The Relationship Between Submaximal Activity of the Lumbar Extensor Muscles and Lumbar Posteroanterior Stiffness.

Key Words: Lumbar spine, Lumbar stiffness, Muscle activation, Physical therapy.

Posteroanterior (PA) forces are commonly applied to the lumbar spine by physical therapists in the assessment of patients with low back pain (LBP). When applying PA forces, physical therapists assess the level of pain, spinal stiffness, and muscle activity.[1-3]

For a long time, practitioners of manual therapy have linked increased muscle activity to the lack of mobility observed in people with LBP.[1,4,5] Limited movement assessed using PA forces is often described as increased stiffness.[6] Increased PA stiffness has been associated with LBP.[6] Latimer et al[6] tested the lumbar PA responses of people during and after their episode of LBP and compared their PA stiffness with a pain-free control group. They found that the PA stiffness of the people with LBP decreased when the level of their pain was lower.

The relationships among LBP, lumbar muscle activity, and lumbar PA stiffness are not well understood. The assumption that these variables are related underlies some of the treatment decisions made by physical therapists. This relationship was investigated in a pilot study[7] undertaken by 2 of us (DS and ML) that was designed to examine whether people with LBP had different lumbar muscle activity and PA stiffness when a PA force was applied to their lumbar spine than did people without LBP. This study is referred to as a pilot study because only a small number of subjects were included. We observed that some subjects with LBP showed relatively high PA stiffness and demonstrated different patterns of lumbar muscle activity in response to the application of a PA force than did similar subjects without LBP. This finding indicated to us a possible link between muscle activity and PA stiffness.

According to Maitland,[1] the aim of manual assessment of the lumbar spine stiffness in a person with LBP is to reproduce pain. Reflex muscle activity in the low back muscles is thought by some authors[4,8] to occur in response to pain, but data to support this assertion are lacking. Therefore, in theory, stiffness assessment that provokes pain could result in the reflex activation of lumbar extensor muscles. Patients with LBP have greater PA stiffness during an episode of LBP.[6] We contend, therefore, that muscle activity could be responsible for the increases in lumbar PA stiffness in people with LBP. The relationship between the muscle activity that occurs in response to the application of PA forces and lumbar PA stiffness has not yet been fully explored in either patients with LBP or people without LBP.

Maximal voluntary contraction (MVC) of the back extensors substantially altered PA stiffness at L3 in subjects with no known impairment of the low back.[9] Maximal back extensor activity has been found to produce a mean increase in PA stiffness of 350%.[9] Shirley and Lee,[7] based on their study of subjects with LBP, estimated that the amount of muscle activity produced in subjects with LBP during the application of a PA force would be equivalent to around 5% to 10% of MVC. Although it has been established that an MVC of the back extensors will substantially increase PA stiffness,[9] it is not known whether small amounts of muscle activity could also be responsible for increased PA stiffness. Our hypothesis is that small amounts of voluntary muscle activity will result in an increase in lumbar PA stiffness noted during the examination of patients. The aim of our study, therefore, was to determine the effect of different levels of voluntary back muscle activity on lumbar PA stiffness in subjects without low back impairment.

Method

The design of the study involved measurement of lumbar PA stiffness at L4 under 5 conditions of muscle activity. The conditions were 0%, 10%, 30%, 50%, and 100% of the force produced during an MVC of the subjects' back muscles. We believe that the amounts of muscle activity observed in people with LBP in response to PA forces are likely to be small percentages of MVC. We therefore focused on lower percentages because we were particularly interested in whether small amounts of muscle activity were capable of altering PA stiffness. The subjects performed these levels of muscle activity in a random order. For each level of muscle activity, the surface electromyographic (sEMG) activity of the lumbar extensors was recorded, in addition to the force generated by the subject, while the lumbar PA stiffness was measured. The L4 level was chosen because it is commonly found to be stiff and painful during assessment with PA pressures.[1]

Subjects

Twenty subjects without LBP participated in this study. The subject, s (16 female, 4 male) had a mean age of 34 years (SD=5.6, range=26-45). Subjects without LBP were selected so that an MVC could be tested. Subjects with LBP may not be able to perform an MVC without exacerbation of their symptoms. Therefore, it would be difficult to ascertain whether a maximal contraction was achieved, and quantification of smaller contractions would be difficult.

