Objective: The Committee for Mycobacterial Examinations has programmed and implemented the external quality assessment for Mycobacterium tuberculosis drug susceptibility testing by proficiency testing panel to the commercial laboratories that perform drug susceptibility testing more than 20 cases monthly.
Method: It utilized 20 strains, exactly pairs of 10 strains, of Mycobacterium tuberculosis that were evaluated for the resistance pattern in the WHO/IUATLD supra-national laboratory network (SRLN), and the concordant results within SRLN were employed as the gold standard of the test. Each laboratory performed the drug susceptibility testing with routine method including the preparation of bacterial suspension.
Results: As a result, the sensitivity for isoniazid (INH) was 95% on an average, ranging from 29% to 100%. The specificity and efficiency for INH were 100% and 97% (50-100), respectively. Additionally, the reproducibility for INH was 99% (90-100). Similarly, the sensitivity, specificity, efficiency and reproducibility for rifampicin (RFP) were 95% (63-100), 99% (67-100), 97% (70-100) and 98% (90-100), respectively. The sensitivity, specificity, efficiency and reproducibility for streptomycin (SM) were 66% (33-100), 94% (64-100), 86% (65-100) and 96% (80-100), respectively. For EB, the sensitivity, specificity, efficiency and reproducibility were 96% (33-100), 74% (57-86), 81% (60-90) and 96% (80-100), respectively.
Discussion: The required standard for satisfactory results of sensitivity, specificity and reproducibility is 95%, and the results for INH and RFP were excellent on an average. However, several laboratories showed unsatisfactory results, especially in sensitivity. Almost all the discordant results for INH and RFP were false-susceptible comparing to the gold standard. For INH, one pair of tested strain showed discordant results in many evaluated laboratories and seemed to reflect the relatively low concordance even in SRLN. On the other hand, the reason for false-susceptible results for RFP was unclear. Several laboratories showed discordant results in one strain pair so that it might be a technical mistake. The sensitivity of SM was remarkably low and it was due to many false-susceptible results, which was twice the number of false-resistance. It was due to the character of tested strain partially considering relatively low concordance in SRLN. However, there was no persuasive explanation for the inconsistency in other strains that showed highly concordant results in SRLN. The difference of critical drug concentration in Lowenstein-Jensen and 1% Ogawa medium might contribute to the discrepancies. The specificity of EB was low and it was due to many false-resistant results. About a specific pair of tested strain, almost 100% of laboratories showed concordant false-resistant results whereas SRLN showed perfectly concordant susceptible result. If this strain was considered as resistant and another strain that showed less than 70% concordance in the tested laboratories was excluded, the sensitivity, specificity and efficiency would improve to 96%, 97% and 97%, respectively. The proficiency testing is a part of external quality assessment system and requires following corrective actions. A permanent quality improvement and maintenance will be expected through the spiral circulation of quality assurance activities.