Abstract
BACKGROUND: Sputum microscopy is the only diagnostic for tuberculosis (TB) available at peripheral levels of health service in resource-poor countries. Its sensitivity is reduced in high HIV-prevalence settings. Sodium hypochlorite (NaOCl) specimen sedimentation prior microscopy and light-emitting diode (LED)-fluorescence microscopy (FM) can individually improve performance of microscopy. This study aimed to evaluate the performance of combined LED-FM and NaOCl sputum sedimentation for TB detection at peripheral level of health services. METHODS: A prospective study was conducted in an urban health clinic in Nairobi, Kenya. Three sputum specimens were collected over 2 days from consecutive TB suspects. Smears were prepared and stained with auramine O and Ziehl-Neelsen (ZN) methods. Bleach (3.5%) was added to the remaining specimen before overnight sedimentation at room temperature. Auramine O staining was performed on smears of sediment. A 4(th) specimen was collected for TB culture. Auramine smears were read under the same microscope as used for ZN smears, but equipped with the LED FluoLED™ fluorescence illuminator. RESULTS: 497 patients were included, and 1394 specimens collected. The yield of positive specimen was significantly increased after NaOCl sedimentation (24.9%) compared to direct LED-FM (20.6%) and direct ZN (20.3%). In detecting smear-positive patients, sensitivity was 78.5% for LED-FM after NaOCl sedimentation compared to 73.2% and 72.0% for direct LED-FM (P = 0.06) and direct ZN (P = 0.06), respectively. Specificity was 87.8% for LED-FM after NaOCl sedimentation compared to 96.7% and 95.9% for direct LED-FM (P<0.01) and direct ZN (P<0.01), respectively. Inter-reading agreement (kappa = 0.7) and technicians' acceptability were good. CONCLUSION: NaOCl sedimentation did not improve the performance of LED-FM in the diagnosis of pulmonary TB at peripheral health service level.