Arkansas ARChoices RUGs resource allocation
Released: January 2016
ARChoices was an algorithmic system introduced by The Arkansas Department of Human Services (DHS) in January 2016 to determine Medicaid benefits for over 8,000 elderly and people with severe disabilities.
Until January 2016, The Arkansas Department of Human Services (DHS) had relied upon nurses to assess home and community care eligibility using a lengthy questionnaire called ARPath.
Its replacement, ARChoices, handled both types of care by sorting people into twenty-three so-called resource utilisation groups ('RUGs'), which the DHS argued was a fairer, more objective, and less costly system.
While some patients seemed to benefit from the new system, 47% discovered their weekly allotment of hours had decreased, sometimes significantly, resulting in poorer health and mental health amongst those impacted.
The change also had major financial ramifications for some patients, notably those with serious disabilities such as cerebral palsy.
In May 2018, a US circuit judge ordered that Arkansas authorities stop using ARChoices.
With the DHS unwilling to update its algorithm, explain how it worked, or give independent experts access to its data, code or model, seven patients sued the department.
Represented by Legal Aid of Arkansas, the patients won their case on the basis that the DHS had failed to give them adequate notice of the change of system and that their benefits may fall.
However, the judge's decision that the DHS stop using the new algorithm, listen to external stakeholders and undergo a legislative review process was appealed on the basis that emergency rule-making was required to ensure the programme's continuity.
The manoeuvre was described by the US circuit judge as 'a manufactured emergency'.
Shortly afterwards, Arkansas replaced RUGs with the ARIA system.
Operator: Arkansas Department of Human Services; Center for Information Management
Developer: University of Michigan; Brant Fries
Sector: Govt - welfare; Govt - health
Purpose: Assess care resource requirements
Technology: Resource allocation algorithm
Issue: Accuracy/reliability; Bias/discrimination - disability, age
Transparency: Governance; Complaints/appeals; Black box; Legal
Arkansas Legislative Council Administrative Rules and Regulations SubCommittee (2018). ARChoices 1-18; Resource Utilization Groups (RUGS) Overview (pdf)
Arkansas Access to Justice (2017). Suit Filed Over Computer Program Making Medicaid Cuts
George Washington University. ETI AI Litigation Database: case details
AI Now Institute (2019) Litigating Algorithms 2019 US Report (pdf)
News, commentary, analysis
Published: March 2022