Study finds Optum risk algorithm understated Black patients' health needs
A 2019 study revealed that Optum's health risk algorithm discriminated against Black patients by substituting health costs for actual health needs. This resulted in a systemic underestimation of risk for Black patients, which limited their access to specialized care management.
Records by entity: Optum
An Optum health risk algorithm systematically underestimated the health needs of Black patients by using healthcare spending as a proxy for health needs.
The algorithm used health costs as a proxy for health needs, reflecting systemic spending disparities rather than actual patient health.
Key facts
- What
- A 2019 study revealed that Optum's health risk algorithm discriminated against Black patients by substituting health costs for actual health needs.
- Incident date
- Oct 1, 2019
- Who
- Optum
- Failure mode
- Policy Violation
- AI surface
- Algorithmic Decision
- Severity
- High
What happened
An algorithm used by Optum to identify patients for high-risk care management exhibited significant racial bias. The system assigned lower risk scores to Black patients compared to white patients with similar health profiles. This caused many Black patients to be excluded from critical care programs they qualified for based on their health status.
What broke inside the model
The model used health care spending as a proxy for health needs. Because of systemic inequities in spending, the model incorrectly predicted that Black patients were healthier than white patients with the same level of illness.
What it cost
Sources
- PrimaryDissecting racial bias in an algorithm used to manage the health of populationsscience.org
- PressNew York insurance regulator to probe Optum algorithm for racial biasfiercehealthcare.com
Cite this entry
https://failureindex.ai/failures/optum-health-risk-algorithm-biased-blackAI Failure Index. "Study finds Optum risk algorithm understated Black patients' health needs" (FI-0363). Realm Labs. https://failureindex.ai/failures/optum-health-risk-algorithm-biased-black (indexed Jun 9, 2026).Data fields CC-BY 4.0, prose citation permitted. Incident ID FI-0363. Full dataset at /data.
Note from Realm Labs, the Index steward
How Realm fits
- Prism
- OmniGuard
This entry sits in the index's predictive wing: a system that scores, ranks, perceives, or steers rather than generates. Realm's runtime layer is built for the generative and agentic systems now moving into these same decision seats, where it watches a model's internal state and holds an unsupported claim or an unchecked action before it commits. The control gap on this record, an automated decision that reached people with no runtime check in front of it, is the same gap. The index keeps predictive failures on the record because the pattern carries straight into the systems shipping today.