Janinne Ortega-Montiel, Wajd Alkabbani, Georg Hahn, Jun Liu, Caroline F Byrne, Deborah J Wexler, Elisabetta Patorno, Julie M Paik
In two large real-world cohorts of patients with T2D and CKD, the tree-based scan statistic identified previously known adverse events, but no new safety signals associated with SGLT2 inhibitor use compared with GLP-1RA use.
RATIONALE & OBJECTIVE: Patients with chronic kidney disease (CKD) are often underrepresented in clinical trials evaluating the safety profile of newer medication classes for type 2 diabetes (T2D). Given the high burden of comorbidities and polypharmacy in this population, further safety research is needed. This study aimed to detect potential adverse events associated with sodium/glucose cotransporter 2 (SGLT2) inhibitors in patients with T2D and CKD, using a tree-based scan statistic approach.
STUDY DESIGN: New-user active comparator cohort study.
SETTING & PARTICIPANTS: Medicare Fee-for-service data (October 2016 - December 2020) and the Merative MarketScan commercial claims database (October 2016 - December 2022) used to identify patients aged ≥65 years (Medicare) or ≥18 years (MarketScan) with T2D and CKD (Stages III-IV).
EXPOSURES: New use of SGLT2 inhibitors or glucagon-like peptide-1 receptor agonists (GLP-1RA).
OUTCOMES: ICD-10 coded adverse outcome events, in outpatient, inpatient, or emergency department settings.
ANALYTIC APPROACH: After 1:1 propensity-score matching, a hierarchical data-mining surveillance method using ICD-10 outcomes, a tree-based scan statistic, to identify signals for incident adverse events.
RESULTS: We identified 22,328 matched pairs in Medicare (mean age, 74 years; 45% female) and 8,918 matched pairs in MarketScan (mean age, 65 years; 42% female). In both cohorts, based on outpatient diagnosis codes, SGLT2 inhibitor initiation showed safety signals for genital infections in both females and males (p < 0.001). However, the tree-based scan statistic did not detect any safety signals based on inpatient/ED diagnosis codes.
LIMITATIONS: Residual confounding, potential misclassification, and limited power.
CONCLUSIONS: In two large real-world cohorts of patients with T2D and CKD, the tree-based scan statistic identified previously known adverse events, but no new safety signals associated with SGLT2 inhibitor use compared with GLP-1RA use.