Abstract Background Professional guidelines recommend an HbA1c 8.5% within 3 years from their last HbA1c were included in analysis. Patients were grouped as improved glycemic control (last HbA1c ≤ 8.5%) or lack of improvement (last HbA1c > 8.5%). Multivariate logistic regression analysis was performed to assess independent predictors of lack of improvement in glycemic control. Results Out of 2,232 patients who met the inclusion criteria, 1,383 had an improvement in HbA1c while 849 did not. In the fully adjusted model, independent predictors of lack of improvement included: younger age (odds ratio, 0.89 per 1-SD [12 years]; 95% CI, 0.79–1.00), female gender (1.30, 1.08–1.56), presence of hypertension (1.29, 1.08–1.55), belonging to Black race (1.32, 1.04–1.68, White as reference), living in low income area (1.86,1.28–2.68, high income area as reference), and insurance coverage other than Medicare (1.32, 1.05–1.66). Presence of current smoking was associated with a paradoxical improvement in HbA1c (0.69, 0.47—0.99). In a subgroup analysis, comparing those with all subsequent HbA1c values > 8.5% (N = 444) to those with all subsequent HbA1c values