Sure - that seems to be a given. However, there is a problem of practicality here I think. How would you suggest obtaining accurate numbers of how many people are HIV-infected in Africa? I think you might find some major obstacles of practicality in testing them all, not least of which would be the expense. If you test sample numbers, enough to get an estimate of how many people are infected, you can know roughly how big the problem is, and how much medication is required based on that estimate. The actual testing could then perhaps be done at the time of distribution, saving a lot of money overall - money that could go towards saving lives. As I understand it, HIV testign can be done much quicker now than say 10-20 years ago, when it took weeks to get a result.

It seems to me a case of avoiding some bureaucracy.