Bryan Lau: Dealing With Missing variables in Subset of Cohorts Within ECHO and Issues of Heterogeneity and Representativeness

ECHO Discovery Summary

Bryan Lau, PhD, MHS, ScM of Johns Hopkins University discussed issues of missing information within subsets of ECHO cohorts.

Throughout the presentation, Dr. Lau explained how cohort consortia around specific diseases are more likely to collect similar information, and thus have fewer missing data, than those based on more varied populations. He noted that since ECHO is composed of cohort studies from various populations of pregnant women and children, there are differences in types of data collected among cohorts, especially data the cohorts collected before ECHO started. Problems can arise when ECHO attempts to combine these different data because some cohorts are missing data that others have.

To resolve this challenge, Dr. Lau suggested that statistical methods to fill in missing data need to take into account the extent to which cohorts are actually drawing from the same underlying population. Additionally, he noted that to avoid bias when filling in missing data, researchers can use statistical algorithms to cluster together cohorts with similar populations and variables.

Speaker:

Bryan Lau, PhD, MHS, ScM

Johns Hopkins School of Public Health

Speaker Bio:

Bryan Lau, is a Professor in the Department of Epidemiology of the Johns Hopkins Bloomberg School of Public Health with joint appointments in the Department of Medicine and Department of Oncology in the Johns Hopkins School of Medicine. Dr. Lau earned his ScM in epidemiology in 1999 and his MHS in biostatistics in 1999 and subsequently earned his PhD in epidemiology in 2004.

Date: Wednesday, December 8th, 1 to 2pm ET