The effect of hamstring muscle stretching on standing posture and on lumbar and hip motions during forward bending.

Key Words: Hamstring muscle; Hip motion; Kinesiology/biomechanics, lower extremity; Lumbar motion; Posture; Stretching.

Rothman and Simeone[1] stated that postural or mechanical disturbances are the most common cause of low back pain (LBP). Certain postural[2,3] and mobility[4,5] characteristics may be related to the presence or absence of LBP. Clinicians commonly try to correct these postural and movement faults by prescribing therapeutic exercise. Stretching of the hamstring muscles appears to be commonly used and advocated.[6]

Cailliet[2] contended that alteration of the normal relationship among the alignment of the spine, the position of the pelvis, and the length of the muscles attaching to the spine and pelvis contributes to the development of LBP. Kendall and McCreary[6] argued that individuals with "flat" backs (reduced lumbar curvature) while standing tend to have short hamstring muscles. They presumed that short hamstring muscles rotate the pelvis posteriorly, resulting in a concurrent reduction of lumbar lordosis. They suggested stretching the short hamstring muscles to correct the faulty alignment.[6(pp288-289)] The effect of such stretching has not been evaluated.

Toppenberg and Bullock[7] examined the interrelationships of spinal curves, pelvic tilt, and muscle lengths (abdominal, erector spinae, iliopsoas, gluteal, rectus femoris, and hamstring muscles) in asymptomatic female subjects. Although they found no relationship between pelvic tilt and lumbar curvature, longer abdominal muscles and shorter erector spinae muscles were associated with an increased lumbar curve. Of the lower-extremity muscle groups studied, only the length of the hamstring muscles was related to the lumbar curve, and it was negatively related (ie, shorter hamstring muscles were associated with a greater degree of lumbar lordosis). Heino et al[8] examined the relationships among hip extension range of motion (ROM), standing pelvic tilt, standing lumbar lordosis, and abdominal muscle performance. No correlation was found among any of these variables. Flint[9] found that lumbar lordosis was unrelated to hip and trunk flexibility.

Many of the tasks that occur during either work or everyday activity require trunk flexion. Forward bending is a complex movement of combined lumbar and hip motion. Short hamstring muscles, because of their attachments to the posterior leg and to the ischial tuberosity, may limit hip flexion ROM. Several authors[5,10,11] have suggested that restricted hip motion is coupled with excessive lumbar motion. Stretching short hamstring muscles to increase hip flexion, therefore, may affect lumbar motion during forward bending. Excessive lumbar motion would increase tensile loads on the spine and thus could result in LBP. According to Cailliet,[2] short hamstring muscles cause back pain due to their influence on lumbopelvic rhythm during forward bending.

Mayer and associates[12] measured lumbar and pelvic motion during forward bending using the two-inclinometer technique. They followed the progress of a group of patients with chronic LBP undergoing intensive ROM training and concluded that subjects with LBP had less overall flexion (65[degrees]) than did subjects without LBP (122[degrees]), with the percentage of lumbar flexion to overall flexion decreased (63%-43%). After 3 weeks in the program, the ranges of both lumbar and pelvic motion were increased; the percentage of lumbar to overall flexion increased (43%-71%), whereas the percentage of pelvic motion decreased (57%-29%). Unfortunately, the details of the exercise program were not described.

Based on the studies described, standing lumbopelvic posture has not been clearly related to muscle length. Furthermore, there are insufficient data to determine the effect of hamstring muscle stretching on standing lumbar and pelvic posture or the relative amounts of lumbar and hip motion during forward bending.

The purposes of this study were (1) to examine the effects of stretching on hamstring muscle length, standing lumbar lordosis, standing pelvic inclination, and the relative amount of lumbar and hip motion during trunk forward bending and (2) to determine whether there is a correlation between hamstring muscle length and standing posture or between hamstring muscle length and the relative amounts of lumbar and hip motion during forward bending.

Method

Subjects

The subjects were 39 volunteers without known musculoskeletal impairment of their spine or lower extremities and with tight hamstring muscles on the right side. A subject's hamstring muscles were considered tight (de, in a shortened position) if there was straight leg raising (SLR) of [less than or equal to]70 degrees.[13,14] The subjects had no current knee, hip, or back pain and were not engaged in exercise programs designed to strengthen or lengthen the hamstring or low back muscles. Prior to participation, subjects signed a consent form that informed them of the nature of the study. The subjects' age, height, weight, gender, and any history of LBP were recorded; these characteristics are shown in Table 1.

[TABULAR DATA 1 OMITTED]

To ensure equal distribution of hamstring muscle tightness, the subjects were stratified into two groups based on their degree of hamstring muscle tightness as measured by the SLR test (SLR between 60[degrees] and 70[degrees], SLR less than 60[degrees]). They then were randomly assigned to either a control group (n=20) or a stretching group (n=19). Both groups were measured before and after a 3-week period.

Measurements

Measurements of lumbar and hip position were obtained using the Metrecom,(*) an electromechanical three-dimensional digitizer. Landmarks on the lumbar spine, pelvis, and thigh were located and marked with the digitizing probe of the Metrecom according to the model shown in the Figure. The Metrecom software yields x, y, and z Cartesian coordinates of all digitized landmarks, which are stored in a file for subsequent data processing. From these landmarks, vectors and angles were calculated to represent lumbar and hip positions at rest and during partial and full forward bending according to the model shown in the Figure.

