AIDS Deaths

[jackolney]

Mildly confused by this issue... Why are AIDS deaths peaking after UNAIDS data? And the sudden drop in 2004 (Clearly ART roll-out), but why so sudden?

[jackolney]

The answer lies here:

void Cd4Test::Execute()
{
    UpdateDaly(pPerson,GetTime());
    ChargePreArtClinicVisit(pPerson);
    ChargePreArtClinicCd4Test(pPerson);
    pPerson->SetEverCd4TestState(true);
    pPerson->SetInCareState(true,GetTime());
    if(immediateArtFlag)
        ScheduleImmediateArt(pPerson,GetTime());
    else if(pPerson->GetCurrentWho() > 2 || theRng->Sample(0.05))
        FastTrackArt(pPerson,GetTime());
    else if(ReceiveCd4TestResult(pPerson,GetTime()))
        ScheduleCd4TestResult(pPerson,GetTime());
    else
        SchedulePreArtResultDropout(pPerson,GetTime());
};

The important part is here:

if(pPerson->GetCurrentWho() > 2 || theRng->Sample(0.05))
    FastTrackArt(pPerson,GetTime());

5% of people getting a CD4 test (regardless of their CD4 count OR WHO stage start ART immediately. This was to match ART initiation data from AMPATH, and the likely situation that many people are put on ART due to other reasons aside from a direct CD4 / WHO stage diagnosis.

This explains the large jump in people starting ART in 2004. This does exceed UNAIDS estimates but falls in line with AMPATH estimates from '07 onwards. This large jump in people initiating ART in 2004, explains the sudden dip in AIDS deaths seen in the model.