Methods for retrospectively improving race/ethnicity data quality: a scoping review

Area of Science:

• Health Informatics
• Data Science
• Public Health

Background:

• Accurate race and ethnicity data are vital for identifying health disparities and informing equitable healthcare policy.
• Secondary data sets often have incomplete or inaccurate race/ethnicity classifications, potentially misrepresenting underserved populations.
• Retrospective methods to improve race/ethnicity data quality are essential for robust health research.

Purpose of the Study:

• To conduct a scoping review of methods used to retrospectively improve race and ethnicity classification in secondary data sets.
• To identify and categorize different approaches for enhancing demographic data quality.
• To analyze the characteristics, strengths, and limitations of existing methods.

Main Methods:

• Systematic search of MEDLINE, Embase, and Web of Science Core Collection databases.
• Screening of 2,441 abstracts and full-text review of 453 articles, including 120 in the final analysis.
• Narrative analysis of study characteristics and categorization of identified methods.

Main Results:

• Six primary method types were identified: name lists (23%), name algorithms (46%), machine learning (12%), expert review (8%), data linkage (8%), and other (5%).
• The majority of studies focused on classifying Asian (47%) and White (43%) populations.
• Validation evaluations were present in 72% of the included articles.

Conclusions:

• Name algorithms are the most common method for improving race/ethnicity data, but innovative approaches are needed for better subgroup identification.
• Strengths, limitations, and potential harms of various methods require careful consideration.
• Accurate, disaggregated race/ethnicity data are critical for effective policymaking and healthcare interventions.