What is the optimal approach for using a direct amplification test in the routine diagnosis of pulmonary tuberculosis? A preliminary assessment.

Abstract

The aim of this study was to determine the most appropriate strategy for the rapid diagnosis of pulmonary tuberculosis (PTB) using a nucleic acid amplification (NAA) test. This was a prospective study of 128 adult patients in whom respiratory secretions were tested for Mycobacterium tuberculosis by the AMPLICOR assay. The basis for starting PTB treatment was noted for each patient. The optimal approach was determined by using Bayes' theorem to compare different combinations of pretest probability, smear results with the AMPLICOR test. The incidence of PTB was 15.6%. In only one patient was treatment for PTB commenced because of a positive AMPLICOR result. The rest were managed according to the conventional approach which relied upon clinical judgment and direct smear. The optimal approach was to treat patients with high or intermediate pretest risk for PTB who returned positive AMPLICOR tests. The overall accuracies of the conventional approach, AMPLICOR test and optimal approach were 89.8, 95.3 and 96.1%, respectively. This small study suggests that NAA testing be limited to patients with high or intermediate pretest risk of PTB. In this group, positive results demand treatment while the management of those with negative results still relies on clinical judgment.