Objective: To update a previously reported meta-analysis of evidence regarding the diagnostic accuracy and performance characteristics of commercial and non-commercial phage-based assays for the detection of rifampicin (RMP) resistant tuberculosis (TB). DESIGN AND OUTCOMES: We conducted a systematic review and meta-analysis of test accuracy using bivariate random effects regression and hierarchical summary receiver operating characteristics (HSROC) analysis. Tests included the commercial FASTPlaque assays, luciferase reporter phage (LRP) assays, and in-house phage amplification tests. Sensitivity and specificity for RMP resistance were the main outcomes.
Results: By updating previous literature searches, a total of 31 studies (with 3085 specimens) were included in this meta-analysis. Evaluations of commercial phage amplification assays yielded more variable estimates of sensitivity (range 81-100%) and specificity (range 73-100%) compared to evaluations of in-house amplification assays (sensitivity range 88-100%, specificity range 84-100%). LRP evaluations yielded the most consistent estimates of diagnostic accuracy, with seven of eight studies reporting 100% sensitivity and four of eight reporting 100% specificity. Estimates of accuracy failed to capture a major failing of the commercial assay, i.e., the rate of contaminated and indeterminate results. These ranged from 3% to 36% in studies looking at direct detection of RMP resistance from patient specimens (mean 20%).
Conclusion: Phage-based assays will require further development to maximise interpretable results and reduce technical failures. Once technical issues are resolved, impact on patient-important outcomes and cost-effectiveness need to be determined to inform policy for widespread use.