Leveraging Electronic Health Records for Phenotyping - Translational Informatics: Realizing the Promise of Knowledge-Driven Healthcare - page 64

concrete resources are in addition to the methods and lessons learned that were
published in the scientifi c literature.
Other groups have also used EHRs for genotype-phenotype studies. The
Pharmacogenomics Research Network (PGRN) includes sites that use EHR data for
phenotypes, though the use of EHRs is not defi ning of the goals as is eMERGE. The
Kaiser Permanente Research Program on Genes, Environment and Health used
EHRs for phenotypic information linked to genotypes [ 26 ]. While not having the
breadth of institutional participation or depth of use of EHR data, these projects
demonstrate the interest in leveraging EHR data for genotype-phenotype research
extends beyond the initial demonstration projects.
The Integrating Informatics and Biology at the Bedside (i2b2) is particularly
important to leveraging EHR data for phenotypes because it provides a platform
for storing, linking and querying data from EHRs directly by researchers. It also
allows cross-institution queries for connected research institutions. Data extracted
from EHRs are loaded into the i2b2 data model. Individuals can then create queries
based on the organization of the clinical data. It allows linking of genotype infor-
mation to support GWAS. It also includes a natural language processing engine to
extract fi ndings from text reports. Perhaps the greatest indication of importance for
i2b2 is its broad use - it has been adopted by over 60 academic health centers
internationally [ 39 ].
The Measurement to Understand the Reclassifi cation of Disease of Cabarrus/
Kannapolis (MURDOCK) Study in North Carolina represents a slightly different
approach to creating an infrastructure for genotype-phenotype studies. It focuses on
creating a comprehensive database of phenotype information along with biospeci-
mens, and is non-disease specifi c. Rather than recruiting biospecimens at one time
with consent to link to EHR data, MURDOCK enrolls the whole individual. Subjects
consent to provide biospecimens, complete annual surveys, link data to their EHRs,
and be contacted for participation in other studies. In addition, environmental and
geospacial data are collected so that non-clinical factors can be measured for their
impact on health and treatment effectiveness. The MURDOCK study is a good
example demonstrating how the extraction of phenotype data from EHRs can be
extended beyond the genotype, and to fully patient-centered and personalized
research.

4.3.5 Projects Leveraging EHRs for Comparative Effectiveness

Some projects have focused on leveraging EHR data for secondary analysis in com-
parative effectiveness and patient-centered outcomes research, rather than specifi -
cally on GWAS or other genetic studies. In 2010, the Agency for Healthcare
Research and Quality (AHRQ) funded multiple large projects designed to create
data infrastructures using EHR data to perform comparative effectiveness research.
Like the eMERGE network, each of these projects collected data for a specifi c pop-
ulation, focusing on a particular disease. Unlike eMERGE, they each included data