Scand J Med Sci Sports 2010: 20: 136144 doi: 10.1111/j.1600-0838.2009.00926.

& 2009 John Wiley & Sons A/S

Regular stretch does not increase muscle extensibility: a randomized controlled trial
M. Ben1,2, L. A Harvey1,2
Rehabilitation Studies Unit, Northern Clinical School, Faculty of Medicine, University of Sydney, Sydney, Australia, 2Physiotherapy Department, Royal Rehabilitation Centre Sydney, Sydney, Australia
1

Corresponding author: Dr Lisa Harvey, Rehabilitation Studies Unit, University of Sydney, PO Box 6, Ryde, NSW, Australia 1680. E-mail: l.harvey@usyd.edu.au
Accepted for publication 8 January 2009

The purpose of this randomized controlled trial was to determine whether regular stretch increases hamstring muscle extensibility. Sixty healthy individuals were randomly allocated to an experimental or a control group. The experimental group attended supervised stretch sessions where they stretched the hamstring muscles of a randomly allocated leg for 30 min ve times a week for 6 weeks. The control group did not attend any stretch sessions during this period. No participants withdrew from the study. A purpose built device was used to measure passive hip exion during a straight leg-raise manoeuvre with the application of a standardized and non-standardized stretch torque. The

stretch intervention did not increase passive hip exion when measured with a standardized stretch torque [mean treatment eect 5 11; 95% condence interval (CI) 5 3 to 21]. It did, however, increase passive hip exion when measured without a standardized stretch torque (mean treatment eect 5 101, 95% CI 5 6141). Six weeks of sustained 30-min daily stretch does not increase the extensibility of the hamstring muscle of healthy individuals. It does, however, improve stretch tolerance leading to increased joint range of motion without any actual improvements in muscle extensibility.

Stretches are routinely incorporated into physical training programs (Bandy & Irion, 1994; Bandy et al., 1997, 1998; Chan et al., 2001; Gajdosik et al., 2005). There is good evidence from animal studies to support the widespread use of stretch. These studies indicate that the muscles of rabbits, cats, guinea pigs and rats are adaptable and remodel in response to prolonged sustained stretch (Tabary et al., 1972; Tardieu et al., 1977; Huet de laTour et al., 1979; Spector et al., 1982; Herbert & Balnave, 1993). In these studies, stretch is typically administered continuously through the application of plaster casts for days and even weeks, although in one notable study stretch was administered for as little as 15 min a day (Williams, 1990). These studies indicate that sustained stretch increases the number of sarcomeres in series and stimulates rearrangement of collagen (Goldspink et al., 1974; Huet de laTour et al., 1979). These changes demonstrate the adaptable nature of muscle tissue necessary for optimal force production and usually reverse once the stimulus is removed (Timson, 1990). Associated with the structural and morphological adaptations are changes in the passive mechanical properties of muscles, notably changes in extensibility (Tabary et al., 1972; Goldspink et al., 1974). Changes in muscle extensibility are reected

by changes in joint angle with the application of a standardized torque. It is not clear whether the muscles of humans respond similarly to the muscles of small animals to the application of continuous sustained stretch. Several studies on humans provide initial evidence about the eectiveness of stretch (Gajdosik, 1991; Lentell et al., 1992; Bandy & Irion, 1994; Bandy et al., 1997, 1998; Chan et al., 2001; Draper et al., 2004; Gajdosik et al., 2005). Interestingly, unlike the animal trials, the majority of studies in humans administer stretch for as little as 30 s a day. However, these ndings need to be interpreted with caution. Often changes in extensibility are measured immediately upon removal of stretch and may primarily reect viscoelastic deformation (Bohannon, 1984; Kirsch et al., 1995; Magnusson et al., 1996c). These studies do not provide evidence about the lasting eects of stretch. In addition, the majority of trials in this area do not blind assessors nor adhere to the design features of clinical trials important for minimizing bias (Schulz et al., 1995; Schulz & Grimes, 2006). Perhaps of more concern though is the failure of trials to measure joint angle with a known torque (Starring et al., 1988; Godges et al., 1993; Bandy & Irion, 1994; Bandy et al., 1997). It is important to

