Modelling the impact of prevention strategies on cervical cancer incidence in South Africa

van Schalkwyk, Catherina

Modelling the impact of prevention strategies on cervical cancer incidence in South Africa

Doctoral Thesis

2021

Abstract

Background: In 2019, the World Health Organisation called for the elimination of cervical cancer as a public health concern. In South Africa, despite having a national screening policy in place since 2000, diagnosed cervical cancer incidence has shown no signs of decline. Since 2014, girls aged 9 have been vaccinated against HPV infection using the bivalent vaccine, with high coverage. However, due to the long delay between HPV infection and progression to cancer, the impact that vaccination will have on cervical cancer incidence will be unobservable in the near future. This thesis sets out to quantify this impact using a mathematical model, and will estimate the impact of scaling up current cancer prevention strategies, as well as proposed alternative strategies. Methods: This research extends a previously developed individual-based model for HIV to include infection with 13 high-risk HPV types and progression to cervical cancer. HPV infection and cervical disease parameters were calibrated to a wide range of South African data sources using a likelihood based approach. In the process of developing an appropriate model for cervical cancer incidence in South Africa, important aspects related to HIV/HPV co-infection dynamics, the natural history of HPV and the current and historic levels of cervical cancer prevention in the Western Cape were investigated. The calibrated and validated model was used to estimate the impact of current and proposed alternative prevention strategies on cervical cancer incidence in the next century. Findings: Using a model structure that does not include a biological transmission co-factor, we show that simulated associations between HIV and HPV transmission are similar to corresponding empirical estimates and therefore these associations may result from residual confounding by sexual behavioural factors and network-level effects. Using simulated vaccine trials, we show that viral latency and reactivation of latent infections is necessary in the natural history of HPV to match results from empirical trials. The model's screening algorithm reflects findings from the Western Cape's public health sector – low levels of screening coverage and linkage to treatment facilities, and poor adherence to screening schedules. The model matches stable trends in diagnosed cervical cancer incidence in South Africa, but it estimates increases in cervical cancer incidence over the last number of years (due to increased life expectancy of women on ART), which will result in sharp increases in diagnoses. While decreasing HIV prevalence and HPV vaccination will substantially reduce cervical cancer incidence in the long term, improvements in South Africa's current screening strategy, as well as switching to new screening technologies, will have significant impact in the short term. Conclusions: This thesis presents an epidemiological model of cervical cancer in South Africa – the first to dynamically simulate infection with both HIV and HPV at national level. It allows for estimation of the impact of both HIV and cervical cancer prevention on cancer incidence, and provides the opportunity to identify the vaccination and screening strategies with the greatest public health significance.

Keywords

Public Health

Reference:

Collections

PhD / Doctoral

Full item page