Multiattribute decision analysis of clinical errors: a case study of computerized arrhythmia detectors.

Abstract

Two types of errors are observed in the process of medical decision making: false positives and false negatives. For various detection criteria these errors are usually plotted to form detection curves. The problem of identifying optimum detection criteria that minimize the impact of these errors is addressed. Usually this involves trade-offs that are determined by the attitudes of the medical decision maker, types of disease, and clinical settings. Utility theory is the theoretical technique developed to quantify subjective elements of decision making. Techniques of utility theory were employed to experimentally obtained data on errors committed by an automated arrhythmia monitor. Two cardiologists were interviewed to assess their utility functions and a multiattribute decision-analysis technique was used. From the utility functions the best detection criteria for the automated arrhythmia detector were determined. This research also highlights various behaviors in medical setting: risk neutral, risk averse , and risk seeking. The effects of various diseases and clinical settings on decision making were also shown.