Treating Chronic Low Back Pain

Full text


Treating Chronic Low Back Pain

EMG Biofeedback Training During Movement



Key Words: Biofeedback, Electromyography, Lumbosacral region, Pain, Physical therapy.

Low back disorders limit a patient's functional and vocational activities and create a major problem when costs for disability evaluation and compensation pay-ments are considered.1 Clinical approaches to the

management of chronic low back pain patients in-clude: exercise, surgery, chemotherapy, analgesic blocks, and transcutaneous electrical nerve stimula-tion.2, 3 These treatments, however, often fail to bring

relief; therefore, clinicians need to seek better meth-ods of attending to this problem.4

Electromyographic biofeedback training may be a very effective treatment method for muscle reeduca-tion.5 Although EMG biofeedback has been used in

conjunction with relaxation training to reduce muscle activity in patients with chronic low back pain,6 to

date, the use of this modality to induce voluntary changes in their standing positions has not been explored. Therefore, we selected a patient with chronic back pain in order to study the effects of EMG biofeedback training to his low back muscula-ture during stationary positions (standing and sitting) and trunk movements. The purpose of this report is to demonstrate that voluntarily induced increases and

Ms. Jones was a graduate student, Graduate Programs in Physical Therapy, Emory University School of Medicine, Atlanta, GA. She is currently Clinical Physical Therapist, Massachusetts General Hos-pital, Fruit St, Boston, MA 02114.

This work was completed by Ms. Jones in partial fulfillment of the requirements for the degree of Master of Medical Science.

Dr. Wolf is Associate Professor, Department of Rehabilitation Medicine; Assistant Professor, Departments of Anatomy and Surgery and School of Allied Health Professions; and Coordinator of Bio-feedback Research, Emory Regional Research and Training Center, Emory University School of Medicine.

This work was supported in part by Grant No. 15-P-56808/4-14 from Rehabilitation Services Administration, Department of Health, Education, and Welfare.

Requests for reprints should be addressed to Steven L. Wolf, PhD, RPT, Center for Rehabilitation Medicine, 1441 Clifton Rd NE, Atlanta, GA 30322.

This article was submitted December 18,1978, and accepted August 30, 1979.

decreases in paraspinal muscle EMG activity that quantitatively approach those values obtained during specific activities, performed by an age- and sex-matched population without back disorders, may be associated with increases in mobility and decreases in the subjective pain state of a patient with chronic back pain. Specifically, we report decreases in EMG activity recorded from single-channel surface elec-trode pairs during movement and increases in verte-bral and hip-joint range of motion, compared to pretraining base-line measurements.

METHOD Subject

The patient was a 35-year-old man with chronic low back pain of undetermined cause and of 24 months' duration. The patient complained of local-ized low back pain and right hip pain that increased 1) as the day progressed, 2) in stressful situations, 3) after prolonged sitting, and 4) after strenuous exercise. Past clinical management had included moist-heat applications, a series of six lumbar sympathetic blocks, and biweekly manipulations by a chiroprac-tor. The patient had also been on a two-year program of daily William's flexion exercises, excluding the supine straight leg raises. He performed regular knee flexor stretching exercises as well as-resisted exercises to hip abductor and extensor muscles. These proce-dures were continued during the full length of this study.


We saw the patient for EMG biofeedback training three times a week for five weeks. We evaluated muscle activity, joint range of motion, and pain 11 times: at the second training session during each of the first five weeks, once a week for the second five


weeks, and once after an additional five-week inter-val.


Range-of-motion measurements were taken before

each evaluation session. These measurements in-cluded 1) straight leg raising in the supine position, 2) lateral trunk flexion and rotation (to the left and right) in the standing position, and 3) forward trunk flexion and extension in the standing position. All goniometric measurements were performed by one therapist in accordance with procedures established by the American Academy of Orthopaedic Surgeons.7

To pick up EMG activity Hewlett-Packard (No. 14245B) disposable surface electrodes* were placed bilaterally at the L3-4 and L4-5 interspaces, 3 cm from the midline formed by the spinous processes (Fig. 1). A ground electrode was placed 3 cm from the L2-3 right interspace. Single-channel EMG activ-ity was recorded sequentially from the left, right, upper, and lower electrode pairs, using the technique described by Wolf and associates.8 Thus, recordings

