Even if you accept the party line on HIV testing, take the numbers and do the math the average "HIV positive" is twice as likely to be false-positive as true-positive.*
The law of positive predictive value is about as arguable as 2+2=4. It is a simple statement of fact that has profound implications for people subjected to HIV testing. A test that claims to be 99% or even 99.5% accurate sounds good. However, any test, regardless of its accuracy, performs badly in a population where very few people are expected to be infected.
There are two ways that people commonly have an "HIV test":
HIV tests in the doctors office or clinic
Office-based screening tests for HIV are believed to be extremely sensitive and specific. The most common tests, the enzyme-linked immunosorbent assay (ELISA) and the Western blot assay, when used in combination, are said to be accurate more than 99% of the time. Let us accept this highly optimistic claim for the moment and apply the laws of positive predictive value:
The prevalence of HIV is estimated to be about 0.5% in the general population. Let us assume that pregnant women fall into this category of risk and, as in Ontario, are routinely tested for HIV. Of 100,000 pregnant women screened 500 will be HIV-infected (prevalence of 0.5%); 495 of the 500 will have positive screening tests (sensitivity of 99%). However, there will be 995 positive tests in these pregnant women who are not HIV-infected (specificity of 99%). This results in a positive predictive value of 33% (495 true positives among 1,490 total "positives"). Of every three women testing HIV-positive, two are certain to be false positive. It is worth noting that with 995 false-positives only 5 false-negatives are predicted to occur. (Figure 1.)
Figure 1. Expected results of office-based HIV screening tests. Disease prevalence of 0.5% among the general population is assumed.
The ramifications of routine screening of people in the general population is disaster in the making. Pregnant women, for instance, have to contemplate abortion or whether to subject themselves and their child to the toxic effects of AZT. If they resist those options, they may lose custody of their child once it is born.
Home testing for HIV
The same math applies - except that the specificity and sensitivity of the do-it-yourself test is only claimed to be 90%. The positive predictive value is now only 4.3%; for every 100 positive home HIV tests in the usual risk population, 95 or 96 will actually be falsely positive. A positive result on these "home tests" are meant to be "confirmed" with the series of tests as described in office-based testing. But there is no way of knowing how many positive home testers make the trip to the doctors office or clinic to seek "confirmation". It would be reasonable to assume that many never do as they chose to test in complete anonymity in the first place. Certainly the package inserts don't alert people to the fact that their chances of reacting false positive are an average more than 9 to 1.
Figure 2. Expected results of home HIV tests. Disease prevalence of 0.5% among the general population is assumed.
A 1992 National Health Interview Survey(1) asked respondents whether or not they would use home HIV testing if it were available. The respondents "at risk" status for HIV infection was assessed, and results showed that 42% of those at risk and 29% of all respondents stated they would be very or somewhat likely of use the home HIV test.
AIDS Update 1999, a college level text book, spells out the real risk:
Former Senator Lawton Chiles of Florida, at an AIDS conference in 1987, told of a tragic example from the early days of blood screening in Florida. Of 22 blood donors who were told they were HIV-positive by the ELISA test, seven committed suicide. There continue to be false positive reactions among blood donors and low-level risk populations because of the low prevalence of HIV infection in such populations.(2)
Percentage of the population outside of AIDS risk groups predicted to be HIV infected: 0.006% (based on screening the U.S. blood supply).Let's apply the law of positive predictive value and these statistics to a population of 100,000 people.
100,000 X 0.006% = 6100 FALSE POSITIVES FOR EVERY 6 TRUE POSITIVES
If the accuracy slips even to 99%, as is sure to happen with the home test kits, we get:
from: Mass HIV Testing: A Disaster in the Making by Christine Johnson
2. Stine G.J, AIDS UPDATE 1999, Prentice Hall