The Impact of Health Care Algorithms on Racial and Ethnic Disparities : A Systematic Review

Area of Science:

• Health Informatics
• Health Equity Research
• Algorithmic Bias Studies

Background:

• Growing concerns exist regarding the influence of health care algorithms on racial and ethnic disparities.
• Existing evidence on this impact is fragmented and requires synthesis.

Approach:

• A systematic review of studies published between January 1, 2011, and September 30, 2023, was conducted.
• Studies were screened for their examination of algorithms' effects on disparities and the efficacy of mitigation strategies.
• Risk of bias was assessed using ROBINS-I and CARE-CPM equity extension.

Key Points:

• Sixty-three studies revealed that algorithms can reduce, perpetuate, or exacerbate racial and ethnic disparities.
• Examples include the kidney allocation system (reducing disparities) and severity-of-illness scores (exacerbating disparities).
• Seven distinct mitigation strategies were identified, including variable adjustment and population-specific thresholds.

Conclusions:

• Health care algorithms' impact on disparities is heterogeneous and not solely dependent on the explicit use of race.
• Algorithm intentionality and implementation critically influence outcomes and potential tradeoffs.
• Further research is needed to understand context-specific effects and optimize algorithmic fairness.