were always made from two reference electrodes at the L3-4 interspaces (upper pair, Fig. 1), the L4-5 interspaces (lower pair, Fig. 1), or the left or right side (L3-4 and L4-5 pair, Fig. 1). Because only one chan-nel (electrode pair) of EMG data could be processed at a time, each quantification of EMG activity for each movement had to be repeated four times, with electrode input leads rearranged accordingly. Intere-lectrode distances between the upper pair or lower pair was approximately 6 cm; the left- or right-side pair, approximately 3 cm. Superimposing these elec-trode configurations over dissected cadaver material indicated that the lower electrode pair rested on the fascial cleft separating the more medially located multifidus muscle from the longissimus. The upper pair was more clearly positioned on the longissimus muscle. Although the probability of recording EMG activity from the more lateral latissimus dorsi muscle would seem remote because of its dense aponeurotic component at these vertebral levels and because of its distance from the recording electrodes, the possibility of incorporating EMG activity pickup from this mus-cle or other musmus-cle groups such as the transversospin-alis complex (for example, rotatores longus or brevis muscles) cannot be excluded.

Recordings were made during the following activ-ities: quiet sitting, quiet standing, trunk flexion and extension (standing), lateral rotation to the left and right (sitting and standing), and deep-knee bending

Fig. 1. Location of recording electrodes on the back. Spinous processes of L1-L5 marked at midline; iliac crests outlined; G, ground electrode; S1, first sacral spinous process.

(stooping). The therapist manually stabilized the pa-tient's pelvis during rotatory movements. The patient was required to take a full 10 seconds to complete each activity. This time interval was defined by a high-frequency, short-duration tone from an audio amplifier and was selected because it was the shortest duration available for EMG data integration by the biofeedback device.

The raw EMG signal was amplified through a Cyborg J33 Unit† (gain = 1000) with special adap-tations to increase input impedance. The raw EMG signal was relayed to a Cyborg P600 processor (root-mean square integration, sampling rate, 10 times per second over a 10-second interval). These integrals, based upon area per unit time of rectified EMG data, were converted to digital values that, in turn, were displayed on a light-emitting diode array (Cyborg Q880). Both the raw and integrated EMG signals were recorded on frequency modulation tape and displayed on an oscilloscope.

The evaluation procedures outlined above are iden-tical to those used in a previous study designed to obtain normative EMG and joint-mobility data.8

Therefore, cues for appropriate training procedures


(see below) could be derived by comparing quanti-tative values recorded on this patient to those previ-ously recorded from men of a similar age.

Additionally, for this patient, pain was evaluated using two methods. The Pain Intensity Rating Scale was used that involves placing a mark along a 100-mm line with a descriptor stating "no pain" at the extreme left and one stating "pain as bad as it could be" at the extreme right. There are no intervening pain descriptors along the line. The pain score is simply a percentage based upon the ratio of number of millimeters that a mark is placed from the extreme left to 100.9 Our patient subjectively rated his pain in

this manner before and after each evaluation session. The McGill Pain Questionnaire was also answered by the patient before each of the 11 evaluations.10


During the five weeks of training, the patient at-tempted to assume body positions that increased sub-jective pain (see below). When compared to norma-tive data for the same age and sex, the patient's integrated EMG activity, recorded from any of the four electrode pairs, was greater than normal.8 The

primary objective for training, then, was to decrease activity of low back muscles. Because EMG activity levels recorded from the right side of the back were always greater than those recorded from the left side for most 10-second movement patterns, a second objective was to develop symmetrical activity, that is, to train the patient both to lower the quantified muscle activity from both sides of the back and to make these EMG activity levels of equal magnitude. Visual feedback was provided by a meter on the P600 processor device. Progressive deflections of the needle to the right indicated increased EMG activity. In addition, muscle responses could be "shaped" by using a threshold setting that indicated when contin-uous integrated EMG activity levels were above (red light) or below (green light) a preset level. Audio feedback was given simultaneously from an amplifier in the form of "clicks," the frequency of which in-creased with increasing levels of integrated EMG activity. Throughout all training procedures, the elec-trode array used during evaluations (Fig. 1) was maintained and the specific electrode pair was chosen based upon the activity to be undertaken. For exam-ple, during movements involving primarily a rotatory component, electrodes on either the left or right side of the back were used. On the other hand, EMG activity during movements primarily of trunk flexion or extension was monitored with the upper or lower electrode pairs. The patient was always informed about which electrode pair was being used.

To attain a decreased EMG level of muscle activity for a specific body position, the patient developed a sequence of readjustments. Because many of the ad-justments were minor, they were impossible to mea-sure accurately. Specifically, he was trained to reduce activity of the back muscles while performing poste-rior pelvic tilts, knee bending, stooping, head posi-tional changes, and anterior movements of the trunk in relation to the pelvis.