136

Regular stretch does not increase muscle extensibility

relate joint angle to torque because joint angle is a direct function of applied torque. If torque is not measured and joint angle is solely determined by participants feelings of discomfort, increases in joint angle following stretch programs may reect changes in participants willingness to tolerate discomfort rather than underlying changes in extensibility (Halbertsma & Goeken, 1994; Halbertsma et al., 1996; Magnusson et al., 1996b; Bjorklund et al., 2001; Folpp et al., 2006). Several recent trials, including those on people with dierent types of disabilities, suggest that stretch does not increase muscle extensibility (Halbertsma & Goeken, 1994; Halbertsma et al., 1996; Magnusson et al., 1996b; Harvey et al., 2000, 2003a, 2006; Bjorklund et al., 2001; Lannin & Herbert, 2003; Ben et al., 2005; Moseley et al., 2005; Turton & Britton, 2005; Refshauge et al., 2006; Lannin et al., 2007) but rather merely improves peoples tolerance to the discomfort associated with stretch. This leads to increased joint range of motion without any actual improvements in muscle extensibility (Halbertsma & Goeken, 1994; Halbertsma et al., 1996; Magnusson et al., 1996a, b; Bjorklund et al., 2001; Folpp et al., 2006). However, it is possible that trials have not administered stretch for long enough to induce lasting changes in extensibility. Most trials, particularly those in able-bodied individuals administer stretch for just a couple of minutes a day. In addition, one trial used a withinsubjects design (Folpp et al., 2006). This is problematic because the eects of stretch could be missed if stretch administered to one leg had an eect on the untreated contralateral leg. Therefore, the aim of this study was to further investigate this issue and determine whether a stretch program administered for 30 min a day over 6 weeks increases hamstring muscle extensibility or merely changes participants tolerance to the discomfort associated with stretch. The secondary aim was to determine whether stretching one leg had any eects on the contralateral untreated leg. For this purpose a combination of a within- and between-subjects design was used where some participants stretched one leg and other participants stretched neither leg. It was hypothesized that if the eects of stretch are unilateral then stretch applied to one leg would have no eect on the contralateral untreated leg. Methods
Participants
Sixty people participated in the study. All participants were sta from a Sydney hospital and aged over 18 years. They were enrolled into the study by the principal investigator. The use of hospital sta provided a sample representative of generally active, healthy people and avoided the inclusion of professional ballerinas, gymnasts or other sports personnel. Participants were excluded if they were able to place the palms of their hands at on the oor in the modied ngertip-tooor test (Gauvin et al., 1990). This criterion was used as a crude way to exclude individuals with extensible hamstring muscles. Also excluded were those with back and/or lower limb pain exacerbated by hamstring stretches and those unable to tolerate the measurement procedure. All participants were asked to avoid taking up new exercise-related activities for the duration of the study period although they were permitted to continue with their current exercise regimes. A minimal between-group dierence of 51 in hip exion with the application of a standardized torque was selected a priori as a clinically worthwhile treatment eect. This is consistent with other investigators (Harvey et al., 2003a; Youdas et al., 2003; Moseley et al., 2005; Folpp et al., 2006; Refshauge et al., 2006). Power calculations indicated that a sample size of 60 people would be sucient to provide 95% probability of detecting a 51 change in hip exion, assuming standard deviation of 51, (Folpp et al., 2006) a of 0.05 and loss to follow-up of 10%. The study received ethical approval from Royal Rehabilitation Centre Sydney and University of Sydney Human Research ethics committees. Informed consent was obtained from all participants. All applicable institutional and governmental regulations concerning the ethical use of human volunteers were followed during the course of this research. The trial was registered before its commencement with Australian Clinical Trial Registry (registration number ACTRN012605000281695).

Study design
A single-blind randomized controlled trial using a combination of a within- and between-subjects design was undertaken. The participants were randomly allocated to a control or experimental group (between-subject component of the trial). Participants in the control group received no stretch intervention to either leg. Those in the experimental group, however, had each of their legs further randomized to either a stretch or nonstretch group (within-subject component of the trial). Thus, three groups were formed which for simplicity will be referred to as (i) control group, (ii) experimental-stretch group and (iii) experimental-non-stretch group (see Fig. 1). A computer-generated randomization schedule was produced by a person independent of the study. The allocation was concealed by sequentially numbered opaque envelopes which were opened by the principal investigator following initial measurements. Participants were considered to have entered the trial at this point. The within-subject component of the trial examined the eectiveness of stretch on passive hip exion with and without a standardized stretch torque. Each subject acted as their own control thereby minimizing between-group variability due to personal factors such as sport participation. However, a limitation of a within-subject design is that it does not account for the possibility of contralateral treatment eects. That is, stretch applied to one leg may increase passive hip exion in the untreated contralateral leg. In this scenario, a within-subjects design would fail to detect a treatment eect because both legs would improve with the intervention. The between-subject component of the trial was used to check for this possibility.