The subjects were recruited from the population of staff and students at the Faculty of Health Sciences of The University of Sydney, Lidcombe, New South Wales, Australia. Subjects were included if' they were not experiencing LBP, had no history of LBP in the 6 months prior to the study, and did not have high blood pressure (diastolic blood pressure below 90 mm Hg). Informed consent was obtained before subjects were admitted to the study.

Equipment

Posteroanterior stiffness in the lumbar spine was measured using the Spinal Physiotherapy Simulator (SPS).[10] For the purpose of this study, posteroanterior stiffness was defined as the gradient of the force displacement curve between 20 and 100 N. The SPS was designed to assess PA stiffness by applying a predetermined force to the spinous process of a lumbar vertebra. The SPS measures the force applied and the resulting displacement of the skin surface over the spinous process. The force applied and the resulting displacement can then be used to calculate stiffness. The SPS was tested on 11 human subjects without LBP and demonstrated good test-retest reliability (intraclass correlation coefficient [2,1]=.88) for measuring PA stiffness on human subjects at L3.[10] The SPS has demonstrated accuracy within 1% of the true value for measuring stiffness of an elastic beam.[10] In order to measure PA stiffness, the indentor of the SPS was positioned over the L4 spinous process (Fig. 1). The L4 spinous process was located by manual palpation after first identifying LS, using anatomical landmarks and counting upward.[5] The indentor is the part of the device that makes contact with the skin over the spinous process in a way similar to the contact of the physical therapist's hand when manually applying a PA pressure. Above the indentor is a load cell (XTRAN S1W 250N(*)) that measures the applied force, and there are 2 linear potentiometers([dagger]) that measure the displacement of the skin surface. A motor drives the indentor up and down in a rhythmical, oscillating manner similar to the way a physical therapist applies a mobilization technique.

The indentor was angled at 4.5 degrees to vertical in a caudal direction. Data were collected via a personal computer using an analog-to-digital converter (Data Translation DT2801A([double dagger])) with a sampling rate of 100 samples per second for 30 seconds. To ensure that activity of the lumbar extensor muscles resulted in an isometric contraction, subjects were prevented from extending their lumbar spine. Extension of the lumbar spine was prevented by restraining the subjects at the level of T4 and the pelvis. The T4 spinous process was located by manual palpation, using anatomical landmarks to identify T1 and counting downward) The subjects were restrained at T4 by a using padded steel plate positioned over T4 and connected via a steel frame to the table. The pelvis was restrained by a belt that was placed around the pelvis and the table. In a study with a different protocol involving measurement of isometric lumbar extension force while the subjects' pelvis and upper thoracic spine were restrained, good reliability (r = .8) was demonstrated for testing isometric force in 0 degrees of extension.[11] One hundred thirty-six subjects without orthopedic problems or medical conditions participated in that study. No information on slope or intercept were reported.

Procedure

Following the initial screening, the subjects lay prone on the testing apparatus, and the L4 and T4 spinous processes were palpated and marked. The padded steel plate was positioned in contact with the skin over the T4 spinous process. The padding deformed only slightly under the load, which we believe ensured a near-isometric contraction. A load cell attached above the steel plate recorded the force generated by the back extensors. The output of the load cell was displayed on an oscilloscope and recorded on a computer. In the prone position, the subjects were first asked to produce an isometric MVC of their back extensors, and the force generated was recorded. Values of 10%, 30%, 50%, and 100% of MVC were then calculated from this force. The subjects were then asked to push against the plate with forces that were equal to these percentages of their MVC. The oscilloscope provided feedback on the required level of force while the subjects attempted to maintain this level of force. A line was marked on the oscilloscope, and the subjects were asked to push so that the force signal just reached the target line. Subjects were also required to lie at rest without any activity of their back extensors for a period of data collection, and this condition was regarded as 0% of MVG. Subjects were asked to hold each contraction for a period of 10 seconds and were given a practice attempt for the MVC at each required level of force.