Prior to digitization, the spinous processes of the first lumbar vertebra (L-1) and second sacral segment (S-2), the right anterior superior iliac spine (ASIS), the right greater trochanter, and the right lateral femoral epicondyle were identified through palpation and marked with adhesive markers. The examiner (YL) located the spinous profess of S-2 by palpating the posterior superior iliac spines and moving her thumbs medially to the midline of the subject's sacrum. The examiner then located the L-1 spinous process by palpating and moving her thumbs superiorly six levels above the subject's S-2 spinous process.

A plastic marking apparatus with a point at each supporting base was made to improve digitization of the S-2 and L-1 spinous processes. First, this plastic apparatus was placed along the subject's spine and centered exactly over the L-1 vertebra. Both upper and lower points of the plastic apparatus, which represented points 2.54 cm (1 in) above and below L-1, respectively, were then digitized and used to form a vector that was tangent to the lumbar curve at L-1. This same procedure was used for S-2. The plastic apparatus was centered over the S-2 vertebra, and points above and below S-2 were used to form a vector that was tangent to the curve at S-2 (Figure). The points obtained by digitizing the greater trochanter and lateral femoral epicondyle were used to form a vector representing the thigh. In the resting standing position, the ASIS was digitized and a vector was calculated between S-2 and the ASIS. Subsequently, three angles were obtained by a computer program that calculates the angle between two vectors using these coordinates. The lumbar angle was formed by the vector at L-1 and the vector at S-2. The hip angle was formed by the vector at S-2 and the vector representing the thigh, and the pelvic inclination was the angle formed by the vector from S-2 to the ASIS with respect to the horizontal. All angles were based on the sagittal-plane projections of the relevant vectors.

Hamstring muscle length was characterized using two methods. First, the SLR method described by Ekstrand et all5 was used. With the subject positioned supine on a table, an inclinometer was placed on the leg 5 cm below the inferior border of the patella in line with the tibia. The examiner placed one hand under the calcaneus of the tested leg while palpating and maintaining the knee joint in extension with the other hand. The examiner then moved the leg vertically until the subject first felt a stretching sensation in the posterior thigh area. At this point, the inclinometer value was read. The SLR test was repeated two more times. These three measurements were averaged, and the resulting mean SLR angle was used to reflect the hamstring muscle length.

Hamstring muscle length also was characterized using the active-knee-extension (AKE) test described by Gajdosik and Lusin.[16] The subject was positioned supine with the posterior thigh held firmly against a vertical board that was attached to the table so that the thigh was stabilized at a 90-degree angle with the trunk. This hip position was maintained while an inclinometer was placed on the leg 5 cm below the inferior border of the patella in line with the tibia. The subject then slowly and actively extended the leg while maintaining the thigh against the vertical board. The lower extremity not being tested was maintained flat on the table. The range of restricted extension in the knee joint was measured by reading the inclinometer (de, total knee extension was recorded as 0[degrees]). The AKE test was repeated two more times, and these three measurements were averaged. The resulting mean AKE angle was used to reflect hamstring muscle length.

Procedure

Once each subject had read and signed the informed consent statement, measurements of hamstring muscle length were performed as described. Subjects were instructed to stretch their hamstring muscles five times, holding each stretch for 10 seconds as a warm-up prior to measurement of hamstring muscle length. Stretching was performed with the subjects positioned supine by actively extending the knee with the hip flexed. The SLR measurement method was used first, followed by the AKE method.

Next, subjects were asked to remove their shoes and stand comfortably with their feet about shoulder width apart. Each subject stood on a piece of paper taped to floor in the middle of the Metrecom measuring field. To ensure that all measurements were taken in the same position, the outline of the subject's feet was drawn and the subject was asked to stand in the tracings for each subsequent measurement. To facilitate digitization, adhesive markers were placed on L-1, S-2, the ASIS, the right greater trochanter, and the right lateral femoral epicondyle. The superior border of each patella also was marked because subjects used this landmark during the test of partial forward bending.

Anatomic landmarks were then digitized as described with the subjects resting in a standing position. To determine the amount of lumbar and hip motion during the early and late phases of trunk flexion, the subjects were instructed to bend forward slowly into partial and full forward-bending positions. The partial forward-bending position was defined by the point at which the subjects' fingertips reached the top of patella. The subjects were instructed to "bend forward as far as you can while keeping your knees straight" to achieve the full forward-bending position. Subjects were asked to hold the partial and full forward-bending positions while the landmarks were digitized. Subjects were permitted to move back into an upright position between the partial and full forward-bending positions. Digitization took about 10 seconds in each position.

Early forward bending was defined as the interval between the resting standing position and the partial forward-bending position. Late forward bending was defined as the interval between the partial forward-bending position and the fully forward-bent position. Total forward bending was defined as the interval between the resting standing position and the fully forward-bent position.

When the subjects were in a partial or full forward-bending position, the same points were digitized as in the neutral standing position, except that the ASIS was not digitized in the partial or full forward-bending positions. Changes in position between standing and partial forward bending represented motion during early forward bending. Changes in position between partial and full forward bending represented motion during late forward bending. Changes in angle between standing and full forward bending represented motion during total forward bending. All measurements were taken by the same investigator (YL).

To decrease the error associated with each Metrecom measurement, the entire measurement procedure was repeated by the same investigator in both pretest and posttest sessions. For each session, the average of two measurements was computed and used as the value for that session. The intersession reliability of all measurements was evaluated by comparing the average value from the pretest session with the average value from the posttest session for subjects in the control group. Twenty paired measurements, therefore, were used to determine intersession reliability over a 3-week period. Intraclass correlation coefficients (ICC[3,k]), as described by Shrout and Fleiss,[17] were used to examine reliability for each measurement.

