Stretching to prevent injury: does sports science support this traditional warm up?

Does stretching really lower the risk of injury?

Many athletes engage in at least some form of stretching before exercise, at least partly in the belief that loosening their muscles and connective tissues will lower their risk of injury. Such faith in stretching seems logical: after all, shouldn’t loose muscles be less susceptible to excessive strain during workouts, compared to sinews that are as tight as violin strings?

Although the idea that stretching can help promote injury-free training is widely accepted among athletes and coaches, scientific support for the notion has been quite modest, and in fact one key study actually linked stretching with a higher frequency of muscle and connective-tissue problems.

Before we take a close look at that investigation and several other key studies on stretching, we should bear in mind that stretching research has traditionally fallen into two key categories: (1) randomized, controlled trials, and (2) cohort studies. In the former, the somewhat homogeneous individuals taking part in the study are randomly divided into two groups, and one group then stretches regularly as a normal part of training while the other group goes ‘cold turkey’. This is actually the best way to get a feeling for stretching’s efficacy, because it tends to eliminate various forms of bias which could significantly distort the results of the research.

Cohort studies are usually easier to carry out, however, and as a result they appear much more often in the scientific literature. In cohort-style research, scientists would simply follow athletes who routinely do or do not stretch over time to determine which group had a higher incidence of injury. Naturally, such cohort investigations are susceptible to considerable bias.

For example, let’s say you that you were an exercise scientist and that you began tracking all the players in the Premier football division, keeping close tabs on their injuries. Of course, at the beginning of your cohort research you would have each athlete fill out a questionnaire in which he described his stretching habits in detail. You could then categorize the players as regular stretchers or non-stretchers (and perhaps eliminate from consideration individuals who stretched rather sporadically) and simply record the number of significant injuries sustained by members of the two groups over the course of a season. You might get fancier still and sub-divide the two groups into strikers, midfield players, and defenders if you wanted, to see whether team position had an effect on injury rates or interacted with stretching in some way. At any rate, at season’s end you would have a very good feeling for whether stretchers or non-stretchers were injured more frequently.

Was it stretching – or something else?

Trouble would arise, however, when you tried to interpret your results. For example, let’s say that you determined that athletes who stretched in a systematic and regular manner had significantly fewer injuries, compared to individuals who didn’t stretch. What would this actually mean?

In other words, was it really the regular stretching that kept injuries at bay – or some other factor? Since these particular footballers indicated that they were quite scrupulous about their stretching, it’s possible that they were also doing other things which decreased their risk of getting hurt. Perhaps they were warming up more thoroughly before training sessions and matches, for example, a practice which might also lower injury incidence. Or perhaps, given that they were so serious about stretching, they were also more careful about their diets, ingesting more protein, carbohydrate, vitamins, minerals, and antioxidants, compared to individuals who ignored stretching. Possibly, they were getting more sleep or even carrying out slightly more effective strength training, too. As you can see, many potential sources of bias could be present, and so it would be very difficult to proclaim that stretching was the ‘magical injury-fighting pill’ which kept these athletes out of trouble.

Further problems

If you found that your stretchers and non-stretchers had similar injury rates, it might be because stretching was ineffective as an injury-prevention technique, but it might also be true that stretching actually lowered injury rates in ‘tight’ soccer players who sensed they needed stretching, making their injury rates comparable with the flexible people who didn’t feel a need to stretch. And if non-stretchers had a lower frequency of injury, you might conclude that stretching actually increased the chance of muscle trauma, but it might also be the case that the timing of stretching was wrong, an improper form of stretching was utilized, or that stretching did lower injury rates – but not enough to make stretchers as injury-resistant as the naturally lax athletes (an easy-to-see bias in these cohort studies is that many individuals who stretch probably do so for a reason – because their muscles are unusually tight; individuals who don’t routinely stretch may avoid stretching because they are naturally flexible, and it may be these intrinsic factors (tightness and flexibility) which ultimately determine injury rates, not the selection or avoidance of stretching). An additional worrisome factor is that athletes might have been somewhat dishonest about their stretching habits (since you had undertaken a cohort study and had no direct control over how the athletes actually stretched and trained, there would be little way of knowing whether the players were actually doing what they said they were doing). If enough athletes distorted their stretching habits, your results would also be distorted.

