Analysis of Serological Surveillance Data: Patterns of Blood Borne Infection and Heterogeneity in People who Inject Drugs

Harris, Ross Jeremy (2019). Analysis of Serological Surveillance Data: Patterns of Blood Borne Infection and Heterogeneity in People who Inject Drugs. PhD thesis The Open University.



A large proportion of blood-borne viruses (BBV) are transmitted via injecting drug use. Understanding patterns of risk and monitoring trends over time in people who inject drugs (PWID) is therefore a crucial part of developing public health policy.

Risks of infection with hepatitis C and B (HCV, HBV) and HIV in PWID are investigated using serial cross-sectional surveillance data, in particular via force of infection (FOI) models. Standard models are extended to include a wealth of covariate data. Individual variability is considered via bivariate shared frailty models and correlations between infections. Gamma, inverse Gaussian and time-varying frailty models are fitted to investigate how variability in risk evolves throughout injecting career. Finally, models are extended to the trivariate case and different forms of component frailty models are proposed.

Recent initiates were found to be at high risk of infection, in particular in London and the North West. Subsequently the FOI is broadly constant, and similar across different regions. Frailty models indicate that there is substantial individual variability in risk, although this declines over the course of injecting career. Females have higher overall risks of HCV and HBV, but are less heterogeneous than males. Including covariate data on demographics and a number of risk factors only resulted in modest reductions in frailty variance. Correlations between HCV-HBV and HBV-HIV associations were stronger than for HCV-HIV; trivariate models including additional pairwise components for HCV-HBV and HBV-HIV provided an improvement in model fit compared to a shared frailty model.

Many of the results in this thesis point towards greater variation in risk at initiation, potentially due to the varied circumstances in which individuals start injecting, followed by more comparable risks in those with established injecting behaviour. The relative importance of injecting and sexual risk may explain patterns of risk and correlation in the three infections.

Viewing alternatives

Download history


Public Attention

Altmetrics from Altmetric

Number of Citations

Citations from Dimensions

Item Actions