The intersession reliability values (ICCs) are presented in Table 2. The intersession ICCs for the SLR and AKE tests were .99. The ICCs for the measures of lumbar and hip postures and motions using the Metrecom ranged from .79 to .95.

Table 2.
Intersession Reliability Values (n = 20)

Measurement                    ICC(a) (3,k)

SLR(b) angle                     .99
AKE(c) angle                     .99
Standing lumbar angle            .92
Standing pelvic inclination      .91
Early lumbar motion              .85
Early hip motion                 .88
Late lumbar motion               .79
Late hip motion                  .85
Total lumbar motion              .92
Total hip motion                 .95
(a) ICC=intraclass correlation coefficient.
(b) SLR=straight leg raise.
(c) AKE=active knee extension.

Stretching Regimen

The stretching group stretched daily for 3 weeks. Each subject was instructed in the stretching technique immediately following the pretest measurements and also was given written instructions. The regimen consisted of slow, static stretching for 15 seconds, followed by a 15-second rest, 10 times daily. The subject was instructed to lie supine between a doorway with hips positioned slightly in)front of the doorframe. The subject raised one lower extremity and rested it against the doorframe while keeping the knee extended and the other lower extremity flat on the floor. To increase the stretch, the subject was instructed to slide the buttocks closer to the doorframe or to lift the limb away from the door frame. The subject was encouraged to stretch to a sensation of mild discomfort in an effort to achieve maximal stretching. After stretching was completed with one limb, the contralateral limb was stretched. Each subject maintained a daily log to document compliance.

Data Analysis

The equivalence of the stretching and control groups prior to the study was checked by conducting independent t tests on age, weight, and height and by conducting the nonparametric Mann-Whitney tests on gender and history of LBP.

A multivariate analysis of variance (MANOVA) was used to examine group differences for dependent variables. The MANOVA was used because the dependent variables may be correlated. In addition, the use of one test of significance for all dependent variables in multivariate analysis, rather than multiple separate tests for each dependent variable in univariate analysis (F tests), can decrease the probability of a Type I error. Therefore, Hotelling's [T.sup.2], which is a multivariate extension of Student's t test, was used for comparison of the two groups. Subsequently, F tests were used to compare group means for each dependent variable in follow-up analysis. The acceptable Type I error was set at P [less than or equal to].05. Pearson product-moment correlation coefficients also were used to show the association between hamstring muscle length and all other dependent variables.

Results

The descriptive characteristics of the subjects in each group are shown in Table 1. There were no differences between the two groups in age, weight, height, gender, or history of LBP.

Results of the MANOVA (Hotelling's [T.sup.2])for group differences showed a difference between groups following the stretching program (P<.05). Thus, in follow-up analysis, univariate F tests were used to compare group means for posts/retching variables. Table 3 shows the descriptive values and F-test results for SLR and lumbar and pelvic angles in a standing position. There was an increase of 9 degrees in SLR angle and a decrease of 12 degrees in AKE angle, both reflecting increased hamstring muscle length after stretching. No change occurred in lumbar lordosis and pelvic inclination in a resting position. The data relating to lumbar and hip motion during forward bending are shown in Table 4. There was no change in lumbar and hip motions, or in the ratio of lumbar to hip motion during early forward bending. During late forward bending, group differences were found only in the amount of hip motion (P<.05). There was an increase of 5.1 degrees in hip motion during late forward bending in the stretching group. The stretching group also had an increase of 8 degrees in hip motion for total forward bending (P<.05).

[TABULAR DATA 3-4 OMITTED]

Because not all subjects in the stretching group demonstrated increases in hamstring muscle length, we decided to do a post hoc analysis on those subjects with increases in AKE angle greater than 10 degrees (11/19). Paired t tests were performed to compare pretest and posttest values in these 11 subjects. A paired t test showed that there was no change in standing posture after stretching. There was an increase of 3.9 degrees (P=.026) in hip motion during early forward bending. The mean decrease of 2.9 degrees in early lumbar motion after the stretching period was not a significant difference. There was an increase of 9.7 degrees in hip motion during late forward bending after stretching (P<.001) and a 13.5-degree increase in hip motion during total forward bending (P<.001). There were no differences found in lumbar motion during the late phase of forward bending.

The results of the Pearson product-moment correlations of hamstring muscle length (AKE test) with standing posture and lumbar and hip motions during forward bending are presented in Table 5. The correlations between AKE test results and standing lumbar angle and pelvic angle were not significant. A number of correlations between AKE test results and measured motions during forward bending emerged. The AKE test results were correlated with lumbar motion and the ratio of lumbar to hip motion during late forward bending. That is, as hamstring muscle length increased, lumbar motion and the ratio of lumbar to hip motion during late forward bending decreased. In addition, significant correlations of AKE test results with hip motion and the ratio of lumbar to hip motion during total forward bending existed (r= -.55 and .50, respectively). As hamstring muscle length increased, as indicated by the AKE test, hip motion increased and the ratio of lumbar to hip motion decreased during total forward bending.

