Performance of frontloading for smear microscopy in the diagnosis of pulmonary tuberculosis: a cross-sectional study at a referral hospital in Uganda.

Abstract

OBJECTIVE: To compare the performance of frontloading and the standard WHO method for diagnosis of pulmonary TB at Mulago Hospital in order to validate the technique in this setting. METHODS: This was a cross-sectional study in which 229 adult (≥18 years) TB suspects were consecutively enrolled. Suspects submitted three sputum samples as follows: at initial presentation, one hour after the first sample, and the next morning. The first and next morning samples formed the standard WHO method, while the first and the one hour later samples formed the frontloading method. Sample processing was by the standard N-acetyl L-cystein (NALC)-NaOH method, and fluorescent microscopy was done for both methods, while cultures of the first sample on Lowenstein-Jensen slants acted as a gold standard. The sensitivity, specificity and predictive values for the WHO standard and frontloading methods were compared. RESULTS: The sensitivity of both the frontloading and standard schemes was 91.1% while their specificities were 86.2% and 91.7% respectively. There was excellent agreement between the diagnostic capacity of the two methods (kappa statistic = 0.87, P<0.0001). The positive predictive value for the frontloading scheme was 87.2% and that for the standard approach was 91.9%, while the negative predictive values were 90.4% and 90.9%, respectively. Among the HIV positive patients, frontloading identified 59/79 (74.7%) culture positive samples while the standard approach identified 55/79 (69.6%). In the HIV sero-negative category, on the other hand, front-loading identified 48/110 (43.6%) culture positive samples compared to 45/110 (40.9%) by the standard approach. CONCLUSION: Frontloading based on smear examination of two same-day sputum samples has a similar performance to the current standard method and would not be associated with any significant missed diagnosis. It may therefore be advocated for use in our setting so as to reduce time to completion of diagnosis and patient loss to follow-up.