Structural remodeling of the lumbar multifidus, thoracolumbar fascia and lateral abdominal wall perimuscular connective tissues: Medium-term test-retest reliability of ultrasound measures

Introduction Growing interest is being paid to the lumbar multifidus (LM) intramuscular fatty infiltrations and fibrosis that are secondary to low back pain as well to the remodeling of perimuscular connective tissues (fasciae) such as the thoracolumbar fascia and fascia sheets separating the abdominal wall muscles. Magnetic resonance imaging and computed tomography have traditionally been used but rehabilitative ultrasound imaging (RUSI) is much more affordable and practical, which can accelerate research and clinical applications on this topic. The aim of this study was to test the medium-term (8 weeks) test-retest reliability of the corresponding RUSI measures. Methods Thirty-four participants with non-acute LBP and 30 healthy controls performed a RUSI assessment before and after an 8-week time interval. LM echogenicity was quantified to assess fatty infiltrations and fibrosis while fasciae were quantified with thickness measures. Relative and absolute reliability were estimated using the generalizability theory as a framework, allowing to partition the different sources of error. Results Overall, the reliability findings were quite acceptable, with negligible systematic effects. Excellent relative reliability was reached in half of the investigated RUSI measures, particularly when averaging measures across trials. However, neither relative, nor absolute reliability results support the use of these RUSI measurements on an individual basis (e.g. clinical applications) but they are useful on a group basis (e.g. research applications). Discussion The different sources of error were distributed unequally across RUSI measures, pointing to different measurement strategies to mitigate the underlying errors. Conclusions The use of the generalizability theory allowed identifying the sources of error of the different RUSI measures. For each category of measure, depending of the distribution of errors, it was possible to recommend specific measurement strategies to mitigate them.