[TABULAR DATA 5 OMITTED]

Discussion

Effect of Hamstring Muscle Stretching on Hamstring Muscle Length and Standing Posture

Both the SLR and AKE tests were used to assess hamstring muscle length before and after a stretching program. Both measures showed that hamstring muscle lengthening occurred in the stretching group. We chose to use both methods because we were unsure which method would be most sensitive. Although both tests indicated changes, we preferred the AKE test because motion at the lumbar spine and pelvis is less likely to influence measurement of the knee angle measured during the AKE test.

The F tests indicated that there was no difference between standing lumbar lordosis and pelvic inclination in the control and stretching groups. This finding does not support the theory that short hamstring muscles cause posterior pelvic inclination and a concomitant reduction in the lumbar lordosis.[6]

The data from this study do not support a direct relationship between hamstring muscle length and standing lumbar and pelvic posture, and the data agree with other research findings of no correlation between postural alignment and muscle length.[8,9] Furthermore, Toppenberg and Bullock[7] found that hamstring muscle length was negatively related to the lumbar curve. The same relationship was noted in our study, but it was not significant. These data do not support Kendall and McCreary's[6] assumption that individuals with flat backs or reduced lumbar curves in a standing position tend to have short hamstring muscles. It is possible that if we had included subjects with longer hamstring muscles (SLR >70[degrees]), and thus had a broader range of hamstring lengths in our sample, we may have found a relationship between hamstring muscle length and lumbar curve. We believe, however, that if hamstring muscle length is related to standing lumbar posture, the stretching group would have shown a change in standing lumbar posture. We are not aware of data that demonstrate a strong relationship between hamstring muscle length and lumbar posture in a standing position.

The degree to which the hamstring muscles act on the pelvis probably depends on the knee and hip angles and the inherent muscle length. Even short hamstring muscles may be placed on slack when standing with the hips in a neutral position and the knees in extension; in this position, a change in hamstring muscle length may cause no change in pelvic or lumbar posture. Therefore, this slack position may be the reason lengthened hamstring muscles caused no changes in standing posture. Another reason may be that posture is dependent on multiple factors (eg, muscle length, lumbosacral angles, postural awareness, motor programming) and will not change significantly with a change in any one variable. In any case, these data suggest that lengthening the hamstring muscles will not change standing postural alignment.

Effect of Hamstring Muscle Stretching on Lumbar and Hip Motion During Forward Bending

Stretching the hamstring muscles resulted in greater hip motion but did not cause less lumbar motion during forward bending. In the 11 subjects in the stretching group who had an increase in hamstring muscle flexibility of greater than 10 degrees, hip motion during early forward bending increased significantly after stretching. Because subjects were instructed to achieve the same partial forward-bending position, the total range of early forward bending should be the same at both the prestretching and posts/retching sessions. Greater hip motion, therefore, should be coupled with less lumbar motion during early forward bending. Lumbar motion was reduced in the same 11 subjects, even though the amount was not significant. Lumbar motion during total forward bending also decreased in these same 11 subjects, although again the amount was not significant. One possible explanation for the lack of significance was sample size. These data, therefore, neither confirm nor disprove tine' idea that an increase in the amount of hip flexion during forward bending results in a decrease of lumbar motion; these data do, however, indicate a trend. A follow-up study, with either more subjects or greater increases in hamstring muscle flexibility, is warranted.

In this study, shorter hamstring muscle length was correlated with an increase in the ratio of lumbar to hip flexion during both late and total forward bending. That is, shorter hamstring muscle length is associated with greater relative lumbar motion during late forward bending.

Hamstring muscle length, however, was not correlated with total lumbar motion during full forward bending. This finding is consistent with those of other studies in which low degrees of correlation between hamstring muscle length and spinal motion during forward flexion were found.[5,18]

Even though some of the subjects (15/39 or 38.5%) had experienced LBP before the study, most subjects (24/39 or 61.5%) had no history of LBP. As a result, application of these data to patients with LBP should be limited. We compared the pretest data for hamstring muscle length, standing posture, and forward bending motion variables between subjects with and without a history of LBP. Subjects with a history of LBP had slightly tighter hamstring muscles, but there was no evidence that either standing posture or forward bending differed between subjects in these two categories. We believed that subjects without LBP were appropriate for the questions posed in this study because most authors claim that hamstring muscle tightness is a factor associated with posture and abnormal motion regardless of whether a history of LBP or current LBP exists.[2,6]

Our results agree with those of some other studies. Biering-Sorensen[18] found diminished hamstring muscle flexibility as a residual sign of low back problems. Day et al19 found there was no difference in lumbar lordosis and pelvic inclination between subjects with and without low back dysfunction in a relaxed standing posture. Thus, attempts to alter standing lumbar and pelvic posture for individuals with a history of LBP are questionable. Results of studies addressing the relationship between spinal mobility and a history of back pain have been varied. Some researchers[12,18,20] have reported a general decrease in spinal mobility associated with a history of back problems, whereas Howes and Isdale[21] have reported increased spinal mobility. Our study showed that a history of back pain had no effect on spinal flexion ROM. The number of subjects with a past history of LBP (n=15), however, was limited in our study. In recent work, we compared subjects with and without a history of LBP and did not find differences between these two groups in the amount of lumbar or hip motion during forward bending.[22]

Further research could be directed toward studying subjects who have chronic LBP. Studying greater numbers of subjects who demonstrate a more dramatic increase in hamstring muscle flexibility after a stretching program also may reinforce and clarify our findings. Our method involved essentially static measurements. Future studies, utilizing continuous sampling techniques for dynamic movement, also may provide greater insight into the effects of a stretching program.

