EVALUATION OF MEASUREMENT STRATEGIES TO INCREASE THE RELIABILITY OF EMG SPECTRAL INDICES TO ASSESS BACK MUSCLE FATIGUE Larivière C. 1 , Arsenault A.B. 2 , Gravel D. 2 , Gagnon D. 3 , Loisel P. 4 1 Research Center, Montreal Rehabilitation Institute, Montreal, Quebec, Canada; 2 School of Rehabilitation, U. of Montreal; 3 Faculty of Physical Education & 4 Dept. of Surgery, U. of Sherbrooke. INTRODUCTION • Poor back extensor muscle endurance is a predictor of first-time occurrence of low-back pain [1]; • Back muscle fatigue can be assessed through the power spectral analysis of the electromyographic (EMG) signal; • However, load sharing between back muscle synergists [2] can seriously affect the reliability of EMG spectral indices; • A triaxial dynamometer was developed to control asymmetric efforts (axial rotation) during the fatiguing extension task [3] but measurement strategies must be considered to further increase the reliability of these EMG indices. OBJECTIVES • To evaluate measurement strategies to maximize the reliability of EMG indices to assess back muscle fatigue. • To evaluate the reliability of EMG indices corresponding to chronic low back pain (CLBP) patients. This has been done only once. [4] METHODS Subjects: Forty male volunteers [20 controls and 20 patients with CLBP, Table 1] were assessed on 3 sessions at least 2 days apart within 2 weeks. Table 1. Characteristics of the control and CLBP subjects DISCUSSION • Control subjects vs CLBP patients reliability results: Surprisingly, similar reliability was observed for CLBP patients and control subjects. This was not expected according to previous results on CLBP patients. [4] This could be explained the considerably different measurement protocol used in the present study. • Reliability results corresponding to EMG indices obtained on each muscle individually: Reliability was at best acceptable and measurement strategies must be considered to improve it. Guidelines (measurement strategies) to improve reliability: • The assessment of muscle more medially located relative to the spine is a good choice from a reliability point of view as demonstrated by others. [7] • The averaging of measures across muscles (bilaterally or across all muscles) allows better reliability results but information concerning left- right differences or muscle specific information is lost. However, according to the present results, left-right differences cannot be assessed reliably. Averaging across bilateral muscles appears sufficient to get acceptable reliability (especially for the medially located muscles) allowing comparisons between muscle groups. • The use of multiple electrodes allows the computation of the most fatiguable muscle that showed the best reliability results. This EMG index has been shown as a more stable and valid measure than measures taken on any back muscle. [8] • The averaging of measures from 2 fatigue tests is beneficial (reliability improvement of about 13% on average) and is feasible with this protocol since the fatigue task is not performed until exhaustion. However, the recovery period between the tests must be sufficient to allow complete muscle recovery before averaging the corresponding EMG measures. A rest period between 10 and 15 min should be sufficient in this context. CONCLUSION The present study revealed that (1) reliable EMG indices can be obtained for both healthy subjects and CLBP patients populations, (2) better reliability results are attained with the medial muscles of the back comparatively to the more lateral muscles and (3) the reliability of EMG indices might be increased by averaging measurements across bilateral muscles, across electrodes positioned at different vertebral levels and by averaging the measurements of 2 fatigue tests performed within a session. REFERENCES 1. Biering-Sorensen F. Spine 1984; 9:106-19. 2. van Dieen JH et al. Eur J Appl Physiol 1993; 66:70-5. 3. Lariviere C et al. Clin Biomech (in Press) 2000. 4. Moffroid M et al. JOSPT 1994; 20:81-7. 5. Shrout PE, Fleiss JL. Psychol Bull 1979; 86:420-8. 6. Shavelson RJ, Webb NM. Generalizability theory. SAGE Publ., 1991. 7. Peach JP et al. J Electromyogr Kinesiol 1998; 8:403-10. 