To broaden access to HIV viral load monitoring (VLM), the use of blood samples from dried blood spots (DBS) or point-of-care (POC) devices, could be of great help in settings where plasma is not easily accessible. The variety of assays available makes the choice complex. This systematic review and meta-analysis aims to estimate the sensitivity and specificity of DBS and POC devices to identify patients in virological failure using World Health Organization (WHO) recommendations (viral load ≥1000 copies/mL), compared with plasma, for the assays currently available. Four databases were searched for articles, and two reviewers independently identified articles reporting sensitivity and specificity of DBS and/or POC to identify patients in virological failure. We excluded articles that used other thresholds as well as articles with a total number of participants below 50 to avoid reporting bias. Heterogeneity and factors associated with assays' performances were assessed by I2 statistics and metaregression. The protocol of this review follows the PRISMA guidelines. Out of 941 articles, 47 were included: 32 DBS evaluations and 16 POC evaluations. Overall, when using DBS, the Abbott RT HIV-1, Roche CAP-CTM, NucliSENS BioMerieux and Aptima assays presented sensitivity and specificity exceeding 85%, but reported results were highly heterogeneous. Factors associated with better performances were high volume of blood and the use of the same assay for DBS and plasma VLM. Regarding the POC devices, SAMBA I, SAMBA II, and GeneXpert devices presented high sensitivity and specificity exceeding 90%, with less heterogeneity. DBS is suitable VLM, but performances can vary greatly depending on the protocols, and should be performed in trained centers. POC is suitable for VLM with less risk of heterogeneity but is more intensive in costs and logistics.
Keywords: dried blood spots; point of care; sensitivity; specificity; viral load; virological failure.