During testing with the SPS, subjects lay prone on the table of the testing apparatus with their arms by their sides and their head resting on a forehead support. A belt was placed around the subjects' pelvis and strapped to the table to provide stability. Subjects were asked hold their breath at the end of a normal expiration while maintaining an isometric contraction for the duration of 5 cycles of force application (ie, 10 seconds). The cycles were applied at a frequency of 0.5 Hz. Five loading cycles at 0.5 Hz were used to maintain reliability and to allow direct comparison with previous studies using similar methods.[6,12] It is also within the hypothesized optimum range of cycles used by physical therapists in manual assessment of PA stiffness.[1]

For each target percentage of MVC, the mean force-displacement curve was calculated by averaging the middle 3 loading cycles. These cycles were used because this period of data collection was where the T4 force was most stable. A regression line was fitted to the linear portion of the mean curve between 20 and 100 N. Force-displacement curves are usually linear within this range.[10] The PA stiffness value (coefficient K) is the slope of the regression line fitted to the force-displacement curve between 20 and 100 N and was calculated for each level of percentage of MVG.

The sEMGs of the muscles adjacent to the lumbar spine were measured during stiffness testing at all levels of muscle activity. Two channels of sEMG were used, recording from the left and right sides. Two self-adhesive Medi-Trace pellet electrodes([sections]) were attached 4 cm lateral to the L4 spinous process on each side, with an inter-electrode distance of 3 cm. The muscles underlying this region of the spine are the lumbar erector spinae muscles.[13] A reference electrode was placed over the sacrum. Similar methods of electrode placement have been used to record activity of the lumbar paraspinal muscles.[14,15] The sEMG signals initially underwent analog processing in which the raw sEMG signal was filtered (bandwidth=8-500 Hz) and amplified (Medelec AA6 Mk II,([parallel]) common mode rejection rate=10,000:1, input impedance=500 M[Omega]), and the mean value was obtained (Medelec 16([parallel])) using a 50-millisecond time constant. The processing described occurred with analog signals before they were converted to digital signals (Data Translator DT2801A) and stored in the computer. Alter analog-to-digital conversion, the signal was termed the "initial digital sEMG."

Digital processing of the initial digital sEMG signal was subsequently carried out using Sigma Plot software.(#) After importing the initial digital sEMG signal into Sigma Plot software, it was first processed by performing a running average with a time window of 2 seconds. The running average of the initial digital sEMG signal was then used to calculate the sEMG. The sEMG was a mean value for the time interval that corresponded to the subjects' contraction of their back extensors during the measurement of PA stiffness (ie, the running average sEMG was averaged over the time that the spinous process was being loaded from 20 to 100 N). The sEMG, was calculated for each target level (percentage of MVC). In order to compare muscle activity among subjects, the sEMG was normalized by expressing the level of activity achieved during each target level as a percentage of the MVC.

To normalize the data, the lowest sEMG value recorded from the running average of the initial digital sEMG signal for each test was first identified and subtracted from the sEMG at each target level (percentage of MVG). This procedure was performed so that the sEMG value used in data analysis reflected the amount of muscle activity resulting from voluntary contraction of the muscle. This value was termed the "corrected sEMG value." Each corrected sEMG value was then expressed as a percentage of the sEMG value recorded during the MVG and was called the "normalized sEMG value." A similar method for normalization of sEMG signals has been used in other work for analysis of sEMG activity of the trunk muscles.[14] A normalized sEMG value was calculated from the data collected from both the right and left sides. The final figure used in data analysis was an average of the normalized sEMG values recorded from right and left lumbar extensors. This value is described as a percentage of maximum sEMG (Table).

