OBJECTIVES: To evaluate variation in characteristics and outcomes of T2DM patients treated with canagliflozin across states as represented in a US claims database.

METHODS: This hypothesis-generating analysis used administrative claims data from a large US health plan for commercial and Medicare Advantage enrollees with T2DM who filled canagliflozin between 04/2013–08/2014. 12 months pre-canagliflozin (baseline) and post-canagliflozin (follow-up) data were required. A1C was assessed in patients with available data and baseline A1C ≥7.0%. 15 states were analyzed and data from among the most populous states (Arizona/Florida/Missouri/Texas) are presented.

RESULTS: Missouri stood out for high insulin use, and contrasted from Texas in terms of proportion non-white (19.7% vs. 45.9%, respectively). Texas had the lowest proportion aged ≥65years. Arizona had the highest proportion of females and patients with diabetes with complications, while Florida had the lowest baseline total cost of care. Missouri had 1.5-1.7x the mean baseline insulin costs, highest proportion on bolus/basal insulin at baseline and follow-up, and lowest proportion using ≥3 follow-up anti-hyperglycemic agents (including canagliflozin) versus other states. Baseline mean A1C levels were similar across states (range 8.81%-9.03%). Missouri had the highest proportion with baseline A1C 8%–9%, but the lowest proportions between 7%–8% or >9%. After starting canagliflozin, Missouri (where canagliflozin is used more frequently after insulin) had the lowest proportion between 6%–7% (14.0%), lowest change in A1C (–0.60%), lowest proportion achieving ≥1% A1C reduction (38.6%), and lowest attainment of A1C <8% versus other states.

CONCLUSIONS: Descriptive analysis within this large national dataset suggests that Missouri, with the highest use of baseline and follow-up insulin and corresponding treatment costs, does not appear to achieve incremental A1C benefit over other states analyzed. This hypothesis needs to be tested controlling for patient baseline clinical and demographic differences and exploring the influence of uncovered ethnic differences on outcomes.