Spatial Heterogeneity of Viral Suppression and Viral Rebound Patterns among ART Patients in Zimbabwe from 2004 to 2017: A Bayesian Mixed Effects Multistate Model - Pages 113-98
Zvifadzo Matsena Zingoni, Tobias F. Chirwa, Jim Todd and Eustasius Musenge
Published: 20 December 2019
Abstract: Augmenting the global efforts towards HIV control and prevention, spatial modelling helps identify areas with poor viral suppression to inform programme planning. This study aims to describe the spatial viral suppression and viral rebound trajectories among ART patients. This is the first application of the fully Bayesian geoadditive semiparametric multistate Markov models to account for unobserved geographical heterogeneity. Time-varying log-baseline effects of the transition intensities and non-linear effects of continuous covariates were estimated as smoothed functions of time using penalised splines. Non-parametric effects of fixed covariates and frailty effects to account for individual variability were also considered. Viral load was the preferred marker for better prediction of HIV/AIDS disease progression; therefore, a three staged model was proposed bases on two viral load transient states defined by undetectable viral cut-off limits and death as the third absorbing state. Model application was based on the routinely collected individual-level data of ART patients from the Zimbabwe national ART programme. Amongst 18,150 participants, both the log-baseline transition rates of attaining undetectable viral suppression and attaining a viral rebound increased with increase in ART duration. Viral rebound transition was significantly prevalent among patients living on the long-distance truck route region (Matabeleland North province) which borders with Botswana and Zambia. Interventions which address health literacy and misconceptions over ART benefits and the gravity of attaining and sustaining viral suppression are a priority in the fight of HIV to increase patients’ life expectancy and lower HIV transmission.
Keywords: Bayesian estimation, multistate Markov models, spatial heterogeneity, viral suppression, viral rebound.