Table. Posteroanterior (PA) Stiffness, T4 Force, and Surface Electromyographic (sEMG) Values at the Different Levels of Maximal Voluntary Contraction (MVC)

                      PA Stiffness
                      (N/mm)         Percentage of Increase in
Target Level                         Mean Stiffness From
(Percentage of MVC)   X       SD     Resting Mean Value

Rest                  14.8    5.2
10% of MVC            17.5    4.0    11.8
30% of MVC            21.9    5.4    41.2
50% of MVC            26.3    8.8    75.2
100% of MVC           29.6    7.3    91.5


                      Percentage of
                      Maximum T4
                      Force           Percentage of
Target Level                          Maximum sEMG
(Percentage of MVC)   X       SD      X       SD

Rest                   0.48    1.6     6.9    8.9
10% of MVC            11.8     4.3    25.4   14.3
30% of MVC            31.7     5.4    44.2   16.3
50% of MVC            50.3     5.8    59.6   22.9
100% of MVC           92.3    11.8   110.1   45.8

The forces measured by the load cell positioned over T4 were expressed as percentages of the force achieved during the MVC. These forces at the target level (percentage of MVC) were calculated by averaging the force over the time corresponding to the calculation of stiffness and sEMG and subtracting the minimal value for the period of data collection. A minimal value was calculated because positioning the padded steel plate over the T4 spinous process caused the load cell to register a force in some cases. The force due to positioning of the plate was termed the "minimal value." The minimal value was subtracted from the average T4 force so that the resulting value, which was used to describe the force at the target level (percentage of MVC), would reflect the actual force exerted against the plate due to muscle activity. The minimal value was obtained by averaging data collected over 0.5 second in a part of the test where the lowest force was recorded, which, ideally, was before activity of the back extensors occurred. If this was not possible, data averaged over 0.5 second from lowest force recorded during the period of data collection were used. In some cases, there was no 0.5 second of lowest force before activity of the back extensors occurred because the subjects started producing muscle activity as soon as they were informed that data collection was commencing. In these cases, the lowest force occurred after muscle activity ceased, and 0.5 second of low force was recorded after muscle activity ceased.

Data Analysis

The stiffness data for each target level (percentage of MVC) were analyzed using a Friedman one-way analysis of variance (ANOVA) for a repeated-measures design (Sigma Stat(#)) to examine for differences in stiffness among the different levels of muscle activity. A nonparametric test (Kolmorogov-Smirnov test for normality) (Sigma Stat(#)) was used because the data were not normally distributed. A post hoc (Student-Newman-Keuls) (Sigma Stat(#)) analysis was performed to isolate the levels that differed from the other levels.

Results

The mean values of PA stiffness, percentage of maximum sEMG, and percentage of maximum T4 force at the different levels of back muscle activity are given in the Table. The mean values and standard deviations for stiffness are plotted against mean percentage of maximum sEMG and are illustrated in Figure 2. A linear regression analysis was performed to fit a line of best fit to the mean values. The regression analysis showed a linear relationship between percentage of maximum sEMG and PA stiffness ([r.sup.2]=.92). The equation of the regression line for these data was:

[Figure 2 ILLUSTRATION OMITTED]

Lumbar PA Stiffness

=0.15 (Target Percentage of sEMG) + 14.7

A difference between the mean PA stiffness values at each target level (percentage of MVC) (Table; P [is less than] .0001) was demonstrated by the Friedman nonparametric ANOVA. In addition, there was a difference among median stiffness values at all levels of activity (P [is less than] .05), as indicated by the Student-Neuman-Keuls post hoc analysis.

A close linear relationship ([r.sup.2]=.99) was observed between the mean percentage of maximum T4 force produced by the back extensors and the mean percentage of maximum sEMG values. The regression equation for these data was:

(Percentage of Maximum T4 Force)

=0.95 (Percentage of Maximum sEMG) x 10.38

The T4 force values are presented in the Table.

Discussion

Our results indicate that even small amounts of back muscle activity can increase lumbar PA stiffness. The finding that voluntary submaximal isometric activity of the trunk extensor muscles increases lumbar PA stiffness is important because clinicians frequently report increased activity in this muscle group during their manual examination of a patient.[1,5] Maximal voluntary contractions of the back extensors can alter lumbar PA stiffness,[9] and the results of our study support that finding.

