An operational study comparing microscopes and staining variations for tuberculosis LED FM.

Setting: Tuberculosis control projects, Damien Foundation Bangladesh.

Objectives: To compare transmitted fluorescence (Olympus CX21™/FRAEN FluoLED™) with epi-fluorescence (Zeiss Primostar iLED™) light-emitting diode microscopes (LED-FM) and various auramine staining and destaining/counterstaining techniques for the detection of acid-fast bacilli.

Design: Multicentre blinded reading of routine smears on both types of microscopes using different staining techniques in multiple phases. LED-FM rechecking of discordant series with and without restaining to calculate operating characteristics.

Results: Among 64 874 smears, both instruments detected 9.6% positives. Compared to the standard technique, the stronger auramine-O formulation did not perform better. Thiazine red counterstaining tended to yield more false-positive as well as false-negative errors. Combined destaining/counterstaining (sensitivity 93%, positive predictive value [PPV] 98%) proved significantly less effective. Both destaining with 1% hydrochloric acid (HCl) and 10% alcohol and the standard 0.5% HCl and 70-95% alcohol were equally accurate (sensitivity 95-96%, PPV 99%). The sturdiness of the microscopes in field conditions was sub-optimal: only 5/16 instruments did not break down.

Conclusions: Both microscopes performed equally well. The standard staining technique is as good as the more complicated and expensive variations. A destaining solution containing only 10% alcohol works perfectly well. The inferior quality of LED-FM microscope components is an obstacle to FM expansion.