Musculoskeletal diagnostic ultrasound imaging for thickness measurement of four principal muscles of the cervical spine -a reliability and agreement study

To improve the quality of the study, the methodology employed was tailored according to proposed recommendations in the literature [48‐50]. It has been highly recommended for reliability studies that they reflect the circumstances in which they would like the results to be generalised [50]. This study included a representative sample from a typical clinical setting in primary care. The sample size was generally higher than that used in previous studies. The current study included a wider age range and was thought to be large enough to represent a variety of different subject types, as well as subjects with and without neck pain. Only females were included, which made the population more homogenous, an important criterion for reliability studies.

Unlike most previous studies both the ICC and the B&A test with LoA were used. An advantage of the B&A plot is that the graph provides a representation of the magnitude of the degree of agreement. One can easily identify bias, outliers, and other relationships between the variance in measure [51].

The small thickness of these muscles may have amplified errors, thus influencing the variability of measurements [33]. With lack of variability, measurements might have fallen within a restricted range that could also have affected the ICC [52]. Four raters were available for the inter-rater part of this study, thus allowing one to yield a more precise reliability estimate. This were considered a strength of this study, even though the numbers of subjects used was thought to have a greater impact on the accuracy of the results than the number of raters [48]. Owing to the use of more than two repeated measurements, calculations were more complex. As a result, the sample size needed to be large enough, preferably greater than 50, to allow the B&A’s LoA to be estimated and to avoid the CI becoming too wide [51]. Even though this study included more subjects than most of the previous studies in this topic area, the sample may still not have been sufficiently large enough when 4 raters were included [47]. Reliable results of small muscles have also reported to be challenging using MRI, despite it being regarded as the gold standard [40]. However, no validated method existed to quantify atrophic changes and fatty infiltration with MSK-DUSI, as the Goutallier classification system on MRI [53].

As the cervical muscles are complex and anatomy may vary between individuals, differences in consistent anatomical landmarks represented a challenge. Measurements were only taken at one spinal level, considered consistent for each muscle. The images were two dimensional (2D), so the entire muscle could not be visualised. It was also challenging to reproduce the muscle image in the exact same plane. There were issues regarding accurate documentation of tissue boundaries and anatomical landmarks. Either because the transition between the different muscles layers was blurred or because of thickened fascia and aponeuroses were difficult to distinguish from each other. A cause for this might be muscular degeneration where decreased water content and increased fat and fibrous content may give a greater echogenicity and change in the architectural features of muscles [28, 54]. These changes could have affected the interpretation of the images in this study, as several images had to be excluded, especially for the deep Dce and the Rcpm. Rankin [55, 56] have also reported the same difficulties in image interpretation. Degenerative changes as osteophytes could have developed in the cervical spine of the subjects and made the bony landmarks more difficult to define on the ultrasound images. Along the superior border of Lcol, is a fascial layer containing the superior ganglia and in some patients this layer also contained a blood vessel. Similarly in the transverse view of the Sscap, a vein was sometimes seen lying between the Sscap and the Sscerv/Cxmult (most likely the deep cervical vein - a branch from the vertebral vein). This was an important consideration, as these vessels could have easily been mistaken as being part of the muscle in the longitudinal view, particularly as Doppler was not standardly used. To help counteract this, the transverse image was taken to help define these borders for the Sscap only. Despite this consideration, several of the images were reported to have uncertain muscle borders. None of these were removed from the analysis. For all the muscles in this study it was decided to measure the APD, as measuring muscle thickness tend to yield lower levels of measurement error compared to CSA [50]. For the Lcol, it may be challenging to define its medial border, due to the shadowing of the trachea, when measuring CSA or its lateral dimensions [40]. On the other hand, the muscle might not have been captured at its thickest part or the exact same location, as the APD was measured on longitudinal images. Transverse images may visualise this better, but it is thought to be more difficult to confirm the exact levels where the measurements are taken on transverse images. The Rcpm was captured in transverse plane, in order to allow comparison with a previous study by Lin et al. (2009) [57]. Longitudinal images might have improved the identification of muscle borders for this muscle. The Dce was measured using the APD and as a group. Although this differed from the methodology utilised in previous studies [39, 58, 59], this decision was based on recommendations made by these studies. It was found to be near impossible to distinguish this group of muscles individually, both on a pre-study cadaver investigation and on MSK-DUSI. However, the Dce was captured transversely. Using a longitudinal view would not have captured this muscle completely due to its oblique course and varying angulation of the muscle fascicles [21]. The Sscap has been described as a complicated muscle, due to tendinous inscriptions and internal aponeuroses that interrupt fascicles and can ultimately lead to underestimation of the CSA of the muscle [60]. Its boomerang shape made it difficult to outline the borders of the whole muscle, especially lateral to its aponeurosis. According to Stokes et al. (2007), longitudinal images may be easier to interpret than transverse views, both for measuring muscle thickness and for providing biofeedback of potential changes in the muscle during contraction [33]. Ideally in conclusion, an orthogonal view (both longitudinal and transverse views) for all muscles should be used in order to enable optimal visualisation.

