Are changes in pain induced by myofeedback training related to changes in muscle activation patterns in patients with work-related myalgia?

The objective of this explorative study was to investigate to what extent changes in perceived pain, induced by myofeedback training, are correlated to changes in muscle activation patterns. Thirty subjects with work-related myalgia received myofeedback training. Before (T(0)), directly after (T(1)) and 4 weeks or, in a subset of patients, 3 months after (T(2)) this training, surface electromyography (sEMG) measurements of the upper trapezius muscle were performed during standardized computer tasks; a typing and a stress task. Besides this, visual analogue scales (VAS) were filled in to assess the levels of pain in the neck and shoulders. From the sEMG, root mean square (RMS) and relative rest time (RRT, i.e. the percentage of time RMS is below a certain threshold) were used for data analysis. The relationships between RRT, RMS and VAS at T(0) as well as for the changes between T(1)-T(0) and T(2)-T(0) were investigated using Spearman correlation coefficients. The results revealed no significant correlations between VAS and RMS both at baseline (range R = -0.22 to 0.17) and for the observed changes (range R = -0.33 to 0.32). Also, for VAS and RRT, low correlations were found for baseline (range R = -0.27 to 0.21) and for changes between T(1)-T(0) (range R = -0.02 to 0.38). However, for the changes between T(2)-T(0), correlation coefficients for the VAS for the shoulder and the RRT of the right trapezius during both the typing and stress tasks were significant at the P = 0.05 level, whereas the correlation coefficients for the VAS for the neck and both the left and right trapezii during the stress task approached significance (P = 0.05 and P = 0.1, respectively). These results suggest that decreases in pain observed at long term follow up after myofeedback training might occur as a result of an increased ability to relax but not as a result of decreased muscle activation level. However, the largest correlation found was 0.6. This means that the maximal explained variance (R (2)) is low (36%), and that there are also other processes than the changes in muscle activation that contribute to changes in perceived pain.