Joint modelling of extremely longitudinal measurements and competing survival outcomes in HIV-infected patients

BackgroundThe CD4 T-lymphocyte (CD4) count is a critical biomarker for HIV disease progression and immune health, that is essential for guiding treatment decisions. However, quantitative assessment of the impact of rapid CD4 decline on AIDS progression and pre-AIDS death remains unclear. MethodsThis study includes 11,647 HIV-positive patients from the Zhejiang Provincial Center for Disease Control and Prevention (CDC) from 2005 to 2017. The longitudinal trajectory of extremely low CD4 counts is captured by a generalized Pareto (GP) model, jointly analyzed with two competing events (AIDS progression and pre-AIDS death) via Weibull spatial survival models. The survival submodels are linked through a shared linear predictor in the GP submodel, providing insights into the effect of extremely low CD4 counts on competing outcomes. The model is implemented using the efficient R-INLA (integrated nested Laplace approximations) approach, with predictive performance assessed using survival Receiver Operating Characteristic (ROC) curves and integrated Area Under the Curve (iAUC) values. ResultsMale individuals, those with homosexual transmission, older age at diagnosis, and extended delays in starting initial antiviral therapy (ART) are more likely to experience sharper reduction in CD4 counts. This decline, along with female gender and late ART initiation, increases the risk of both AIDS progression and pre-AIDS death, with a more pronounced effect on the former. The considerate spatial survival frailty structure with the shared GP longitudinal model for the sharp decline of CD4 counts enhances the predictive accuracy for both outcomes in comparison with the Gaussian longitudinal submodel for CD4 counts (iAUC: 0.9183 vs 0.7688 (AIDS progression) and 0.8901 vs 0.6634 (pre-AIDS death)). ConclusionOur study confirms that demographic factors, route of infection, diagnosis related factors, and extremely low CD4 counts have a significant impact on AIDS progression and pre-AIDS death. These findings underscore the importance of developing effective, evidence-based strategies to mitigate the risk of HIV/AIDS.