Reduction of the rate of false-positive cultures of Mycobacterium tuberculosis in a laboratory with a high culture positivity rate

Our laboratory, engaged in a prospective study of adult pulmonary tuberculosis, processed on average 1186 sputum samples per year for the detection of Mycobacterium tuberculosis (M. tuberculosis). Approximately 55% of all sputum samples were culturepositive. The study protocol required that all patients had their M. tuberculosis isolates DNA fingerprinted at diagnosis, and at subsequent time points if the patients either failed treatment or presented again with tuberculosis. Over a 22-month period, there were 14 apparent treatment failures from 109 patients who had completed 6 months of therapy. Only two of these were true treatment failures, while the other 12 had DNA fingerprints that were different from those obtained at diagnosis. It was concluded that these 12 cultures represented episodes of laboratory crosscontamination. Retrospective DNA fingerprinting of patient isolates was done so that each patient had at least two independent isolates fingerprinted. This survey revealed that 7.3% of DNA fingerprints were discordant. Falsepositive cultures with discordant DNA fingerprints generally arose late in chemotherapy and the isolates were usually coprocessed with other strongly smearpositive sputum samples. Simple modifications of laboratory procedures were made, and over a following 10.5-month period the falsepositive rate was reduced to 2.1%. These modifications did not increase the workload or the cost of processing samples and can thus be used successfully by any laboratory, and particularly by those in resourcepoor settings.