Conclusions

This study has provided information basic to the understanding of the normal relationships between hamstring muscle length, standing posture, and motions during forward bending. This information can act as a guide to clinicians when designing treatment programs.

The data gathered in this study suggest that a 3-week program of hamstring muscle stretching (1) will not alter standing lumbar and pelvic postures, (2) will produce greater forward bending as a result of increased motion at the hips, and (3) may alter the pattern of lumbar and hip motion during forward bending.

(*) Faro Technologies Inc, 125 Technology Park, Lake Mary, FL 32746.

References

[1] Rothman RH, Simeone FA. The Spine, Volume 1. Philadelphia, Pa: WB Saunders Co; 1975:56-58.

[2] Cailliet R. Low Back Pain Syndrome. 3rd ed. Philadelphia, Pa: FA Davis Co; 1980.

[3] Willis TA. Anatomical variations and roentgenographic appearance of the low back in relation to sciatic pain. J Bone Joint Surg. 1941;23: 410-416.

[4] Frymoyer JM, Hanley E, Howe J. Disc excision and spine fusion in the management of lumbar disc disease: a minimum ten-year follow-up. Spine. 1978;3:1-6.

[5] Mellin C. Correlations of hip mobility with degree of back pain and lumbar spinal mobility in chronic low-back pain patients. Spine. 1988; 13:668-670.

[6] Muscle function in relation to posture. In: Kendall FP, McCreary EK, eds. Muscles: Testing and Function. 3rd ed. Baltimore, Md: Williams & Wilkins; 1983:269-320.

[7] Toppenberg R, Bullock M. The interrelation of spinal curves, pelvic tilt and muscle length in the adolescent female. Australian Journal of Physiotherapy. 1986;32:6-12.

[8] Heino JG, Godges JJ, Carter CL. Relationship between hip extension range of motion and posture alignment. J Orthop Sports Phys Ther. 1990;12:243-247.

[9] Flint MM. Lumbar posture: a study of roentgenographic measurement and the influence of flexibility and strength. Res Q. 1963;34:1522.

[10] Thurston AJ. Spinal and pelvic kinematics in osteoarthrosis of the hip joint. Spine. 1985;10:467-471.

[11] Offierski CM, Macnab I. Hip-spine syndrome. Spine. 1983;8:316-321.

[12] Mayer TG, Tencer AF, Kristoferson S, Mooney V. Use of noninvasive techniques for quantification of spinal range-of-motion in normal subjects and chronic low-back dysfunction patients. Spine. 1984;9:588-595.

[13] Markos PD. Ipsilateral and contralateral effects of proprioceptive neuromuscular facilitation techniques on hip motion and electromyographic activity. Phys Ther. 1979;59:1366-1373.

[14] Medeiros JM, Smidt GL, Burmeister LF, Soderberg GL. The influence of isometric exercise and passive stretch on hip joint motion. Phys Ther. 1977;57:518-523.

[15] Ekstrand J, Wiktorsson M, Oberg B, Gillquist J. Lower extremity goniometric measurements: a study to determine their reliability. Arch Phys Med Rehabil. 1982;63:171-175.

[16] Gajdosik R, Lusin G. Hamstring muscle tightness: reliability of an active-knee-extension-test. Phys Ther. 1983;63:1085-1090.

[17] Shrout PE, Fleiss JL. Intraclass correlations: uses in assessing rater reliability. Psychol Bull. 1979;86:420-428.

[18] Biering-Sorensen F. Physical measurements as risk indicators for low-back trouble over a one-year period. Spine. 1984;9:106-119.

[19] Day JW, Smidt GL, Lehmann T. Effect of pelvic tilt on standing posture. Phys Ther. 1984;64:510-516.

[20] Burton AK, Tillotson KM, Troup JDG. Variation in lumbar sagittal mobility with low-back trouble. Spine. 1989;14:584-590.

[21] Howes RG, Isdale IC. The loose back: an unrecognized syndrome. Rheum Phys Med. 1971;11:72-77.

[22] Esola MA, McClure PW, Fitzgerald GK, Siegler S. Analysis of lumbar spine and hip motion during forward bending in subjects with and without a history of significant low back pain. Spine. 1996;21:71-78.

* Invited Commentaries

Following are two invited commentaries on "The Effect of Hamstring Muscle Stretching on Standing Posture and on Lumbar and Hip Motions During Forward Bending."

Li and colleagues are to be commended on their interesting article examining the relationships among hamstring muscle length, standing pelvic alignment, and motion at the lumbar spine and hip during forward bending. These relationships, their effect on function, and changes resulting from exercises are critical to substantiating the role of physical therapy in treatment. My questions and comments about this study pertain to the methodology and the suggested implications of the findings.

One purpose of this study was to examine an assumption attributed to Kendall that short hamstring muscles will cause posterior pelvic tilt and a flat-back standing alignment. Although the investigators classified their sample by using the straight-leg-raising (SLR) test advocated by Kendall, they failed to adhere to the guidelines she advocates for accuracy. She emphasizes the importance of using a hard table, making sure the lumbar spine is flat, accommodating for any hip flexor shortness, and holding the contralateral thigh down, while performing the SLR.[1(pp148-153)] The methods described by Li et al did not address any of these important criteria for accurate testing. Kendall also recommends that the SLR be taken to the point of resistance and a feeling of slight discomfort. Failure to stabilize the contralateral lower extremity to prohibit pelvic and lumbar motion and failure to stretch to the point of resistance raise further questions about accurate assessment of hamstring muscle length. Certainly, the reliability reported by the investigators is impressive, but the issue is accuracy not reliability.