8. Mannion AF, Dolan P. Spine 1994; 19:1223-9. ACKNOWLEDGMENTS Christian Larivière was supported by a post-doctoral fellowship from the Institut de recherche en santé et en sécurité du travail (IRSST) of Québec. This project is presently funded also by IRSST. • Both EMG indices were computed for each muscle (N = 8), each muscle pair (mean of left and right homologous muscles) and for the back muscles as a whole (mean of all eight muscles). The steepest decline of MFslp among all muscles, that is corresponding to the most fatiguable muscle, was also retained as an EMG fatigue index. Finally, absolute differences between bilateral muscles was computed for each index to obtain additional EMG indices sensitive to left-right differences. Statistics: Intra-class correlation coefficient (ICC) ([5]: type 2,1) and standard error of measurement (SEM expressed as a percentage of the grand mean across days) were calculated for each EMG index, each muscle combination and each group of subjects. Generalizability theory [6] was applied to predict the reliability of EMG indices corresponding to the mean of 2 fatigue trials performed in the same session. RESULTS • The reliability of EMG indices is comparable between groups (Table 2) • EMG indices for individual muscles (Table 2, section #1): Reliability is moderate • EMG indices for bilateral muscles (Table 2, section #2): Reliability is improve comparatively to EMG indices computed on each muscle individually • EMG indices computed from multiple electrodes (Table 2, section #3): Excellent reliability was observed especially for the MFslp results corresponding to the most fatiguable muscle. • Medial vs lateral muscles (Table 2): The reliability of EMG indices of medially located muscles (MU-L5, LO-L1) is better than from more laterally located muscles (IL-L3, LO- T10) (see electrode sites Fig. 1) • Averaging 2 fatigue tasks in the same session (Table 2, section #4): The averaged scores show that this measurement strategy is beneficial • The EMG indices based on left-right differences between bilateral muscles: Not reliable with ICCs lower than 0.52 and SEMs greater than 234%. Table 2. Reliability results of EMG indices for both groups of subjects Measures and Tasks: • Surface EMG was collected (2048 Hz) from 8 back muscles (Fig. 1) while the subjects performed, in a static dynamometer (Fig. 2), two maximal voluntary contractions (MVC) and 2 static trunk extension tasks at 75% MVC separated by a 60 s rest period: (1) a 30 s fatigue task and (2) a 5 s recovery task • Moments at L5/S1 were displayed in real time as visual feedback to control the main effort (75% MVC in extension) while minimizing axial rotation efforts Fig. 1. EMG recording Fig 2. Dynamometer Data analysis: • From the EMG signal of each muscle, series of 250 ms windows of data, 25% overlapped, were taken from the 30 s data of the 75% MVC fatigue and of the 5 s recovery tasks. • For each window, the median frequency (MF) was computed from spectral analysis (512 points, Hamming window processing, fast Fourier transform). • Linear regressions were applied to the time-series of MF to compute the fatigue (MFslp) and recovery (REC) EMG indices as illustrated in Fig. 3. Fig 3. Computation of EMG fatigue (MFslp) and recovery (REC) indices Control subjects (n = 20) CLBP subjects (n = 20) Unpaired t-test Mean (SD) Mean (SD P value Age (yr) 38 (13) 41 (14) 0.51 Height (m) 1.75 (0.04) 1.77 (0.08) 0.34 Mass (kg) 73 (9) 80 (13) 0.06 BMI (kg/m 2 ) 23.8 (2.5) 25.6 (3.7) 0.09 Oswestry (%)* / / 18 (14) / Pain intensity (cm)* † / / 1.9 (2.4) / Back pain duration (mo) / / 126 (236) / *At the first day of testing; † assesed with a 10 cm VAS * : see methods for details concerning the nomenclature used to identify the electrode sites (Fig. 1) and the EMG indices of fatigue; †: Mean value of the left and right (L&R) electrode sites for a given muscle group; ‡: Mean value of all electrode sites; ¶: muscle showing the highest increase of fatigue ; : Mean and standard deviation (SD) of all the EMG indices of the table (N = 14); : Mean and standard deviation (SD) of the mean of 2 measures (anticipated from the generalizability theory) for all the EMG indices (N = 14). Control subjects (n = 20) CLBP patients (n = 20) Muscle * MFslp REC MFslp REC ICC SEM (%) ICC SEM (%) ICC SEM (%) ICC SEM (%) MU-L5-L 0.55 31 0.60 28 0.62 43 0.69 46 MU-L5-R 0.76 22 0.75 25 0.76 34 0.81 31 IL-L3-L 0.59 47 0.54 49 0.65 49 0.62 63 IL-L3-R 0.56 42 0.67 45 0.48 60 0.58 56 LO-L1-L 0.77 29 0.66 39 0.76 38 0.82 36 LO-L1-R 0.73 36 0.70 41 0.58 48 0.60 55 LO-T10-L 0.59 71 0.71 57 0.49 87 0.55 72 LO-T10-R 0.26 68 0.15 85 0.43 83 0.49 73 MU-L5 L&R † 0.69 25 0.71 24 0.75 33 0.85 27 IL-L3 L&R 0.61 40 0.63 42 0.61 49 0.67 50 LO-L1 L&R 0.82 25 0.75 33 0.77 34 0.81 35 LO-T10 L&R 0.49 60 0.56 58 0.57 64 0.60 54 Back muscles ‡ 0.78 23 0.77 25 0.79 30 0.88 24 Most fatiguable ¶ 0.93 21 / / 0.79 26 / / Mean (1 measure) 0.64 38 0.63 43 0.65 48 0.69 48 SD (1 measure)  0.15 17 0.16 17 0.13 19 0.13 16 Mean (2 measures) 0.77 27 0.76 30 0.78 34 0.81 34 SD (2 measures)  0.13 12 0.16 12 0.10 13 0.09 11 #1 #2 #3 #4