Setting: Microscopy Centres, Velliyur Tuberculosis Unit, Tiruvallur District, Tamil Nadu, India.
Procedures: Twelve microscopy centre laboratory technicians examined 41978 direct sputum smears for acid-fast bacilli. Senior Tuberculosis Laboratory Supervisors (STLS) checked all positive smears (4696) and 10% of negative smears (4776) in an unblinded fashion as per Revised National Tuberculosis Control Programme (RNTCP) guidelines. Ten per cent of the positive and negative slides and another 10% of unchecked negative slides were selected systematically for blinded rereading at the Tuberculosis Research Centre (TRC); 422 slides were reread without and with restaining.
Results: Unblinded checking by STLS of the smears read by the laboratory technicians yielded 95 to 100% agreement. Blinded rereading at the TRC revealed that false-negative errors were greater among the laboratory technicians (2-7%) than the STLS (0-3%). Restaining and blinded rereading of slides reduced false-positive errors from 27% to 7%.
Conclusions: Blinded rereading at the reference centre facilitates assessment of laboratory technicians and STLS. Restaining before rereading the smears was found to be useful for precise estimation of false-positive errors.