Investigating Reports of HIV Incidence Decline

© UNAIDS Global Report 2013

    
The 2012 UNAIDS Report on the Global AIDS Epidemic reported large declines in HIV incidence in a number of countries, both in Africa and globally. While these estimates have been promoted as evidence for the effectiveness of the international response to HIV/AIDS, in many settings these observed reductions were not coupled with changes in risk behaviour or large program efforts or scale up of ART, which may have been anticipated a priori. As such, the HIV Modelling Consortium were asked by the Steering Committee to explore potential mechanisms that (1) could reconcile the apparent discrepancy between these, or (2) could potentially lead to spurious or exaggerated estimates of declines in HIV incidence.

 

Background

UNAIDS have a requirement to report on the status of the HIV epidemic for the United Nations Political declarations on HIV/AIDS and Millennium Development Goals. In order to provide estimates on incidence and prevalence at a country level UNAIDS use the Spectrum and Estimation and Projection Package (EPP) suite of models developed by Futures Institute. The Spectrum and EPP programs are used to estimate key HIV indicators based on HIV surveillance and surveys, programme statistics and epidemic patterns.  Each year UNAIDS publish the latest estimates for these key indicators in their Global Report. The 2012 report estimated that 25 countries experienced a greater than 50% decline in HIV incidence between 2001 and 2011 (Table 1), which is very substantial. While these figures may appear very reassuring it is important to recognize that incidence estimation is inherently difficult in HIV and reporting of these figures may require more explanation of the uncertainty. 

Approach

The Secretariat, in conjunction with external collaborators, identified existing research on the area and developed and conducted some preliminary investigations into several hypotheses. This work, in addition to that of numerous other research groups, was presented at a meeting in London in September 2013. This meeting has been held at the hypothesis gathering stage: to identify whether the incidence declines are real or not and what factors could be attributable for these observations.

Meeting aim: To review and examine reports of HIV incidence declines at a country level

1a) Can we be confident that estimated incidence declines are real?

1b) What might explain lack of association between estimated incidence and individuals' risk of HIV infection?

2) How should this affect messaging/communication on UNAIDS estimates?

3) What development of methods is possible to increase confidence?