As you can see, cohort studies have lots of problems! Incidentally, there are four cohort studies concerning stretching and injury in recent, peer-reviewed, high-quality scientific journals, and not a single one of these investigations documents a protective effect for stretching. It would be easy to conclude from these four works that stretching has little impact on injury, but consider the above comments about cohort studies before you make that kind of ‘leap of faith’.

What happened in Australia

Randomized controlled trials might help clear up the picture, but unfortunately there are only two such studies in the literature. In one of them, 1093 male army recruits were randomly divided into stretch and control groups (‘Effects of Flexibility and Stretching on Injury Risk in Army Recruits,’ Australian Journal of Physiotherapy, vol. 44, pp. 165-172, 1998). Recruits in the stretching group unkinked their calf muscles during their warm-ups, while control-group subjects did not. At the end of the experimental period, the total frequency of five different types of lower-leg injury in the stretch group was 4.2 per cent, compared with 4.6 per cent for the control group, but the difference was not statistically significant.

In the second study, 1538 male Australian army recruits were randomly allocated to stretch and control groups (‘A Randomized Trial of Preexercise Stretching for Prevention of Lower-Limb Injury,’ Medicine and Science in Sports and Exercise, vol. 32(2), pp. 271-277, 2000). In case you are wondering why army recruits seem to be the sole providers of randomized, controlled information about stretching and injury, bear in mind that such novitiates exercise rather intensely during their basic training, their rate of lower-limb injury is very high, and they also represent a captive audience on whom experimental manipulations can be easily imposed and monitored. They are thus ‘perfect subjects’ (from a pure research standpoint), but bear in mind that army neophytes are not necessarily in great shape – nor do they train like typical endurance athletes, and thus the results may not apply very well to experienced athletes with superior functional strength and overall fitness.

Marching, running, swimming – you name it

In this second study, recruits in the stretch group statically loosened six key lower-limb muscle groups (gastrocnemius, soleus, hamstrings, quads, hip adductors, and hip flexors) every other day – prior to training – for 20 seconds each. The fitness, age (which ranged from 17 to 35 years), height, weight, body-mass index (BMI), and day of enlistment were recorded carefully for all subjects (the first five factors have been associated with injury risk in various pieces of scientific research; day of enlistment was included because – anecdotally – Australian-army-recruits’ injury rates seem to rise later in the year). The training lasted for 11 weeks and included 40 training sessions (about four per week) totalling 50 hours of exercise. Training activities included marching while carrying a rifle and backpack (10 hours), running over distances ranging from four to eight kilometres (10.5 hours), negotiating obstacle courses (12.5 hours), carrying out circuit training (7.5 hours), swimming – plus pool-side press-ups and sit-ups (four hours), and battle training (wrestling, log lifting, fireman’s-lift training, and shoulder rolls), which added on another 5.5 hours. Stretch-group members interspersed four minutes of light jogging and side-stepping with their stretching routines (this generally meant 40 seconds of jogging and side-stepping in between the 20-second stretches). Control-group recruits did carry out the warm-up jogging and side-stepping but completed no static stretching at all.

So who got hurt?

Over the course of the 11-week study (which incorporated a total of 60,013 hours of training), there were 175 lower-limb injuries in the control group and 158 maladies in the stretched recruits. The overall injury rate was 5.5 injuries per 1000 hours of physical training, or one injury every 181 hours (by the way, this is not dissimilar to the rate of injury commonly observed in regular endurance runners). Although the total number of injuries was slightly higher in the control group, regression analysis revealed that there was no statistically significant difference in injury rates between the groups, either for skeletal or soft-tissue injuries.

