Endurance training of trunk extensor muscles.

Key Words: Electromyography; Exercise, strengthening; Fatigue; Muscle performance, trunk; Neck and trunk, general.

The study of the endurance of low back muscles is important to our understanding of how best to manage low back pain. Biering-Sorensen[1] demonstrated that the time a subject can maintain a horizontal, unsupported posture (a measure of mechanical capability and willingness) is a predictor for first-time occurrence of low back pain in men. This test is now known as the Sorensen Test (Fig. 1). Other investigators[2-4] have also considered endurance of the erector spinae muscles to be a contributing factor to idiopathic low back pain. Furthermore, lack of muscular endurance has been shown to be related to decreased proprioceptive awareness[5] and decreased productivity in the workplace.[6]

Electromyographic recordings are used to characterize physiologic aspects of muscle fatigue. One such measure is the change in the frequency of the recorded signals that occurs with fatigue, as described by Roy et al,[7] whose laboratory has explored clinical uses of a fatigue monitor, based on the principle of the median frequency (MF) shift during a fatiguing isometric contraction.

In physical therapy, muscular endurance is assessed less frequently than muscle strength or muscle length, and exercise protocols to increase muscular endurance are less well detailed. The purpose of this study was to determine the effectiveness of a particular exercise protocol for increasing endurance, as measured by changes in duration and changes in the shift of the MF.

Literature Review

Time, as a measure of endurance of low back musculature, varies with intensity of loading and with posture. On average, healthy subjects maintain the Sorensen Test position for a few minutes.[1] Jorgensen and Nicolaisen[8] compared two different methods for measuring trunk endurance: (1) the Sorensen Test and (2) requiring standing subjects to sustain an isometric extension contraction that was 60% of their maximal voluntary capacity (MVC).[8] The Sorensen Test was found to be more variable than the other method within a heterogeneous population of male postal workers but not in a homogeneous student group. In describing the Sorensen Test, Jorgensen and Nicolaisen reported a reliability coefficient (r) of .89 when the testing was stopped at 4 minutes, but they later recommended that the test be continued until the subject stopped the test. The advantage of the Sorensen Test is the avoidance of testing of maximal effort to establish any percentage of MVC, which is a distinct advantage when testing in the presence of pain. Smidt and Blanpied[9] reported that the Sorensen Test, which loads each individual by proportionate body weight, is approximately equivalent to 52% of MVC.

Because pain or fear may be a limiting factor with any voluntary testing, the addition of an EMG test provides a measure not affected by pain or motivation.

Using a sustained contraction of the erector spinae muscle, Carmichael et al[10] reported that the decay rate of the MF of the myoelectric power spectrum was correlated with the isometric holding time (r=.88) and had a test-retest reliability (r) of .79. Figure 2 demonstrates the decay of the MF in a subject performing the Sorenson Test. Although isometric fatiguing contractions may produce a nonlinear shift in the MF over time, the shift occurring during performance of the Sorensen Test has been shown to be linear over a large portion of the recording.[4,11] In Figure 2, the slope of the shift in MF was calculated by linear regression and the extent of the linearity is reported in terms of the coefficient of determination ([R.sup.2]). The intent of Figure 2 is to clarify our measurements. The reader should note that the initial MF here was higher than that of most of our subjects, which creates a steeper slope of the decay of the MF than was typically observed in this study.

The frequency content of the EMG constitutes a spectrum, of which the component frequencies reflect the recorded shapes of the active motor units, which in turn reflect the type of electrodes and distance from the recording electrodes, interelectrode spacing, possible synchrony with other motor units, and depolarization time along the muscle membrane.[12,13] The depolarization time along the muscle fibers' membranes will in turn reflect the muscle fibers' cross-sectional area, metabolic environment, and temperature.[14]

The MF is that value that evenly divides the area under the spectral power plot as a function of frequency and is a more reliable statistic than either the mode or the mean frequency. The MF is dependent on all the factors discussed above that applied to frequency, as well as the type of muscle being investigated,[13] the exertion of the effort,[4] and the duration of the contraction.[15] As an example, Lindstrom et al[16] reported that during a constant-force contraction of the biceps muscle in the unfatigued state, the motor units were in a range of frequencies between 20 and 170 Hz, with the MF located at 40 Hz. In the 1920s, investigators[17] recognized that the MF during a sustained fatiguing isometric contraction shifted or compressed toward lower frequencies over time. Although many investigators have utilized the decay of the MF as a measure of fatigue, it is still not clear whether this fatigue-related compression of the frequency spectrum of the EMG reflects a decrease in muscle fiber conduction velocity,[18] synchronization of motor units resulting in signals of longer duration,[19] selective fatigue of fast-twitch motor units,[20] changes in central and peripheral mechanisms,[21] or a combination of these factors.

