Modelling paediatric HIV burden and assessing progress towards global elimination targets
File(s)
Author(s)
Walters, Magdalene
Type
Thesis or dissertation
Abstract
Eliminating vertical HIV transmission is a global priority. As of 2025, elimination has primarily been achieved in countries with low HIV prevalence. Progress towards elimination targets in countries with high HIV prevalence can be assessed using the Spectrum AIDS Impact Module (Spectrum-AIM), a deterministic HIV model. National HIV programmes often cannot diagnose as many children living with HIV as Spectrum-AIM estimates exist, raising questions about the accuracy of Spectrum-AIM estimates. This thesis critically examines sources of uncertainty in Spectrum-AIM estimates of paediatric HIV incidence. First, I examined the number of infants at risk of HIV infection by identifying when data on and estimates of HIV prevalence and treatment coverage among pregnant women diverge from expected patterns using heuristic and statistical approaches. Second, I examined the impact of newer, more effective antiretroviral regimens and differentiated service delivery on vertical transmission through systematic review and meta-regression. Third, I decomposed uncertainty in Spectrum-AIM paediatric HIV incidence estimates to determine which parameters contribute the most uncertainty to paediatric HIV incidence in Spectrum-AIM. I incorporated revised uncertainty estimates from Aims 1 and 2 and published uncertainty estimates of breastfeeding duration and maternal treatment retention. Results indicate uncertainty about perinatal HIV incidence is lower than estimated by Spectrum-AIM due to low uncertainty in both HIV prevalence among pregnant women and transmission probabilities among treated women. Conversely, uncertainty about paediatric incidence during breastfeeding is higher than estimated by Spectrum-AIM, predominantly due to uncertain maternal treatment retention and transmission probabilities among women not on treatment. These results indicate that national programmes’ inability to identify as many children living with HIV as estimated by Spectrum-AIM reflects difficulty in testing infants born to women not engaged in prevention services rather than model overestimation. Better quantification of maternal treatment retention is essential for understanding progress towards elimination of vertical HIV transmission.
Version
Open Access
Date Issued
2025-10-20
Date Awarded
2026-06-01
Copyright Statement
Attribution-NonCommercial 4.0 International Licence (CC BY-NC)
License URL
Advisor
Imai-Eaton, Jeffrey
Boily, Marie-Claude
Sponsor
National Institutes of Health
Grant Number
1R01AI152721-01A1
Publisher Department
School of Public Health
Publisher Institution
Imperial College London
Qualification Level
Doctoral
Qualification Name
Doctor of Philosophy (PhD)