The objective of this study is to investigate the proportion of potentially low-value knee MRI in Norway and to provide an estimate of the related costs.

Register study based on conditional data extraction and analysis of data from Control and Reimbursement of Healthcare Claims registry in Norway.

MRI in public specialist healthcare with universal health coverage (Norway).

48 212 MRIs for 41 456 unique patients and 45 946 reimbursement claims.

Proportion of MRIs of the knee that (1) did not have a relevant tentative diagnosis prior to the knee MRI, (2) did not have a relevant alternative image of the knee before the MRI or (3) did not have a relevant code from the specialist care within 6 months after the MRI, and those that had combinations of 1, 2 and 3. Estimated costs for those that had combinations of 1, 2 and 3.

Very few patients (6.4%) had a relevant diagnosis code or prior imaging examination when having the MRI and only 14.6% got a knee-related diagnosis code from the specialist care within 6 months after the MRI. 21.8% of the patients had knee X-ray, CT or ultrasound within 6 months before the MRI. Between 58% and 85% of patients having knee MRIs in Norway have no relevant examinations or diagnoses six months prior to or after the MRI examination. These examinations are unlikely to benefit patients and they correspond to between 24 108 and 35 416 MRIs at a cost of 6.7–9.8 million per year.

A substantial proportion of MRIs of the knee in Norway have no relevant examinations or diagnoses before or after the MRI and are potentially of low value. Reducing low-value MRIs could free resources for high-value imaging, reduce waiting times, improve the quality of care and increase patient safety and professional integrity.