1Health Outcomes Research Using Time-to-Event Analytic Approaches

Health outcome events may be characterized as morbidity such as disease or injury or may be the result of a particular health care procedure or medical intervention. The occurrence of these events is likely dependent on differing risk sets among those with and without the event. Health outcome researchers strive to identify at-risk populations by providing quantitative evidence allow more informed decisions by practitioners and policy makers. The focus of this tutorial will be on time-to-event analyses pertaining to occurrence of health outcomes. Specifically, this tutorial will discuss multicollinearity, confounding, and interaction investigation steps in multivariable model building. Cox’s Proportional Hazards Regression will be employed to investigate for differences between event probabilities over time for two or more cohorts while adjusting for other influential covariates. Finally, interpretation of the hazard ratios and graphing the probability of outcome over time will be discussed. This tutorial is appropriate for intermediate level statisticians, epidemiologists, and data analysts