Kondraske et al,[4] using Jorgensen and Nicolaisen's methodology,[8] studied the decay of the MF of the erector spinae muscles in healthy male subjects and reported that the rate of decay was linear, with correlations between time and the MF as recorded by surface electrodes ranging between .92 and .97. The slope of the decay of the MF in their subjects averaged -0.25 [+ or -] 0.09 Hz during testing for a contraction that was 50% of MVC and that led to fatigue in 98 [+ or -] 17 seconds.[4] In a cross-sectional design, Roy et al[7] reported differences in the initial MF at an L-5 electrode site between male subjects with and without low back pain, when exerting 80% of MVC with the trunk extensors.

The hypotheses of our study were that a 6-week exercise program would produce increases in isometric endurance times, as measured by a modified Sorensen Test, and that the rate of decay of the MF would decrease in an exercise group as compared with a control group. We believed that the initial MF would not change, but that it would take longer to arrive at the final MF, resulting in a less precipitous decay rate of the MF, thus demonstrating the trained muscles' capability to utilize higher-frequency motor unit activity for a longer duration.

Method

Subjects

Twenty-eight healthy female students were randomly selected from a volunteer population of 51 female physical therapy undergraduate students. A homogeneous group was studied because of reported differences of age and gender in endurance, fiber type, and fiber size.[22,23] Subjects who had pain on back extension or who reported any recent history (less than 2 months) of low back pain were excluded because of reported effects of changes in MF measurements obtained from subjects with low back pain.[7] All volunteers, after consenting to participate in the study, completed a survey questionnaire about leisure exercise habits. The number of hours per week of structured physical activity per subject was totaled, and a median was calculated so that subjects could be stratified into high- or low-activity groups. From each of these groups stratified by activity level, subjects were randomly assigned to an exercise group or to a control group, so that the two groups of 14 subjects each would have a strong likelihood of being homogeneous on factors of height, weight, initial endurance time, initial measures of MF, motivation, and compliance. Table 1 summarizes certain baseline values for the two groups. [TABULAR DATA 1 OMITTED]

Testing Procedure

An examiner, who was not one of the authors, performed all of the testing. Subjects were tested initially at the beginning of the study and again after 3 and 6 weeks. The Sorenson Test was performed by having the subjects assume the posture as shown in Figure 1. Subjects were instructed to hold the given position as long as possible. The test was concluded when the subject ceased holding the test position and dropped below the horizontal position. There were no visible timing devices, and the examiner was blinded as to which subjects were in which groups.

While the Sorensen Test was being administered, surface EMG records were obtained by recording bilaterally over the erector spinae muscles approximately 3 cm to either side of the inferior tip of the L-3 spinous process. Electrode placement was based on previous work that noted the position and orientation of longissimus muscle fibers at the L-3 level[24] and demonstrated that there was a steeper decline of the MF in the longissimus muscle at L-3 than in the multifidus muscle while the Sorensen Test was being administered.[25] Prior to commencing the test, but while the subject remained in the prone position, the skin was prepared by light abrasion and alcohol wiping. Two pairs of 12.5-mm-diameter silver-silver chloride electrodes were aligned parallel to the palpable mass of muscle fibers, with the inferior electrode at the L-3 level and with a separation of 25 mm between electrode pair centers. Resistance at the electrode-skin interface was checked periodically and found to be less than 50 k[Omega]. The electrode leads were taped to the iliac crests to guard against movement artifacts and then connected to a differential amplifier. A common ground electrode was placed over the T-12 spinous process. Two tests were performed 15 minutes apart at week 0 to determine repeatability of all measures.

