Comparative Analysis of SGLT2i and DPP4i Drugs on AKI Risk

SGLT2i (dapagliflozin, empagliflozin, and canagliflozin) and DPP4i (alogliptin, linagliptin, sitagliptin, saxagliptin, and vildagliptin) were analyzed for drug type, quantity, dose, dispensing date, and days of drug supply. The primary outcomes were the incidences of AKI and AKI-D in the propensity score–matched cohort. AKI diagnosis was based on ICD-9-CM and ICD-10-CM diagnostic codes (eTable 1 in Supplement 1), while AKI-D diagnosis also required dialysis treatment during the same hospitalization. The dialysis treatment procedure codes are shown in eTable 1 in Supplement 1. The codes used to identify AKI were validated in our database, with a positive predictive value of 98.5% and a negative predictive value of 74.0%.^{14 (link)} The accuracy of acute dialysis procedure coding has also been validated, with a positive predictive value of 98%,^{15 (link)} as accurate procedure codes are necessary for reimbursement in Taiwan.
Different diseases interact with AKI and possibly aggravate it. Likewise, AKI can induce injury in these distant organs. These are grouped as AKI with heart disease, sepsis, respiratory failure, and shock. These 4 diseases were chosen because they are the most common contributors to AKI.^{16 (link)} These diseases were diagnosed based on ICD-9-CM and ICD-10-CM diagnostic codes (eTable 1 in Supplement 1) during the AKI hospitalization. AKI prognosis was also analyzed. We considered advanced CKD (defined as CKD stages 4 and 5 by ICD-10-CM diagnostic codes), ESKD (confirmed by the registry of catastrophic illness), or death that occurred within 90 days of AKI hospitalization.