OBJECTIVES: To study health care resource utilization and costs associated with long-term non-adherence to imatinib in CML patients. METHODS: Two large administrative claims databases were combined (MarketScan and Ingenix Impact, January 2002-July 2008) to identify patients diagnosed with CML (ICD-9 code 205.1x). Patients with ≥2 imatinib prescriptions and continuous enrollment (≥6 months prior to and ≥1month post index date) were selected. Patients were followed for up to 3 years after the index date. A longitudinal retrospective open-cohort design was used to measure patients’ adherence to imatinib repeatedly over time. Imatinib treatment periods were divided into 90-day intervals. Treatment intervals were categorized as adherent (MPR ≥85%) or non-adherent (MPR<85%). Patients’ health care utilization and costs were compared between adherent and non-adherent intervals. Multivariate regression models were used to compare rates of inpatient admissions, outpatient visits, and emergency room visits, controlling for clinical and demographic characteristics. Additional regression models including past cumulative MPR were used to assess the long-term impact of non-adherence. RESULTS: For the 1877 CML patients in the study, there were 6175 adherent and 3163 non-adherent intervals. Only 34% of patients were fully adherent throughout the observation period. During non-adherent intervals, patients incurred significantly more inpatient (IRR=2.76, p<0.001), ER (IRR=1.25, p=0.021), and outpatient (IRR=1.09 p=0.001) visits. Though non-adherence was associated with lower pharmacy cost ($3053 p<0.001), this difference was outweighed by higher medical costs ($4531 p<0.001), resulting in a net cost increase ($1477 (p<0.001). Patients who were adherent throughout their observation period incurred an average cost of $11,759 per interval, compared to $13,773 for patients who were not always adherent. In patients who had not always been adherent (past cumulative MPR<85%) an adherent interval cost $1,239 (p=.002) more, while another non-adherent interval cost $2122 (p<0.001) more compared to an adherent interval in patients who had been adherent. CONCLUSIONS: Imatinib non-adherence is associated with long-term negative economic consequences.