I was interested that Li and colleagues used the measurements from the SLR to group subjects but did not use this same measure for correlational analysis. The second measure of active knee extension (AKE) was used for analysis, which the investigators explained by the statement, "... we preferred the AKE test because motion at the lumbar spine and pelvis is less likely to influence measurement of the knee angle measured during the AKE test." Such a statement suggests that the SLR was influenced by motion at these segments. If this is the case, then the AKE should have been used to classify the subjects or a correlation between the SLR and AKE measurements would have established their relationship.

I am confused by the values in Table 4. The lumbar/hip ratios for early and late phases do not seem accurate when the means for the separate values are calculated, although those for total motion are correct based on the mean calculations.

The lack of correlation between hamstring muscle length and standing alignment measurements is not surprising. Such a correlation implies a very precise relationship between the length of the hamstring muscles and the angles of the pelvis and the spine. Such a precise relationship cannot be expected for anatomical, physiological, and biomechanical reasons. Finding such a correlation is not a prerequisite to finding short hamstring muscles in individuals with a flat back. The data in Table 3 indicate that very few subjects had a flat back before stretching and that no subjects had a flat back after stretching. Scatter plots of the distributions for hamstring muscle length and lumbar and pelvic angles would have been helpful to the reader. Although these correlations were not significant, the sample size was very small, particularly considering that less than 60% of the subjects showed a change in hamstring muscle length. Furthermore, that change was only 10 degrees. The percentage of change for standing pelvic inclination was 80%, which suggests a possible effect if a suitable sample size was used. The investigators should have performed a power analysis to determine a suitable sample size, particularly when such strong conclusions are made.

I do not believe, nor do I believe that Florence Kendall thinks, that in the standing position the hamstring muscle tension is enough to pull the pelvis posteriorly and flatten the lumbar spine. In my opinion, if Kendall believes that the length of the hamstring muscles permits 80 degrees of hip flexion with the lumbar spine flat,[1(p148)] she certainly does not contend that the hamstring muscles that permit 60 degrees of hip flexion, the mean in this study, would posteriorly tilt the pelvis and flatten the back when the hip is extended as in standing.

A more reasonable hypothesis is that an individual who has short hamstring muscles, particularly one who sits with the knees extended and leans forward, would repeatedly stretch the lumbar spine into flexion, increasing its flexibility. Van Wingerden et al,[2] based on an unpublished study, contend that in the sitting position with the knee angle 15 degrees, there is more lumbar motion during the first third of the movement than when the knee angle is 45 degrees. Furthermore, this difference was not evident during forward bending. I also believe that many individuals, particularly men, have a flat back because of structural variation and that the relationship between the posterior superior iliac spine and the anterior superior iliac spine varies because of structural differences, not just soft tissue tension properties, a belief shared by Kendall.[1(p275)] As discussed by Kendall,[1(p273)] the counterbalancing tensions from many muscles are important factors in alignment. The resting tension or stiffness (change in tension/unit change in length) of muscle has not been shown to be correlated with the overall length of a muscle but with the specific length of a defined muscle. Variation is possible because stiffness is a function of the series elastic components arising from intracellular and extracellular contractile proteins and crossbridge weak binding,[3] all of which are affected by hypertrophy or conversely by atrophy.

A useful finding in this study is that in subjects who showed an increase in hamstring muscle length, even a minimal one, there was increased hip motion and a trend to decreased lumbar motion during the forward bend. This type of change could be useful in those patients with low back syndrome whose pain is related to flexion of the lumbar spine. Certainly, it is not surprising that there were no differences between subjects who had a history of back pain and those without a history of back pain. Back pain cannot be considered to have a single mechanical source nor common contributing factors such as short hamstring muscles. I also consider it a giant leap to make a statement such as "attempts to alter standing lumbar and pelvic posture for individuals with a history of LBP are questionable." Anyone who has experience with patients who have low back pain while standing in a marked lumbar lordosis, or with patients with spinal stenosis who are immediately relieved of their pain by contracting their abdominal muscles posteriorly and tilting their pelvis, knows this statement is naive. It is also not substantiated by this study.

A reasonable and useful way to use the information that Kendall has provided in her publications is, as intended, as clues to possible relationships between alignment and muscle characteristics. The patient with back pain and a flat-back posture should be carefully examined for hamstring muscle length as a possible contributing factor in causing pain associated with lumbar flexion, whereas the patient with back pain and a lumbar lordosis is less likely to have short hamstring muscles as a contributing factor.

Shirley A Sahrmann, PhD, PT, FAPTA Associate Professor Program in Physical Therapy Washington University School of Medicine 660 S Euclid Ave St Louis, MO 631 10

References

[1] Kendall FP, McCreary EK. Muscles: Testing and Function. 3rd ed. Baltimore, Md: Williams & Wilkins; 1983.

[2] van Wingerden JP, Vleeming A, Stam, HJ, Stoeckart R. Interaction of spine and legs: influence of hamstring tension on lumbo-pelvic rhythm. In: Proceedings of the 2nd International World Congress on Low Back Pain: The Integrated Function of the Spine and SI Joints. 1996.

[3] Ettema GJC, Huijing PA. Series elastic properties of rat skeletal muscle: distinction of series elastic components and some implications. Neth J Zool. 1993;43:306-325.

