Methods: We searched PUBMED, EMBASE, SCOPUS and other databases through August 2015 for relevant diagnostic accuracy studies published as full-text in English, French or Spanish. Cross-sectional and cohort studies that provided original 2x2 tables on the accuracy of commercialized rapid tests, i.e., results in <30 minutes, compared to RT-PCR as reference standard in children and adults with suspected influenza were included. Two reviewers independently screened citations and extracted data. Study quality and risk of bias were assessed using QUADAS-2. For each of the 3 types of index tests, Bayesian bivariate random effect models were used to calculate separate summary accuracy estimates with their 95% credible intervals (CrI).
Results: We included 102 articles comprising 141 studies of individual index tests. 124 studies evaluated traditional RIDTs, 12 DIAs (BD Veritor; Quidel Sofia) and 5 rapid NAATs (Alere; Cobas Liat). Pooled sensitivities and specificities (and 95% CrI) for influenza A were: 53.6% (47.9%-59.2%) and 99.4% (99.1%-99.4%) for traditional RIDTs, 80.7% (73.6%-87.0%) and 98.2% (97.0%-99.0%) for DIAs, and 87.5% (77.2%-94.4%) and 98.9% (97.0%-99.8%) for rapid NAATs. Pooled sensitivities and specificities for influenza B were: 48.0% (36.0%-60.4%) and 99.8% (99.7%-99.9%) for traditional RIDTs, 79.1% (63.4%-90.0%) and 97.9% (94.7%-99.2%) for DIAs, and 94.0% (83.2%-98.7%) and 99.3% (98.0%-99.9%) for rapid NAATs.
Conclusion: Novel DIAs and rapid NAATs demonstrate markedly higher sensitivities for influenza A and B compared to traditional RIDTs, with equally high specificities.
I. Schiller, None
G. C. Gore, None
C. Chartrand, None
N. Dendukuri, None
J. Papenburg, Becton, Dickenson & Co.: Investigator , Research grant
RPS Diagnostics: Scientific Advisor , n/a
AbbVie: Scientific Advisor , n/a