Although the limited stretching regime had no effect on risk of injury in these Australian recruits, there were three factors which were decent predictors of injury – day of enlistment, age, and performance on the ’20-metre shuttle-run test’. The later in the year a recruit enlisted, the greater was his risk of injury, for reasons unknown. The older the recruit, the higher his chance of spraining an ankle, ripping up some thigh muscles, or stirring up a bout of significant knee pain (for obvious reasons). The best predictor of all, however, was fitness: the fitter the recruit (at least, as measured by performance during the 20-metre shuttle-run test), the smaller the chance of getting hurt, both for bony injuries and soft-tissue damage. Incidentally, in case you are concerned about whether the 20-metre shuttle run represents a reasonable method for determining fitness, research has indicated that it is indeed a reliable and valid indicator of both maximal aerobic capacity and running ability (‘Validity of the 20-Meter Shuttle Run Test with 1 Min Stages to Predict VO2max in Adults,’ Canadian Journal of Sport Science, vol. 14 pp. 21-26, 1989).

How the shuttle run works

At this point, you might be surprised about how hard it is to detect any injury-preventing properties for stretching (all four cohort studies and both randomized, controlled investigations have failed to indicate that stretching is good for injury prophylaxis) – and you might well wonder why the 20-metre shuttle-run test is such a great foreteller of injury (you might even be wondering what the test actually is). To understand the value of the shuttle run, bear in mind that it is a test in which athletes must repeatedly run back and forth between two lines which are exactly 20 metres apart. As they bop back and forth, they must touch one of the 20-metre lines at approximately the same time that a sound signal emanates from a pre-recorded tape. Challengingly, the frequency of this sound signal increases in such a way that running speed between the lines must be increased by .5 kilometres per hour (8.33 metres per second) every minute (the starting speed is set at 8.5 kilometres per hour). The test ends when an athlete is no longer able to keep up with the required pace (‘The Multistage 20 Metre Shuttle Run Test for Aerobic Fitness,’ Journal of Sports Sciences, vol. 6, pp. 93-101, 1988).

While the initial speed of 8.5 kilometres per hour is rather modest (it equates with 11.3 minutes per 1600 metres or 2:50 per 400 metres), the pace picks up fairly quickly, and the athlete who is fit and nimble enough to keep scrambling back and forth for 20 minutes can reach a speed of 20 kilometres per hour (333 metres per minute, or 4:48 per 1600 metres) before falling into a heap. As mentioned, an athlete’s score on this test is a very reliable predictor of VO2max.

Coordination of movements is important

To understand why the shuttle test works as an injury predictor, note that athletes who fare very well on the shuttle run must have (1) a decent aerobic capacity (so that they can exercise continuously and at a fairly high intensity during the test), (2) good strength (so that they are less vulnerable to fatigue as the test progresses and so that they can accomplish all the braking and turning near the 20-metre lines without becoming completely exhausted), and (3) excellent coordination (to carry out the turn-arounds, accelerations, and decelerations without losing time due to sloppy movements). Of course, aerobic capacity is a reflection of overall muscle health. Good strength translates into better protection of bones, joints, ligaments, tendons, and even muscles during movement, which should decrease the risk of getting hurt. Outstanding coordination should have the same effect, because it would automatically reduce the stress placed on muscles and joints during activity (without coordination, a joint might be put through too-large a range of motion, or a muscle group might be forced to exert excessive force to correct the movement of a body part which had moved too far beyond its usual position). Studies of individuals recovering from low-back pain and injury reveal that often improvement in the coordination of movements involving the low back is more important than merely upgrading the strength of the low back for alleviating symptoms.

While this big Australian study suggests that stretching is a pretty ineffective way to lower the risk of injury (and that if you want to avoid getting hurt, high fitness, strength, and coordination are paramount factors), a number of criticisms can be raised. First, the stretches were somewhat short in duration (just 20 seconds). While there is no real consensus about how long individual stretches should be carried out, many athletes take considerably longer than 20 seconds to unkink tight parts of their body (stretching durations from 30 to 60 seconds are commonly recommended).

