Sequential decision making with continuous disease states and measurements: I. Theory

Area of Science:

• Decision Analysis
• Biostatistics
• Health Economics

Background:

• Continuous health states (e.g., blood pressure) require effective monitoring strategies.
• Current decision-making for testing and treatment can be suboptimal.
• Assessing the utility of continuous health state measurements is complex.

Purpose of the Study:

• To develop a mathematical framework for optimizing decisions regarding continuous health state testing.
• To compare the expected utility of testing versus immediate treatment or withholding treatment.
• To establish a sequential decision-making process for ongoing health state monitoring.

Main Methods:

• Decision analysis framework assuming normally distributed health states and measurement variability.
• Application of Bayes' theorem to derive expected utility of testing.
• Development of a sequential decision-making process with test/treat thresholds.
• Extension of the conjugate-normal-linear model for correlated observations.

Main Results:

• Derived an expression for the expected utility of performing a continuous test.
• Established a sequential decision-making process with readily calculable thresholds.
• Demonstrated a method for stepwise optimal results using a running average of measurements.
• Extended the model to handle correlated observations within a single visit.

Conclusions:

• The developed decision analysis provides a mathematical basis for optimizing continuous health state testing.
• Sequential decision rules enable stepwise optimal management of continuous health variables.
• The methodology is applicable to various continuous health states, including blood pressure monitoring.