Relative Accuracy of the Screening Tests When Absolute Determination of the Disease Status is Not Feasible

摘要: Comparison of the accuracy a new screening test to that standard one can be implemented by administering both tests group asymptomatic subjects for which disease status determined using gold (GS) test. Nevertheless, GS may too costly or invasive hence unethical administer all study subjects, including those who screen negative on tests. When this is case, relative two estimated when randomized paired positive (RPSP) design used collect data. However, contains cells with missing data, thus likelihood function not available. The objective demonstrate parsimonious way estimating determination screened and was conducted due ethical concerns. Markov Chain Monte Carlo simulation technique parsimoniously address aforementioned shortcoming RPSP subjective approach estimation used. Multiple data imputation Gibbs sampler performed. point interval estimates measures rates are computed treated be: