Estimation of infection prevalence and sensitivity in a stratified two‐stage sampling design employing highly specific diagnostic tests when there is no gold standard