Spatiotemporal modelling and mapping of cervical cancer incidence among HIV positive women in South Africa: a nationwide study

• Main Authors: Dhokotera Tafadzwa, Riou Julien, Bartels Lina, Rohner Eliane, Chammartin Frederique, Johnson Leigh, Singh Elvira, Olago Victor, Sengayi-Muchengeti Mazvita, Egger Matthias, Bohlius Julia, Konstantinoudis Garyfallos
• Format: Article
• Language: English
• Published: BMC 2021-06-01
• Series: International Journal of Health Geographics
• Online Access: https://doi.org/10.1186/s12942-021-00283-z

Abstract Background Disparities in invasive cervical cancer (ICC) incidence exist globally, particularly in HIV positive women who are at elevated risk compared to HIV negative women. We aimed to determine the spatial, temporal, and spatiotemporal incidence of ICC and the potential risk factors among HIV positive women in South Africa. Methods We included ICC cases in women diagnosed with HIV from the South African HIV cancer match study during 2004–2014. We used the Thembisa model, a mathematical model of the South African HIV epidemic to estimate women diagnosed with HIV per municipality, age group and calendar year. We fitted Bayesian hierarchical models, using a reparameterization of the Besag-York-Mollié to capture spatial autocorrelation, to estimate the spatiotemporal distribution of ICC incidence among women diagnosed with HIV. We also examined the association of deprivation, access to health (using the number of health facilities per municipality) and urbanicity with ICC incidence. We corrected our estimates to account for ICC case underascertainment, missing data and data errors. Results We included 17,821 ICC cases and demonstrated a decreasing trend in ICC incidence, from 306 to 312 in 2004 and from 160 to 191 in 2014 per 100,000 person-years across all municipalities and corrections. The spatial relative rate (RR) ranged from 0.27 to 4.43 in the model without any covariates. In the model adjusting for covariates, the most affluent municipalities had a RR of 3.18 (95% Credible Interval 1.82, 5.57) compared to the least affluent ones, and municipalities with better access to health care had a RR of 1.52 (1.03, 2.27) compared to municipalities with worse access to health. Conclusions The results show an increased incidence of cervical cancer in affluent municipalities and in those with more health facilities. This is likely driven by better access to health care in more affluent areas. More efforts should be made to ensure equitable access to health services, including mitigating physical barriers, such as transportation to health centres and strengthening of screening programmes.