Pretest Probability in Ophthalmology: Preserving Diagnostic Calibration While Moving Beyond Race

Area of Science:

• Medical Diagnostics
• Bayesian Reasoning
• Health Equity

Background:

• Race is a social construct, not a biological determinant, but often used as a proxy in clinical practice.
• Current recommendations on race/ethnicity in medicine may inadvertently impact diagnostic accuracy.
• Ophthalmology examples illustrate risks of inaccurate baseline disease probability estimation.

Purpose of the Study:

• To evaluate consequences of removing race-based proxies in clinical reasoning.
• To assess the impact on diagnostic calibration using a Bayesian framework.
• To explore alternatives for precise causal predictors.

Main Methods:

• Narrative review and conceptual analysis of Bayesian diagnostic reasoning.
• Application to current recommendations on race/ethnicity in medicine and ophthalmology.
• Examination of the French experience with race/ethnicity data restrictions.

Main Results:

• Suppressing race-based proxies without precise causal substitutes can degrade diagnostic calibration.
• Uniform diagnostic thresholds across diverse populations may lead to inefficient testing and delayed diagnoses.
• Inaccurate pretest probability estimation can result in unnecessary testing and avoidable morbidity.

Conclusions:

• The aim is accurate, equitable diagnostic calibration, transitioning to mechanism-based predictors.
• Precise variables (genetic, environmental, structural) should replace imprecise proxies.
• Eliminating proxies without replacements risks impairing clinical reasoning and diagnostic outcomes.