Incidence Estimation Methodology
Reliable incidence estimates are key to monitoring the spread of HIV.
Estimation methods can be broadly categorised as follows:
- Measuring incidence during the prospective follow-up of an initially uninfected cohort, directly observing changes in infection status between follow up visits.
- Estimating incidence by fitting dynamical models (such as the UNAIDS SPECTRUM model, and the ASSA AIDS and Demographic model) to a range of available demographic and epidemiological data
- Inferring incidence from cross-sectional surveys testing for biomarkers of ‘recent infection’, formalising the intuition that a high prevalence of ‘recent infection’ correlates to a high recent incidence
- Inferring incidence by fitting prevalence data at multiple time points to fundamental demographic population renewal equations, given differential mortality as an additional input, but without reliance on assumptions, such as about population mixing and functional forms for post infection survival, as typically found in ‘whole population’ models.
The cohort approach is often referred to as the ‘gold standard’ for incidence estimation, but in practice, enrolled cohorts present many challenges, including high cost and potential for bias.
The use of dynamical models to estimate parameters such as incidence involves qualitative assumptions and potentially onerous data requirements.
SACEMA focuses primarily on the use of