An Internally Consistent Model of HIV Burden for Countries with Vital Registration and Case Notification Data
MetadataShow full item record
This thesis aims to provide a novel, internally consistent approach to estimating HIV morbidity and mortality in countries with both vital registration and case notification data. In most high-income countries, case notifications and deaths recorded in vital registration serve as the primary data sources for monitoring the course of the HIV epidemic. However, past modelling efforts have encountered difficulty reconciling these two sources; achieving a good fit to diagnosis data has resulted in underestimating mortality, and vice versa. These incongruities have led to disconnected approaches to modelling HIV burden, producing morbidity and mortality estimates that are not internally consistent. This study extended an existing modelling framework, the Estimation and Projection Package Age/Sex Model (EPP-ASM), to develop internally consistent estimates of HIV burden in countries with case notification and high-quality vital registration data. EPP-ASM was modified to carry out Bayesian inference on both HIV transmission dynamics and an adjustment to off-antiretroviral therapy (ART) HIV mortality rate. Three structural assumptions on the relationship between off-ART HIV mortality rate and ART coverage were considered. Vital registration data stratified by age and sex and case notifications were incorporated into the likelihood calculation within the model. EPP-ASM was fit for the Netherlands and Australia, and in-sample predictive validity was assessed using coverage of posterior predictive intervals and mean absolute error. EPP-ASM with no adjustment to off-ART mortality rate demonstrated the best model performance, and out-performed the Cohort Incidence Bias Adjustment method used by the Global Burden of Disease study. Future investigation of adjustments to HIV mortality rate and extension to additional locations could provide further insight into reconciliation of diagnosis and mortality data.
- Global health