Outcome measures
The two primary outcome measures were passive hip exion with and without the application of a standardized hip exor torque. The standardized torque was 18 Nm and was selected because it was the largest torque all participants were likely to

137

Ben & Harvey

60 participants recruited (120 legs)

Baseline measurements

Concealed random allocation to:

BETWEEN-SUBJECT COMPONENT

Control group, n=60 legs

Experimental group, n=60 legs

Concealed random allocation to:

WITHIN-SUBJECT COMPONENT

Stretch group, n=30 legs

Non-stretch group, n=30 legs

All legs received control or experimental interventions according to group allocation

Post-intervention measurements n= 120 legs

Fig. 1. Flow of participants and study design.

tolerate (Folpp et al., 2006). It was important to select a torque that all participants could tolerate so results between participants could be compared. The non-standardized torque was determined by participants. That is, passive hip exion was measured with the highest torque participants were willing to tolerate as is commonly done in clinical practice and clinical trials (Halbertsma et al., 1996; Magnusson et al., 1996b; Folpp et al., 2006). The secondary outcome measure was pain intensity at the point of the highest tolerated stretch torque and was used to determine participants perceptions of discomfort. Participants were asked to rate their pain on an 11-point visual analogue scale (VAS), with zero indicating no pain and ten indicating the worst possible pain (Bijur et al., 2001). Passive hip exion was measured during a straight leg raise manoeuvre with a device previously described and tested for reliability (ICC 5 0.98; see Fig. 2) (Harvey et al., 2003b). The device has three important features, namely it provides a way of stabilizing the knee in extension, ensures the hip moves in a sagittal plane and provides a way of quantifying the applied hip exor torque. Furthermore, the device negates the torque due to the mass of the leg by a counterweight system. In this way, the applied hip exor torque is solely directed at stretching the hamstring muscles and not inuenced by the mass of a participants leg.

Measurement procedure
Assessors, blinded to group allocation, tested the participants on two occasions at a physiotherapy gymnasium within a

Sydney hospital. Initial measurements were taken 17 days before the beginning of the study. The nal measurements were taken at the completion of the 6-week study period. All experimental participants were tested at least 24 h and not more than 4 days (mean 5 2 days, SD 5 1 day) after the last stretch intervention. This was important for ensuring measurements reected the lasting and not transient eects of the stretch intervention. Measurement procedures always followed the same protocol. Participants were prevented from seeing their legs during testing by a screen placed across their chests. This helped ensure that participants responded to sensation rather than visual cues of leg position. The right leg was measured before the left leg. Initially a standardized hip exor torque of 18 Nm was applied for 3 min to exhaust viscoelastic tissue deformation (Bohannon, 1984; Kirsch et al., 1995; Magnusson et al., 1996c). Following measurements with the standardized 18 Nm torque, passive hip exion was again measured but with a non-standardized torque. For this part of the test, the stretch torque was increased from the original 18 Nm by 6 Nm increments every 30 s. However, between each 6 Nm increment, small random increases and decreases in torque were applied. These were included to minimize participants ability to predict the testing protocol. Once participants indicated that they were not willing to tolerate further increases in torque, the last 6 Nm increment was removed and the torque was increased by 1.5 Nm every 10 s. Participants were again requested to indicate when they had reached their maximal stretch tolerance at which point passive hip exion and the

138

Regular stretch does not increase muscle extensibility

to the control group did not perform any hamstring stretches during the study period, nor have any contact with research sta.

Data reduction and analysis

Results from the within-subject component of the trial were used to determine the eectiveness of the stretch intervention. Thus the experimental-stretch legs were compared with the experimental-non-stretch legs. Mean changes in hip exion angle with and without a standardized torque from initial to nal measures were calculated for the stretched and nonstretched legs. The t-distribution was used to estimate 95% condence intervals (CI) for the dierences in change scores. A positive change reected a relative increase in passive hip exion with the stretch intervention. Similarly, mean changes in pain intensity at the highest torque tolerated by participants for each leg were calculated. A positive change in pain intensity scores indicated participants increased willingness to tolerate higher levels of pain following the stretch intervention. Results from the between-subject component of the trial were analyzed in the same way but used to determine whether stretch administered to one leg had contralateral eects on the other non-stretched leg. Thus the experimental-non-stretch leg was compared with the control legs (mean values were calculated from the measurements of both control legs and used for the analyses). The principles of intention-to-treat were used for all analyses. That is, all participants data were analyzed according to their initial group allocation (Pocock, 1983).

Fig. 2. Device used to assess passive hip exion with and without a standardized stretch torque. A wheel with an attached leg splint (S) was secured to the side of the bed. Weights (W) were hung from the rim of the wheel to produce a hip exion torque. The torque due to weight of the splinted leg was eliminated by using a counterweight (CW) (image reproduced with permission from http://www.physiothera pyexercises.com.au).