Signal Analysis

A short cable connected each bipolar electrode pair to battery-powered differential amplifiers having an input impedance of [10.sup.8] [Omega], a gain of 1,650, and a common mode rejection ratio of 70 dB. These amplifiers were AC coupled and had universal filters with a high-pass frequency of 15 Hz (-3-dB point) and a flat response ([+ or -]0.5 dB) to 2,000 Hz. A spectrum analyzer(*) demonstrated that 95% of the recorded signals were less than 200 Hz.

Raw EMG signals were recorded on an FM cassette tape recorder([dagger]) with a bandwidth of 0 Hz to 1.25 kHz ([+ or -]1.0 dB) and a signal-to-noise ratio of 40 dB. Subsequently, the signals were processed through an antialiasing filter of the sixth order, set at 220 Hz, so virtually no signals above this frequency were analyzed. Data were then sampled at a rate of 512 Hz with a sweep rate of 1,024 points fast-Fourier transforms, so that 2 seconds of data were sampled in each 6-second time block. Other sweep rates were compared, and it was determined that the selected sampling rate provided an accurate reflection of the activity. The filtered signals were processed through a DEC 1123 minicomputer([double dagger]) for which an algorithm calculated the MF every 6 seconds between 20 and 200 HZ.(26)

A software program, which was developed on site, was used to determine the MF measurements. This software program performed a linear regression on the data points, thus producing a line of best fit or slope measure, a coefficient of determination to indicate the degree of linearity, and a coefficient of variability about the line.

Design and Analysis

The exercise group subjects trained twice a day for 6 weeks using a specific protocol. They were reevaluated biweekly, although subjects knew the progression and could advance on their own if they were ready. Exercise group subjects maintained a daily record so that compliance could be ascertained. Control group subjects also kept a record of activity so that lack of training could be documented.

Figure 3 demonstrates the progression used for training. All subjects began the study using the first test position. As soon as the subjects could hold a given position for 10 seconds and perform 25 repetitions (with a 3-second rest between efforts), they progressed themselves to the next position, which increased the torque of the spinal extensors. Reevaluation occurred twice a week and consisted of one of the investigators (MTM) checking that the exercises were being done correctly and recording the progressions. Only one subject progressed as far as the fifth posture (Fig. 3e). The posture shown in Figure 3b had to be modified for five subjects. We found that keeping the elbows and shoulders to just 90 degrees of flexion and abduction, respectively, improved cervical posture and decreased the occurrence of headaches, which were reported by three subjects. The self-reported daily records indicated that no exercise group subject missed more than eight of all exercise sessions (84 possible) and that no subject in the control group changed her activity pattern during the study. The data were coded, entered, and analyzed by a Statistical Analysis System([section]) package.[27] Because of incomplete data at some of the test sessions, we used paired t tests to account for a full sample size.

Results

Repeatability Measures

Table 2 gives the values for the two pretest measures for both the exercise group and the control group subjects on all measures of interest in this study. The two tests were 15 minutes apart, and subjects were allowed to move around, although the electrodes were not removed. [TABULAR DATA 2 OMITTED]

Repeatability (r) of the Sorensen Test time was established during the first test-retest session to be .87. Average holding times at the second testing (15 minutes after the first) were on average 23.14 and 27.86 seconds less in the control and experimental groups, respectively, but these reductions were not statistically significant at the 5% level, based on a paired t test of the time differences within each group. The first testing results were used thereafter as the pretest measure in all cases. The MF measurements at the start and end of the second Sorensen Test were not significantly different from those of the first test, which is consistent with the reported time of 5 minutes for the reconstitution of the frequency spectrum after a fatiguing contraction.[28]

Others have reported consistency in the MF measurements of the right and left sides of the back extensor muscles,[22] and although some investigators average right- and left-side values, we only analyzed data from the right side, as the EMG recording from the left side was only complete on the final test. We report right- and left-side values at week 6.

