Determining Diagnosis Date of Diabetes Using Structured Electronic Health Record (EHR) Data: The SEARCH for Diabetes in Youth Study (Preprint)

This is the authors' abstract. We don't add key points for this paper.

BACKGROUND Manual chart review remains the primary method to identify diabetes status, diabetes type, and date of onset. Previous work determined that increasingly available structured electronic health record (EHR) data can accurately determine diabetes status and type. Validated algorithms to determine date of diabetes diagnosis are lacking. OBJECTIVE To validate two EHR-based algorithms to determine date of diabetes diagnosis to make incidence surveillance more efficient and sustainable. METHODS A rule-based ICD-10 algorithm identified youth with diabetes among three children’s hospitals in Ohio, Washington, and Colorado participating in the SEARCH for Diabetes in Youth Study. Two date of diagnosis algorithms were compared to the chart-reviewed gold standard among cases detected in the EHR from 2009 through 2017: year of occurrence of second ICD-9 or ICD-10 diabetes code (ICD code), and year of first occurrence of any of the following criteria: ICD diabetes code, elevated glucose, elevated HbA1c, or diabetes medication (multiple-criteria). RESULTS Among 3777 cases, the algorithms demonstrated high agreement with true diagnosis year and differed in classification (p=.006): 86.5% agreement for ICD code algorithm and 85.9% agreement for multiple-criteria algorithm. Agreement was high for both type 1 and type 2 cases for the ICD code algorithm. Performance improved over time. CONCLUSIONS Year of occurrence of second ICD diabetes code yields an accurate diagnosis date within these pediatric hospital systems leading to increased efficiency of surveillance methods. CLINICALTRIAL not applicable