Twenty-one studies with a total of 1,778 samples were included in the review. There were 14 studies of amplification-based assays and 7 studies of luciferase reporter phage assays.
Eight of the 21 included studies used random or consecutive sampling methods. Twelve studies reported blinded interpretation of either the phage-based assay or the reference standard. None of the studies had the potential for verification bias. The included studies reported the proportion of indeterminate results or contaminated samples: this ranged from zero to 17%.
Accuracy of phage amplification assays (14 studies).
Most studies used culture isolates; 2 studies directly applied FASTPlaque-TB assays to sputum specimens. For the commercial kits (8 studies), the sensitivity ranged from 81 to 100% and the specificity from 73%to 100%; the AUC, derived from the summary ROC curve, was 0.99 (SE=0.0105) and the Q* index was 0.95 (SE=0.0232). Of the 2 studies that assayed sputum samples directly, one reported sensitivity and specificity values of 100% and 99%, respectively, while the other reported a sensitivity of 86% and a specificity of 73%. For the in-house assays (6 studies), the sensitivity ranged from 97 to 100% and the specificity from 84 to 100%; the AUC, derived from the summary ROC curve, was 0.99 (SE=0.0016) and the Q* index was 0.98 (SE=0.0056).
Accuracy of luciferase reporter phage assays (7 studies).
All of these studies were of in-house assays applied to culture isolates. With the exception of one study, which had a sensitivity of 92%, all of the studies had sensitivity estimates of 100%. The specificity estimates ranged from 89 to 100%. The AUC, derived from the summary ROC curve, was 0.98 (SE=0.0042) and the Q* index was 0.95 (SE=0.0085).
RIF resistance as a marker for multi-drug resistant TB (7 studies).
Seven studies determined the number of RIF-resistant samples that were also isoniazid-resistant (the criteria for multi-drug resistant TB). On average, 96% of RIF-resistant samples were also isoniazid-resistant.