Laboratory implementation of the polymerase chain reaction for confirmation of pulmonary tuberculosis.

Decision analysis methods were used to compare four mycobacteriology laboratory strategies with respect to time to confirmation and exclusion of smear-positive and smear-negative cases of pulmonary tuberculosis. Strategies assessed included the following: (i) polymerase chain reaction (PCR) on all respiratory specimens; (ii) PCR on smear-positive specimens and on the broth of vials for other specimens attaining a growth index >10 in a radiometric culture detection system; (iii) PCR on smear-positive specimens only; and (iv) radiometric culture detection, with DNA probe for species identification of vials attaining a growth index >999. Strategies i and ii had predicted average times to confirm cases of 5 and 7.6 days, respectively, and remained within 3 days of each other over a broad range of PCR performance with smear-negative specimens. In contrast, case-confirmation times using strategies iii and iv were 10.4 and 15.3 days, respectively. Only 10% of specimens were processed by PCR in strategy ii. Times to confirm smear-negative cases were comparable for strategies i and ii when PCR sensitivity was <40% with these specimens. Times to exclude pulmonary tuberculosis were similar for all strategies. Given the current suboptimal performance of PCR with smear-negative specimens, strategy ii offers accelerated case confirmation with limited PCR usage.