Natural Experiment: "Gentle Nudge" Intervention, Eliminating Phone Call Requirement

Overview

Back pain costs the U.S. over $100 billion annually, and much of this spending is wasteful due to the overuse of advanced diagnostic imaging. Despite prominent clinical guidelines and the nationally recognized Choosing Wisely campaign discouraging use of costly and low value imaging, magnetic resonance imaging (MRI) and computed tomography (CT) studies remain frequently overused. Real-time electronic clinical decision support (CDS) at the point of care has been increasingly emphasized as an important strategy to improve the value of back pain management; however, studies suggest that CDS at best only modestly influences practice patterns. The aim is to implement a behavioral economic-based intervention in the ED to promote the use of CDS system.

Despite prominent clinical guidelines and the nationally recognized Choosing Wisely campaign discouraging use of costly and low value imaging, magnetic resonance imaging (MRI) and computed tomography (CT) studies remain frequently overused. Few studies have rigorously investigated the causes of CDS' limited influence on care as well as interventions to enhance CDS' impact on reducing low value imaging. The implementation of CDS to reduce low value MRI and CT imaging studies for back pain at a large safety net health system was monitored. The CDS systems was integrated into the electronic health record system. Clinicians answered several questions and select from a list of basic imaging indications and CDS provides an American College of Radiology Appropriateness Criteria score. Appropriate scores ranged from 7-9, borderline scores ranging from 4-6, and inappropriate scores ranging from 1-3 (clinicians are encouraged to cancel inappropriate orders). Electronic order data on imaging studies was pulled using CPT billable data and imaging studies were categorized as appropriate, canceled, changed, and unscored orders. Early observation of CDS implementation revealed that LAC+USC Medical Center (one of 16 sites) had high percentages of unscored orders. At this intervention site, a "nudge" grounded in behavioral economic theory (nudge=gentle incentive, preserving freedom of choice) was provided where clinicians ordering medium or high scoring studies could bypass the usual mandatory phone call to radiology. This natural experiment was evaluated using a quasi-experimental difference-in-differences (DinD) analysis to measure whether high scores increased and unscored studies decreased at the intervention site vs. 15 control sites over time. Generalized linear regression models were used that accounted for clustering by practice site and adjusting for patient and clinician characteristics.