Early potential safety signals for gliptins and gliflozins using real-world pharmacy data compared to spontaneous reporting

Abstract

Background Dipeptidyl peptidase 4 inhibitors (DPP4i, gliptins) and sodium-glucose co-transporter 2 inhibitors (SGLT2i, gliflozins) are two oral antidiabetic drug classes that have been widely used in recent decades. Their pharmacological actions have led to the identification of new adverse drug reactions and contraindications. In this context, ambulatory real-world data (RWD) analysis has the potential to detect early signals of suspected adverse drug reactions (ADR) that complement or anticipate risk events. Methods We conducted an exploratory pharmacovigilance study based on a multicentre observational cross-sectional survey with retrospective 6-month recall in community pharmacies among patients treated with DPP4i or SGLT2i, using metformin as a control. Candidate drug–event signals identified in the pharmacy dataset collected in 2021 were compared with spontaneous reporting data from the Spanish Agency of Medicines and Medical Devices repository in 2022 and 2024. A validated adaptation of the Bayesian Confidence Propagation Neural Network (BCPNN) methodology was applied in all analyses. Results Exploratory signals identified in the community pharmacy dataset included sitagliptin–dry mouth (FDR 0.059), empagliflozin–asthenia (0.067), linagliptin–bone fracture (0.077), linagliptin–renal impairment (0.082), dapagliflozin–hyperglycemia (0.088), canagliflozin–pruritus (0.092), dapagliflozin–urinary tract infection (0.096), and alogliptin–exanthematic eruptions (0.100). Some of these findings were already described in the Summaries of Product Characteristics (SmPC), whereas others were not and later showed greater convergence with spontaneous reporting patterns in 2024. These findings should be interpreted as exploratory statistical signals for follow-up, particularly when based on small numbers of reports. Conclusions Community pharmacy real-world data may contribute to the early identification of potential safety signals that complement spontaneous reporting systems. Signals detected near the exploratory threshold should be interpreted cautiously and considered candidates for monitoring and further investigation rather than confirmed adverse drug reactions.