All training was done with the patient in the stand-ing position, because prolonged standstand-ing in job-re-lated activities exacerbated his pain and supine and brief standing positions had little effect upon this pain. For example, the patient experienced consider-able pain during what he termed prolonged "easy standing." Electromyographic recordings taken se-quentially from each of the four electrode-pair com-binations revealed: 1) a failure to achieve electrical silence and 2) substantially more activity from the upper electrode pair than from the other three elec-trode-pair combinations (lower, left, or right). By practicing the pelvic tilt during "easy standing," the patient could learn to reduce integrated EMG values recorded from the upper electrode pair. Isometric contractions of the gluteus maximus muscles in both the standing position and while achieving standing from a sitting position were also effective in reducing EMG activity recorded from the upper electrode pair. In these and other situations the patient's observation of reduced integrated EMG activity from the visual feedback display helped to reinforce the effects of voluntary body readjustments or isolated contractions on muscle activity.

Fig. 2. Example of increase in straight-leg-raising range of motion during the evaluation sessions. Movements measured from left (◊) and right (●) hip joints in the supine position.


Fig. 3. Example of reduced EMG activity during a stoop-ing movement, based upon root-mean square integration and digital conversion over a 10-second movement inter-val. Recordings made separately from upper (◊), lower

(♦), left (○), and right (●) reference electrode pairs.

The patient was briefly instructed in the use of a minitrainer (Basmajian/Emory Myotrainer‡) to rein-force the training learned in the clinic. The portable minitrainer was then given to the patient to use exclusively in his work environment from the seventh to the tenth week. The threshold detector on the minitrainer was set so that auditory feedback was provided only when the EMG levels became elevated beyond the reduced integrated levels achieved in the clinic using the Cyborg equipment. These levels were determined and preset by the therapist so that the patient knew at what value to set the threshold detec-tor when performing specific maneuvers at work. The patient was asked to associate body positional changes undertaken during working conditions with those movements made in the clinic. For example, when leaning forward to pick up an object in the standing position, the patient would perform a pos-terior pelvic tilt in order to maintain reduced activity of low back muscles.

There is limited value in comparing the threshold levels of the Cyborg unit used in the clinic with the minitrainer used at work. Although identical elec-trodes were used with both units, the methods of EMG integration differ between the two devices. Nonetheless, the patient was able to maintain the reduced EMG levels attained during training in the clinic while at work, so the discrepancy in the elec-tronic processing of EMG data may not be an impor-tant factor.

‡ Karlin Instruments, Advanced Health Systems, Inc, 54 E South

Temple, Salt Lake City, UT 84111.


At final evaluation, marked increases were noted in joint range of motion. Although all range-of-mo-tion measurements were made by the same clinician, we did not try to validate these recordings. Range of motion changes for straight leg raising (Fig. 2) follow a course similar to that seen in lateral trunk flexion (left and right) and lateral trunk rotation (left and right) for standing posture.

By the end of the fourth week of training, the patient demonstrated marked decreases in the activity of low back muscles during stooping movements (Fig. 3). A decrease in EMG activity was also typical during hip flexion and extension and trunk extension from a stooped posture. This reduced activity per-sisted at the 15-week follow-up, thus suggesting that the training procedure was effective in changing the use of some back musculature.

Pain measurement using the McGill Pain Ques-tionnaire dropped substantially from a range of 2.0 to 2.5 during the first eight visits to a range of 1.0 to 1.7 during the last three visits (Fig. 4). The Pain Intensity Rating Scale showed progressive decreases in both pretreatment and posttreatment scores during the first seven weeks, followed by a transient increase during the seventh to ninth week (Fig. 5). Thereafter, pain-intensity ratings continued to decrease.

At the final evaluation (week 15), the patient's comment suggested that his drug intake had been reduced substantially. At the beginning of the treat-ment, the patient had been taking 75 mg of diazepam (Valium®) and 650 mg of propoxyphene

hydrochlo-ride (Darvon®) each month. The patient had reduced

intake by 50 percent by the beginning of the fourth week of training. After 15 weeks, he estimated that he was only taking one or two 5-mg tablets of diaze-pam each week.

Fig. 4. Responses to the McGill Pain Questionnaire at each evaluation, demonstrating reduced pain-intensity values over time.


Fig. 5. Measurements of pain-intensity rating before (•) and after (○) each evaluation.


Other investigators have suggested that, to reduce activity of low back muscles, patients with chronic low back pain might benefit from EMG biofeedback training while lying down.6 Such improvement is

presumably due to a reduction in stress and anxiety associated with the chronic pain state. To our knowl-edge, no one has yet used EMG biofeedback to retrain back muscles of these patients during trunk and lower extremity movements.

