Latent class analysis of patients’ background factors affecting the risk of specific adverse drug reactions to dipeptidyl peptidase 4 inhibitors

Background and purpose: Spontaneous reporting is widely used to identify adverse drug reactions (ADRs), but relatively little is known about the relationships between specific ADRs and background factors of affected patients. Here, we applied latent class analysis (LCA) to identify background factors associated with different ADRs in type 2 diabetes patients treated with dipeptidyl peptidase 4 (DPP-4) inhibitors, using the Japanese Adverse Drug Event Report (JADER) database. Materials and methods: Patients using only a DPP-4 inhibitor who encountered ADRs were selected from the JADER database up to April 2019 (N = 3,577). LCA was employed to classify these cases based on underlying diseases and lifestyle factors (alcohol, tobacco, diet, and exercise) and to identify characteristic ADRs in each class. The optimum number of classes was determined by selecting the model with the lowest value of the Bayesian information criterion (BIC). Results: A six-class model had the lowest BIC, and these classes were characterized by specific background factors and ADRs. For example, one class included diabetes complications, while another class included exercise and diet as background factors. Increased risk of a specific ADR(s), such as pancreatitis or pemphigoid, was found in each class. The nine DPP-4 inhibitors were not uniformly distributed among the classes, though individual classes included patients receiving different inhibitors. Conclusion: Our findings indicate that characteristic background factors of patients experiencing specific DPP-4 inhibitor-induced ADRs reported in the JADER database are different and can be classified by LCA. This methodology may be useful for predicting ADRs not detected during drug development.