OBJECTIVES: Type 2 diabetes mellitus (T2DM) is a major risk factor for cardiovascular disease (CVD). Treatment with sodium-glucose cotransporter-2 inhibitor (SGLT-2i) has been recently associated with lower incidence of CVD including myocardial infarction (MI), compared to other antidiabetic drugs (Kosiborod et al, 2018). We compared the incidence of MI among classes of oral antidiabetic drugs (OADs).

METHODS: We used medical and drug claims data and insurance exposure data in multiple health plans in USA (2011-2016). Patients having ≥1 claim of OAD (dipeptidyl peptidase-4 inhibitor [DPP-4i], glucagon-like peptide-1 receptor antagonist [GLP-1Ra], or SGLT-2i, defined by NDC), having ≥1 diagnosis of T2DM, aged ≥40 in index date (the earliest date of OAD prescription) and having ≥6 months of wash-out period of OAD before index date, were analyzed. A prediction model for incidence of MI (coded 410.xx or I21 by ICD9 or ICD10 code) was developed with deep learning using dummy variables of 200 of frequently observed health care procedures and 200 of frequently observed diagnoses within 6 months before index date, age, and sex as explanatory valuables. Proportional hazard analysis for incidence of MI was conducted using the probability obtained by the prediction model and class of OAD as explanatory valuables.

RESULTS: Among 1,4309,499 T2DM patients aged ≥40, 304,085 patients were extracted for the analysis. Number of patients (mean ± standard deviation [STD] of age, percentage of female) prescribed with DPP-4i, GLP-1Ra, or SGLT-2i in index date was 149,411 (58.1 ± 9.6, 45.9%), 79,664 (55.0 ± 7.9, 53.7%), or 75,010 (55.5 ± 7.8, 44.3%), respectively. Hazard ratio for MI compared to prescription with DPP-4i was 0.819 (p < 0.0001) with GLP-1Ra, and 0.863 (p = 0.0006) with SGLT-2i.

CONCLUSIONS: Lower risk of MI was suggested for patients prescribed with GLP-1Ra or SGLT-2i compared to those prescribed with DPP-4i.