In a previous study by Lee et al,[9] the mean increase in PA stiffness under MVC was 350%, whereas the mean increase in PA stiffness at MVC in the current study was 92%. There are a number of reasons why this difference may have occurred. We measured PA stiffness at L4, whereas Lee et al[9] measured PA stiffness at L3. Spinal PA stiffness varies depending on the vertebral level being tested.[16] The effect of muscle activity may also be dependent on vertebral level due to the ability of some muscles (eg, multifidus muscle) to exert control very precisely on a specific vertebral level.[17] The force-generating ability of subjects is another factor that may have contributed to the difference in results between our study and that of Lee et al.[9] In the study by Lee et al,[9] there was a higher ratio of male subjects to female subjects than in our study, where the majority of subjects were female. It is possible that the subjects in the study by Lee et al[9] were stronger and produced higher absolute T4 forces. The mean PA stiffness value at MVC in the study by Lee et al[9] was 50.9 N/mm, which was boosted by 3 subjects who had particularly high stiffness values, whereas the mean stiffness value at MVC in our study was 29.6 N/mm.

The T4 force was recorded to provide visual feedback to the subjects while activating their back muscles. The close linear relationship between mean percentage of maximum T4 force and the mean percentage of maximum sEMG indicates that analysis of sEMG data is appropriate to reflect the target forces that a subject is asked to match with feedback from the oscilloscope. The sEMG value was not used to give visual feedback because it is difficult to determine the various percentages of MVC from a raw signal on an oscilloscope, whereas the percentage of maximum force was easy to determine. The sEMG value was used in the analysis of muscle activity at the various target levels (percentages of MVC) because the aim of this study was to determine the effect of activity of the lumbar extensor muscles on lumbar PA stiffness.

We believe that the sEMG signal primarily reflects activity of the muscles immediately underlying the electrodes (eg, erector spinae muscles).[13] The electrodes may have recorded activity of other muscles that were recruited while the subjects were attempting to use their back extensors, for example, the deeper or more medial extensors (eg, multifidus muscle) at that level or the erector spinae muscles at other levels. The erector spinae muscles are likely to be the main extensors at this level, and, even if activity of other muscles is recorded, their activity is likely to be closely related to the lumbar erector spinae muscles.[13]

During a voluntary contraction of the back extensors, it is possible that a physiological extension of the lumbar spine could occur, which could result in an increase in lumbar PA stiffness.[18] We believe that such an increase in PA stiffness would probably be small compared with the increases that we observed. The subjects were positioned with their arms by their sides and with their head in a neutral position. The thoracic spine was restricted by a plate at T4. A belt was also placed around their pelvis, and the pelvis was strapped to the table for stability. Although some degree of spinal extension may have occurred in this position, it is our opinion that the effect of the increased extension would have been minimal compared with the effect of muscle activity.

Contraction of the spinal extensor muscles could increase PA stiffness by their action at individual vertebral levels. Activity of these muscles causes the vertebrae to rotate posteriorly (that is, in the direction of extension) in the sagittal plane.[13] This rotation increases resistance to anterior shear[19] and, consequently, probably increases PA stiffness of the vertebral column. Thus, when a PA force is applied during muscle activity, it is likely that the resistance to anterior vertebral movement will result in increased PA stiffness, as shown in our study.

A degree of variability among subjects in the PA stiffness values was observed at each target level (percentage of MVC). All subjects in this study, however, demonstrated an increase in PA stiffness with increasing muscle activity. In the resting state, the mean PA stiffness at L4 was 14.8 N/mm (SD=5.2). These values are similar to those reported by other researchers who have used this method of testing for PA stiffness of the lumbar spine. Lee et al[9] reported PA stiffness at L4 to be 17.5 N/mm, and Latimer et al[6] reported a PA stiffness value of 14.84 N/mm (SD=3.46) for subjects without LBP, although these authors did not specify the vertebral level. Therefore, our findings for the resting state are consistent with those of other studies. The PA stiffness at MVC in our study was 29.6 N/mm (SD=7.3) compared with 50.9 N/mm (SD=23.4) observed by Lee et al.[9] Therefore, the variability at MVC in our study was less than that reported by Lee et al.[9] In our study, the standard deviation gradually increased as stiffness increased, so the larger variability observed in the study by Lee et al[9] is possibly a function of the larger stiffness values obtained and the fact that 3 subjects' stiffness values were substantially greater than those of the rest of the sample.

