Identification of atypical pediatric diabetes mellitus cases using electronic medical records

Authors: ASTUDILLO, MARCELA - Baylor College Of Medicine; WINTER, WILLIAM - University Of Florida; BILLINGS, LIANA - Northshore University Health System; KREIENKAMP, RAYMOND - Harvard Medical School; BALASUBRAMANYAM, ASHOK - Baylor College Of Medicine; REDONDO, MARIA - Baylor College Of Medicine; TOSUR, MUSTAFA - Baylor College Of Medicine; SISLEY, STEPHANIE - Baylor College Of Medicine

Submitted to: BMJ Open Diabetes Research & Care
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 10/4/2024
Publication Date: 11/7/2024
Citation: Astudillo, M.F., Winter, W.E., Billings, L.K., Kreienkamp, R., Balasubramanyam, A., Redondo, M.J., Tosur, M., Sisley, S. 2024. Identification of atypical pediatric diabetes mellitus cases using electronic medical records. BMJ Open Diabetes Research & Care. 12. Article e004471. https://doi.org/10.1136/bmjdrc-2024-004471.
DOI: https://doi.org/10.1136/bmjdrc-2024-004471

Interpretive Summary: Some children have a type of diabetes that doesn’t fit the usual patterns, and it can be hard to find these cases. Researchers in Houston, TX tested two ways to spot unusual types of diabetes by using hospital records. The first way used a checklist to rule out common types of diabetes. Out of 100 kids, 6 were found to have an unusual kind. The second way used computer searches to find kids with three special types of diabetes: diabetes with no clear type, type 2 diabetes in kids younger than 10, and type 1 diabetes without typical antibodies. Using this method, small groups of children with each of these unusual types were found. These children came from many different racial and ethnic backgrounds. The study showed that it’s possible to find kids with unusual diabetes by using hospital records in smart ways. These results could help doctors better understand and treat these rare cases in the future.

Technical Abstract: There are no established methods to identify children with atypical diabetes for further study. We aimed to develop strategies to systematically ascertain cases of atypical pediatric diabetes using electronic medical records (EMR). We tested two strategies in a large pediatric hospital in the USA. Strategy 1: we designed a questionnaire to rule out typical diabetes and applied it to the EMR of 100 youth with diabetes. Strategy 2: we built three electronic queries to generate reports of three atypical pediatric diabetes phenotypes: unknown type, type 2 diabetes (T2D) diagnosed <10 years old and autoantibody-negative type 1 diabetes (AbNegT1D). Strategy 1 identified six cases (6%) of atypical diabetes (mean diagnosis age=11+or- 2.6 years, 16.6% men, 33% non-Hispanic white (NHW) and 66.6% Hispanic). Strategy 2: unknown diabetes type: n=68 (1%) out of 6676 patients with diabetes; mean diagnosis age=12.6 +or- 3.3 years, 32.8% men, 23.8% NHW, 47.6% Hispanic, 25.4% African American (AA), 3.2% other. T2D <10 years old: n=64 (6.6%) out of 1142 patients with T2D; mean diagnosis age=8.6 +or- 1.6 years, 20.3% men, 4.7% NHW, 65.6% Hispanic, 28.1% AA, 1.6% other. AbNegT1D: n=38 (5.6%) out of 680 patients with new onset T1D; mean diagnosis age=11.3 +or- 3.8 years; 57.9% men, 50% NHW, 19.4% Hispanic, 22.3% AA, 8.3% other. In sum, we identified 1%-6.6% of atypical diabetes cases in a pediatric diabetes population with high racial and ethnic diversity using systematic review of the EMR. Better identification of these cases using unbiased approaches may advance precision diabetes.