Context: False-positive cultures for Mycobacterium tuberculosis have been found in nearly all DNA fingerprinting studies, but the effectiveness of interventions to reduce cross-contamination has not been evaluated.
Objective: To evaluate whether changes in laboratory policies and procedures reduced the rate of false-positive cultures.
Design: Retrospective study of isolates with matching DNA fingerprints.
Setting: A mycobacteriology laboratory serving an urban tuberculosis control program and public hospital system.
Patients: All M tuberculosis isolates processed from July 1994 to December 1999.
Methods: Isolates were fingerprinted using IS6110; pTBN12 was used to fingerprint isolates having fewer than 6 copies of IS6110. We further evaluated all patients having only one positive culture whose DNA fingerprint matched that of another isolate processed in the laboratory within 42 days.
Interventions: We changed laboratory policy to reduce the number of smear-positive specimens processed and changed laboratory procedures to minimize the risk of cross-contamination during batch processing.
Main outcome measure: The rate of false-positive cultures.
Results: Of 13 940 specimens processed during the study period, 630 (4.5%) from 184 patients and 48 laboratory proficiency specimens grew M tuberculosis. There were no cases (0/184) of probable or definite cross-contamination, compared with the 4% rate (8/199) identified in our previous study (P =.008). We also fingerprinted a convenience sample of isolates from other laboratories in Denver; 13.6% (3/22) of these were false-positive, a rate similar to the 11.9% rate (5/42) identified for other laboratories in our previous study (P =.84).
Conclusions: Laboratory cross-contamination decreased significantly after relatively simple, inexpensive changes in laboratory policies and practices. Cross-contamination continued to occur in other laboratories in Denver.