Background/Aims: With Group Health’s commitment to Affordable Excellence, Group Health has implemented multifaceted interventions since 2011 around helping clinicians become aware of their health services use relative to their colleagues, with the goal of increasing quality and reducing waste. This quality improvement initiative aims to measure changes in physician medical resource use and identify physician practice characteristics associated with decreasing use of low-value services.

Methods: We calculated physician ordering rates for specialty referrals, prescriptions, high-end imaging, and lab tests for 197 practicing primary care physicians with panels greater than 500 in 2012 and 2013. Our primary outcomes were the differences between 2013 and 2012 ordering rates per 1,000 patients adjusted for age, sex and clinical complexity. Multiple regression models were used to examine associations between outcomes and physician years in practice and FTE-adjusted panel size. Models also were adjusted for gender and ethnicity.

Results: In 2012, mean (with standard deviation) primary care physician ordering rates for referrals, prescriptions, imaging and labs were 336.1 (142.7), 7,838.2 (1,601.3), 19.8 (10.6) and 1,805.7 (647.5), respectively. Between 2013 and 2012, ordering decreased across all services; mean change in referrals, prescriptions, imaging and labs was -89.6 (125.7), -1,255.5 (1,509.5), -5.3 (8.9) and -435.9 (562.1), respectively. For referral, imaging and lab test orders, reduction was greatest in physicians newer to practice. Also for lab test orders, physicians with smaller panels had greater reductions. We did not observe any significant practice characteristics associated with change in prescriptions.

Discussion: Group Health physicians with fewer years practicing and smaller panel sizes were associated with larger decreases in resource use from 2013 to 2012. The larger decrease in resource use in newer physicians may be associated with a learning curve early in practice and may represent an important time for early feedback and intervention. Since the multifaceted intervention also may have influenced physician resource use, more understanding of what accounts for reductions in resource use is useful for future development of interventions.