Effect of exenatide, pen insulin, and vial insulin on patient outcomes: a retrospective database analysis of persistence and first-year costs in a commercially insured population.

OBJECTIVE: We compared health care costs and medication persistence for patients with type 2 diabetes initiating treatment using exenatide, pen insulin, or vial insulin. METHODS: Commercial health plan data (2004-2008) were used to identify episodes of antidiabetic drug therapy, which were then classified according to treatment history: first observed treatment, restarting a previous therapy (90-day gap in all treatment), switching therapy, and augmentation therapy. Three time periods were defined for each episode: the month in which the episode was initiated (index month), 6 months before the index month (preindex period), and 12 months after the index month (postindex period). All exenatide and insulin episodes were selected for this analysis of persistence and first-year costs. Multivariate statistical methods were adjusted for demographic characteristics, drug use history, previous medical care use, comorbid medical conditions, and prescription drug profile. Several sensitivity analyses were conducted. RESULTS: A total of 213,701 episodes of antidiabetic drug therapy were identified, of which 7031 patients were initiated using exenatide, 21,011 used vial insulin, and 422 used pen insulin. Time to all-cause discontinuation (TTAD) was measured for the index drug and all diabetic-related drugs. Pen insulin was discontinued 91 days earlier than exenatide, whereas vial insulin was continued 18 days longer than exenatide. Patients using pen insulin discontinued all antidiabetic drugs 34 days earlier than patients on exenatide, whereas patients using exenatide and vial insulin exhibited similar TTAD for all drugs. Exenatide use was estimated to significantly reduce medical costs of the first posttreatment year sufficient to offset higher prescription drug costs. These results were confirmed using propensity score matching estimation and were robust across episode type. CONCLUSIONS: Patients initiating drug therapy using exenatide might incur lower posttreatment costs than similar patients who initiated treatment using insulin.