Table 5.
Accuracy of MTBDRsl for detection of extensively drug-resistant TB, by reference standard and type of testing, direct comparisons
Type of testing |
Number of participants (studies) |
Number of resistant cases (TPs + FNs) |
Pooled sensitivity (95% CI) |
Pooled specificity (95% CI) |
Culture testing | ||||
Indirect | 880 (8 studies) | 173 | 70.9% (42.9 to 88.8) | 98.8% (96.1, 99.6) |
Direct | 644 (3 studies) | 50 | * | * |
Genetic testing | ||||
Indirect | 541 (3 studies) | 66 | 100.0% (94.6 to 100.0)‡ | 97.5% (95.6 to 98.7)‡ |
Direct | 0 (0 studies) | 0 | Not applicable | Not applicable |
Genetic and culture testing | ||||
Indirect | 435 (2 studies) | 102 | 58.8% (49.1 to 67.9) | 98.8% (96.8 to 99.5) |
Direct | 0 (0 studies) | 0 | Not applicable | Not applicable |
Culture followed by genetic testing of discordant results | ||||
Indirect | 0 (0 studies) | 0 | Not applicable | Not applicable |
Direct | 516 (1 study) | 26 | 92.3% (74.9 to 99.1) | 99.6% (98.5 to 100.0) |
We observed considerable heterogeneity and did not pool results.
We observed little heterogeneity and determined summary estimates of sensitivity and specificity separately using a fixed-effect model.
TP = true positive; FN = false negative.