Ordering and appropriateness of lumbar spine MRI for low back pain in a digital musculoskeletal integrated practice unit
Mary I. O’Connor, Jim Fiechtl, Ryan A. GrantBackground
Despite established appropriate use criteria for magnetic resonance imaging (MRI) of the lumbar spine in patients with low back pain, MRI ordering is often inconsistent with these guidelines, resulting in poor use of healthcare resources and potentially in unnecessary surgical or interventional procedures. In the in-person setting, the rate of inappropriate MRI in low back pain patients ranges from 28.5% to 30.7%. While diagnostic uncertainty may be greater in telemedicine, where no in-person examination is possible, specific analysis of the rate and appropriateness of MRI ordering for low back pain in telemedicine is lacking.
Methods
The aim of this retrospective cohort study was to determine the rate and appropriateness of MRI ordering of the lumbar spine in adults evaluated for low back pain by musculoskeletal physicians and advanced practice providers in a digital musculoskeletal Integrated Practice Unit (MSK-IPU) model and to analyze clinical outcomes. All patients evaluated between January 1 and December 31, 2024, were included, with data extracted through June 30, 2025, allowing at least 6 months of potential follow-up. Each lumbar spine MRI ordered was classified for appropriateness using the American College of Radiology (ACR) Appropriateness Criteria. Clinical outcomes were pain and physical function.
Results
Medical records of 1,156 patients were analyzed. Lumbar spine MRI was ordered in 32 patients (2.77%; 95% CI 1.90–3.89%), lower than published population benchmarks, which measure imaging received rather than ordered. Five patients (0.43%) were referred for spine surgery consultation. Ordering was classified inappropriate in 3 of 32 cases (9.4%; 95% CI 2.0–25.0%) by the more conservative of two independent reviewers, below reported rates of 28.5%–30.7%. Among the 972 patients with paired pain scores, mean scores fell from 5.08 to 2.99 (p < 0.001); 706 (72.6%) recorded any reduction and 549 (56.5%) a reduction of at least 2 points. Of the 705 with paired Single Assessment Numeric Evaluation (SANE) assessments, scores improved from 51.19 to 72.30 (p < 0.001), a 41.24% increase over baseline; 490 (69.5%) improved by any amount and 366 (51.9%) by at least 15 points.
Discussion
In this cohort managed within a digital musculoskeletal IPU, lumbar spine MRI was ordered infrequently, most orders met appropriate use criteria, and pain and function improved over follow-up. Without an in-person comparator or a measure of diagnostic uncertainty, these findings describe what was observed in one telemedicine practice and do not establish that telemedicine delivery caused the low ordering rate.