Monitoring the epidemic of Human Immunodeficiency Virus (HIV) infection plays a vital role in tracking the leading edge of HIV transmission and designing intervention programs both at individual and population level. At individual level, it is imperative to identify newly infected individuals to reduce onwards transmission. At population level, knowledge on the HIV incidence is essential to monitor the spread of the epidemic and plan/evaluate HIV prevention programs. This dissertation will examine the way in which biomarker data can be used to monitor the HIV epidemic. There are two primary aims of this thesis: a) to investigate the use of biomarkers in quantifying the recency of HIV infection at individual level and b) to estimate quantities such as mean window period and testing rate that are the building blocks for estimating HIV incidence at population level. We apply and further develop existing statistical methods to answer the research questions of interest. At individual level, we investigate the use of one or more biomarkers to quantify the recency of HIV infection. We propose a novel approach to make probabilistic statements on the recency of HIV infection by combining the knowledge on the growth of such biomarkers with observations from a newly diagnosed individual. Univariate and bivariate non-linear mixed-effects models are implemented in a fully Bayesian framework. A simulation study is conducted to investigate the biomarkers’ features that affect the accuracy of the estimation of recency. The research findings suggest that rapidly evolving biomarkers of antibody response, such as LAg Avidity, provide reliable estimates of the probability of recency. The proposed methods are applied to a panel of individuals for whom information on various biomarkers is given along with an estimated date of detectable infection. At population level, we focus on estimating two fundamental ingredients, the mean window period and the HIV testing rate, required for estimating HIV incidence using biomarker data. We compare commonly used statistical methods and explore the use of multi-state models in estimating the mean window period of the fourth generation Architect Avidity. We further investigate the factors that are associated with the probability of having an HIV test and the HIV testing rate using surveillance data. Logistic and count regression models using the Generalized Estimating Equations (GEE) approach are employed to make inference at population level about the probability of testing and HIV testing rate respectively.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:744932 |
| Date | January 2018 |
| Creators | Koulai, Loumpiana |
| Contributors | De Angelis, Daniela ; Presanis, Anne |
| Publisher | University of Cambridge |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
| Source | https://www.repository.cam.ac.uk/handle/1810/277038 |
Page generated in 0.0019 seconds