Summary

Introduction
To round out the tools needed to evaluate and validate various exercise programs designed to help workers with chronic low back pain return to work, tools for evaluating lumbar deficiencies must be implemented and assessed. To evaluate the functioning of the deep trunk muscles, ultrasound measurements have been developed. The aims of this study were (1) to import this expertise into our research team, (2) to develop the analysis tools needed to extract the required measurements from the images produced, and (3) to evaluate the INTRA- and INTER-evaluator reliability of these measurements.

Methodology
Fifteen healthy subjects and 15 with chronic low back pain were evaluated in three sessions. The first two sessions (a.m. and p.m.) were conducted with two different evaluators, allowing an evaluation of INTER-evaluator reliability. The third session was held 7 to 14 days after the last session and was conducted with Evaluator No. 1, allowing an evaluation of INTRA-evaluator reliability. In each session the subjects performed four standardized tasks of low intensity and automatically recruiting certain deep trunk muscles: two soliciting the lumbar multifidus (LM) and two soliciting the transversus abdominis (TrA). The tasks were repeated with the conventional method to position the ultrasound probe and with methods (different for the two muscle groups) having the potential to control variations in the probe repositioning, in order to improve the reliability of the measurements. Muscle thickness was measured on an image taken while at rest and another taken when the contraction had been completed. These measurements, and others derived from them, were analyzed for reliability.

Results
As expected, the INTRA-evaluator reliability was higher than the INTER-evaluator reliability, confirming that it is preferable to use the same evaluator to compare results, both for research and in the clinic. Some measurements showed excellent reliability while others showed moderate or poor reliability; thus it is possible to select the most promising ones. It was shown that various automated tasks can produce comparable results in terms of reliability, which leaves more of a margin for future users. The methods for controlling variations in probe position led to a significant improvement (P > 0.05) in measurement reliability, but the effects were marginal (< 2%). The study report makes recommendations as to certain strategies for improving reliability.

Discussion and conclusion
The measurements obtained are comparable to those published elsewhere by other researchers, thus demonstrating the success of this technique as used by our team. Based on the results, it is possible to state which measurements (variables) should be retained and which ones rejected for future use, whether for research or clinical applications; however, the usefulness of this type of measurement remains to be demonstrated before it can justify a change in clinical practices. Although the present results are acceptable for research applications, other methods must be proposed and evaluated in order to better standardize evaluation protocols and thus reduce the variations inherent in this type of measurement.