Online Program

Defining type-II diabetes with electronic medical record data

Tuesday, November 5, 2013

Marie Lynn Miranda, PhD, Children's Environmental Health Initiative, University of Michigan, Ann Arbor, MI
Ben Strauss, MS, Children's Environmental Health Initiative, Durham, NC
Pamela Maxson, PhD, Duke Center for Community and Population Health Improvement, Duke University, Durham, NC
Background Through a grant from the Bristol-Myers Squibb Foundation, the Durham Diabetes Coalition is working to reduce the burden of diabetes and understand disparities in Durham County, North Carolina. As a first step, it was necessary to identify patients in the Duke patient warehouse who have type-II diabetes. ICD-9 diagnosis codes alone are often insufficient for identifying all patients with type-II diabetes, thus multiple criteria based on electronic health record (EHR) data have been identified.

Objective We explore four methods to define diabetes: those based on SUPREME, EMERGE, AHRQ bundle, and our own derived definition. We seek to determine what the primary differences are among the definitions, what patients are being captured or excluded by each, and which definition allows us to use EHRs in intervention programs targeted at type-II diabetes.

Methods Using five years of EHR data for the entire Duke University Health System patient population, we identify patients with type-II diabetes based on each of the four definitions. Agreement in diagnosis assignment across definitions is explored using Cohen's kappa. The demographic distribution of the patient populations identified as having type-II diabetes under each of the four definitions is also compared.

Results Each disease definition identified diabetes patients missed using ICD-9 codes alone and exhibited varying degrees of overlap with the other definitions. Patient populations with diabetes as identified by each definition varied in terms of demographics and health. We describe the four definitions and compare the yield of patients for each, characterizing the nature of patient diagnosis for those definitions that differentiate patients.

Discussion Our work provides a phenotype comparison that can inform a standard definition of type-II diabetes. Our methods and approach leverage EHR clinical and laboratory data to improve disease identification and understanding, and can be used successfully with many disease endpoints.

Learning Areas:

Chronic disease management and prevention

Learning Objectives:
Assess the utility of using electronic medical record data for identifying patient populations of interest Compare four approaches to diagnosing type-II diabetes using medical records Discuss the importance of a standard definition for type-II diabetes

Keyword(s): Diabetes, Information Systems

Presenting author's disclosure statement:

Qualified on the content I am responsible for because: I am the lead GIS analyst on this project and have worked extensively with clinicians to define diabetes in various ways and implement the definitions in our clinical datasets.
Any relevant financial relationships? No

I agree to comply with the American Public Health Association Conflict of Interest and Commercial Support Guidelines, and to disclose to the participants any off-label or experimental uses of a commercial product or service discussed in my presentation.