The magnitude of increases in PA stiffness observed in our study may or may not be clinically relevant. The changes in PA stiffness resulting from levels of muscle activity that might be observed in subjects with LBP may or may not be detectable by manual palpation procedures. Our results indicate that a 10% increase in muscle activity will lead to a mean increase in lumbar PA stiffness of 11.8% (Table). The mean threshold for stiffness discrimination when assessing linear elastic springs is 11% for physical therapists,[20] although this value may be higher for more complex force displacement relationships. That is, physical therapists can detect changes in stiffness as low as 11% in models, but whether this is true in patients is not known. The increases in stiffness resulting from large increases in muscle activity (eg, 30%, 50%, and 100% of MVC), in our opinion, should be easy to detect manually, as they are well above the stiffness discrimination threshold for most physical therapists. Some physical therapists have an extremely low stiffness discrimination threshold with elastic springs[19] and, theoretically, may be able to manually detect the increases in stiffness resulting from muscle activity as low as 10% of MVC. It is also possible that the increases in muscle activity (~5%-10% of MVC) observed in some patients with LBP by Shirley and Lee[7] could result in increases in lumbar PA stiffness that some physical therapists would be able to detect during manual stiffness assessment. Further investigation is needed to better establish the level of lumbar extensor muscle activity that occurs in patients with LBP in response to applied PA forces and to determine whether the pattern and extent of muscle activity in a voluntary contraction are similar to the pattern and extent of muscle activity in patients with LBP.

We investigated the ability of a voluntary isometric contraction of the lumbar extensor muscles to alter PA stiffness and determined that there is a linear-relationship between voluntary use of these muscles and lumbar PA stiffness. People with LBP may have increased activity of their lumbar muscles in response to the application of a PA force.[7] There is not necessarily any parallel in the pattern and extent of response between the voluntary activity produced in the our study and any muscle activity that may occur in response to an applied PA force in a patient population. Therefore, the degree to which our results can be applied to patients with LBP has not yet been established. One possible difference is that the activity occurring in response to an applied PA force in the patient with LBP will be more localized than during voluntary activity of the trunk extensors.

Conclusions

Voluntary activity of the back extensor muscles resulted in an increase in lumbar PA stiffness. This increase in lumbar PA stiffness was observed at all levels of activity that we examined. We think it is possible that the levels of muscle activity achieved with 10% to 30% of MVC are similar to the amount of muscle activity that is thought to occur in response to the application of a PA force to the lumbar spine in patients with LBP. We suggest, therefore, that a relationship may exist between lumbar muscle activity and increased PA stiffness as measured by the manual assessment of PA forces. Of potential clinical importance is the fact that even low levels of activity produce an increase in PA stiffness. That is, clinicians should be aware that increased PA stiffness may not be due to the passive properties of the spine, but rather may be due to increases in muscle activity.

(*) Applied Measurement Australia Pty Ltd, 124 Rowe St, Eastwood, New South Wales, Australia 2122.

([dagger]) Tsushin Kogyo Co Ltd, 322 Ichinotsubo, Nakahara-ku, Kawasaki, Kanagawa 211, Japan, and Penny and Giles, Somerford Road, Christchurch, Dorset DH23 3RS, England.

([double dagger]) Data Translation Inc, 100 Locke Dr, Marlboro, MA 01752-1192.

([sections]) Graphic Controls Corp, Medical Products Div, PO Box 1274, Buffalo, NY 14240.

([parallel]) Medelec Ltd, Old Woking, Surrey 9JU, England.

(#) Jandel Corp, PO Box 7005, San Rafael, CA 94912-7005.

References

[1] Maitland GD. Vertebral Manipulation. 5th ed. London, England: Butterworth & Co (Publishers) Ltd; 1986.

[2] Magarey M. Selection of passive treatment techniques. In: Proceedings of the Fourth Biennial Conference of the MTAA. Brisbane, Queensland, Australia: MTAA; 1985:298-320.

