Paper plus: Do warm-up exercises prevent leg injuries?

Leanne Tite considers a study that asks whether warm-up exercises reduce injury. She also looks in depth at the weaknesses of this cluster randomised controlled trial

This month's paper is Olsen O-E, Myklebust G, Engebretsen L, Holme I, Bahr R. Exercises to prevent lower limb injuries in youth sports: cluster randomised controlled trial. BMJ 2005;330:449-52. You can read it by going to studentbmj.com and clicking on the link.

Why do the study?

Although regular exercise has obvious health benefits, it also carries a risk of injury for people at all levels of ability. In fact, the authors report that 10-19% of all injuries seen in hospital casualty departments in Scandinavia result from participation in sports, with knee and ankle injuries the most common. Such injuries can have long term consequences, sometimes requiring surgery and long term care, or leading later to other long term medical problems, such as osteoarthritis.

Prevention is always better than cure, and finding ways to prevent sports injuries using simple warm-up exercises makes a lot of sense. Some scientific studies have already evaluated, or tested, the effectiveness of warm-up exercises in preventing sports injuries and have reported promising results. None of these studies gives conclusive information, however, because they were not randomised controlled trials and so lacked an important methodological procedure that would give the findings greater certainty. The aim of the current study was to build upon these previous studies by using a more sound research design to assess the effectiveness of a warm-up programme in preventing sports injuries.

HANS NELEMAN/PHOTONICA

How did the researchers design the study?

The study is described as a “cluster randomised controlled trial.” Essentially, it is a standard randomised controlled trial design but the clusters refer to the fact that participants in the study are clustered into natural groupings (that is, the handball clubs in which they participate). This is of little interest to anyone except a statistician, because the kind of statistics usually used to analyse the study data state that all the participants in the study must be independent. But in a study in which the participants come in natural groups, as in this study, they are not considered to be truly independent, from a statistical point of view, and a different kind of statistical procedure is used to take account of the clusters.

The researchers approached all 145 handball clubs for players of 15-17 years of age in central and eastern Norway. In total, 123 clubs agreed to participate. The next step was to allocate clubs to the different experimental groups: the intervention group, who would use the exercise programme being evaluated for injury prevention, and the control group, who would not use the exercise programme but to whom the intervention group would be compared with to look for differences in injuries (the outcome measure). Importantly, clubs were allocated to each experimental group randomly so that any differences between clubs would also be randomly distributed and would not systematically affect the outcome for either the control or intervention group. In this study, “block randomisation” was also used. This means that the clubs were first put into blocks or groups so that each block contained clubs with an equal balance of the different kind of clubs taking part in the study. In other words, each block was matched for the level of ability of the players, their region, the number of players, and sex of the players in the club, to ensure that these characteristics were evenly spread across the control and intervention groups.

HÅKON T SØNDERLAND

All clubs were followed for one season, and clubs in the intervention group were given a special warm-up programme to follow before training and matches, and the control group was told to warm up as normal (no intervention). Physiotherapists then recorded all the injuries in each club for the season, as well as the type of injury and amount of exposure to matches and training sessions. Club coaches also recorded the compliance of the intervention group clubs to the exercise programme.

In the study period and across both the control and intervention groups, 1837 players took part in the study. Among all players, during the season 262 players contracted 298 injuries. Statistical analyses showed that players in the control group had significantly more injuries than players in the intervention group, indicating that the warm-up exercises had been effective in preventing injury. For many types of injury, players in the control group who did not use the warm-up exercise were about twice as likely to be injured than players in the intervention group. In addition, 87% of clubs in the intervention group did the exercise programme as instructed. This is important because if the programme is difficult for clubs to follow for any reason, then no matter how good it is at preventing injury it could not be useful in reducing injury rates on a large scale. Also, this suggests that if all of the intervention clubs had stuck to the programme, the rate of injury prevention in the intervention group might have been even greater.

Was it a good study?

As an undergraduate psychology student I was told by one of my lecturers that every research study has flaws. At the time I found this hard to believe—all those clever people writing unfathomable reports about such wonderful and complicated things—how could they possibly make mistakes? But, several years on, and I can attest for myself that this is indeed the case. Yes, all studies do have flaws; it is just a matter of finding them. And usually they are often something to do with the simple, common-sense things that strike you as a bit dubious the first time you read the article. Hint: if you find yourself trying to explain something away that puzzles you when you read the article, then it is probably a flaw.

One of the problems that struck me with this study is the way injuries were reported. Although physiotherapists were responsible for recording information about the injuries it was about a month after the injury that the players were interviewed (and in some cases up to four months). One of the problems with studies that rely on self reported data is that participants can easily forget about the events that they are describing or recall them inaccurately and so can produce misleading results. A related problem is that all participants were told at the beginning that they were involved in a study. Naturally, the participants in the intervention group would have been aware that their warm-up exercises were different to normal. And both groups were also aware that the study aim was to reduce injuries rates. Simply being aware that they are involved in a study and being observed is often enough to make people change their behaviour, often unknowingly (known as the Hawthorne effect, after a famous sociological study of factory workers). So for both the control and the intervention group, participants could have been more alert to the possibility of injury and perhaps more, or even less, willing to interpret something as an injury and then report it as such to the physiotherapist. Clearly this can have a problematic effect on the validity of the data being collected. Ideally, in a randomised controlled trial, all participants (and also the investigators) are blinded to both the aims of the study and to which study group they are in so that the Hawthorne effect, or the “placebo effect,” is minimised. In some studies, such as this one, however, it is not always possible to blind participants and investigators, for obvious reasons.

On the plus side, however, the study does have good generalisability because participants were selected from a large and representative population (all handball clubs in central and eastern Norway) and nearly all the clubs approached agreed to participate. Generalisability refers to how confidently a researcher can say that the participants actually studied are similar to a larger population of interest, and can therefore argue that the study findings would also apply to that larger population. There are usually limits as to what the findings can be generalised to, however. In this study, the authors give reasons in the discussion section as to why the exercise programme might be useful in preventing injury in sports other than handball. Until further research confirms this speculation, however, generalisability of these findings is limited to the sport studied in this research.