Data from longitudinal studies have suggested that patients infected with HIV-1 might have cardiovascular abnormalities. These abnormalities may be the result of the treatments (AZT, etc.) or the HIV-1 virus itself. In this grant, we will develop methods to analyze longitudinal and survival studies of heart function in AIDS patients. Cardiotoxicity in AIDS patients could be a leading cause of morbidity and mortality. Development of methods leading to increased understanding of the cardiotoxicity could prove invaluable in determining the appropriate time to give cardioprotective therapies. The developed methods will be applied to the Pediatric Pulmonary and Cardiac Complications (P2C2) of Vertically Transmitted HIV Infection Study, which was a large, prospective longitudinal study designed to monitor heart disease and the progression of cardiac abnormalities in children born to HIV-infected women. To truly understand the change in heart function over time, new models and methods need to be developed that take into account multiple measures of heart function (LV mass, end-diastolic dimension, end-systolic dimension, fractional shortening, etc.). The models and associated statistical methods should be such that biases do not result because of higher rates of clinic visits by sicker patients, higher chance of dropout for sicker patients, missed clinic visits due to patients' health conditions, etc. In this grant, we develop appropriate models and methods to deal with these situations. We will examine the following six Specific Aims. Regression models for Longitudinal Discrete and Continuous Outcomes; A likelihood method for Longitudinal Studies with Informative Dropout; A Protective Estimator for Mixed Discrete and Continuous Outcomes Subject to Nonignorable Non-monotone Missingness; Methods for Estimation in Longitudinal Studies with Outcome-Dependent Follow-Up; and Prediction in Multivariate Survival Models; Sensitivity Analysis for the previous six aims.
Showing the most recent 10 out of 16 publications