Results No participants withdrew from the study. The protocol dictated that participants receive 30 stretch interventions over 42 days and that at least 18 of the 30 stretch interventions be supervised. The overall compliance was excellent and the participants received a mean of 29 stretch interventions (SD 5 2) over 41 days (SD 5 9). An average of 24 stretches interventions (SD 5 5) were supervised, and 5 (SD 5 3) were unsupervised for each participant. Participants characteristics are provided in Table 1. There were no statistically signicant between group dierences in passive hip exion with or without a standardized torque at the beginning of the trial (Tables 2 and 3; Figs 3 and 4). Several participants from each group were involved in regular recreational or sporting activities such as walking, jogging, weight training, cycling and various team sports. The level of participation in sport was similar between groups with 23 of 30 participants regularly exercising in the control group, and 24 of 30 participants regularly exercising in the experimental group. The mean changes in passive hip exion angle and VAS scores measured with and without a standardized torque for the three groups are summarized in Tables 2 and 3. The within-subject analyses (Figs 3 and 4; Table 2) revealed that there was no between-group (i.e., between experimental-stretch and experimental-non-stretch

corresponding torque were recorded. At this point participants were also asked to rate intensity of their pain.

Intervention
Participants in the experimental group were required to stretch the hamstring muscles of the allocated leg continuously for 30 min a day, ve times a week for 6 weeks. At least three and often ve stretches were supervised by one of the study investigators. For this purpose, participants were required to attend a physiotherapy gymnasium within a hospital. The fourth and fth sessions were not always directly supervised. However, compliance with the stretch program was monitored with diaries to record the number of supervised and unsupervised sessions. Diaries were checked at each supervised stretching session. Stretches were self-administered by participants. They were instructed to sit on a chair in an upright position with their backs supported and one leg raised on a height-adjustable examination bed. The knee was extended and the height of the examination bed was adjusted by participants with a handheld electric control. Participants were encouraged to raise the bed as high as possible throughout each 30-min stretch session in order to induce the largest stretch they were willing to tolerate. If the participants became uncomfortable, they were instructed to lower the bed slightly to reduce the stretch, or reposition themselves in the chair. Torques applied during each stretch session were not measured or standardized, however this mimics clinical practice. Participants allocated

139

Ben & Harvey

legs) dierence in the change scores of passive hip exion angle with the application of a standardized stretch torque (mean treatment eect was 11; 95% CI 5 3 to 21 degrees). This indicates no improvement in hamstring muscle extensibility. However, there was a dierence between the groups in passive hip exion angle with the application of non-standardized stretch torque (mean treatment eect was 101; 95% CI 5 6 141). The corresponding mean increase in the applied torque was 10 Nm (95% CI 5 515 Nm). This was not accompanied by any between-group dierence in pain intensity (mean change in VAS 5 0; 95% CI 5 01). These results indicate that stretch improves tolerance to the discomfort associated with stretch enabling participants to endure larger stretch torques for the same perceived level of pain. The between-subject analyses (Figs 3 and 4; Table 3) revealed no between-group dierence (i.e., between experimental-non-stretch and control groups) of change scores in passive hip exion angle with the application of a standardized torque (mean treatment eect was 11; 95% CI 5 5 to 31) and non-standardized torque (mean treatment eect was 41; 95% CI 5 8 to 01). In addition, there were no between-group dierence for pain intensity with the non-standardized stretch torque (mean change in pain score 5 0; 95% CI 5 1 to 1). These results indicate that stretch has no contralateral treatment eects. That is, stretch administered to one leg does not have an eect on the opposite untreated leg.

Discussion Thirty minutes of sustained stretch administered daily for 6 weeks does not increase muscle extensibility in healthy, active, able-bodied individuals. It does, however, improve participants tolerance to the discomfort associated with stretch allowing them to tolerate larger stretch torques. There was no change in pain intensity scores at the point of highest tolerated torque for any group following the 6week trial period. This indicates that the stretch intervention did not simply enable participants to endure higher levels of pain. Rather, the stretch intervention enabled participants to tolerate higher levels of stretch torque for the same amount of perceived pain. The eects of the stretch intervention on tolerance are only evident in the treated leg with no improvements seen in the contralateral untreated leg. The failure to demonstrate a treatment eect on muscle extensibility cannot be explained by insucient subject numbers because the 95% CI is narrow (95% CI 5 3 to 21) and falls well short of the minimally worthwhile treatment eect (i.e. 51). Nor can the negative results be explained by poor compliance. Participants were directly supervised 80% of the time. Participants performed the remaining 20% of stretch sessions in their own time but

Table 1. Number of males and females and mean (SD) age (years), height (cm), weight (kg) and distance between ngers and oor in toetouch test (cm) of participants

Control group (n 5 30 participants)

Experimental group (n 5 30 participants)

Gender Males Females Age (years) Height (cm) Weight (kg) Toe-touch distance (cm) Participants involved in sport and recreational activities

7 23 39 (12) 169 (8) 70 (12) 3 (10) 23

35 168 74 5

9 21 (12) (10) (17) (12) 24

Number of participants involved in regular sport and recreational activities is also included.

