Chenzi Liu, Daojun Yu
Pathogen-specific analyses indicated high diagnostic accuracy for rapid integrated molecular testing of influenza A, influenza B, RSV, and SARS-CoV-2. Because platforms, specimens, populations, settings, designs, and reference methods varied, the overall estimates and HSROC area should be interpreted only as exploratory summaries. Most studies were laboratory based, and direct clinical or economic benefits were not established.
BACKGROUND: Rapid integrated molecular tests are automated sample-to-answer nucleic acid amplification systems that provide respiratory virus results within approximately 120 min. Because their use ranges from genuine near-patient sites to rapid laboratory workflows, testing environments must be clearly classified.
METHODS: PubMed, Embase, and the Cochrane Library were searched from inception through June 2026 without language restrictions. Diagnostic-accuracy studies evaluating integrated rapid molecular assays for influenza A, influenza B, respiratory syncytial virus (RSV), or SARS-CoV-2 were eligible. For the primary analysis, one prespecified platform and one non-nested cohort were selected from each study, and target-specific 2 × 2 counts were aggregated into one study-level composite record. Sensitivity and specificity were jointly synthesized using a bivariate random-effects model. The overall analysis was considered exploratory, with pathogen-specific findings prioritized. Pathogen- and setting-specific subgroup analyses, leave-one-study-out analyses, QUADAS-2 assessment, and Deeks' funnel-plot asymmetry testing were performed.
RESULTS: Of 1,337 identified records, 43 underwent full-text assessment and 23 were excluded. Twenty studies contributed 20 independent study-level records. Sensitivity and specificity were 0.970 and 0.987 for influenza A, 0.961 and 0.992 for influenza B, 0.955 and 0.992 for RSV, and 0.950 and 0.992 for SARS-CoV-2, respectively. The exploratory overall analysis yielded a sensitivity of 0.971 [95% confidence interval (CI), 0.955-0.982], specificity of 0.991 (95% CI, 0.980-0.996), and HSROC area under the curve of 0.957. Five studies were conducted in genuine near-patient settings and 15 in laboratory settings. Leave-one-study-out estimates ranged from 0.968 to 0.974 for sensitivity and 0.989 to 0.992 for specificity. Deeks' test showed no statistically significant funnel-plot asymmetry (P = 0.269).
CONCLUSIONS: Pathogen-specific analyses indicated high diagnostic accuracy for rapid integrated molecular testing of influenza A, influenza B, RSV, and SARS-CoV-2. Because platforms, specimens, populations, settings, designs, and reference methods varied, the overall estimates and HSROC area should be interpreted only as exploratory summaries. Most studies were laboratory based, and direct clinical or economic benefits were not established.