A microsimulation-based framework for mitigating societal bias in primary care data.

Area of Science:

• Health Informatics
• Biostatistics
• Public Health

Background:

• Societal biases in healthcare data can perpetuate inequities.
• Reassessing race-based criteria requires understanding their impact on disease progression.

Purpose of the Study:

• To propose a novel microsimulation framework to reduce societal bias in primary care data.
• To study the effect of race-based criteria on kidney function decline.

Main Methods:

• Developed an individual-level simulation model for kidney function decline (eGFR).
• Simulated eGFR trajectories, incorporating factors like hypertension and diabetes.
• Applied a Bayesian calibration to estimate eGFR decline rates.

Main Results:

• Without race adjustment, Black individuals qualified for diagnosis earlier than with adjustment.
• This difference was most pronounced in earlier disease stages.
• No significant differences in life expectancy were observed between scenarios.

Conclusions:

• The data transformation framework can inform policy changes by representing data generation processes.
• This approach can mitigate bias in various health datasets.
• Explicitly modeling data generation is key to understanding and addressing bias.