Table 2. Results of the within-subjects component of the study showing effects of stretch on muscle extensibility and stretch tolerance following the 6-week stretch program

Experimental-stretch group Pre Post Change

Experimental-non-stretch group Pre Post Change

Between-group difference in change Mean (SD) 95% CI*

Standardized torque (muscle extensibility) Hip exion angle (1) 53 (11) Non-standard torque (stretch tolerance) Hip exion angle (1) 86 (15) Highest tolerated torque (Nm) 54 (17) VAS score 7 (2)

57 (12) 97 (17) 65 (25) 7 (2)

4 (9) 10 (11) 11 (14) 0 (2)

55 (15) 88 (14) 53 (18) 7 (2)

59 (13) 88 (14) 54 (19) 8 (1)

4 (7) 0 (9) 1 (11) 1 (2)

1 (8) 10 (11) 10 (14) 0 (1)

3 to 2 614 515 01

Mean (SD) changes in hip flexion angle (1) with the application of a standardized and non-standardized torques are provided for the experimental -stretch and experimental-non-stretch groups. Mean (SD) change in highest tolerated torque (Nm) and in VAS scores is also provided for the two groups. *Mean and 95% CI data are rounded to the nearest integer. For this reason the 95% CIs do not appear to be perfectly symmetrical about the mean. CI, confidence interval; VAS, visual analogue scale.

140

Regular stretch does not increase muscle extensibility

Table 3. Results of the between-subjects component of the study showing contralateral effects of stretch on muscle extensibility and stretch tolerance following the six-week stretch program

Experimental-non-stretch group Pre Post Change

Control group Pre Post

Between-group difference in change Change Mean (SD) 95% CI*

Standardized torque (muscle extensibility) Hip exion angle (1) 55 (15) Non-standard torque (stretch tolerance) Hip exion angle (1) 88 (14) Highest tolerated torque (Nm) 53 (18) VAS score 7 (2)

59 (13) 88 (14) 54 (19) 8 (1)

4 (7) 0 (9) 1 (11) 1 (2)

54 (11) 60 (13) 6 (8) 92 (15) 96 (16) 4 (7) 56 (16) 61 (19) 5 (11) 7 (2) 7 (2) 0(2)

1 (7) 4 (8) 4 (11) 0 (1)

5 to 3 8 to 0 10 to 2 1 to 1

Mean (SD) changes in hip flexion angle (1) with the application of a standardized and non-standardized torques are provided for the experimental-nonstretch and control groups. Mean (SD) change in highest tolerated torque (Nm) and in VAS scores is also provided for the two groups. *Mean and 95% CI data are rounded to the nearest integer. For this reason the 95% CIs do not appear to be perfectly symmetrical about the mean. CI, confidence interval; VAS, visual analogue scale.

120 100 Hip flexion angle (degrees) 80 60 40 20 0

Pre Post

exp-stretch

exp-non-stretch

control

Fig. 3. Mean (SD) dierences in hip exion angle with application of standardized torque for the experimentalstretch, experimental-non-stretch and control groups.

Fig. 4. Mean (SD) dierences in hip exion angle with application of non- standardized torque for the experimental-stretch, experimental-non-stretch and control groups.

compliance was monitored weekly with diaries by the investigators. Thus, compliance with the stretch program was excellent and probably better than what is typically achieved in the community setting. It is, however, possible that the failure to demonstrate a treatment eect reects the low torque used to test muscle extensibility. The low torque was used to ensure all participants would tolerate it and thereby enable between-group comparisons. However, the results may have been dierent if extensibility had been measured with a higher torque. Future studies could explore this possibility by restricting inclusion to those able to tolerate larger torques. The conclusions made from the results of this study are in stark contrast to the majority of previous studies in able-bodied individuals (Toft et al., 1989; Gajdosik, 1991; Bandy & Irion, 1994; Worrell et al., 1994; Bandy et al., 1997; Chan et al., 2001), which conclude that stretch administered for as little as 30 s a day increases muscle extensibility (Bandy & Irion,

1994; Bandy et al., 1997). This may merely reect a failure to standardize torque when measuring joint angle (Starring et al., 1988; Godges et al., 1993; Bandy & Irion, 1994; Bandy et al., 1997). That is, often joint angle is measured at what investigators deem to be end of range. However, end of range is commonly determined by participants feelings of discomfort at the time of testing and does not necessarily reect muscle extensibility. Studies reporting increases in end of range may therefore merely reect changes in participants willingness to tolerate discomfort. Several recent studies support this explanation and have highlighted the importance of making a distinction between stretch tolerance and muscle extensibility (Halbertsma et al., 1996; Magnusson et al., 1996a, b; Bjorklund et al., 2001; Folpp et al., 2006). The conicting results may be also be due to failure to use non-blinded assessors, a control group (Toft et al., 1989), and concealed allocation (Toft et al., 1989; Gajdosik, 1991; Lentell et al., 1992; Bandy et al., 1997; Chan et al., 2001;

