Table 3.
Accuracy of MTBDRsl for detection of resistance to fluoroquinolones and second-line injectable drugs, by reference standard and type of testing, indirect comparisons
| Pooled sensitivity (95% CI) | Pooled specificity (95% CI) |
Pooled sensitivity (95% CI) |
Pooled specificity (95% CI) |
Pooled sensitivity P value1 |
Pooled specificity P value1 |
|
Fluoroquinolone, culture, indirect testing (16 studies, 1766 participants) |
Fluoroquinolone, culture, direct testing (7 studies, 1033 participants) |
||||
| 83.1% (78.7 to 86.7) | 97.7% (94.3 to 99.1) | 85.1% (71.9 to 92.7) | 98.2% (96.8 to 99.0) | 0.670 | 0.293 |
|
Fluoroquinolone, culture, indirect testing (16 studies, 1766 participants) |
Fluoroquinolone, genetic sequencing, indirect testing (7 studies, 974 participants) |
||||
| 83.1% (78.7 to 86.7) | 97.7 (94.3 to 99.1) | 99.3% (85.9 to 100) | 99.7% (92.0 to 100) | < 0.001 | 0.663 |
|
Fluoroquinolone, culture, indirect testing (16 studies, 1766 participants) |
Fluoroquinolone, genetic sequencing and culture, indirect testing (7 studies, 1211 participants) |
||||
| 83.1% (78.7 to 86.7) | 97.7% (94.3 to 99.1) | 82.0 (77.7 to 85.6) | 99.8 (98.5 to 100) | 0.983 | < 0.001 |
|
Ofloxacin, culture, indirect testing (11 studies, 1544 participants) |
Moxifloxacin, culture, indirect testing (4 studies, 222 participants) |
||||
| 82.9% (79.5, 85.9) | 98.2% (96.1, 99.1) | 91.4% (64.7 to 98.4) | 90.6% (79.3 to 96.1) | 0.239 | 0.061 |
|
Second-line injectable drugs, culture, indirect testing (14 studies, 1637 participants) |
Second-line injectable drugs, culture, direct testing (6 studies, 947 participants) |
||||
| 76.9% (61.1 to 87.6) | 99.5% (97.1 to 99.9) | 94.4% (25.2 to 99.9) | 98.2% (88.9 to 99.7) | 0.451 | 0.005 |
|
Second-line injectable drugs, culture, indirect testing (14 studies, 1637 participants) |
Second-line injectable drugs, indirect, genetic sequencing (6 studies, 873 participants) |
||||
| 76.9% (61.1 to 87.6) | 99.5% (97.1 to 99.9) | 97.0% (77.0 to 99.7) | 99.5% (94.5 to 100) | 0.047* | 0.935* |
|
Second-line injectable drugs, culture, indirect testing (14 studies, 1637 participants |
Second-line injectable drugs, genetic sequencing and culture, indirect testing (7 studies, 1491 participants) |
||||
| 76.9% (61.1 to 87.6) | 99.5% (97.1 to 100) | 56.7% (40.8 to 71.3) | 99.9% (99.2 to 100) | 0.340 | 0.003 |
|
Amikacin, indirect, culture (9 studies, 1017 participants) |
Kanamycin, culture, indirect testing (9 studies, 1342 participants) |
||||
| 87.9% (82.1 to 92.0) | 99.5% (97.5 to 99.9) | 66.9% (44.1 to 83.8) | 98.6% (96.1 to 99.5) | 0.006 | 0.262 |
|
Amikacin, culture, indirect testing (9 studies, 1017 participants) |
Capreomycin, culture, indirect testing (10 studies, 1406 participants) |
||||
| 87.9% (82.1 to 92.0) | 99.5% (0.975 to 0.999) | 79.5% (58.3 to 91.4) | 95.8% (93.4 to 97.3) | 0.309* | 0.003* |
|
Kanamycin, culture, indirect testing (9 studies, 1342 participants) |
Capreomycin, culture, indirect testing (10 studies, 1406 participants) |
||||
| 66.9% (44.1 to 83.8) | 98.6% (96.1 to 99.5) | 79.5% (58.3 to 91.4) | 95.8% (93.4 to 97.3) | 0.437 | 0.043 |
|
Amikacin, culture, indirect testing (9 studies, 1017 participants) |
Amikacin, culture, direct testing (4 studies, 803 participants) |
||||
| 87.9% (82.1 to 92.0) | 99.5% (0.975 to 0.999) | 97.3% (55.1 to 99.9) | 99.3% (92.3 to 99.9) | 0.739 | 0.035 |
*
An indirect comparison uses all studies. Indirect statistical comparisons for the purpose of determining pooled accuracy estimates are not to be confused with indirect MTBDRsl testing which involves testing of the culture isolate.
*Indicates the model allowed the variances of the random effects to be associated with the covariate.
1
Likelihood ratio test for evidence of a significant difference between accuracy estimates.