Changes with Training

Six weeks of training resulted in a 22% increase in endurance time for subjects in the exercise group (Fig. 4). The changes in times from 0 to 6 weeks, and also from 0 to 3 weeks, were tested for statistical significance by paired t tests. After 3 weeks, the subjects in the exercise group showed a mean increase of 17%, which was statistically significant (P<.05); after 6 weeks of training, the increase of 22% was also significant (P<.01) (Tab. 3). Subjects in the control group only showed a 1% average change in Sorensen Test holding times after 6 weeks, and the difference was not statistically significant. The standard deviations in test time among subjects were relatively large (initially 35% and 33% of the two group means) and remained so throughout the testing.

Table 3. Changes in Sorensen Test Times
              Exercise Group             Control Group
              X [bar]     SD             X[bar]      SD
Week 0 (s)    205.7       71.3 (n=14)    200.1       66.8 (n=14)
Week 3 (s)    240.0(a)    86.1 (n=12)    203.5       66.7 (n=11)
Week 6 (s)    251.0(b)    90.3 (n=12)    197.5       45.8 (n=13)
(a) P<.05.
(b) P<.01.

Measurements of initial and final MF did not change, on average, in either group over the 6 weeks (Tab. 4), as determined by paired t testing. Table 4 indicates that the measurements at the beginning and ending of each fatiguing session were essentially constant, on average, with a coefficient of variability of less than 12% at baseline. There were no significant differences between right and left sides in either group, as assessed at the final test.

Table 4. Median Frequency Measurements Over Time (in Hertz)
               Exercise Group            Control Group
               X[bar]      SD            X[bar]      SD
Week 0
 Initial       104.9       8.9           106.9       9.1
 Final          81.5       9.8            81.1       8.4
Week 3
 Initial       106.2      11.0           106.9       6.5
 Final          75.6      11.6            82.4       5.6
Week 6
 Initial
  Right side   105.5       9.7           103.8       7.3
  Left side    103.7       9.6           105.0      10.0
 Final
  Right side    78.5      11.0            81.3       10.4
  Left side     78.9      11.5            82.2        7.9

The pattern of the decay in MF during the sustained contractions in this study was found to be linear. The correlations (r) between time and MF ranged between .87 and .90 across test sessions. The average slopes (with standard deviations among subjects) as calculated by linear least-squares regression are presented in Table 5. No significant differences were found in the slopes by group, by time, or by side, based on paired t tests of differences and t tests of mean differences between groups. Note that there is substantial intersubject variation in these slopes, with coefficients of variation ranging from 41% to 70%.

Table 5. Slopes of Decay of Median Frequency
               Exercise Group            Control Group
               X[bar]      SD            X[bar]      SD
Week 0         -.125       .059          -.143       .066
Week 3         -.132       .063          -.133       .054
Week 6
 Right side    -.105       .073          -.129       .055
 Left side     -.110       .073          -.128       .064

Discussion

Progression of loading through postural changes produced increases in endurance time of the back extensors, as measured by the Sorensen Test. These postural progressions increased the load moment on the spine and thereby stressed the erector spinae muscles more. Undoubtedly, other muscles participated in the load sharing during the testing and especially in the training. During the test sessions, with the upper extremities at the subject's sides, the multifidus and erector spinae muscles were probably the most active muscles. During training progressions, however, the thoracic extensors, scapular muscles, and cervical extensors became more active. As the center of gravity moved more cephalad, muscles closer to the new fulcrum might be expected to contribute more. Dul et al[20] theorized that larger muscles take a proportionately greater share of an increasing load, so the training activities probably addressed those trunk muscles most stressed during the testing sessions: the multifidus and the lumbar erector spinae muscles. A possible shortcoming of the study was that we did not measure MF changes of the hamstring or the gluteus muscles or of other extensors that may have contributed to our performance measure (ie, time).

A mechanical and psychological parameter, measured by holding time, did increase in the exercise group, but this change was not accompanied by measurable physiologic changes, which might be inferred had there been a change in the initial or final MF or in the slope of the decline of the MF. Changes in MF with exercise might be attributable to changes in metabolic activity affecting conduction velocity of the sarcolemma or changes in fiber type or fiber size.

Given that the initial and final MF measurements were stable and that the holding time did increase in the exercise group, one would expect to see some change in slope of the decline of the MF (becoming less steep in the exercise group). That the slopes became slightly less steep in both groups, but not significantly so in either group, suggests that the usefulness of the measure of slope change was limited in this study because of its high variability and resultant low sensitivity. Given adequate loading to induce fatigue, MF measurements obtained from more muscle recording sites or from measures of sequential initial MFs during repeated short efforts in the recovery cycle[29] may have provided more sensitive slope measurements than we were able to obtain.