Fifteen seconds as good as two minutes

However, scientific research is not at all kind to the idea that more than 20 seconds are required to loosen up particular muscle groups. For example, in classic scientific research carried out at the Stanford University School of Medicine, 72 men were randomly divided into four groups: Members of one group statically stretched their hip adductors for 15 seconds at a time, individuals in a second group stretched for 45 seconds, and men in a third group attempted to loosen their adductors for two minutes at a time (a fourth, non-stretching group served as a control). As it turned out, there was ultimately no difference in hip-adductor flexibility between the groups, i.e., 15 seconds of stretching was just as effective as two minutes, in terms of untightening the adductor muscles (‘Effect of Duration of Passive Stretch on Hip Abduction Range of Motion,’ The Journal of Orthopaedic and Sports Physical Therapy, vol. 8(8), pp. 409-416, 1987).

A second, somewhat harsher criticism of the controlled Australian research would be that the stretches were completed without a thorough pre-stretching warm-up. Only four minutes of jogging and side-stepping were completed during the whole stretching regime, which meant that the first few stretches were performed with essentially no warm-up at all; warm-up is widely believed to make muscles less viscous and less resistant to elongation, thus lowering the possibility that the stretching activity itself could be somewhat damaging. Thirdly (but not finally), the stretches were carried out before training sessions took place, not after. There has been both anecdotal and research evidence which suggests that stretching is most properly placed at the ends of workouts, instead of the beginnings.

So is stretching better after workouts?

In fact, in a study carried out with 1543 runners who participated in the Honolulu Marathon, exercise physiologist David Lally, PhD. was able to link stretching before workouts with a higher risk of sustaining injury (‘New Study Links Stretching with Higher Injury Rates,’ Running Research News, vol. 10(3), pp. 5-6, 1994).

In Lally’s survey, 47 per cent of all male runners who stretched regularly were injured during a one-year period, while only 33 per cent of male non-stretchers were hurt, a statistically significant difference (an injury was defined as a problem severe enough to disrupt normal training for at least five days). Lally’s work was a cohort affair, so one might argue that those individuals who decided to stretch were also the ones who suffered from the most muscle tightness – and that this tightness caused more injuries, not the stretching. However, this relationship did not apply to female marathoners; in the fairer sex, stretchers and non-stretchers were injured at the same rate. The linkage also did not work for Oriental runners of either sex (the Honolulu Marathon is popular with Japanese endurance athletes); only white men who stretched were particularly susceptible to injuries (the popular film ‘White Men Can’t Stretch’ hit cinemas shortly after Lally’s work was published). The mechanism underlying these observations is unknown, although if the white-male-marathoner group contained a disproportionate number of ‘Type-A’ personalities who were rather impatient while carrying out their stretching (i.e., too quick to force their leg muscles into a stretched-out position), one would expect them to be prone to injury. Research clearly shows that a muscle suffers greater peak tension and is forced to absorb more energy as the rate at which it elongates increases (‘Viscoelastic Properties of Muscle-Tendon Units,’ The American Journal of Sports Medicine, Vol. 18(3), pp. 300-309, 1990).

An adept researcher, Lally was able to control for the possibility that those individuals who had been injured before his study began had taken up stretching as a prophylactic measure (this somewhat answers the concern that stretchers might have been as tight as fiddle strings; if they were really excessively taut, they would surely have had a recent history of injury and would have been excluded from Lally’s analysis). If the stretchers had indeed been recovering ‘wrecks’, Lally’s findings would have been highly biased (one of the best predictors of injury in endurance athletes is a past history of injury). When Lally threw the males with previous injuries out of his study groups, things still looked bad for the stretchers, who had a 33-per cent greater risk of injury, compared to non-stretching runners. The stretched runners did not run more miles than the non-stretched individuals, so higher mileage was not a possible explanation for the stretching-injury phenomenon.