Ten studies, with a total of 3,349 specimens, were included in the review. These included 14 data sets for the detection of rifampicin resistance and 15 data sets for the detection of isoniazid resistance in tuberculosis. Most studies were cross-sectional but only two reported consecutive or random sampling. All studies reported complete verification of GenoType MTBDR assay results with drug-susceptibility testing. Four studies did not report blinded interpretation of either test.
Accuracy for rifampicin resistance:
The pooled estimates of the diagnostic accuracy of the GenoType MTBDR assay for rifampicin resistance were 98.1% (95% CI: 95.9, 99.1) for sensitivity and 98.7% (95% CI: 97.3, 99.4) for specificity. No significant heterogeneity was detected. Results were similar across all assay and specimen types.
Accuracy for isoniazid resistance:
The pooled estimates of the diagnostic accuracy of the GenoType MTBDR assay for isoniazid resistance were 84.3% (95% CI: 76.6, 89.8%) for sensitivity and 99.5% (95% CI: 97.5, 99.9%) for specificity, but sensitivity estimates were highly heterogeneous across studies. Results were similar across all assay and specimen types.