-05-08T17:21:00","guid":{"rendered":"https:\/\/www.dinsmore.com\/?post_type=publications&p=57434"},"modified":"2025-11-24T18:51:57","modified_gmt":"2025-11-24T18:51:57","slug":"new-affordable-care-act-final-rule-prohibits-discriminatory-use-of-ai","status":"publish","type":"publications","link":"https:\/\/www.dinsmore.com\/publications\/new-affordable-care-act-final-rule-prohibits-discriminatory-use-of-ai\/","title":{"rendered":"New Affordable Care Act Final Rule Prohibits Discriminatory Use of AI"},"content":{"rendered":"\n
\"\"\/<\/figure>\n\n\n\n

The Department for Health and Human Services Office for Civil Rights (\u201cOCR\u201d) recently issued a final rule<\/a> under Section 1557 of the Affordable Care Act, which prohibits federally funded health programs from discriminating on the basis of race, color, national origin, sex, age or disability. Consistent with the Supreme Court\u2019s 2020 decision<\/a> in Bostock v. Clayton County<\/em>, the final rule codifies that discrimination \u201con the basis of sex,\u201d under Section 1557, includes discrimination on the basis of sexual orientation and gender identity. The final rule also clarified that Section 1557 protections extend to providers\u2019 use of artificial intelligence, and more specifically the use of \u201cpatient care decision support tools,\u201d and established an ongoing duty for providers to identify and mitigate the risks of any potentially discriminatory tools.<\/p>\n\n\n\n

Under the rule, a \u201cpatient care decision support tool\u201d is \u201can automated or non-automated tool, mechanism, method, technology or combination thereof used by a covered entity to support clinical decision-making in its health programs or activities.\u201d Specific examples of patient care decision support tools include predictive algorithms that assess patient risks of severe health events, as well as tools used to provide medical necessity analyses to approve or deny medical claims. OCR clarified that in using these tools, the unique facts and circumstances of each individual patient must be considered, citing studies where reliance on certain algorithms have resulted in racial and ethnic disparities. Importantly, non-automated tools, such as flowcharts, are included in the above definition. Such tools may result in discrimination if, for instance, they screen out individuals with disabilities such that they are prohibited from accessing equitable care, or covered entities fail to consider individual factors in health assessments.  <\/p>\n\n\n\n

Covered entities now have an ongoing duty to \u201cmake reasonable efforts to identity\u201d its uses of patient care decision support tools that use input variables of race, color, origin, sex, age or disability. The entity then must mitigate the risk of discrimination that may result from any identified tools. OCR says that compliance with this requirement would be measured on a case-by-case basis, and that it would consider factors including:<\/p>\n\n\n\n