Characteristics Associated with the Choice of First Injectable Therapy Among US Patients With Type 2 Diabetes

Abstract

PurposeThe objective of this retrospective observational study was to describe and identify clinical and demographic characteristics associated with the choice of first injectable therapy (glucagon-like peptide-1 receptor agonist [GLP-1-RA] or basal insulin) among patients with type 2 diabetes mellitus (T2DM). MethodsThis analysis included adults naive to injectable therapy with T2DM who initiated a GLP-1-RA or basal insulin (index date) between November 2014 and February 2016 using data from the Practice Fusion Electronic Health Record database. Patients with T2DM, ≥1 office visit between 6 and 18 months before the index date, and with ≥1 glycosylated hemoglobin (HbA1c) result in the 6-month preindex (baseline) period were included. A generalized boosted regression model was used to determine the patient characteristics most influential in the selection of a GLP-1-RA or basal insulin as first injectable therapy. Sensitivity analysis was performed by using bootstrapped logistic regression. FindingsThe study included 3546 and 7507 GLP-1-RA and basal insulin initiators, respectively. At baseline, GLP-1-RA initiators were significantly younger (mean, 58 vs 63 years), had lower HbA1c values (mean, 8.2% vs 9.1%), lower Diabetes Complications Severity Index (DCSI) scores (mean, 1.0 vs 1.7), and a higher body mass index (BMI) (mean, 36 vs 33 kg/m2) compared with basal insulin initiators. Variables selected by using the generalized boosted regression model with the highest relative importance (≥5%) in the selection of GLP-1-RA or basal insulin were HbA1c level (20.43%), BMI (17.73%), age (12.21%), prior prescription of a sodium-glucose cotransporter-2 inhibitor (9.17%), and DCSI score (8.39%). The same variables, as well as race, were selected by using stepwise logistic regression in all the bootstrapped samples. Patients who were older (adjusted odds ratio [OR], 0.975 [95% CI, 0.971–0.979]) and had higher HbA1c values (OR, 0.741 [95% CI, 0.721–0.761]) and DCSI scores (OR, 0.870 [95% CI, 0.848–0.892]) were significantly less likely to be prescribed a GLP-1-RA compared with basal insulin. Patients with higher BMI (OR, 1.046 [95% CI, 1.040–1.053]) and previous prescription of sodium-glucose cotransporter-2 inhibitors (OR, 2.633 [95% CI, 2.224–2.982]) were significantly more likely to be prescribed a GLP-1-RA. ImplicationsThe clinically relevant differences observed between the 2 patient populations suggest that GLP-1-RAs and basal insulin are prescribed to different types of patients with T2DM. Examining patients’ demographic and clinical characteristics may be important in assisting physicians in the choice of patient-centered injectable treatment regimens.