This study was conducted in accordance with the current regulations on ethics in research and personal data protection and was approved by the institutional ethics committee (Research Ethics Committee Jimenez Diaz Foundation. TFG013-21_FJD). Study data was collected and managed using our institution’s research electronic data capture tools. For the present study, the Department of Information Technology provided the data from the patients attended in outpatient clinics for LBP who had had two or more MRIs of the lumbar spine performed in a time interval equal to or less than two years, regardless of the reason for which it was requested, in the period between January 1st, 2015, and December 31st, 2019.

Patients who for any reason, could not underwent MR imaging (i.e. no attendance to the appointment) and those whose images were not appropriately saved to the hospital’s database were excluded from the study. Since there were patients that had had more than two MRI during the study period, these patients were divided in such a way that each MRI and its successive one could be analyzed independently by comparing the first and the immediately following one.

MR scans were performed using models 1,5 T MAGNETOM Aera SIEMENS MR. The following sequences were applied to L1-S1 levels: sagittal T2-weighted images, sagittal T1-weighted images, sagittal proton density T2-weighted images and axial T1- and T2-weighted images parallel to the individual disc spaces at each level between L1 and S1 with a slice thickness of 3.5 mm.

Each MRI was independently assessed by a radiologist specialized in musculoskeletal pathology and an orthopedic surgeon with exclusive dedication to the spine, according to the protocol previously defined by the authors, and classified by the occurrence, absence, or severity of the following seven findings on the most affected segment:

1.

Endplate changes according to Modic et al. [4]

2.

Degenerative disc disease according to Pfirrmann et al. [5]

3.

Foraminal stenosis according to Lee et al. [6]

4.

Morphology and size of the disc herniation according to the Mysliwiec et al. MSU classification [7].

5.

Spondilolisthesis and grading according to Myerding et al. [8]

6.

Severity of central canal stenosis according to Schizas et al. [9]

7.

Facet degeneration as described by Weishaupt et al. [10].

In case of discrepancy between the evaluators, a third independent radiological evaluation was performed, and a consensus was reached. The MRI evaluation process was based on the review of the first MRI of each patient, and it was compared with the data from the second and subsequent MRIs to assess the progression of degenerative changes. Prior to analysis, a scoring system was determined for each selected MRI finding. Babinska et al. [1] designed a scoring system according to five different classifications, assigning each degenerative finding between 0 and 2 points according to its severity, up to a maximum of 10 points. A modified classification system of Babinska et al. was used to assess the degree of severity of the findings for each imaging test. By using the modified Babinska classification it was possible to score the severity of the findings, and therefore the comparison and assessment of their progression over time (Table 1).

Table 1 MRI scoring system

Disc degeneration was assessed on sagittal T2-weighted images using a five-level scoring system conceived by Pfirrmann et al. [5] that assessed signal intensity changes in the disc, its internal structure, homogeneity, and height. Grade 1 was scored 0 points, which represents a regular disc; grades 2 and 3 were scored with 1 point where the disc structure is homogeneous and the height is normal or slightly decreased; grades 4 and 5 with a heterogeneous disc structure, a hypointense signal, loss of distinction between nucleus and annulus, and decreased disc height was scored with 2 points.

Endplate changes were classified according to the description by Modic et al. [4]. Babinska et al. rated type 1 higher than types 2 and 3 according to the possible significant relationship expressed in the literature between pain intensity in patients with low back pain and type 1 Modic changes [11,12,13].

The occurrence of foraminal stenosis was classified according to the grading proposed by Lee et al. [6] where Grade 0 (normal) was scored 0 points; Grade 1 (mild grade of foraminal stenosis) with obliteration of the perineural fat surrounding the nerve root in a vertical or transverse direction, was scored with 1 point; and grades 2 and 3, moderate and severe degrees of foraminal stenosis respectively, were scored with 2 points.

To represent the size of the herniated disc, each lesion is classified as 1, 2 or 3 according to the MSU Classification by Mysliwiec et al. [7], depending on the size and location of the herniated disc at the level of maximum extrusion related to the interfacet line. Giving zero points in the absence of disc herniation; one point if the herniated disc extends to or less than 50% of the distance from the non-herniated posterior aspect of the disc to the interfacet line (size 1); and two points if it extends more than 50% of that distance (size 2) or if the hernia extends completely beyond the interfacet line (size 3).

Spondylolisthesis was assessed according to the grades described by Meyerding et at. [8]. Attending to the natural history of grade I degenerative slips of no progression over time they were scored 1 point; however, due to their greater predilection for progression, grades II-IV were scored 2 points [14, 15].

Severity of central canal stenosis according to the classification of Schizas et al. [9] is based on the CSF/rootlet ratio as seen on axial T2 images. In Grade A stenosis (0 points), CSF is clearly visible within the dural sac; with Grade B stenosis (1 point) the rootlets occupy the entire dural sac but can still be individualized; in grades C and D stenosis (2 points) no rootlets are recognizable.

In facet degeneration by rating the severity of osteoarthritis as described by Weishaupt et al. [10] a normal facet joint space was scored 0 points; narrowing of the facet joint space less than 2 mm and/or small osteophytes and/or mild hypertrophy of the articular process was scored with 1 point; moderate and/or sever changes with subarticular bone erosions and/or subchondral cysts was scored 2 points.

The added number of points according to the modified Babinska classification represented the severity of the degenerative changes assessed on MRI. The scoring system ranged from 0 to 14 points with a higher score representing more severe changes. After obtaining the score for each MRI, we proceeded to compare the score of the first MRI of each patient with the subsequent one to assess the degrees of change in the findings between consecutive imaging tests. Radiological changes were defined as increases in the severity of the findings on the most affected segment. The electronic medical records annotations were carefully reviewed to extract information on the clinical reason for the MRI request and to record any changes in the treatment plan after the MRI was performed (Table 2).

Table 2 General prevalence of findings in all MRIs

Statistical analysis. A database was created in IBM SPSS 12.0 where the field work was registered. Statistical analysis was performed by an independent team of statisticians specifically engaged for this purpose. For the qualitative variables, a Pearson’s Chi2 test and a Fisher’s exact test were carried out; while for the quantitative variables, a mean difference test was carried out. The null hypothesis (‘‘variables are independent’’) was rejected in cases in which p values were under 0.05.