Whether to Retest the Lipids of HIV-Infected Patients: How Much Does Fasting Bias Matter?

Authors:
Xiting Yang;Joseph B. Kadane;Heidi M. Crane;Mari M. Kitahata
Affiliations:
Department of Statistics, Carnegie Mellon University, Pittsburgh, Pennsylvania 15213;Department of Statistics, Carnegie Mellon University, Pittsburgh, Pennsylvania 15213;Department of Medicine, University of Washington, Seattle, Washington 98104;Department of Medicine, University of Washington, Seattle, Washington 98104
Venue:
Decision Analysis
Year:
2011

Citing 3
Cited 2

Clinical Applications in the Decision Analysis Literature

Decision Analysis
The Influence of Influence Diagrams in Medicine

Decision Analysis
Analysis of the Biological Clock Decision

Decision Analysis

From the Editors---Deterrence, Multiattribute Utility, and Probability and Bayes' Updating

Decision Analysis
From the Editors---Probability Scoring Rules, Ambiguity, Multiattribute Terrorist Utility, and Sensitivity Analysis

Decision Analysis

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper aids in the diagnosis and treatment of lipid abnormalities by enhancing understanding of lipid values among HIV-infected patients. An elicited loss function highlights the importance of true low-density lipoprotein (LDL) cholesterol. In clinical settings, total cholesterol, high-density lipoprotein cholesterol, and triglycerides are often measured. Interpreting the resulting values can be problematic because of uncertainty due to the unknown period of fasting before the patient's blood was drawn. This results in uncertainty in the LDL cholesterol values, which are often calculated from other lipid values rather than measured directly. To model true LDL cholesterol, a four-level Bayesian hierarchical model is analyzed using Markov chain Monte Carlo techniques and elicited prior distributions. In turn, this yields expected-loss-minimizing treatment decisions for individual patients.