Adjusting for Immortal Time Bias in Studies Assessing the Association Between Polygenic Risk and Residual Lifetime Risk

• Activity Number: 417 - Recent advancement on life time data analysis
• Type: Contributed
• Date/Time: Thursday, August 12, 2021 : 2:00 PM to 3:50 PM
• Sponsor: Lifetime Data Science Section
• Author(s): Qiuxi Huang* and Sarah Conner and Favel Mondesir and Ludovic Trinquart
• Companies: Boston University School of Public Health and Boston University School of Public Health and Boston University School of Public Health and Department of Biostatistics, Boston University School of Public Health
• Keywords: Immortal time bias; Polygenic risk score; Lifetime risk; Inverse-probability-of-censoring weighting

Abstract

In studies assessing the association between polygenic risk scores (PRS) and incident disease risk, the ascertainment of PRS can sometimes occur long after participants became initially at risk. If only participants who are alive and free of the disease at PRS ascertainment are selected for the analysis, immortal time bias may be introduced. We considered 4 different methods to estimate the association between PRS and the residual lifetime risk, according to different definitions of the risk sets and the use of inverse-probability-of-censoring weighting (IPCW). In simulation studies, we assessed the mean bias and mean squared error of these methods and of an oracle method under 16 unique scenarios. The IPCW method had the best performance. We illustrated all methods for the association between a PRS for atrial fibrillation (AF), derived using 986 AF-associated single-nucleotide polymorphisms, and the residual lifetime risk of AF after 55 years in the Framingham Heart Study. Studies assessing the association between genetic predisposition and residual lifetime risk must be analyzed properly to avoid immortal time bias.