High proportion of positive HIV tests are false in low risk populations.
In low risk populations, the proportion of false positive HIV tests is high, on the order of 50%.
- Assume the tests are 99.9% specific as advertised, ignoring the disclaimers on the test kits. That means there would be only 1 false positive in 1000 tests.
- Also assume the prevalence of HIV in the general population is 1 in 1000.
- To visualize, it may help to think of 1000 marbles in a tin, of which one is red (the false positive) and one is purple (the true positive, a person with antibodies (what is actually checked for) that react with selected HIV proteins). The rest are some other color.
- Now test everyone, which results in two positive tests (the red and purple marbles).
- Therefore, 50% of the positive test results are wrong.
- Probability is a tricky subject, but that really is the correct answer.
- In a high risk population (imagine 500 purple marbles), the proportion of false positives encountered, as a percentage of all positive tests, would be much smaller. In this example, 1 out of 501, or .2%
So, if universal HIV testing is implemented, as some propose, with half the positive tests beng false, what recourse do those wrongly stigmatized have? No small matter, as those testing positive are told to have only protected sex, meaning they will be childless, terminating their future families. In the case of Africa, 25 million or so families.
The answer is, not much in the way of recourse. The isolation of HIV from patients is at best difficult, expensive, and rarely done. In an Adelaide court case, some contended it has never been done.
Some suggest using the PCR (also called viral load) to detect the presence of a claimed HIV protein, but the PCR is prohibited from use in diagnosis (for adults), and has a number of concerns, including a scientific paper which found that PCR results did not correlate with immune system status.
Another suggestion is to retest the sample. That approach works for dice - a second roll is unlikely to come up snake eyes again. In dice however, assuming fair dice, each roll or trial is independent of the prior one. For retesting to be valid, each trial must be independent.
In contrast, retesting the same sample, or even a second sample from the same person, is not independent - the same antibody that confused the first test is likely to confuse the second, being still present in the sample.
Back to home page.