The Development and Validation of a Synthetic Cardiovascular Cohort

• Activity Number: 608 - Novel Methods for Longitudinal Analysis in Large Cohort Studies
• Type: Topic Contributed
• Date/Time: Thursday, August 3, 2017 : 8:30 AM to 10:20 AM
• Sponsor: Biometrics Section
• Author(s): Norrina Allen and Hongyan Ning and Juned Siddique and Donald Lloyd-Jones and Amy Krefman*
• Companies: Northwestern University and Northwestern University and Northwestern University and Northwestern University and Northwestern University
• Keywords: synthetic cohort ; imputation

Abstract

In this study we developed and validated a synthetic cohort approach to examine CV risk factors from age 18-90 years of age. We used the Lifetime Risk Pooling Project including participants of CARDIA, MESA, Framingham, ARIC, JHS and CHS. Individuals' demographic data, traditional CV risk profile and outcomes at each exam were included. To generate a complete set of values from age 18 to 90 for each participant, we multiply imputed the participant's CV risk profile and outcomes based on a joint multi-level imputation model via the jomo package in R. To validate our imputed values, we removed the observed CARDIA data ages 18-30, MESA data ages 50-59 and CHS data ages 80-90. We then imputed the CV risk profile of these deleted values and compared imputed and observed values. We included 41,387 participants (avg follow-up time 20 yrs). Among our validation sample, imputed CV risk factor levels were consistent with observed values at both younger and older ages or BMI, SBP, DBP, glucose and total cholesterol as were the rates of medication usage, smoking and events. This synthetic cohort approach provides valid and unbiased estimates of CV risk factors across the lifespan.