Traditionally, endurance is thought of as the time for sustaining a nonstationary activity, and increased endurance is believed to develop through light, repetitive loading. Trunk extensors, however, which are thought to serve a postural function, exhibit a high proportion of fatigue-resistant fibers[23] and have been shown to fatigue more with a stationary activity (ie, sitting) as compared with a nonstationary activity (ie, vibration during sitting).[30] Therefore, it is reasonable to strive to increase endurance through postural exercises, as was done in this study. To obtain physiologic changes (measured by MF shift), as well as mechanical changes, however, it may be necessary to exercise more vigorously, for longer periods of time, or with less healthy individuals.

Effective training did not alter the prefatigue or the postfatigue MF of the muscles being tested. The findings of this study demonstrated a stable initial and final MF, which means that recruited motor units continued to be active longer before reaching a critical level when subjects reported fatigue and ceased activity. Accepting Stulen and DeLuca's argument that motor units do not just "drop out,"[18] one could postulate that exercise altered some aspect of the vascular system by which conduction velocity of the sarcolemma was maintained over longer time periods. There is no reason to suspect any change of fiber size as a result of this training. Such a change would presumably produce an increase in initial (prefatigue) MF associated with a change in conduction velocity because of the increase in the muscle fiber's diameter.[31]

Based on constancy of final MF measurements at fatigue, there may be a critical frequency at which subjects perceived fatigue. Because ischemia is related to perceived discomfort[32] and perception of exertion has been reported to be related to the characteristics of the frequency spectrum, the changes in vascularity as a result of endurance training may be the direct measure of interest, both for understanding training effects and for understanding performance capability.

Conclusions

We have described a method for increasing isometric endurance of trunk extensor muscles in healthy young female subjects. The exercise protocol was effective after 3 weeks (P<.05) and after 6 weeks (P<.01), compared with a control group. Subjects were stratified by fitness levels, as determined by a questionnaire. We used a modified Sorensen Test to assess endurance of the trunk extensor muscles.

We were not able to document significant physiologic changes accompanying the increases in isometric endurance using the technique of monitoring the change of the MF of the surface EMG during the fatiguing contraction, despite stable values for the initial and final MFs at all test times. The change of the MF (slope) as measured in this study was linear, repeatable, and efficient, but not sensitive enough to demonstrate any physiologic changes that may have accompanied the increases observed in isometric endurance time. (*) Model 5820A, Wavetech Rockland Inc, Rockleigh, NJ 07647. ([dagger]) Type R71, TEAC, Tokyo, Japan. ([double dagger]) Digital Equipment Corp, 146 Main St, Maynard, MA 01754. ([section]) SAS Institute Inc, PO Box 8000, Cary, NC 27511.

