OBJECTIVES : To examine treatment persistence and medication switching patterns across different therapeutic classes (metformin, sulfonylurea, meglitinide, thiazolidinedione, Glucagon-like peptide-1 receptor agonists (GLP1 RAs), and dipeptidyl peptidase-4 (DPP4) inhibitors) in patients with type 2 diabetes mellitus (T2DM).

METHODS : This was a retrospective study using a large national claims database (2007-2014). Persistence was defined as the time between initiation and discontinuation of the initial therapy. Each patient was followed-up from the day after the index date until either discontinuation of their initial therapy (during the follow-up period) or the end of follow-up (i.e., six months from the index date). Discontinuation was defined as either a drug gap of 45 days or more following the end of the last prescription supply or a switch to an alternative antidiabetic drug during the follow-up period. We used Kaplan-Meier (KM) curves to assess the time to discontinuation, and Cox proportional hazard models to determine the likelihood of non-persistence.

RESULTS : There were 301,853 patients with T2DM initiating antidiabetic agents between 2007 and 2014. The KM curves showed that the differences in persistence were significant across different therapeutic classes. Metformin initiators were most persistent to their first treatment (mean persistence time = 89.32±26.65 days) while GLP1 RA initiators were least persistent (mean persistence time = 84.28±23.68 days). Results from Cox regression analyses showed that the likelihood of non-persistence was significantly lower for metformin (HR=0.66, 95% CI: 0.62-0.71). However, the likelihood of non-persistence was significantly higher for sulfonylurea (HR=1.07, 95% CI: 1.00-1.15), meglitinide (HR=1.27, 95% CI: 1.00-1.60), thiazolidinedione (HR=1.47, 95% CI: 1.33-1.62), GLP1 RAs (HR=1.28, 95% CI: 1.10-1.50), and DPP4 inhibitors (HR=1.49, 95% CI: 1.31-1.60).

CONCLUSIONS : Persistence was greater with metformin compared with other antidiabetic agents. The risk of discontinuation was also lower with metformin. Future studies are warranted to determine factors associated with the differences in persistence and the risk of discontinuation.