That last sentence in the excerpt means laboratory evidence. So much for the claim that these diagnoses in Africa are just presumptive diagnoses based only on clinical presentation, i.e. appears ill, malnourished, fever, etc.
Well, not exactly. Until the new millennium (or in some cases, the late 1990s), the UN-favored methodology for obtaining AIDS rates in Africa was to use AIDS rate from so-called sentinel testing of pregnant women who visited health clinics as a proxy for the country's rate. National random voluntary blood or saliva testing is a fairly new phenomenon.
The problem with the old methodology is that most clinics are in urban areas, where the AIDS rate is higher, pregnant women are at a higher risk of getting AIDS since they obviously haven't been abstinent or used condoms, and African women tend to be more at risk for AIDS than African men for some reason. So the AIDS rate in many African countries was overstated, often by a factor of about 1.5x-2x, and in Sierra Leone's case, by a factor of 7x.
More info here:
How AIDS in Africa Was Overstated (Washington Post, Craig Timberg 4/6/06)
I made that comment because if I don't read that HIV is a harmless virus, ala Duesberg, then I get comments that the prevalence of HIV/AIDS is overestimated because of the lack of HIV serology, and that the real numbers are obscured by TB, malaria, etc. Kenya has some real numbers, now.