Article Title

Choosing Wisely: Using the EHR to Identify Variability in Provider Ordering Behavior for High-End Imaging of the Head

Publication Date



low-value care, Choosing Wisely


Background/Aims: Over the last three years, a team of Group Health researchers and clinical leaders has worked collaboratively on projects to reduce low-value care. The team investigates, tests and implements metrics and interventions to improve care and reduce costs. The team’s work is informed by the national Choosing Wisely campaign, which works with specialty medical societies to generate lists of low-value clinical activities. Group Health participates in the Washington State Choosing Wisely Task Force, which has identified and defined 10 high-priority recommendations on topics such as imaging, screening and antibiotic use.

Methods: Our team is developing methods to assess variability in clinician ordering behavior with regard to the Task Force recommendation to avoid high-end imaging for uncomplicated headache. Claims data, as in the Group Health virtual data warehouse (VDW), generally do not identify the ordering provider for radiology procedures, which typically have different ordering and performing providers. To understand ordering patterns, we are instead using clinical data to identify ordering provider, and to detect both completed and cancelled orders. We examine all electronic health record (EHR) orders with a CPT code denoting head computed tomography or magnetic resonance imaging, and identify orders with an ICD code signifying uncomplicated headache. All orders, whether completed or not, should be accompanied by a justifying diagnosis, although the specifics of radiology ordering add significant challenges to identifying these diagnoses. For completed orders, we gather additional diagnosis information from EHR billing data, which (unlike claims data) still identifies the ordering provider.

Results: We will compare the sensitivity of EHR ordering data, EHR billing data, and claims-based VDW data in identifying ordering of high-end imaging for headache. Achieving high specificity may not be feasible for this measure, given the resources required for chart review and programming exclusionary criteria. However, since our main goal at this point is to document variability between providers, high specificity for any individual provider is less critical.

Discussion: Once we validate measures for headache imaging, we will distribute data to clinics and providers for review and discussion, and develop measures for other imaging-related Task Force recommendations. We also will investigate the feasibility of adding ordering provider to Group Health VDW data for services provided in Group Health clinics.




March 31st, 2015


April 28th, 2015