Li and colleagues' study attempted to answer three questions: (1) Does hamstring muscle stretching affect measured hamstring muscle length? (2) Does hamstring muscle stretching affect resting lumbopelvic posture? and (3) Does hamstring muscle stretching affect the relative contribution of the lumbar spine and hip flexion to forward bending motion? They examined a group of asymptomatic individuals with hamstring muscle "tightness." The subjects were randomly assigned to group that participated in a 3-week stretching program or to a control group that did not stretch. Standing lumbopelvic posture and lumbar and hip motions during forward bending were measured before and after the intervention. The answers to the questions, based on the data, were "yes," "no," and "maybe." These answers are interesting and, in the case of the second question, predictable.

The most interesting and clinically important finding of this study is that an uncomplicated and easily applied home stretching program actually increased hamstring muscle length, as indicated by an increase of more than 10 degrees in the straight-leg-raising measure, in the majority of subjects in the intervention group. In addition, in those subjects for whom the stretching program was effective, there was a trend for increased hip motion to be matched by a corresponding decrease in lumbar motion.

The lack of a relationship between changes in resting posture and changes in hamstring muscle length is not surprising, given the brevity of the intervention. As Kendall et al state, "Good posture is a good habit...,"[1] and habits are difficult to change. I, too, am skeptical about effects of muscle length and strength on resting posture. I would not, however, expect the acute effects of an exercise program to change something as stereotypical as posture in 3 weeks. Perhaps resting posture can be affected by changes in muscle length, but it would seem that attempts to change the posture would have to accompany the stretching intervention.

Caution needs to be applied in extrapolating these data to the following questions: (1) Is resting lumbopelvic posture related to hamstring muscle length? and (2) Is the relative contribution of lumbar spine and hip flexion to forward bending motion related to hamstring muscle length? These questions cannot be answered by this study. The sample population was small, and there is insufficient statistical power to conclude that the lack of a statistical difference would hold up with a larger sample size. The study only included subjects with hamstring muscle "tightness" and was not representative of the breadth of hamstring muscle length values that would be necessary to establish normative relationships. The authors identified these limitations and yet concluded (in the abstract) that "[t]he results of this study suggest that there is no relationship between hamstring muscle length and resting lumbopelvic posture [emphasis added]." The results of this study suggest only that a hamstring muscle stretching program alone has no effect on standing lumbopelvic posture. In a standing posture, the hamstring muscles are in a shortened position at the hip. Although I concur that it is unlikely that shortening or lengthening of a biarticular muscle such as the hamstrings would result in changes in lumbopelvic posture, this study did not test, nor can it provide information about, that relationship.

Lynn Snyder-Mackler, ScD, PT Associate Professor Department of Physical Therapy University of Delaware Newark, DE 19716

References

[1] Kendall FP, McCreary EK, Provance PG. Muscles: Testing and Function. 4th ed. Baltimore, Md: Williams & Wilkins; 1993:3.

* Author Response

Dr Sahrmann suggests that our method of straight-leg-raising (SLR) testing may have been flawed because we did not use a "hard table" and did not stabilize the contralateral lower extremity as described by Kendall. We are unaware of any data that support the biases written in Kendall's text[1] and shared by Dr Sahrmann or any methods that provide more accurate (valid) data than the data obtained by our method. We believe that our method reflects a theoretically sound technique that is commonly used by clinicians. Our reliability data were highly repeatable over the time interval used in our study. Dr Sahrmann also states that we failed to "stretch to resistance." We operationally defined the endpoint for the test as the point at which the subject felt a "stretching sensation," and based on our reliability findings, this is probably a reasonable way to define an endpoint.

Although the relationship between the SLR and active-knee-extension (AKE) tests was not the focus of our study, Dr Sahrmann raises the issue of differences between the two tests. At the beginning of the study, we were not certain which test might yield the most meaningful information and felt it was a simple matter to perform both tests. We compared both the SLR and AKE tests with regard to postural and motion variables and found essentially the same relationships. We chose to report only the AKE findings for simplicity. We had also assessed the correlation between the SLR and AKE tests, which was .72. A relatively modest correlation of .72 ([r.sup.2]=.52) suggests that the tests are not measuring exactly the same thing. The amount of change in each measure after the stretching period was similar, and the correlation between each of these measures of hamstring muscle length and the various postural and motion variables was also quite similar. Our bias is that the subtle movements of the pelvis and lumbar spine during the SLR test may contribute to the differences between the SLR and the AKE tests; however, we question the ability of examiners to reliably detect and measure these subtle movements. Another obvious difference is that the SLR is a passive test and the AKE is dependent on quadriceps femoris muscle activity.

The apparent discrepancy for the lumbar/hip ratios in Table 4 alluded to by Dr Sahrmann is a function of the methods used to calculate the mean values for lumbar/hip ratio. We chose to use the mean of the individual ratios calculated for each subject rather than calculate a single ratio based on group means of lumbar and hip motion. We believe that this measurement is more representative of the variable of interest. Simply dividing the mean amount of lumbar motion for a group by the mean hip motion for the group would obscure the variation present among individuals in the lumbar/hip ratio.

