SETTING: A tuberculosis referral hospital in Canada.
OBJECTIVE: To determine the validity of acid-fast (AFB) smears of gastric a
spirates (GA) in the diagnosis of pulmonary tuberculosis, and to assess the
prevalence of nontuberculous mycobacteria (NTM) in GA isolates from such p
atients.
DESIGN: A retrospective case review of our experience with AFB smears (Kiny
oun) and cultures of GA and sputum over a 3-year period.
RESULTS: From 1994 to 1996 inclusive, 1155 GA were performed in 889 patient
s. Mycobacteria were cultured from 109 (9%) GA. Thirteen of these were posi
tive on smear (sensitivity 19%). All GA that were positive on smear were cu
lture positive for Mycobacterium tuberculosis. There were no false positive
smears (specificity 100%). The sensitivity and specificity of the sputum s
mear mere 45% and 99%, respectively. Of the 96 culture positive, smear nega
tive GA, 54 grew M. tuberculosis and 42 grew an NTM. Of 13 patients who had
sputum and GA studied coincidentally, and in whom the sputum was both smea
r and culture positive, the GA culture was positive in 13 and the smear was
positive in eight (66%).
CONCLUSION: AFB smear of GA is a relatively insensitive but highly specific
indicator of pulmonary tuberculosis warranting institution of antitubercul
osis treatment. Gastric AFB smear positivity appears to reflect a high baci
llary burden within the respiratory tract.