Send to

Choose Destination
Lancet HIV. 2018 Jun;5(6):e277-e290. doi: 10.1016/S2352-3018(18)30044-4. Epub 2018 Apr 24.

Reliability of HIV rapid diagnostic tests for self-testing compared with testing by health-care workers: a systematic review and meta-analysis.

Author information

Department of HIV, World Health Organization, Geneva, Switzerland. Electronic address:
Department of HIV, World Health Organization, Geneva, Switzerland; Clinical Research Department, London School of Hygiene & Tropical Medicine, London, UK.
Department of HIV, World Health Organization, Geneva, Switzerland.
Department of Essential Medicines and Health Products, World Health Organization, Geneva, Switzerland.
PSI Zimbabwe, Harare, Zimbabwe.



The ability of individuals to use HIV self-tests correctly is debated. To inform the 2016 WHO recommendation on HIV self-testing, we assessed the reliability and performance of HIV rapid diagnostic tests when used by self-testers.


In this systematic review and meta-analysis, we searched PubMed, PopLine, and Embase, conference abstracts, and additional grey literature between Jan 1, 1995, and April 30, 2016, for observational and experimental studies reporting on HIV self-testing performance. We excluded studies evaluating home specimen collection because patients did not interpret their own test results. We extracted data independently, using standardised extraction forms. Outcomes of interest were agreement between self-testers and health-care workers, sensitivity, and specificity. We calculated κ to establish the level of agreement and pooled κ estimates using a random-effects model, by approach (directly assisted or unassisted) and type of specimen (blood or oral fluid). We examined heterogeneity with the I2 statistic.


25 studies met inclusion criteria (22 to 5662 participants). Quality assessment with QUADAS-2 showed studies had low risk of bias and incomplete reporting in accordance with the STARD checklist. Raw proportion of agreement ranged from 85·4% to 100%, and reported κ ranged from fair (κ 0·277, p<0·001) to almost perfect (κ 0·99, n=25). Pooled κ suggested almost perfect agreement for both types of approaches (directly assisted 0·98, 95% CI 0·96-0·99 and unassisted 0·97, 0·96-0·98; I2=34·5%, 0-97·8). Excluding two outliers, sensitivity and specificity was higher for blood-based rapid diagnostic tests (4/16) compared with oral fluid rapid diagnostic tests (13/16). The most common error that affected test performance was incorrect specimen collection (oral swab or finger prick). Study limitations included the use of different reference standards and no disaggregation of results by individuals taking antiretrovirals.


Self-testers can reliably and accurately do HIV rapid diagnostic tests, as compared with trained health-care workers. Errors in performance might be reduced through the improvement of rapid diagnostic tests for self-testing, particularly to make sample collection easier and to simplify instructions for use.


The Bill & Melinda Gates Foundation and Unitaid.

Supplemental Content

Full text links

Icon for Elsevier Science Icon for PubMed Central
Loading ...
Support Center