Electronic health records can be helpful tools for identifying phenotypes for genetic and other research, according to a new study published in the Journal of the American Medical Informatics Association.
While EHRs contain many phenotypes, such as specific traits or the presence of disease, the primary use for such tools is patient care, not genetics research. Because of that, such information is recorded inconsistently in different formats.
The researchers, from Group Health Research Institute in Seattle and elsewhere, sought to determine how accurately EHRs can identify large numbers of phenotypes for potential genetics research. They analyzed 51 different algorithms of 13 distinct phenotypes, using information from the five members of the Electronic Medical Records and Genomics (eMERGE) Network created by the National Human Genome Research Institute to use EHRs in genomics studies.
They found that that they were able to validate the EHR-derived algorithms across all five sites, with almost three-fourths of them yielding values of 90 percent or greater. Some algorithms, such as for type 2 diabetes, were more prone to error than others, and needed further review. The researchers also made some suggestions on how to improve the accuracy of the information.
"EMRs cannot capture all nuances of patient-provider interactions, but they are extremely useful resources for well designed, informative clinical studies," the researchers said. "Accurate EMR capture of diagnosis, laboratory, and medication data, supplemented with text-mining tools and [natural language processing], can provide excellent phenotype data for genomic studies."
Studies have shown that EHRs can not only improve direct patient care, but also can be helpful for secondary uses such as research and quality management.
To learn more:
- read the abstract