Comparison of algorithms for selective use of nucleic-acid probes for identification of Mycobacterium tuberculosis from BACTEC 12B bottles.

We retrospectively compared the sensitivity of two approaches, a time-to-detection algorithm and the presence of serpentine cords of acid-fast bacilli, for discriminating between BACTEC 12B cultures containing either Mycobacterium tuberculosis complex (MTB) or Mycobacterium avium complex (MAC). From January 1996 through March 1997 a total of 217 of 2089 respiratory specimens received in our laboratory were positive in the BACTEC 12B radiometric culture system for either MTB (120 specimens) or MAC (97 specimens). Use of a previously published time-to-positivity algorithm would have resulted in the correct use of the MTB probe on 109 of 120 cultures (91% sensitivity), and the MAC probe on 52 of 97 cultures (54% sensitivity). The presence of serpentine cords was detected in 58 of 120 cultures containing MTB (48%), and in 3 of 97 (3%) cultures containing MAC. Using a combination of time to positivity and cord formation to determine initial probe selection would have resulted in first use of the MTB probe in 116 of 120 (97%) instances in which MTB was present in the culture. In only 49 of 97 (51%) cultures, however, from which MAC was recovered would the correct probe have been selected. These results indicate that limiting the initial use of the MTB probe to those cultures that are either identified by the time-to-detection algorithm or demonstrate serpentine cords on acid-fast smear would eliminate a considerable amount of unnecessary probe use without compromising the efficiency of identification of isolates of MTB.