Both Dr Sahrmann and Dr Snyder-Mackler suggest that our sample size was too small to make any statements about the relationship between posture and hamstring muscle length and that a larger sample size may have produced correlation coefficients that were statistically significant. Although this may be the case, our findings agree with those of other investigators,[2,3] who did not find a relationship between lumbopelvic posture and lower-extremity flexibility measures. Toppenberg and Bullock[4] found that short hamstring muscles were associated with increased lumbar lordosis, which is opposite that predicted by Kendall.[1] The correlation between hamstring muscle length and pelvic inclination in our study was .277. This value would be statistically different from 0 with a sample size of 50 ([[alpha].sub.2]=.05); however, we question the meaningfulness of such a weak relationship. The coefficient of determination ([r.sup.2]) would be .08, implying that only 8% of the variance in pelvic inclination could be explained by a change in hamstring muscle length. Portney and Watkins indicated that the statistical significance of a correlation coefficient is very biased by sample size, stating:

With a sufficient increase in sample size, almost any

observed correlation value will be statistically significant,

even if it is so small as to be a meaningless indicator of

association. Low correlations should not be discussed as

clinically important just because they have achieved statistical

significance. Such interpretations should be made only

on the basis of the magnitude of the correlation

coefficient.[5(pp444)]

Likewise, Dr Sahrmann suggests that the 2.5-degree increase (80%) in pelvic inclination may be statistically significant with a larger sample size. Using the effect size obtained in our study for pelvic inclination (0.25), our calculations suggest that a sample size of 123 would be required to show a difference with [[alpha].sub.2]=.05 and power (1 - [beta])=.8, assuming a repeated-measures design. We question the meaningfulness of a 2.5-degree change and continue to question the relationship between posture and hamstring muscle length until other, more convincing, data are available. Thompson[6] has persuasively argued that with enough subjects, researchers can find statistically significant differences (low probability values) between almost any sets of data, yet low probability values (ie, P<.05) in no way ensure the differences are meaningful. He and others have argued that effect size is a better indicator of the differences between groups than are tests of statistical significance, primarily because effect size is not biased by sample size.

Contrary to Dr Sahrmann's implication, Kendall, describing the standing posture of individuals with the flat-back type of posture, states that "the most constant finding is a tightness of the hamstring muscles which pulls the pelvis into a posterior pelvic tilt."(p285)] We believe that those who hold strongly to the belief that muscle length and posture are related, and teach the principles underlying this belief, have the responsibility to produce data to support this contention.

Dr Sahrmann points out that it is often helpful to teach patients with low back pain to alter their posture by conscious, active movements to alleviate symptoms, and we agree wholeheartedly. Perhaps our statement that "attempts to alter standing lumbar and pelvic posture for individuals with low back pain are questionable" was misleading. Obviously, we studied only the effect of hamstring muscle stretching on standing posture, and our statement should be interpreted in that context. For example, improvement in symptoms is commonly observed after correcting a lateral shift posture in patients with low back pain. We did not mean to imply that any attempts at altering lumbar or pelvic position to alleviate symptoms are wrong or misguided.

Dr Snyder-Mackler brings up the interesting point that posture is stereotypical and probably not amenable to change within a 3-week period. We agree with her; however, 3 weeks is probably close to the amount of time therapists typically have to work closely with a patient in attempts to alter posture. If the goal is to truly permanently alter standing lumbar or pelvic posture, how long is required?

Finally, our study was an attempt to investigate kinematic mechanisms associated with a commonly used treatment technique and did not address the usefulness of hamstring muscle stretching as a treatment. In addition to the urgent need for outcome type studies that document efficacy of treatment, we must continue to try to understand mechanisms associated with our treatments. This should lead to refinement of existing theories, development of new theories, and, ultimately, more scientifically based practice.

Yenchen Li, PT Philip W McClure, PT, OCS Neal Pratt, PhD, PT

References

[1] Kendall FP, McCreary EK, eds. Muscles: Testing and Function. 3rd ed. Baltimore, Md: Williams & Wilkins; 1983.

[2] Heino JG, Godges JJ, Carter CL. Relationship between hip extension range of motion and posture alignment. J Orthop Sports Phys Ther. 1990;12:243-247.

[3] Flint MM. Lumbar posture: a study of roentgenographic measurement and the influence of flexibility and strength. Res Q. 1963;34:15-22.

[4] Toppenberg R, Bullock M. The interrelation of spinal curves, pelvic tilt, and muscle length in the adolescent female. Australian Journal of Physiotherapy. 1986;32:6-12.

[5] Portney L, Watkins MP. Foundations of Clinical Research: Applications to Practice. East Norwalk, Conn: Appleton & Lange; 1993.

[6] Thompson B. AERA editorial policies regarding statistical significance testing: three suggested reforms. Educational Researcher. 1996;25: 26-30.

Y Li, PT, is Physical Therapist, Chang Gung Memorial Hospital, Taiwan, Republic of China. She was a student in the Department of Orthopedic Surgery and Rehabilitation, Hahnemann University, Philadelphia, PA, 19102, when this study was completed in partial fulfillment of the requirements for her Master of Science degree.

PW McClure, PT, OCS, is Assistant Professor, Department of Physical Therapy, Medical College of Pennsylvania and Hahnemann University, MS 502, Broad and Vine Streets, Philadelphia, PA 19102 (USA) ([e-mail]). Address correspondence to Mr McClure.

N Pratt, PhD, PT, is Professor, Departments of Physical Therapy and Anatomy, Medical College of Pennsylvania and Hahnemann University, and Director, Orthopaedic Physical Therapy Program, Medical College of Pennsylvania and Hahnemann University.

This study was approved by the Medical College of Pennsylvania and Hahnemann University Institutional Review Board.

This article was submitted July 5, 1995, and was accepted February 27, 1996.