Background: Nearly 80% of adults will experience non-specific low back pain at some point in their life.(1) In the absence of concerning features such as saddle anesthesia, imaging studies are not indicated and can lead to a downstream cascade of services which can result in undue physical, emotional, and financial harm to the patient.(2-6) As such, the American College of Radiology (ACR), the American College of Physicians (ACP), and the American College of Emergency Physicians (ACEP) all recommend against any imaging for low back pain lasting less than six weeks and without red flag symptoms.(7-13)Despite these recommendations, inappropriate imaging studies continue to be ordered. Prior interventions addressing this have focused on education, referral restrictions, and clinical decision support (CDS) tools within the electronic health record (EHR).(14-18) These interventions have shown varying degrees of success but have limited generalizability because they were implemented at single institutions and/or within single departments.

Purpose: Using CDS, we set out to reduce inappropriate imaging for the evaluation of low back pain across the largest safety net hospital system in the United States. Changing provider behavior within resource-limited settings requires overcoming a unique set of challenges that have been amplified since the COVID-19 pandemic. Interventions that do not add to resource-strain but change practice are especially valuable in this setting.

Description: This was a quality improvement (QI) initiative that took place at New York City Health and Hospitals (NYC H+H). It was led by the System High Value Care Council and included input from subject matter experts in hospital medicine, emergency medicine and neuroradiology. We identified computed tomography (CT) of the lumbar spine without (WO) contrast, magnetic resonance imaging (MRI) of the lumbar spine WO contrast, and lumbar radiography as the most common imaging modalities used to evaluate back pain. We redesigned each order to include an advisory statement recommending against inappropriate imaging within the order title as well as in the order instructions. We also condensed the existing list of 24 indications within the order to better reflect ACR recommendations. These changes were applied to all inpatient and outpatient orders across the system.We analyzed ordering patterns per 1000 inpatient days and per 1000 patient emergency department (ED) visits pre-intervention (June 2020 – July 2022) and post-intervention (July 2022 – November 2022) among all patients that were hospitalized, inpatients, and those that were seen in and discharged from the ED, respectively.We saw a significant decrease in the volume of all our target orders. Inpatient and ED CT lumbar spine orders decreased by 13% (p< 0.001) and 7% (p=0.022), respectively. Lumbar radiography decreased by 48% (p< 0.001) in the inpatient setting and by 31% (p< 0.001) in the ED. Inpatient and ED MRI volume decreased by 14% (p=0.022) and 27% (p=0.004).

Conclusions: Our intervention successfully decreased lumbar imaging across both inpatient and outpatient settings in the largest safety net system in the US. This initiative is unique in its scope (applying across all settings) and reliance on CDS alone (minimizing resource utilization) to affect change. This success, despite the diversity of our facilities, suggests that our initiative can be applied across a broad range of settings and can produce more sustainable results.