Intra-examiner and inter-examiner reliability of rehabilitative ultrasound imaging for lumbar multifidus and anterolateral abdominal muscles in females with recurrent low back pain: an observational, cross-sectional study

The present study had a cross-sectional, observational, single-group design, and was conducted between March and July 2020. Prior to the study, an approval of the ethics committee was obtained (decision no. IR.IUMS.REC.1398.1368).

The sample size was determined a priori, based on a previous study conducted by Koppenhaver et al. ⁽⁹⁾ The researchers examined the intra-rater and inter-rater reliability of RUSI in determining thickness of the TrA and LM muscles in non-specific LBP. The lowest intraclass correlation coefficient (ICC) with 95% CI observed in that study was 0.80 for measuring the TrA muscle thickness in the contraction condition⁽⁹⁾. The sample size was estimated using the ‘sampicc’ command in Stata software. The null hypothesis was selected to be equal to 0.40, power at 80%, type I error at 5%, and three repetitions for each measurement. The result of sampling showed that the participation of at least 10 participants with RLBP is required. In order to allow for 30% attrition, the sample size was increased to 14.28, and thus 15 participants were recruited.

The sample populations were recruited by convenience sampling. The study followed the principles of the Declaration of Helsinki. Before the registration, the participants completed a written informed consent form. Each participant was checked for inclusion and exclusion criteria. The inclusion criteria were: 1) diagnosis of RLBP (patients experienced LBP that needed medical attention or limited their activities at least twice during the preceding year)⁽¹⁰⁾; 2) pain between 30 and 60 at rest, on 0–100 point numeric pain rating scale (NPRS), where 0 represents no pain and 100 is the worst imaginable pain⁽¹¹⁾, and 3) age between 18 and 50 years. The exclusion criteria included: 1) trauma or injury to the musculoskeletal system; 2) deformity of extremities or lower back and pelvis; 3) rheumatological or neurological diseases; 4) infection; 5) tumors or radicular symptoms; 6) spinal fracture or surgery; 7) pregnancy.

Sonography imaging and examinations were performed by two independent and blinded examiners. Both examiners were physiotherapists, with clinical experience ranging between 5 to 8 years, and 2 to 5 years of experience in sonography.

A convex transducer (C2-8 probe, center frequency: 4.9 MHZ, 128 elements, 51 mmR, B-mode) was used. According to the available evidence, static cross-sectional images acquired from the whole surface of the transducer in the B-mode are adequate for analyzing the structure and diameter of the muscle and its surroundings⁽¹²⁾.

After the recruitment process, the participants were asked to complete a self-reported demographic and history form which included age, gender, body mass, stature, pain intensity, and repetition of the symptoms. Then, sonography imaging of the TrA, IO, EO and LM muscles was performed three times in rest and contraction, and their average was measured. The imaging and measurements of all muscles were performed bilaterally from the left and right side. Two hours later and seven days after the first imaging examination, the examiner A performed it again.

The participants were asked to lie supine, and place their hands on the chest. The hip and knee joints were flexed, and a pillow was placed below the knees to support them (Fig. 1A). The transducer was positioned transversely on the midaxillary line at a point between the lower edge of the ribcage and the superior border of the iliac crest (Fig. 1B)⁽¹³⁾, while a clear picture of the muscle belly and the fascial lines were seen in the center field of view. The fascial lines were hyperechoic (i.e. appeared bright white), and the adjacent muscle tissues were more hypoechoic (appeared darker)⁽¹⁴⁾. To ensure the same conditions in all participants, images of the rest condition were taken at the end of normal exhalation and measured from the thickest region of the muscle belly (Fig. 2A). The thickness of the anterolateral abdominal muscles was also measured in abdominal draw-in maneuver (ADIM) (Fig. 2B)⁽¹⁵⁾. To perform the ADIM, the participants were instructed to take a relaxed breath in and out, hold the breath out, then draw in your lower abdomen without moving your spine, and contract the abdominal muscles by pulling the navel up and in toward the spine⁽⁹⁾.

Fig. 1.

Fig. 2.

For the imaging of the LM muscle at rest, the participants were positioned prone and instructed to place their hands symmetrically next to the trunk (Fig. 3A). The examiners found the L5/S1 zygapophyseal joints by palpation and also on the ultrasound image. Next, they placed the transducer on them longitudinally and recorded parasagittal images (Fig. 3B)⁽¹⁶⁾. According to the evidence, in the parasagittal plane, the zygapophyseal joints and overlying LM muscle bulk at 2 to 3 vertebral levels could be visualized and were thus suitable for measuring the LM muscle thickness⁽¹⁷⁾ (Fig. 4A). After the images were taken at rest, the participants were instructed to flex the elbows at approximately 90° and abduct the shoulders at approximately 120°. Then, they lifted their head, trunk and upper extremities, and held with maximum effort for ultrasound imaging at muscle contraction. A sample ultrasound image at maximum contraction is provided in Fig. 4B ^(18,19).

Fig. 3.

Fig. 4.

All statistical analyses were performed using SPSS version 21.0 (SPSS Inc., IBM Corp., Armonk, NY, USA). The reliability of all measurements was evaluated using the ICC with 95% CIs model (3, k) for the intra-rater and model (2, k) for the inter-rater reliability⁽²⁰⁾. The interpretation for the ICC with 95% CI was as follows: ≤0.20 poor, 0.21–0.40 fair, 0.41–0.60 moderate, 0.61–0.80 good, and 0.81–1.00 excellent⁽²¹⁾. Using the SPSS, the ICC model (2, k) was computed by selecting the options including two-way random, average measure, and absolute agreement. The ICC model (3, k) was also computed by selecting two-way mixed and average measure.

Standard error of measurement (SEM) was used to evaluate the precision of the instrument, and was calculated as follows: pooled SD ×Ö1 – ICC. One SEM represents that the clinicians can be 68% certain that the true measurement value lies within ± 1 SEM from the clinical measurement. Measurement error was also expressed as the SEM%, which can be calculated as SEM/mean × 100. The SEM% shows measurement error independently of the measurement unit.

The minimum detectable change at the 95% confidence level (MDC₉₅) was calculated as Ö2 ×1.96 ×SEM, which shows the magnitude of change that is necessary to provide confidence that a change was not a result of random variation or measurement error. The 95% LOA were also computed as the mean difference ± 1.96 × SD.

First, despite the extensive clinical experience of the examiners, they did not have a comparable experience in sonography. Second, the participants had to perform three contractions for each muscle, and had to maintain the contraction until the correct image was recorded. However, they may gradually get tired and, as a result, the quality of the contractions will be different from normal. Previous studies have found that patients with LBP usually show a decrease in endurance and higher fatigability in the trunk muscles⁽²⁹⁾.