4E-05 - A data mining application tool to detect safety signals of sodium glucose cotransporter 2 inhibitors in patients with type 2 diabetes and chronic kidney disease.

Background: Patients with chronic kidney disease (CKD) have been underrepresented in clinical trials evaluating the safety profile of the newer medication classes used for the treatment of type 2 diabetes (T2D), one of the leading causes of CKD. There is a lack of information with respect to whether a tree-based scan methodology may be useful for signal detection in populations with such a high comorbidity burden and polypharmacy, and thus who are at higher risk for adverse safety events.

Objectives: To screen for potential adverse events among initiators of sodium-glucose cotransporter 2 inhibitors (SGLT2i) vs glucagon-like peptide 1 receptor agonists (GLP-1RA) using a TreeScan software.

Methods: Leveraging two commercial claims databases (IBM MarketScan [October 2016 – December 2022] and CMS Medicare [October 2016 – December 2020]) we identified new users of SGLT2i or GLP-1RA who were ≥18 (Marketscan) or ≥65 (Medicare) years old, with T2D and CKD (Stages 3-4), and who had at least 12 months of continuous health plan enrollment. From this population, we created 1:1 propensity-score (PS) matched cohorts and applied a hierarchical tree-based scan statistic data mining method, with the hierarchical outcome tree constructed based on International Classification of Disease (ICD-10) codes, to identify signals for any potential incident adverse events associated with SGLT2i, compared with GLP-1RA.

Results: We identified two pairwise PS-matched cohorts of 8,918 SGLT2i vs 8,918 GLP-1RA initiators (Marketscan) and 22,328 SGLT2i vs 22,328 GLP-1RA initiators (Medicare). The Medicare population was older than the Marketscan population (mean age 74.0 vs 65.1 years), and there was a higher prevalence of congestive heart failure in the Medicare vs. Marketscan population (25.9% vs 16.6%). We observed safety signals for female and male genital infections in both cohorts (p < 0.001) for SGLT2i initiators when we screened outpatient diagnoses. However, no safety signals were found with SGLT2i when we screened inpatient or emergency department visit diagnoses.

Conclusions: In two large population-based cohorts of patients with T2D and kidney disease, we identified previously known adverse events associated with SGLT2i use, but no other major or new safety signals. These findings suggest that tree-based scan methodology can be used as an additional tool for signal detection in safety surveillance in complex populations with high comorbidity burden and polypharmacy, who may be at higher risk for adverse events.