References

[1] Biering-Sorensen F. Physical measures as risk indicators for low-back trouble over a one-year period. Spine. 1984;9:106-117. [2] Smidt G, Herring T, Amundsen L, et al. Assessment of abdominal and back extensor function. Spine. 1983;8:211-219. [3] Jackson CP, Brown MD. Is there a role for exercise in the treatment of patients with low back pain? Clin Orthop. 1983;179:39-45. [4] Kondraske GV, Deivanayagam S, Carmichael T, et al. Myoelectric spectral analysis and strategies for quantifying trunk muscular fatigue. Arch Phys Med Rehabil. 1987;68:103-110. [5] Skinner HB, Wyatt MP. Effect of fatigue on joint position sense. J Orthop Res. 1986;4:112-118. [6] Gomer FE, Silverstein LD, Berg K, Lassiter D. Changes in electromyographic activity associated with occupational stress and poor performance in the work place. Hum Factors. 1987;29:131-143. [7] Roy SH, DeLuca CJ, Casavant DA. Lumbar muscle fatigue and chronic lower back pain. Spine. 1989;14:992-1001. [8] Jorgensen K, Nicolaisen T. Trunk extensor endurance: determination and relation to low-back trouble. Ergonomics. 1987;30:259-267. [9] Smidt GL, Blanpied PR. Analysis of strength tests and resistive exercises commonly used for low back disorders. Spine. 1987;12:1025-1034. [10] Carmichael TW, Potvin AR, Mooney V. Frequency spectral measures of the myoelectric signal. IEEE Trans Biomed Eng. 1983;30:495. [11] Petrofsky JF. Computer analysis of the surface EMG during isometric exercise. Comput Biol Med 1980; 10:83-95. [12] DeLuca CJ. Myoelectric manifestations of localized muscular fatigue in humans. CRC Critical Reviews in Biomechanical Engineering. 1979;2:251-279. [13] McLeod WD, Nunnally HN, Cantrell PE. Dependence of EMG power spectra on electrode type. IEEE Trans Biomed Eng. 1976;23: 172-175. [14] DeLuca CJ. Physiology and mathematics of myoelectric signals. IEEE Trans Biomed Eng. 1979;26:313-325. [15] Kadefors R, Kaiser E, Petersen I. Dynamic spectrum analysis of myo-potentials and with special reference to muscle fatigue. Electromyography. 1968;8:39-74. [16] Lindstrom L, Magnusson R, Petersen I. Muscle load influence on myo-electric signal characteristics. Scand J Rehabil Med Suppl. 1974;3: 127-148. [17] Muscio B. Is a fatigue test possible? Br J Psychol. 1921;12:31. [18] Stulen FB, DeLuca CJ. Frequency parameters of the myoelectric signal as a measure of muscle conduction velocity. IEEE Trans Biomed Eng. 1981;28:515-523. [19] Milner-Brown HS, Stein RB, Yemm R. Changes in firing rate of human motor units during linearly changing voluntary contractions. J Physiol. 1973;230:371-390. [20] Dul J, Johnson GE, Shiavi R, Townsend MA. Muscular synergism, II: minimum-fatigue criterion for load sharing between synergistic muscles. J Biomech. 1984;17:675-684. [21] Nelson RM. Muscle fatigue: single motor unit activity. Trends in Ergonomics Factors. 1987;4:717-723. [22] Biedermann HJ, Shanks GL, Inglis J. Median frequency estimates of paraspinal muscles: reliability analysis. Electromyogr Clin Neurophysiol 1990;30:83-88. [23] Bagnall KM, Ford DM, McFadden KD, et al. The histochemical composition of human vertebral muscles. Spine. 1984;9:470-473. [24] Haig AH, Moffroid MT, Henry SM, et al. A technique for needle localization in paraspinal muscles with cadaveric confirmation. Muscle Nerve. 1991;14:521-526. [25] Moffroid MT, Haig AH, Hency SM, Haugh LD. Power spectrum analysis of longissimus and multifidus at one vertebral level. Orthopaedic Transactions. 1991;15:303. [26] Seroussi R, Wilder P, Krag MH, Pope MH. The design and use of a microcomputerized real-time muscle fatigue monitor based on the median frequency shift in the electromyographic signal. In: Proceedings of the RESNA 10th Annual Conference, San Jose, California. 1987:314-317. [27] SAS User's Guide: Statistics, Version 6 Edition. Cary, NC: SAS Institute Inc; 1990, [28] Kuorinka I. Restitution of EMG spectrum after muscular fatigue. Eur J Appl Physiol. 1988;57:311-315. [29] Klein AB, Snyder-Mackler L, Roy SH, DeLuca CJ. Comparison of spinal mobility and isometric trunk extensor forces with electromyographic spectral analysis in identifying low back pain. Phys Ther. 1991;71:445-454. [30] Pope MH, Wilder DG, Donnermeyer DD. Muscle fatigue in the static and vibrational seating environment. In: Conference Proceedings of the Advisory Group for Aerospace Research and Development, No. 378, Naples, Italy. 1985:7-11. [31] Salmons S, Henrikkson J. The adaptive response of skeletal muscle to increased use. Muscle Nerve. 1981;4:94-105. [32] Sjogaard G, Kiens B, Jorgensen K, Saltin B. Intramuscular pressure, EMG and blood flow during low-level prolonged static contraction in man. Acta Physiol Scand. 1986; 128:475-484.