For our patient, a reduction in pain and low back muscle activity was accompanied by increases in joint range of motion. These observations support the the-ory that altered muscle activity may be associated with change in subjective pain ratings. This investi-gation indicates that when EMG biofeedback to reed-ucate back muscles is provided during movements correlating with exacerbated pain, the discomfort can become minimal. Our patient substantially reduced low back muscle activity when performing a specific movement and maintained these changes over time, suggesting that a learned response that disrupts pain behavior is involved. Longer follow-up would seem advantageous to both patient and therapist to ensure continuation of the positive effects of training.

This study's encouraging results may be attributed to the tailored treatment approaches to specific pa-tient complaints. Our procedures used for training this patient may not be applicable to all sufferers of low back pain, but they may provide a framework for program design. Another factor that could have influ-enced the patient's subjective pain ratings is the re-lationship between patient and clinician. However, this seems unlikely with a chronic pain sufferer who, for some time, had interacted with many other health practitioners. Furthermore, exercise was probably not

the primary cause for improvement, inasmuch as two years of exercise had produced relatively little change in the perceived pain state.

One might speculate that there are underlying sea-sonal influences on the chronic low back sufferer. For instance, our patient was seen in the spring and throughout the summer. Perhaps if he had been fol-lowed (or treated) during the colder months, his responsiveness to feedback training would not have been as remarkable.

Discretion must be used in comparing techniques for evaluating chronic pain. We noted gross differ-ences in ratings between the McGill Pain Question-naire (Fig. 4) and the Pain Intensity Rating Scale (Fig. 5). For example, during the seventh through the ninth week, these two methods demonstrated opposite trends in pain reports. The greater fluctuations in the ratings from the pain-intensity scale may be attributed to the lack of descriptors, which are incorporated into the McGill Pain Questionnaire. Thus, in using the pain-rating scale, the patient has little guidance to help him assess his pain. Definitive comments about the relative benefits of one pain evaluation scheme over another must await analysis of trends in scoring different tests for a larger sample of patients with low back pain who are receiving EMG biofeedback train-ing.

Every person with low back pain has unique pain related to postures or movements that should be the focus of treatment application. In light of the frequent failure of conventional physical therapy, analgesic blocking, orthopedic bracing, and chiropractic tech-niques to reduce low back pain and to improve function, neuromuscular reeducation may be a real-istic treatment, alternative.


Fifteen sessions of EMG biofeedback training to low back musculature were completed by a patient with chronic low back pain during a five-week inter-val. The training was provided while standing, during trunk movements, and for movement patterns which exacerbated pain. Specific joint ranges of motion and pain ratings were recorded at each session. A mini-trainer worn by the patient in his occupational setting was used to reinforce training. Joint ranges of motion increased over time while variability in integrated EMG was reduced. Subjective pain ratings declined after 11 of 17 sessions and remained "very low" throughout a follow-up period. The patient substan-tially decreased intake of prescription drugs. Fifteen weeks after the first treatment session, the patient was still able to minimize variability in EMG activity for specific body positioning.


Acknowledgments. We wish to thank Jim Fee and

Jim Young of the Cyborg Corporation for the loan of

the biofeedback equipment. We also wish to thank

Thomas Russe for his excellent assistance during the

data collection and evaluation in the study.


1. Kosiak M, Aurelius J: The low back problem. Journal of Occupational Medicine 10:588-593, 1968

2. Blom S, Lemperg R: Electromyographic analysis of lumbar musculature in patients operated on for lumbar rhizotomy. J Neurosurg 26:25-30, 1962

3. Finneson B: Low Back Pain. Philadelphia, J. B. Lippincott Co, 1973, pp 83-246

4. Sweetman BF, Moore C: Monitoring work factors related to low back pain. Postgrad Med J 52:151-155, 1976 5. Baker M, Regenos E, Wolf SL, et al: Developing strategies

for biofeedback: Applications in neurologically handicapped patients. Phys Ther 57:402-408, 1977

6. Seres JL, Newmann Rl: Results of treatment of chronic low back pain at the Portland Pain Center. J Neurosurg 45:32-36, 1976

7. Joint Motion: Method of Measuring and Recording. Chicago, American Academy of Orthopaedic Surgeons, 1965 8. Wolf SL, Basmajian JV, Russe CTC, et al: Normative data on

low back mobility and activity levels: Implications for neuro-muscular re-education. Am J Phys Med 58:217-229, 1979 9. Scott J, Huskisson EC: Graphic representation of pain. Pain

2:175-184, 1976

10. Melzack R: The McGill pain questionnaire: Major properties and scoring methods. Pain 1:277-299, 1975





Related subjects :