From pre-test and post-test probabilities to medical decision making

Area of Science:

• Medical Decision-Making
• Bayesian Statistics
• Health Informatics

Background:

• Modern evidence-based medicine aims for simple tools to integrate quantitative data into clinical decisions.
• The Bayesian Pre-test/Post-test Probability (BPP) framework quantifies diagnostic uncertainty but doesn't fully address decision-making.
• Simple, flexible quantitative approaches for clinical decision-making remain elusive.

Purpose of the Study:

• To extend the BPP framework using Bayesian Decision Theory to incorporate costs for clinical decision-making.
• To develop a simple quantitative framework for binary clinical decisions.

Main Methods:

• Extension of the BPP framework by integrating concepts from Bayesian Decision Theory.
• Development of a quantitative framework for binary decisions (e.g., treat/no-treat).

Main Results:

• A simple quantitative framework for binary clinical decisions was developed.
• A critical value, the decision boundary ($$), was identified, representing the optimal point for action or inaction based on relative costs.
• The framework's utility was demonstrated through bedside case studies and a reanalysis of a study on probability misestimation.

Conclusions:

• The developed approach is a simple, core component of the BPP framework.
• It requires minimal resources (hand-held calculator) and is broadly applicable.
• It is particularly useful for patient-specific decisions with difficult-to-quantify costs and benefits.