Reliability of HIV Rapid Tests in a Clinical Setting is Largely Dependent on User Performance

Abstract

OBJECTIVES
Four widely used HIV rapid tests were evaluated in KwaZulu Natal to determine potential disparities in performance characteristics of rapid tests when used by nurses and counsellors as compared to skilled laboratory staff.
DESIGN
Whole blood samples from 961 pregnant women at 12 clinics that routinely provide Prevention of Mother-to-Child Transmission (PMTCT) services were tested on-site with 4 commercially available HIV rapid tests. Remaining whole blood specimens were tested with ELISA (Virology Laboratory) and 20% (191) of these were also tested with the four rapid tests by qualified laboratory staff.
RESULTS
The sensitivity and specificity of the rapid tests when performed by nurses and compared against the ELISA (Gold Standard) ranged from 92.5%-97.3% and 97.6%-98.2% respectively. Results of all four rapid tests performed by laboratory technicians on 211 samples were concordant with the ELISA results (sensitivity of 100%; 95%CI 95.9-100) and (specificity of 100%; 95%CI 96.5-100).
CONCLUSIONS
All four locally available HIV rapid tests meet international requirements however our data showcase the differences in performance characteristics when used by nursing and counselling staff at clinical sites as compared to skilled laboratory staff. Using such tests in a country of high HIV seroprevalence (30%), with an estimated 500 000 pregnant women testing for HIV annually, an estimated 3 600 (2.4%) and 7 000 (2%) women could be falsely diagnosed as HIV positive and negative respectively. These evaluations highlight the need for ongoing training, supervision and quality control to ensure reliable rapid HIV testing to avoid imminent litigations.

Word count: 247