BACKGROUND: Medical license numbers that identify physicians in pharmacy claims data are used increasingly for both research and quality improvement efforts; however, little is known about how well this information identifies the physician who wrote the prescription. We studied the accuracy of the medical license number in data from a state-run drug benefit plan by assessing its consistency with 2 external sources of data. METHODS: We studied a cohort of new users of osteoporosis medications who participated in Medicare and a state-run pharmaceutical benefit program. The medical license number from the prescription data were merged with the American Medical Association's (AMA) Masterfile to determine if the physician on the pharmacy claim existed in the AMA directory and practiced in the area under study. The prescription data were then merged with Medicare Part B data to determine if the physician on the prescription had an outpatient visit with the patient who received the medication. RESULTS: Of the 40,002 index prescriptions, 38,671 (96.7%) were written by physicians or doctors of osteopathy. Of those, 38,618 (99.9%) could be matched to the AMA Masterfile of which 37,375 (98%) had a local address. Of the AMA-matched prescriptions with a valid Unique Physician Identification Number (UPIN), 28,888 (96.1%) could be matched to Medicare Part B data, indicating that the physician whose license number appeared on the prescription had at least 1 outpatient visit with the patient who received the medication. CONCLUSIONS: The state medical license number in the pharmacy claims dataset studied seems to identify the prescribing physician with a high degree of accuracy.
BACKGROUND: Medical license numbers that identify physicians in pharmacy claims data are used increasingly for both research and quality improvement efforts; however, little is known about how well this information identifies the physician who wrote the prescription. We studied the accuracy of the medical license number in data from a state-run drug benefit plan by assessing its consistency with 2 external sources of data. METHODS: We studied a cohort of new users of osteoporosis medications who participated in Medicare and a state-run pharmaceutical benefit program. The medical license number from the prescription data were merged with the American Medical Association's (AMA) Masterfile to determine if the physician on the pharmacy claim existed in the AMA directory and practiced in the area under study. The prescription data were then merged with Medicare Part B data to determine if the physician on the prescription had an outpatient visit with the patient who received the medication. RESULTS: Of the 40,002 index prescriptions, 38,671 (96.7%) were written by physicians or doctors of osteopathy. Of those, 38,618 (99.9%) could be matched to the AMA Masterfile of which 37,375 (98%) had a local address. Of the AMA-matched prescriptions with a valid Unique Physician Identification Number (UPIN), 28,888 (96.1%) could be matched to Medicare Part B data, indicating that the physician whose license number appeared on the prescription had at least 1 outpatient visit with the patient who received the medication. CONCLUSIONS: The state medical license number in the pharmacy claims dataset studied seems to identify the prescribing physician with a high degree of accuracy.
Authors: Suzanne M Cadarette; Jeffrey N Katz; M Alan Brookhart; Raisa Levin; Margaret R Stedman; Niteesh K Choudhry; Daniel H Solomon Journal: J Rheumatol Date: 2007-12-01 Impact factor: 4.666