141

Ben & Harvey

Draper et al., 2004); factors which can overstate treatment eectiveness (Schulz et al., 1995). Of course, studies that measure joint angle immediately upon removal of stretch will invariably report a treatment eect. These studies demonstrate the transient eects of stretch from viscous deformation; a well-documented characteristic of soft tissues. These eects are not lasting and dissipate soon after the removal of the stretch (Bohannon, 1984; Kirsch et al., 1995; Magnusson et al., 1996c). The ndings of this trial are comparable to a similar study of our own which used a withinsubjects design (Folpp et al., 2006). A within-subjects design is useful as it minimizes noise due to intersubject variability. However, a disadvantage of solely relying on a within-subjects design is that it can conceal a treatment eect if stretching one leg induces changes in extensibility in both legs. The combination of within- and between-subjects design enabled us to rule out this possibility. That is, stretching the hamstring muscles of one leg does not change the extensibility or stretch tolerance of the contralateral, untreated leg. The improvements in stretch tolerance occur specically in the stretched leg. The underlying physiological mechanisms explaining improved stretch tolerance are not known. Presumably stretch has an inuence on some characteristics in the sensory neural pathways (Laessoe & Voigt, 2003). Alternatively, changes in stretch tolerance may merely reect the use of non-blinded participants with strong pre-conceived ideas about the eectiveness of stretch. Unfortunately it is not possible to blind participants in studies of this kind. However, in an attempt to minimize the eects of participants expectations on outcome, their legs were screened from view and eorts were made to minimize the likelihood of participants predicting and remembering the testing protocol. The results of this trial are consistent with studies involving people with disabilities. Nine high-quality randomized controlled trials (mean quality score 5 7/ 10, SD 5 1) have all failed to nd a treatment eect from stretches applied from 30 min up to 12 h per day, over 412 weeks in patients with spinal cord injuries, stroke, ankle fractures and Charcot-MarieTooth disease (Harvey et al., 2000, 2003a, 2006; Lannin et al., 2003, 2007; Ben et al., 2005; Moseley et al., 2005; Turton & Britton, 2005; Refshauge et al., 2006). In all of these studies, joint angle was measured with a standardized stretch torque ensuring the results reect changes in extensibility. In addition, all studies had sucient subject numbers to detect a treatment eect of 51. Of course the muscles of people with neurological disabilities may not respond to stretch in the same way as the muscles of healthy able-bodied individuals. Although the more likely explanation is that muscles are not as responsive to stretch as commonly assumed, regardless of underling pathology. Thirty minutes of stretch was used in this study because animal studies indicate that stretches sustained for longer periods of time (i.e., 24 h per day) are more eective than stretches sustained for short periods of time (i.e., 15 min per day) (Williams, 1990). Thirty minutes was also chosen because a previous study of our own indicated that 20 min of stretch a day administered over 4 weeks was ineective (Folpp et al., 2006). The eectiveness of stretches administered for slightly longer or shorter periods of time each day or stretches administered regularly over many years is not known. It is possible that we are yet to nd the critical dosage of stretch required to change extensibility. Changes in hamstring muscle extensibility were measured with a standardized 18 Nm torque. The torque due the mass of the leg was counteracted by the testing device. However, the torque was not adjusted according to participants mass. Thus, heavier participants received a relatively smaller testing torque when expressed in proportion to their body mass than lighter participants. Standardizing torque to participants mass was unlikely to have yielded dierent ndings. Stretch is often justied on the basis of strong anecdotal evidence. For example, the excellent extensibility of ballerinas, gymnasts and yoga enthusiasts is commonly attributed to their stretching regimes. However, there may be other explanations for these anecdotal observations. For instances, individuals with good pre-morbid extensibility may self-select to these activities and professions. Alternatively, it may be that most of these individuals administer stretches from a very early age when soft tissues are growing and perhaps more responsive to stretch. The results of this study indicate that regular stretch does not change muscle extensibility in healthy able-bodied individuals. It does, however, change perceptions of discomfort associated with stretch resulting in apparent increases in joint range of motion. In some situations, increases in range of motion alone may be therapeutic. For example, an improved ability to touch the toes may help some sports people, with or without underlying changes in hamstring extensibility. However, where the purpose of stretch is to increase extensibility, the results of this study indicate that human muscles are not as responsive to stretch as previously thought. The general misconception that stretch induces lasting improvements in muscle extensibility may be accounted for by the well-documented transient eects of stretch and the observed changes in stretch tolerance.

142

Regular stretch does not increase muscle extensibility

Perspectives Animal studies indicate that muscles are highly adaptable and gain extensibility in response to stretch. However, recent trials involving healthy able-bodied individuals and people with disabilities indicate that human muscles are not as responsive to stretch as the muscles of small animals (Bjorklund et al., 2001; Folpp et al., 2006; Harvey et al., 2006; Refshauge et al., 2006; Lannin et al., 2007). The results of this study further support these ndings. That is, sustained daily 30-min stretch administered to the hamstring muscles of healthy individuals for 6 weeks does not induce any lasting changes in muscle extensibility. It does improve participants tolerance to the discomfort associated with stretch leading to increased joint range of motion without any actual improvements in muscle extensibility.
Key words: stretch, hamstring, extensibility.

