Accuracy

Health Protection Agency evaluation¹ of ten fourth-generation tests found that all except one had a sensitivity of 100% – in other words, all HIV positive people tested were correctly diagnosed.

One test (Abbott Murex HIV Ag/Ab combination) did fail to identify a single HIV-positive sample (of 508 tested). This sample was of HIV-2, and this test was judged to have a sensitivity of 99.8%.

A WHO evaluation of five tests found that each had a sensitivity of 100%.²

A French evaluation of twelve tests found that nine had a sensitivity of 100%. Two tests did fail to identify a single HIV-positive sample (of 669 tested), and were judged to have a sensitivity of 99.8%. These tests were the Vidas Duo and Enzygnost HIV Integral. The samples missed were of subtypes F and C.³

The 10 tests evaluated by the HPA all had a specificity of 99.7% or over – in other words, if 1000 HIV-negative people were tested, 997 would be correctly diagnosed as such, while three people’s samples would be reactive (positive). (However, in practice, all reactive results are checked with confirmatory tests, so individuals would not receive an incorrect diagnosis.)

The WHO evaluation found that the specificity of the kits they assessed was somewhat variable:

• Genscreen Plus HIV Ag/Ab – 98.3%
• Vironostika HIV Uni-Form II Ag/Ab – 99.0%
• Abbott Murex HIV Ag/Ab Combination – 99.3%
• Genedia HIV Ag-Ab – 99.7%
• Enzygnost HIV Integral II – 100%

The French evaluation of twelve fourth-generation kits noted a significant improvement in specificity when comparing the tests launched before and after 2000. The older tests had a specificity of between 99.4% and 99.6%; the newer tests ranged between 99.8% and 100%.³

The HPA evaluation assessed the ability of the tests to detect recent infection. All fourth-generation tests outperformed third-generation tests in this respect. Whereas the best third generation test could detect infection a mean of 9.8 days after HIV RNA was detectable, several tests were able to identify infection several days earlier (e.g. Abbott AXSYM HIV Ag/Ab Combo, 4.2 days after RNA; Genscreen Ultra HIV Ag/Ab, 5.3 days after RNA).

The French study³ found that fourth-generation tests varied a little in their ability to detect recent infection:

• Three tests identified infection less than three days after RNA (Vidas HIV Duo Quick, Vidas HIV Duo Ultra, Architect HIV Combo).
• Seven tests identified infection between three and five days after RNA.
• Two tests had similar performance to third-generation tests (Enzygnost HIV Integral, Vironostika HIV Uniform II).

Levels of p24 start to drop approximately a month after infection, but antibody is usually detectable by this stage. However in some extremely rare cases, p24 may become undetectable before antibodies are detectable, and a fourth-generation test could give a false negative result. Tests for p24 only do exist, and these are more sensitive than the combined tests, so are more likely to show infection at this stage.^{4 3 5}

False negatives usually occur only when someone is tested soon after exposure, or occasionally when a test is insufficiently sensitive to a rare subtype or to HIV-2. Tests are liable to produce a small number of false positive results, regardless of the person testing, but some medical conditions and vaccinations can also contribute to false positives.⁶