[3] Jull G. Examination of the lumbar spine. In: Grieve GP, ed. Modem Manual Therapy of the Vertebral Column. London, England: Churchill Livingstone; 1986:547-560.

[4] Mennell J. Bach Pain Diagnosis and Treatment Using Manipulative Techniques'. Boston, Mass: Little, Brown and Co Inc; 1960.

[5] Grieve GP. Mobilisation of the Spine: Notes' on Examination, Assessment, and Clinical Method. 4th ed. Edinburgh, Scotland: Churchill Livingstone; 1984.

[6] Latimer J, Lee M, Adams R, Moran C. An investigation of the relationship between low back pain and lumbar posteroanterior stiffness. J Manipulative Physiol Ther. 1996; 19:587-591.

[7] Shirley D, Lee M. A preliminary investigation of the relationship between lumber posteroanterior mobility and low back pain. Journal of Manual and Manipulative Therapy. 1993;1:22-25.

[8] Travell JG, Simons DG. Myofascial Pain and Dysfunction: The Trigger Point Manual. Baltimore, Md: Williams & Wilkins; 1983.

[9] Lee M, Esler MA, Mildren J, Herbert R. Effect of extensor muscle activation on the response to lumbar posteroanterior forces. Clinical Biomechanics. 1993;8:115-119.

[10] Lee M, Svensson N. Measurement of stiffness during simulated spinal physiotherapy. Clin Phys Physiol Meets. 1990;11:201-207.

[11] Graves JE, Pollock M, Carpenter DM, et al. Quantitative assessment of full range-of-motion isometric lumbar extension strength. Spine. 1990;15:289-294.

[12] Latimer J, Goodsell MM, Lee M, et al. Evaluation of a new device for measuring responses to posteroanterior forces in a patient population, part 1: reliability testing. Phys Ther. 1996;76:158-165.

[13] Bogduk N, Twomey L. Clinical Anatomy of the Lumbar Spine. 2nd ed. Melbourne, Victoria, Australia: Churchill Livingstone; 1991.

[14] Lavender S, Trafimow J, Andersson G, et al. Trunk muscle activation: the effects of torso flexion, moment direction, and moment magnitude. Spine. 1994;19:771-778.

[15] Sihvonen T, Partanen J, Hanninen O, Soimakallio S. Electric behavior of low back muscles during lumbar pelvic rhythm in low back patients and healthy controls. Arch Phys Med Rehabil. 1991;72: 1080-1087.

[16] Lee M, Liversidge K. Posteroanterior stiffness at three locations in the lumbar spine. J Manipulative Physiol Ther. 1994; 17:511-516.

[17] Aspden RM. Review of the functional anatomy of the spinal ligaments and the lumbar erector spinae muscles. Clin Anat. 1992;5: 372-387.

[18] Edmonston S, Allison G, Gregg S, et al. Effect of position on the posteroanterior stiffness of the lumbar spine. Manual Therapy. 1998; 3(1):21-26.

[19] Potvin JR, Norman RW, McGill SM. Reduction in anterior shear forces on the [L.sub.4]/[L.sub.5] disc by the lumbar musculature. Clinical Biomechanics. 1991;6:88-96.

[20] Maher C, Adams R. A psychophysical evaluation of manual stiffness discrimination. Australian Journal of Physiotherapy. 1995;41:161-167.

D Shirley, PT, GradDipManipTh, is Lecturer, School of Physiotherapy, Faculty of Health Sciences, The University of Sydney, East Street, POB 170, Lidcombe, New South Wales, Australia 2141 ([e-mail]). Address all correspondence to Ms Shirley.

M Lee, MBiomedE, is Lecturer, School of Exercise and Sports Science, Faculty of Health Sciences, The University of Sydney.

E Ellis, PhD, PT, is Senior Lecturer, School of Physiotherapy, Faculty of Health Sciences, The University of Sydney.

This study was approved by The University of Sydney Human Ethics Committee.

A Mechanical Equipment Grant from the Faculty of Health Sciences, The University of Sydney, provided partial funding for this study.

This article was submitted July 10, 1997, and was accepted October 8, 1998.