This article was reported by AIDSMAP.
NAM aidsmap reported that a large, two-part San Francisco study indicated that "fourth-generation" laboratory and rapid HIV tests were more effective than other commonly used HIV tests in diagnosing acute infections, when individuals were more likely to infect others. The study found that HIV antibody tests missed most acute -- recent -- HIV infections and also missed established HIV infections. The study included 21,234 tests that identified 761 HIV infections (3.6-percent prevalence) and 58 acute infections (0.3-percent prevalence).
The study reviewed a range of HIV tests used from 2003 to 2008 to test high-prevalence groups in San Francisco. Clinics supplemented antibody lab or rapid tests with HIV RNA testing (viral load testing), the most accurate method for acute HIV infections. Seeking an alternative to expensive HIV RNA testing, researchers used a variety of methods to retest stored blood plasma samples from the 58 individuals with acute infections.
Key measures included sensitivity (percentage of correct positive results when HIV is present) and specificity (the percentage of correct negative results when HIV is not present). First-generation antibody tests had 92.3-percent sensitivity, missing some established and most acute infections. Third-generation laboratory antibody tests had 96.2-percent sensitivity, but the rapid version was less sensitive. The OraQuick test had "poor results" when individuals used saliva rather than fingerstick blood samples. OraQuick results based on saliva samples missed all acute HIV infections and one of 20 established HIV infections. Fourth-generation laboratory tests had 99.1-percent sensitivity for all infections and 87.3-percent sensitivity for acute infections. Fourth-generation rapid tests had overall 96.6-percent sensitivity, with 54.4-percent sensitivity for acute infections.
The full article, "Performance of Rapid Point-of-Care and Laboratory Tests for Acute and Established HIV Infection in San Francisco," was published online in the journal PLOS One (2013; doi: 10.1371/journal.pone.0080629).