OBJECTIVES: This study examines factors associated with drug treatment changes after an HbA1c test in a privately insured sample of diabetes patients in California, USA. METHODS: Diabetes patients were selected from a physician organization with approximately 1600 physicians and 220,000 patients. Electronic records of HbA1c test results, prescriptions, inpatient and outpatient visits, and enrollment periods from 1/1/1997 through 9/30/2000 were collected for patients enrolled in two affiliated Health Maintenance Organizations. Patients in the final sample had at least one HbA1c test result and were enrolled for 120 days after (observation period) and 12 months before (baseline period) their most recent test. Diabetes comorbidities, drug class, and therapy days (a proxy for compliance/persistence) were determined for the baseline period. A logistic regression tested whether these factors or patient demographics were associated with addition or discontinuation of diabetes drug classes following the test. RESULTS: Of the 7104 diabetes patients in the database, 2643 (37.2%) met all criteria for the final sample. Mean age was 64.6 years (SD=14.3), and 49.9% of the population was female. Patients’ baseline period drug treatments included insulin alone (n=246, 9.3%), sulfonylureas alone (n=649, 24.6%), metformin alone (n=195, 7.4%), sulf-met combination (n=536, 20.3%), other multiple (n=233, 8.8%) and no therapy (n=784, 29.7%). Overall, 599 (22.6%) patients changed treatments. Higher HbA1c levels predicted increased rates of treatment change (Odds Ratio [OR]=1.22 per 1-point HbA1c difference, p < .0001). Relative to patients with no drug treatment, patients receiving oral drug treatment were more likely to change treatments (OR range 14.4-33.1, all p < .0001). Compliance/persistence with oral treatments during the baseline period predicted a lower rate of treatment changes (OR range 0.04-0.09, all p < .0004). CONCLUSIONS: Patients with higher HbA1c test results are more likely to change drug treatment, suggesting appropriate physician responses to inadequate glycemic control. Compliance/persistence appears to affect the decision to change oral treatments.