Performance of LED-Based Fluorescence Microscopy to Diagnose Tuberculosis in a Peripheral Health Centre in Nairobi

BACKGROUND Sputum microscopy is the only tuberculosis (TB) diagnostic available at peripheral levels of care in resource limited countries. Its sensitivity is low, particularly in high HIV prevalence settings. Fluorescence microscopy (FM) can improve performance of microscopy and with the new light emitting diode (LED) technologies could be appropriate for peripheral settings. The study aimed to compare the performance of LED-FM versus Ziehl-Neelsen (ZN) microscopy and to assess feasibility of LED-FM at a low level of care in a high HIV prevalence country. METHODS A prospective study was conducted in an urban health clinic in Nairobi, Kenya. Three sputum specimens were collected over 2 days from suspected TB patients. Each sample was processed with Auramine O and ZN methods and a 4(th) specimen was collected for TB culture reference standard. Auramine smears were read using the same microscope, equipped with the FluoLED™ fluorescence illuminator. Inter-reader agreement, reading time and technicians' acceptability assessed feasibility. RESULTS 497 patients were included and 1394 specimens were collected. The detection yields of LED-FM and ZN microscopy were 20.3% and 20.6% (p = 0.64), respectively. Sensitivity was 73.2% for LED-FM and 72% for ZN microscopy, p = 0.32. It was 96.7% and 95.9% for specificity, p = 0.53. Inter-reader agreement was high (kappa = 0.9). Mean reading time was three times faster than ZN microscopy with very good acceptance by technicians. CONCLUSIONS Although it did not increase sensitivity, the faster reading time combined with very good acceptance and ease of use supports the introduction of LED-FM at the peripheral laboratory level of high TB and HIV burden countries.