Base rates revisited: Assessment strategies for HIV/AIDS

Publisher Website
Google Scholar
Abstract
The efficiency of a diagnostic test is largely determined by the base rate, or prevalence, of disease in the population under study, with the consequence that low prevalence diseases are often difficult to detect. However, a review of clinical decision‐making, from a Bayesian standpoint, indicates that even relatively inefficient measures may be effective when combined in appropriate ways, and when the costs and benefits of detection versus non‐detection are considered. In the case of HIV/AIDS, a number of factors, including low prevalence population characteristics, the tendency to distort critical information, and the horrendous consequences of this disease, severely complicate the decision‐making task. The present paper reevaluates the problem of prediction as it relates to HIV/AIDS by examining the use of multiple tests, the relevance of Bayesian utility theory, and the significance of both immediate and projected costs and benefits.