(305) High Misclassification of Glycogen Storage Disease Type 1a in a US Electronic Medical Records Database – Insights from a Physician Note Review

Background: Glycogen storage disease type 1 (GSD1) is a rare genetic disease characterized by impaired glycogen metabolism consisting of two major subtypes, GSD1a and GSD1b. As diagnosis codes do not differentiate subtypes, this study was initially designed to identify and validate a claims-based algorithm to identify subtypes; however, upon review of physician notes, we noticed that GSD1a, caused by glucose-6-phosphatase deficiency (G6P deficiency), and glucose-6-phosphate dehydrogenase (G6PD) deficiency, a hematologic disorder, were commonly misclassified.

Objectives: To assess the degree of conflation between GSD1a and G6PD deficiency in a large US electronic medical records (EMR) database.

Methods: US patients having ≥1 diagnosis code for GSD1 in Loopback Analytics EMR data (10/2015-12/2022) were eligible. Patients with a diagnosis code for inflammatory bowel disease or neutropenia were excluded. Physician notes were reviewed for 30 randomly selected patients to classify patients into the following categories: chart-confirmed GSD1, chart-confirmed G6PD deficiency with and without clinical correlates (e.g., hemolytic anemia, jaundice, reported allergies to sulfa/fava beans), and undetermined (condition not mentioned or insufficient information in notes to confirm diagnosis). Classification results from the note review were compared to diagnosis codes (GSD1: ICD-10-CM E74.01, G6PD deficiency: ICD-10-CM D55.0, D75.A). Potential misclassification was further evaluated in the IQVIA PharMetrics® Plus US insurance claims database (10/2015-6/2023).

Results: Of all 30 patients reviewed with ≥1 GSD1 diagnosis code, over half (n=16, 53.3%) also had ≥1 diagnosis code for G6PD deficiency. Only 5/30 (16.7%) were confirmed as having GSD1 through notes review and 0/5 of these patients had a G6PD deficiency diagnosis code. Among all 30 patients, 18 (60.0%) were confirmed as having G6PD deficiency in notes despite having ≥1 GSD1 diagnosis code, including 16 (53.3%) with clinical correlates and 2 (6.7%) without; 15/18 had ≥1 G6PD deficiency diagnosis code. Diagnosis was undetermined for 7 (23.3%) patients. Potential misclassification was also observed in the claims database, where 402/1,104 (36.4%) of patients with a GSD1 diagnosis code also had a G6PD deficiency diagnosis code.

Conclusions: A large conflation due to misclassification between G6PD deficiency and GSD1a in a large EMR database was found. Using diagnosis codes to identify patients with GSD1a will result in significant misclassification and physician note review may be necessary to accurately identify patients with GSD1a. Although a query in a large US claims database potentially confirmed this misclassification, further validation is needed.