It is difficult to directly compare previous studies with the current study, primarily due to methodological differences. In these studies, the ICC has often been considered, more often intra-rater reliability of measuring muscle size of various cervical muscles, with a range from 0.60–0.99 [39, 40, 55, 59, 61‐64, 70]. In general, fewer subjects have previously been included, with the recruited subjects often being younger and healthy with no NP [47, 55, 57, 59, 62, 64]. The spinal level investigated has differed in previous studies [15, 40, 55, 58, 59, 61‐63, 65‐67]. Few previous studies have looked at agreement. There have also been issues with blinding not being accounted for [55, 57, 61], and incomplete reporting on reliability [39, 51, 58], which limits the generalisability of their findings.

We are uncertain whether further improvement of measurement procedures or more training will allow the raters to better agree and reduce errors, especially on images featuring difficult anatomical borders unless new technology with use of different ultrasound apparatus applications can improve this. Nevertheless, our study has provided clinicians with a recommendation for an ultrasound scanning protocol and measurement procedure for four different neck muscles. Intra-rater reliability was greater than inter-rater reliability and therefore our recommendation so far would be that the same examiner performs all ultrasound examinations, especially if repeated exams are being performed on each individual. We recommend the use of orthogonal views. Access to a video clip of the scanning may also be useful. More anatomical landmarks than the SPs should be used to identify cervical levels, as described in the methods, as differentiation otherwise may be difficult. However, the validity of this must be investigated further. The levels of investigation may also need to be reconsidered depending on muscle function at different spinal levels. In a study by Skeie et al. (2015), grading the degree of contraction of the lumbar multifidus, not measuring the exact thickness change, was suggested [68]. This may be a more clinically relevant approach that may improve the agreement intervals in future studies. The Sscap and the Dce may be considered a functional unit as they span lateral to medial as the transverso-spinal system and all act as agonists with common neural signals [19, 72]. Hence they could be evaluated with MSK-DUSI as a group, particularly as they are small muscles when considered individually. This may decrease the measurement error, relative to the muscle thickness, and tissue borders may be easier to interpret. If the whole unit is evaluated, the degree of contraction can be categorised. This of course presumes that all layers respond equally in various neck disorders, which may not be the case. Future studies should aim to establish what constitute clinically relevant muscle changes, and to outline acceptable agreement levels for these muscles. As it is recommended that future studies assess other functionally muscular related variables that pertain to muscle morphology [69], it would be interesting to see if MSK-DUSI could be a reliable method in quantifying the degree of muscle atrophy and fat infiltration of cervical muscles, and whether this has any clinical value. Implementation of ultrasound into clinical practice, may in the future, act as an objective tool in the evaluation of neck pain, but can not at present be considered appropriate for clinical use.