Confirmatory tests

In laboratory practice,^{1,2 3 4} nobody should receive an HIV-positive diagnosis without the following steps all being taken:

• A test giving a positive result.
• The test being repeated (either with the same testing equipment or with a more sensitive test) and still giving a positive result.
• A new blood sample being taken and this also giving a positive result.
• Any discrepancy between the test results, or any indeterminate results, being investigated, and further tests carried out if necessary.

The principle of this sequential testing strategy is that the first tests used should have a very high sensitivity. In other words, they detect the largest possible number of samples from those who really do have HIV. For example, a test with a sensitivity of 99.8% will correctly identify 998 of every 1000 HIV-positive people who take it.

However, the test may also incorrectly report as positive a number of people who are not infected at all. These cases will be cleared up by confirmatory testing.

Depending on the laboratory's practice, negative results may be retested with one other assay, or the initial negative result may be considered reliable enough. (Repeat testing after a time interval may, however, be suggested if there is good reason to think there has been a recent exposure to HIV, or there are possible symptoms of seroconversion.)

All laboratories will also retest positive samples using one or two different types of assay to the one first used. The new test kits should work on different principles to the first one, so that they are unlikely to give false positive results to the same samples. Moreover, they should have a higher specificity than the first test. In other words, they should accurately identify the largest number of HIV-negative samples. For example, a test with a specificity of 99.9% will correctly identify 999 of 1000 HIV-negative people.

Only after this sequence of tests has been completed will the doctor who arranged the tests be informed of the results, which remain preliminary. If tests on the first sample were consistent with HIV infection, it is still essential to test a second, follow-up blood sample. This will check against possible clerical errors such as the original sample having been mislabelled.

During confirmatory testing of HIV-positive samples, it is important to identify samples with HIV-2 rather than HIV-1. Whilst almost all HIV tests used in the UK detect both HIV-1 and HIV-2, few distinguish one infection from the other. The prognosis and treatment choices for HIV-2 are different from the more common HIV-1.

The tests used for confirmatory purposes are most commonly fourth-generation tests, although HIV RNA tests and p24 antigen-only tests could be used in cases of suspected recent infection.

When the first test is a rapid test and its result is reactive, some providers confirm the result by immediately testing again with a different rapid test. With this approach, more individuals agree to confirmatory testing and receive their results than if they are required to provide a venous blood sample or to attend a different healthcare facility.⁵,⁶

Moreover, there are some specialised confirmatory tests which can be used to resolve complex cases where a diagnosis is in doubt. These techniques, which detect antibodies, include Western Blots (WB), line immunoassays (LIA) and immunofluorescent antibody assay (IFA). These tests are extremely specific, but tend to be expensive, labour intensive and require special equipment. Western Blots also sometimes give indeterminate results, and have a longer window period than fourth-generation tests.

Tests sometimes give indeterminate, equivocal or difficult to interpret results. This means that the test kit shows a weak positive reaction, perhaps because of an unknown antibody cross-reaction, because of cross-contamination with another person's sample or because of another technical error.

However there may also be indeterminate results in people following recent infection, because their immune response is still developing.

If an indeterminate result is not reproduced in other tests, it is almost certainly simply a false positive reaction. However, if more than one type of test gives an indeterminate result, it will be necessary to retest with several types of test.

If results remain difficult to interpret, a new sample should be taken around two weeks later. If the cause of the indeterminate results was recent infection the tests should now give a more definitive answer.