Hospitalists using EHR-based definitions to identify the characteristics of diabetes in patients should take a closer look at those standards, according to the results of a recent Journal of the American Medical Informatics Association study.
The authors compared eight distinct EHR-based phenotypes, or observable markers, of diabetes, with the gold standard diabetes diagnosis chart created by the American Diabetes Association. They found the accuracy of using EHR-based standards was significantly more variable than the ADA standards.
"Analysis of electronic health data presents many opportunities for generating hypotheses, validating exploratory and predictive models, and testing new types of statistical approaches; however, these opportunities can be adversely affected by the challenges of unstandardized cohort definitions and the complexity of defining diseases," the authors wrote.
According to the researchers, the study highlights a need to distinguish diabetes phenotype definitions and standardize them to avoid generalizing diabetes-type symptoms to other conditions patients may have.