OBJECTIVES: Doctors’ practice and prescribing patterns are based on many factors, some of which are not observable.  We derived doctors’ prescribing patterns from U.S. claims data to show how it might be related with decisions on anticoagulant use for venous thromboembolism (VTE) treatment. METHODS: Based on U.S. claims data, we assigned doctors IDs based on the physician who treated the enrollee for the longest period of time after eliminating any emergency room, laboratory, and radiology services. Physician prescribing patterns were then calculated from prescription drug records. Patients were grouped as compliant and non-compliant to warfarin. RESULTS: We identified the doctors’ prescribing pattern with the percentage of time they prescribed warfarin, injectable anticoagulation, antiplatelet, anti-arrhythmics, rate control drugs and other drugs. We showed that patients were more likely to be compliant to warfarin if their physician’s prescribing pattern favored warfarin.  Patients were less compliant if their physician’s prescribing pattern favored injectable anticoagulation or antiplatelet. There were no effects on compliance if doctors’ prescribing patterns favored anti-arrhythmics or rate control drugs. CONCLUSIONS: Doctors’ prescribing patterns are important factors for patient compliance.  Therefore, failing to control for these patterns in compliance models might lead to omitted variable bias.