Use of non-insulin diabetes medicines after insulin initiation: A retrospective cohort study

Abstract

Background Clinical guidelines recommend that metformin be continued after insulin is initiated among patients with type 2 diabetes, yet little is known regarding how often metformin or other non-insulin diabetes medications are continued in this setting. Methods We conducted a retrospective cohort study to characterize rates and use patterns of six classes of non-insulin diabetes medications: biguanides (metformin), sulfonylureas, thiazolidinediones (TZDs), glucagon-like peptide 1 receptor agonists (GLP1 receptor agonists), dipeptidyl peptidase 4 inhibitors (DPP4 inhibitors), and sodium-glucose co-transporter inhibitors (SGLT2 inhibitors), among patients with type 2 diabetes initiating insulin. We used the 2010-2015 MarketScan Commercial Claims and Encounters data examining 72,971 patients with type 2 diabetes aged 18-65 years old who initiated insulin and had filled a prescription for a non-insulin diabetes medication in the 90 days prior to insulin initiation. Our primary outcome was the proportion of patients refilling the various non-insulin diabetes medications during the first 90 days after insulin initiation. We also used time-to-event analysis to characterize the time to discontinuation of specific medication classes. Results Metformin was the most common non-insulin medication used prior to insulin initiation (N = 53,017, 72.7%), followed by sulfonylureas (N = 25,439, 34.9%) and DPP4 inhibitors (N = 8,540, 11.7%). More than four out of five patients (N = 65,902, 84.7%) refilled prescriptions for any non-insulin diabetes medications within 90 days after insulin initiation. Within that period, metformin remained the most common medication with the highest continuation rate of 84.6%, followed by SGLT2 inhibitors (81.9%) and TZDs (79.3%). Sulfonylureas were the least likely medications to be continued (73.6% continuation) though they remained the second most common medication class used after insulin initiation. The median time to discontinuation varied by therapeutic class from the longest time to discontinuation of 26.4 months among metformin users to the shortest (3.0 months) among SGLT2 inhibitor users. Conclusion While metformin was commonly continued among commercially insured adults starting insulin, rates of continuation of other non-insulin diabetes medications were also high. Further studies are needed to determine the comparative effectiveness and safety of continuing insulin secretagogues and newer diabetes medications after insulin initiation.