Question: In patients waiting for knee surgery, does the use of magnetic resonance imaging (MRI) scans reduce arthroscopy rates and improve clinical outcomes?
Design: Randomized (unclear allocation concealment), blinded (patients, physiotherapists, and radiologists) controlled trial with 6 months of follow-up after surgery or scheduled surgery date.
Setting: A teaching hospital with a core catchment population of approximately 500,000 persons in the United Kingdom.
Patients: 252 patients (mean age 43 y, 68% men) who were on a waiting list for knee arthroscopy (mean duration of symptoms, 46 wk). Exclusion criteria were age <18 years; primary synovial disease, knee infection, or tumor; arthroscopy or MRI of the affected knee in the previous 12 months; contraindication to MRI scanning; or multiple trauma. Follow-up was 81% at 6 weeks and 83% at 6 months.
Intervention: 125 patients received treatment from surgeons who were given orthopaedic clinical notes plus MRI results, and 127 patients received treatment from surgeons who were given orthopaedic clinical notes and a blank MRI card.
Main outcome measures: Arthroscopy rate. Secondary outcomes included EuroQol EQ-5D, Short Form-36, Knee Society Score, and complications.
Main results: Groups did not differ for arthroscopy rates or secondary outcomes (Table). No complication was seen in either group. The study had 90% power at the 5% significance level to detect potential differences across various secondary outcomes. A blinded analysis showed that the likelihood of detecting a clinically significant difference in arthroscopy rates was negligible.
Conclusion: In patients waiting for knee surgery, the use of MRI scans did not reduce the rate of arthroscopy or improve clinical outcomes.
Multiple nonrandomized studies in the literature have shown that MRI is cost-effective before the performance of knee arthroscopy and can decrease the frequency and subsequent need for arthroscopic surgery. This study investigates the role of MRI in reducing arthroscopy rates.
There are several issues with this study that run counter to established clinical practice. First, patients were not evaluated by the treating surgeons prior to being placed on the waiting list for surgery.
Second, half of the patients in the study probably should have been excluded from an MRI efficacy study because their diagnosis was not appropriate for MRI as they had severe degenerative disease or patellar malalignment. This is illustrated by the fact that 51% of the MRI group underwent washout and/or débridement or plica excision. As a general rule, degenerative knees should not undergo arthroscopy at all, and the decision to perform surgery to correct patellar malalignment will not be influenced by MRI findings.
Third, the MRI was done months after the decision to operate had been made and was thus not used as a diagnostic tool. The authors in fact admitted that recruitment at this point invariably led to surgery. This is particularly important because if the surgeons in this study did not endorse the use of preoperative MRI and had no plan to act on the findings, then this would explain why MRI had no benefit. This supposition seems confirmed, given that 16 patients had normal results on MRI and underwent surgery anyway. Surgeons may have operated independent of the MRI findings, and the diagnosis of a plica or degenerative change was used to justify the procedure.
Finally, the primary measured outcome of this study was the rate of arthroscopy. Perhaps if the surgeons had heeded the MRI findings, they might have reduced the number of arthroscopies.
In summary, this study does not appear to refute the long-standing belief that MRI reduces the need for arthroscopy in appropriately selected patients.