The primary goal of public health efforts which aim to control HIV epidemics in the United States and around the world is to diagnose and treat people with HIV infection as soon as possible after seroconversion. The timing of the first antiretroviral therapy (ART) treatment after HIV diagnosis is therefore an important population-level indicator that can be used to measure the effectiveness of HIV care programs and policies at local and national levels. However, there are no population-based estimates of the timeliness of ART initiation in the US because data on the timing of ART initiation cannot feasibly and efficiently be collected as part of routine jurisdictional HIV surveillance activities. In this project, we propose to develop a statistical model for the estimation of the timing of ART initiation following HIV diagnosis. We will use routinely collected, population-based data on laboratory tests from all persons diagnosed with HIV infection, including biomarkers such as viral load (VL) and CD4 count, from the New York City (NYC) Department of Health and Mental Hygiene (DOHMH). We will develop a change- point model in which the VL and CD4 counts are jointly modeled, where the ART initiation time is treated as a random change point that induces simultaneous trajectory changes to the biological process of both VL and CD4 counts. Several data complexities challenge the model development. First, variability in CD4 count or VL monitoring by providers leads to imbalanced VL and CD4 count reporting times at the individual level. Second, the instrument detection limit leads to data censoring for VL. Third, the study population is heterogeneous, such that some individuals start ART right after the diagnosis (test and treat), while others delay treatment (delay treatment) or remain untreated (no treatment). We will develop statistical methods to address these data complications. The methodology will be built upon likelihood and approximated-likelihood inferences coupled with missing data handling and mixture population modeling. Leveraging ART prescription information from Medicaid claims data linked at the individual level to the HIV surveillance data, we will cross-validate our model- based estimations and refine the modeling. Finally, we will disseminate a free R package to share the methods and the toolset with researchers. Following the completion of this project, we hope to substantially improve the ability to obtain reliable population-based estimates on the timing of ART initiation for people living with HIV. We expect the methods and the toolset will be used to inform policies and programs in NYC and other jurisdictions and will help these jurisdictions improve access to HIV prevention and treatment services and facilitate initiatives to better control HIV epidemics.
Using population-based HIV surveillance data, this project aims to develop a change-point model on the estimation of an important health indicator: the timing of antiretroviral therapy initiation after HIV diagnosis. The model will be validated with antiretroviral therapy prescription information, and the methods will be disseminated through a free software package for use by health departments, public health practitioners, and policymakers. Our method and software will be strategically important in influencing policies and programs to expand access to HIV prevention and treatment services and to monitor and inform the ending of HIV epidemics at the local and national level.