Mycobacterial infections: yield of bacillary microscopy in different clinical samples (1975-1988)

BACKGROUND: The yield of microscopy examination as a quick diagnostic test in several pulmonary and nonpulmonary samples referred to the mycobacterial laboratory of a general hospital is reviewed. METHODS: During a 14-year period (1975-1988), 113,836 biological products were investigated. In 9,972 a positive culture for mycobacteria was obtained. For the microscopy examination the auramin technique was used; if positive, acid-alcohol resistance was confirmed by overstaining with the Ziehl-Neelsen technique. The culture was used as the reference method. RESULTS: Microscopic examination was positive in 34% of samples with a positive culture, being 39% for Mycobacterium tuberculosis and 10% for environmental mycobacteria. The overall specificity was 99%, the positive predictive value was 91% and the negative predictive value was 94%. In pleuropulmonary samples the sensitivity ranged from 48% in sputum and 2% in pleural biopsy, with specificity higher than 99%. In nonpulmonary samples, sensitivity, specificity and positive and negative predictive values varied with the type of sample. The false positive rate (positive microscopy with negative culture) was 0.3, and it was shown that 80% of these patients had received previous therapy. In organic fluids (pleural, peritoneal, cerebrospinal), the sensitivity was not greater than 13%. CONCLUSIONS: Sputum, bronchoaspirate and bronchoalveolar lavage were better for the diagnosis of tuberculosis than gastric aspirate. Approximately 1 in each positive microscopy examinations corresponded to environmental mycobacteria. In some nonpulmonary samples with high sensitivity the positive predictive value was low. 80% of the false positive results corresponded to previously treated patients.