Sputum smear microscopy predicting mycobacterial culture in Ontario: A population-based laboratory report

Abstract

RATIONALE: Tuberculosis (TB) and nontuberculous mycobacterial (NTM) pulmonary disease may present similarly but must be swiftly distinguished because of public health implications. The first diagnostic test usually available for suspected pulmonary mycobacterial disease is smear microscopy for acid-fast bacilli (AFB).OBJECTIVES: To measure the diagnostic utility of sputum AFB in Ontario.METHODS: All patients’ first AFB smears submitted to Public Health Ontario Laboratories during 2012–2014 were compared against final cultures. Diagnostic test characteristics were calculated for AFB smears in predicting TB and NTM separately.MEASUREMENTS AND MAIN RESULTS: Among 30 136 patients, AFB smears and cultures were positive in 4.2% (1279) and 17.4% (5233) respectively, comprising TB in 800 (2.7%, AFB-positive 489 [61.1%]) and NTM in 4433 (14.7%, AFB-positive 652 [14.7%]). Test characteristics for TB were: specificity 97.3%, negative predictive value 98.9%, sensitivity 61.1%, and positive predictive value 38.2%. Performance improved somewhat in males: specificity 97.4%, negative predictive value 98.8%, sensitivity 60.1%, positive predictive value 42.0%, and more so in people <60 years old: specificity 98.3%, negative predictive value 98.5%, sensitivity 60.9%, positive predictive value 56.8%. Test characteristics for NTM were: specificity 97.6%, negative predictive value 86.9%, sensitivity 14.7%, positive predictive value 51.0% in the entire population.CONCLUSIONS: Sputum AFB performed poorly in predicting TB because of the large proportion of AFB-positives caused by NTM (51.0%) and the relatively small proportion of AFB-positive TB cases (61.1%). The predictive ability for NTM was also poor. Regardless of age and sex, AFB remained an inadequate predictor of TB and NTM in Ontario.