Introduction & Objectives: Efficient automated EMR review methods to identify people with atypical diabetes (DM) for research are lacking. We aimed to develop a novel Python-based Expeditious Program for EMR Review (PEPPER) and assess its efficiency.
Methods: We extracted the list of 1660 youth (<19 yo) with type 2 DM (T2D) seen between 2019-2022 from EMR to identify candidates with A-β+ (autoantibody negative, preserved β-cell function) Ketosis-prone DM (KPD) for enrollment into Rare and Atypical Diabetes Network (RADIANT). We developed PEPPER to identify diabetic ketoacidosis (DKA) occurrence within 6 months (mo) of diagnosis to prioritize individuals for detailed manual chart review for RADIANT eligibility. We also manually reviewed EMR of 100 youth with T2D to identify DKA occurrence within 6 mo of diagnosis without PEPPER for comparison.
Results: PEPPER identified 110 youth with T2D who had DKA within 6 mo of diagnosis. Twenty-one met the RADIANT A-β+ KPD criteria. The time spent to identify those with T2D and DKA was significantly shorter with PEPPER compared to manual review (13.4 ± 3.9 vs. 26.6 ± 9.4 seconds, p<0.001), translating to 6.2 vs. 12.3 hours to review 1660 charts with and without PEPPER. Both methods yielded identical results, confirming PEPPER’s accuracy.
Conclusion: We developed a novel, efficient and reliable EMR review method that could be used on large cohorts to identify research candidates.
M. Ahmed: None. E.A. Kubota-Mishra: None. A.F. Siller: None. A.E. Davis: None. I. Migacz: None. S. Sisley: Speaker's Bureau; Rhythm Pharmaceuticals, Inc. J. Faruqi: None. Z.I. Saeed: None. S. Ahmed: None. L.H. Philipson: Research Support; Novo Nordisk, Novo Nordisk Foundation, Dompé, Vertex Pharmaceuticals Incorporated, Imcyse. M.J. Redondo: None. A. Balasubramanyam: None. M. Tosur: None.
U54 DK118638 and U54 DK118612 (RADIANT Study Group) and K23-DK129821 (MT)
