Potential for Algorithmic Bias in Clinical Decision Instrument Development.
systematic_review · Level I
Where this comes from
- Record sourced from PubMed, PMID 41372566.
- Also identified by DOI 10.1038/s41746-025-02119-7 and PMC identifier 12708808.
- Licence recorded as CC BY.
- The licence permits redistribution, so the abstract is shown in full and the full text is available from the publisher.
Abstract
Clinical decision instruments (CDIs) face an equity dilemma. They reduce disparities in patient care through data-driven standardization of best practices. However, this standardization may perpetuate bias and inequality within healthcare systems. We perform a quantitative, systematic review to characterize four potential sources of bias in the development of 690 CDIs. We find evidence for potential algorithmic bias in CDI development through various analyses: self-reported participant demographics are skewed-e.g. 73% of participants are White, 55% are male; investigator teams are geographically skewed-e.g. 52% in North America, 31% in Europe; CDIs use predictor variables that may be prone to bias-e.g. 1.9% (13/690) of CDIs use Race and Ethnicity; outcome definitions may introduce bias-e.g. 26% (177/690) of CDIs involve follow-up, which may skew representation based on socioeconomic status. As CDIs become increasingly prominent in medicine, we recommend that these factors are considered during development and clearly conveyed to clinicians.