A new method to isolate local-area practice styles in prescription use as the basis for instrumental variables in...

Knowledge of whether particular prescription drugs for particular conditions are overused or underused is needed to attain a more efficient healthcare system. The portion of geographic variation in prescription drug use representing "practice style" may provide natural experiments as instrumental variables to address this knowledge gap.

To demonstrate a new method to isolate local-area practice style in prescription drug use as instrumental variable in community settings.

A retrospective cohort study using Iowa Medicaid claims files from 1997 to 1999. Patients with continuous enrollment and essential hypertension diagnosis were selected. An approach to measure geographic prescribing patterns at the ZIP code level based on the method of driving area of clinical care (DACC) was introduced. The properties of the DACC-based measure to serve as an instrumental variable were compared with a geographic prescribing measure based on primary care service areas and a measure of individual physician prescribing preferences. The use of thiazide diuretics treatment for the hypertension served as the basis for this study.

Each measure described a significant portion of the variation in the use of thiazide diuretics, and measured patient characteristics were well balanced across patient groups based on each measure. In multivariate analysis, the DACC-based measure had a stronger relationship with thiazide choice for individual patients than either primary care service areas-based or individual physician-based measures.

Treatment variation identified by the DACC method appears suitable as the basis for instrumental variables to assess the comparative effectiveness of treatments in practice.