Reducing Unnecessary Shoulder MRI Examinations Within a Capitated Health Care System: A Potential Role for Shoulder Ultrasound.

<bold>Purpose: </bold>MRI is frequently overused. The aim of this study was to analyze shoulder MRI ordering practices within a capitated health care system and explore the potential effects of shoulder ultrasound substitution.<bold>Methods: </bold>We reviewed medical records of 237 consecutive shoulder MRI examinations performed in 2013 at a Department of Veterans Affairs tertiary care hospital. Using advanced imaging guidelines, we assessed ordering appropriateness of shoulder MRI and estimated the proportion of examinations for which musculoskeletal ultrasound could have been an acceptable substitute, had it been available. We then reviewed MRI findings and assessed if ultrasound with preceding radiograph would have been adequate for diagnosis, based on literature reports of shoulder ultrasound diagnostic performance.<bold>Results: </bold>Of the 237 examinations reviewed, 106 (45%) were deemed to be inappropriately ordered, most commonly because of an absent preceding radiograph (n = 98; 92%). Nonorthopedic providers had a higher frequency of inappropriate ordering (44%) relative to orthopedic specialists (17%) (P = .016; odds ratio = 3.15, 95% confidence interval = 1.24-8.01). In the 237 examinations, ultrasound could have been the indicated advanced imaging modality for 157 (66%), and most of these (133/157; 85%) could have had all relevant pathologies characterized when combined with radiographs. Regardless of indicated modality, ultrasound could have characterized 80% of all cases ordered by nonorthopedic providers and 50% of cases ordered by orthopedic specialists (P = .007).<bold>Conclusions: </bold>Advanced shoulder imaging is often not ordered according to published appropriateness criteria. While nonorthopedic provider orders were more likely to be inappropriate, inappropriateness persisted among orthopedic providers. A combined ultrasound and radiograph evaluation strategy could accurately characterize shoulder pathologies for most cases. [ABSTRACT FROM AUTHOR]