This two-phased study consists of a decision tree model built from real-world claims data to determine factors related to physician choice of 2nd line therapy for type 2 diabetes (T2D) patients after metformin, followed by a physician focus group study to validate results of the data-driven approach. We identified 58patients in the U.S. who initiated sulfonylurea (SU), DPP-4 inhibitor, GLP-1-RA, or insulin after metformin during 2012-2014 from the HealthCore Integrated Research Database. Claims-derived factors, such as demographics, clinical characteristics, and prior healthcare utilization and costs, were selected from a classification and regression tree (CART) model that minimized classification errors. Of factors most predictive of 2nd line drug assignment, patients’ cost sharing for antidiabetic drugs (Relative Importance, RI=1) was 3 times more important than HbA1c level (RI=0.3), and 10 times more important than age, comorbidity, and kidney status (GFR) (RI=0.1). SU was most likely to be assigned to patients with lowest drug spending, moderate HbA1c control, abnormal kidney function, or no comorbid conditions. Similar themes were identified from the four physician focus groups (consisting of 8 endocrinologists and 10 PCPs in NYC and LA) and supported the findings. This study identified a gap in diabetes care where patients often receive SU due to lack of access to other treatment options.


S. Wu: None. A.D. Raval: None. J.J. Stephenson: None. X. Peng: Employee; Self; Eli Lilly and Company. H. Weisman: Employee; Self; Anthem, Inc. D.E. Faries: Employee; Self; Eli Lilly and Company.

Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at