Table_2.xls (9.5 kB)
Prevalence, sensitivities and specificities, positive and negative predictive values (PPV's and NPV's) using culture as gold standard and for two Bayesian latent class models.
dataset
posted on 2010-08-30, 00:54 authored by Direk Limmathurotsakul, Kris Jamsen, Arkhom Arayawichanont, Julie A. Simpson, Lisa J. White, Sue J. Lee, Vanaporn Wuthiekanun, Narisara Chantratita, Allen Cheng, Nicholas P. J. Day, Claudio Verzilli, Sharon J. PeacockModel 0 assumed that for a given patient, each diagnostic test was not correlated. The final model (Model 3) assumed that in infected patients, all serological tests were correlated.
*Values shown are mean estimates with 95% confidence interval.
†Values shown are median estimates with 95% credible interval.
‡Prevalence estimated by the models are for the test population as a whole, since they cannot determine whether a given patient was infected or not infected.