Magnetic resonance imaging (MRI) of a single patient’s lumbar spine at 10 different MRI centers showed marked variability in reported interpretive findings and a high prevalence of interpretive errors, according to a prospective observational study published online ahead of print in The Spine Journal.
“I have worked in the area of spinal imaging for 25 years, and in my judgment, subspecialization is probably the single most important factor in determining the quality of MRI interpretation,” said Richard J. Herzog, MD, FACR, Director of Spinal Imaging at Hospital for Special Surgery and Executive Director of Spreemo Health’s Quality Research Institute (QRI) in New York City.
“The findings demonstrate why spine surgeons should never trust a reading from a radiologist without viewing the MRI themselves,” commented Donald S. Corenman, MD, DC, Orthopaedic Spine Surgeon at The Steadman Clinic in Vail, CO. Unfortunately, other than spine surgeons and neurosurgeons, other providers more likely are dependent upon the radiologist to read the MRI and interpret the findings, Dr. Corenman said.
Single Patient Study
The study patient was a 63-year-old woman with a history of low back pain and right L5 radiculopathy. The patient underwent 10 MRI examinations of the lumbar spine at 10 different regional imaging centers in the New York City area over a 3-week period. The equipment used at these study centers included open 0.3T (1 center), stand-up 0.6T (1 center), closed 1.5T (7 centers), and closed 3.0T (1 center) MRI systems. All centers were accredited by the American College of Radiology and were unaware that the patient was part of this study.
Two reference MRIs were performed at one of the author’s institutions—one immediately before and one immediately after the series of 10 MRIs.
The study was funded by Spreemo Health’s Quality Research Institute.
Low Rate of MRI Sensitivity and High Miss Rate Found
A total of 49 distinct findings were reported in either the body or impression section of the MRI reports. Notably, approximately one-third of the interpretive findings (32.7%) appeared only once among the 10 reports, and not one finding was unanimously reported in all 10 interpretations. The most common finding—anterior spondylolisthesis at L5-S1—was reported in 9 out of 10 examinations.
Compared with the reference MRI examinations, the average rate of true-positive interpretive findings (sensitivity) was 56.4% and average rate of false-negative interpretive findings (miss rate) was 43.6%. These rates varied widely depending on pathology (Table).
Subspecialization May Improve Accuracy of MRI Interpretation
“In large cities across the country, fellowship-trained radiologists, specialized in spinal imaging, are usually available to interpret spinal MRI exams,” Dr. Herzog said, adding that “staff members can call different imaging centers to locate such radiologists.”
“Nonetheless, if there is, in fact, no imaging center in your area with a fellowship-trained radiologist, you can direct your patient to a center that performs the largest number of spine exams,” Dr. Herzog said. “A radiologist who interprets a large number of spinal MRI exams may be an expert in spinal diagnosis, even if he or she has not received fellowship training.”
“Overall, our study has demonstrated that where (and how) a patient obtains his or her MRI exam may have a direct and meaningful impact on his or her diagnosis, subsequent treatment pathway, and ultimate clinical outcome. That is why the clinical and data science teams at Spreemo Health and I are working to develop a nationally recognized benchmark for diagnostic quality that factors in both subspecialization as well other quality measures,” Dr. Herzog said.
A Spine Surgeon’s Perspective
Dr. Corenman said that low Tesla strength MRI magnets generally confer a poor interpretation, and that diagnoses may be missed using MRIs of less than 1T.
“In my practice, if it is not an obvious diagnosis, a new MRI will always be ordered if somebody presents with an MRI of less than 1T,” Dr. Corenman said. In addition, important techniques such as T1 weighted axial and sagittal sequences are sometimes skipped in an effort to contain costs, Dr. Corenman noted.
“Furthermore, there are times that radiologists will miss disorders that are significant,” Dr. Corenman said. “The diagnosis might be obvious by the clinical examination, but the radiologist does not have access to the patient to corroborate these findings. These missed findings might require a call to the radiologist to ask them to correct the report,” Dr. Corenman said.