|
MéTODOS ESTADíSTICOS DESARROLLADOS PARA LA ESTIMACIóN EN LA PREVALENCIA Y PARáMETROS DE DESEMPE O DE TESTS PARA DIAGNóSTICO CLíNICO: UNA REVISIóN DE LITERATURAKeywords: Keywords: Prevalence , Estimation , Sensibility , Specificity , Statistical Methods , Dependence. Abstract: Introduction: to do a review of the bibliography published in statistical methodology specialized journals about the procedures developed to estimate the prevalence and performance of clinical diagnostic tests.Methods: this review includes 52 articles about procedures developed to estimate the prevalence and performance of clinical diagnostic test parameters found in journals specialized in statistical methodology applied to health areas.Results: three important statistical problems were found: the presence of a number of parameters higher than the pieces of information available to do the estimation procedure, known as lack of the ability to identify, the presence of individuals with negative outcome in the screening tests for comparison, whose actual health conditions are unknown (verification bias) and the dependence between test results.Conclusions: a wide range of possibilities to address the three problems that characterized the estimation problem was found. To address the verification bias problem many alternatives consider the use of the latent variables and others use different parameters. Many strategies to address the lack of the ability to identify and the dependence between test outcomes were found too. Most of the proposed alternatives are characterized by their complexity and difficult practical applicability.
|