Search HIVST

Supervised oral HIV self-testing is accurate in rural Kwazulu Natal, South Africa


Author
Martinez-Perez et al.

Publication year
2016

Country

Type of approach
Dual

Type of assistance
Directly assisted

Specimen
Oral-fluid

Study population
General population: General population

Study design
Feasibility/acceptability

Sample size
2198

UNAIDS HIV prevalence (2017)
19.2 [18.4 - 20.0]

Methodology
Cross-sectional study with two government clinic sites and three Medecin Sans Frontieres fixed sites, potential participants were approached in clinic waiting areas, testing involved counselors demonstrating how to use an oral fluid HIV self-test and observing usage. Lay counselors sampled and recruited eligible participants, sought informed consent, and demonstrated the use of an oral fluid-based HIV self-test. Participants used the HIVST in front of the counselor and underwent a blood-based Determine and a Unigold rapid diagnostic test as gold standard for comparison. Primary outcomes were user error rates, inter-rater agreement, sensitivity, specificity, and predictive values.

Summary of findings
Among the 130 people who had never tested before but who self-tested, 30 men (42.8%) and 29 women (48.3%) were in the 18-25 years age group. Of 2181 participants who were unaware of their status, inter-rater agreement between HIVST and standard HIV testing was 99.8% (Kappa 0.9925). Overall, HIVST was correctly performed by 2196 users (user error rate 0.09%). Overall sensitivity for HIVST was 98.7% (95%CI 96.8-99.6) and specificity was 100.0% (95%CI 99.8-100). HIV prevalence in this study was 14.7% (95% CI; 13.2-16.2): 10.6% (95% CI; 13.1-16.1) were women and 3.9% (95% CI: 3.1-4.8) were men.

Acceptability
n/a

Acceptability details
n/a

Willingness to pay
n/a

Willingness to pay details
n/a

Sensitivity
0.987

Specificity
1

Concordance
0.998

HIV positivity
14.9% (327/2187)

Accuracy details
Sensitivity for the OralST was 98.7% (95%CI 96.8-99.6) and specificity was 100% (95%CI 99.8-100). Overall inter-rater agreement was 99.8% (Kappa 0.9925).

Social harm
n/a

Linkage to prevention, care and treatment
All participants with reactive tests received confirmatory testing, and in cases of an HIV diagnosis they were linked to care (n=n/a).


Study status
Completed