Acknowledgements
The Engineering Department of the Royal Rehabilitation Centre Sydney for assisting with the design of the testing device and for manufacturing the testing device.

References
Bandy WD, Irion JM. The eect of time on static stretch on the exibility of the hamstring muscles. Phys Ther 1994: 74: 845852. Bandy WD, Irion JM, Briggler M. The eect of time and frequency of static stretching on exibility of the hamstring muscles. Phys Ther 1997: 77: 10901096. Bandy WD, Irion JM, Briggler M. The eect of static stretch and dynamic range of motion training on the exibility of the hamstring muscles. J Orthop Sports Phys Ther 1998: 27: 295300. Ben M, Harvey LA, Denis S, Glinsky J, Goehl G, Chee S, Herbert RD. Does 12 weeks of regular standing prevent loss of ankle mobility and bone mineral density in people with recent spinal cord injuries? Aust J Physiother 2005: 51: 251256. Bijur PE, Silver W, Gallagher EJ. Reliability of the visual analog scale for measurement of acute pain. Acad Emerg Med 2001: 8: 11531157. Bjorklund M, Hamberg J, Grenshaw AG. Sensory adaptation after a 2-week stretching regimen of the rectus femoris muscle. Arch Phys Med Rehabil 2001: 82: 12451250. Bohannon RW. Eect of repeated eightminute muscle loading on the angle of straight-leg raising. Phys Ther 1984: 64: 491497. Chan SP, Hong Y, Robinson PD. Flexibility and passive resistance of the hamstring of young adults using two dierent static stretching protocols. Scand J Med Sci Sports 2001: 11: 81 86. Draper DO, Castro JL, Feland B, Schulthies S, Eggett D. Shortwave diathermy and prolonged stretching increase hamstring exibility more than prolonged stretching alone. J Orthop Sports Phys Ther 2004: 34: 1320. Folpp H, Deall S, Harvey L, Gwinn T. Can apparent increases in muscle extensibility with regular stretch be explained by changes in tolerance to stretch? Aust J Physiother 2006: 52: 45 50. Gajdosik RL. Eects of static stretching on the maximal length and resistance to passive stretch of short hamstring muscle. J Orthop Sports Phys Ther 1991: 14: 250255. Gajdosik RL, Linden DWV, McNair PJ, Williams AK, Riggin TJ. Eects of an eight-week stretching program on the passive-elastic properties and function of the calf muscles of the older women. Clin Biomech 2005: 20: 973983. Gauvin MG, Riddle DL, Rothstein JM. Reliability of clinical measurements of forward bending using the modied ngertip-to-oor method. Phys Ther 1990: 70: 443447. Godges JJ, MacRae PG, Engelke KA. Eects of exercise on hip range of motion, trunk muscle performance, and gait economy. Phys Ther 1993: 73: 468476. Goldspink G, Tabary C, Tabary JC, Tardieu C, Tardieu G. Eect of denervation on the adaptation of sarcomere number and muscle extensibility to the functional length of the muscle. J Physiol 1974: 236: 733 742. Halbertsma JPK, Goeken LNH. Stretching exercises: eect on passive extensibility and stiness in short hamstrings of healthy subjects. Arch Phys Med Rehabil 1994: 75: 976981. Halbertsma JPK, van Bolhuis AI, Goeken LNH. Sport stretching: eect on passive muscle stiness of short hamstrings. Arch Phys Med Rehabil 1996: 77: 688692. Harvey L, de Jong I, Goehl G, Mardwedel S. Twelve weeks of nightly stretch does not reduce thumb webspace contractures in people with a neurological condition: a randomised controlled trial. Aust J Physiother 2006: 52: 251258. Harvey LA, Batty J, Crosbie J, Poulter S, Herbert RD. A randomized trial assessing the eects of 4 weeks of daily stretching on ankle mobility in patients with spinal cord injuries. Arch Phys Med Rehabil 2000: 81: 13401347. Harvey LA, Byak A, Ostrovskaya M, Glinsky J, Katte L, Herbert RD. Randomised trial of the eects of four weeks of daily stretch on extensibility of hamstring muscles in people with spinal cord injuries. Aust J Physiother 2003a: 49: 176181. Harvey LA, McQuade L, Hawthorne S, Byak A. Quantifying the magnitude of torque physiotherapists apply when stretching the hamstring muscles of people with spinal cord injuries. Arch Phys Med Rehabil 2003b: 84: 14. Herbert RD, Balnave RJ. The eect of positioning of immobilisation on resting length, resting stiness, and weight of the soleus muscle of the rabbit. J Orthop Res 1993: 11: 358366. Huet de laTour E, Tabary JC, Tabary C, Tardieu C. The respective roles of muscle length and muscle tension in sarcomere number adaptation of guinea-pig soleus muscle. J Physiol 1979: 75: 589592. Kirsch RF, Weiss PL, Dannenbaum RM, Kearney RE. Eect of maintained stretch on the range of motion of the human ankle joint. Clin Biomech 1995: 10: 166168.

143

Ben & Harvey

Laessoe U, Voigt M. Modication of stretch tolerance in stooping position. Scand J Med Sci Sports 2003: 13: 16. Lannin NA, Cusick A, McCluskey A, Herbert RD. Eects of splinting on wrist contracture after stroke. A randomized controlled trial. Stroke 2007: 38: 111116. Lannin NA, Herbert RD. Is hand splinting eective for adults following stroke? A systematic review and methodological critique of published research. Clin Rehabil 2003: 17: 807816. Lannin NA, Horsley SA, Herbert RD, McCluskey A, Cusick A. Splinting the hand in the functional position after brain impairment: a randomized controlled trial. Arch Phys Med Rehabil 2003: 84: 297302. Lentell G, Hetherington T, Eagan J, Morgan M. The use of thermal agents to inuence the eectiveness of a lowload prolonged stretch. J Orthop Sports Phys Ther 1992: 16: 200206. Magnusson SP, Simonsen EB, Aagaard P, Dyhre-Poulsen P, McHugh M, Kjaer M. Mechanical and physiological responses to stretching with and without preisometric contraction in human skeletal muscle. Arch Phys Med Rehabil 1996a: 77: 373378. Magnusson SP, Simonsen EB, Aagaard P, Henrik S, Kjaer M. A mechanism for altered exibility in human skeletal muscle. J Physiol 1996b: 497: 291298. Magnusson SP, Simonsen EB, Aagaard P, Kjaer M. Biomechanical responses to repeated stretches in human hamstring muscles in vivo. Am J Sports Med 1996c: 24: 622627. Moseley AM, Herbert RD, Nightingale EJ, Taylor DA, Evans TM, Robertson GJ, Gupta SK, Penn J. Passive stretching does not enhance outcomes in patients with plantar exion contracture after cast immobilization for ankle fracture: a randomized controlled trial. Arch Phys Med Rehabil 2005: 86: 11181126. Pocock S. Clinical trials. A practical approach. Chichester: Wiley, 1983. Refshauge KM, Raymond J, Nicholson G, van den Dolder PA. Night splinting does not increase ankle range of motion in people with Charcot-Marie-Tooth disease: a randomised, cross-over trial. Aust J Physiother 2006: 52: 193199. Schulz K, Chalmers I, Hayes R, Altman D. Empirical evidence of bias. Dimensions of methodological quality associated with estimates of treatment eects in controlled trials. JAMA 1995: 273: 408412. Schulz K, Grimes D. Handbook of essential concepts in clinical research. Edinburgh: Elsevier, 2006. Spector SA, Simard CP, Fournier M, Sternlight E, Edgerton VR. Architectural alterations of rat hindlimb skeletal muscles immobilized at dierent lengths. Exp Neurol 1982: 76: 94110. Starring DT, Gossman MR, Nicholson GG, Lemons J. Comparison of cyclic and sustained passive stretching using a mechanical device to increase resting length of hamstring muscles. Phys Ther 1988: 68: 314320. Tabary JC, Tabary C, Tardieu C, Tardieu G, Goldspink G. Physiological and structural changes in the cats soleus muscle due to immobilization at dierent lengths by plaster cast. J Physiol 1972: 224: 231244. Tardieu C, Tabary J-C, Tabary C, Huet de laTour E. Comparison of the sarcomere number adaptation in young and adult animals. J Physiol 1977: 73: 10451055. Timson F. Evaluation of animal models for the study of exercise-induced muscle enlargement. J Appl Physiol 1990: 69: 19351945. Toft E, Espersen GT, Kalund SK, Sinkjaer T, Hornemann BC. Passive tension of the ankle before and after stretching. Am J Sports Med 1989: 17: 489494. Turton AJ, Britton E. A pilot randomized controlled trial of a daily muscle stretch regime to prevent contractures in the arm after stroke. Clin Rehabil 2005: 19: 600612. Williams PE. Use of intermittent stretch in the prevention of serial sarcomere loss in immobilised muscle. Ann Rheum Dis 1990: 49: 316317. Worrell TW, Smith TL, Winegardner J. Eect of hamstring stretching on hamstring muscle performance. J Orthop Sports Phys Ther 1994: 20: 154159. Youdas JW, Krause DA, Egan KS, Therneau TM, Laskowski ER. The eects of static stretching of the calf muscle-tendon unit on active ankle dorsiexion range of motion. J Orthop Sports Phys Ther 2003: 33: 408417.

144