Results

The search strategy resulted in 4439 potentially relevant articles. A total of 140 abstracts and 56 full text articles were selected for review and, of these, 9 studies met the inclusion criteria10 ,16–22 (figure 1).

Results

The injury definition in the included articles was partially consistent; three articles had used almost identical descriptions where an injury was defined as a condition that forces the athlete to refrain or modify training or competition.20–22 Junge et al10 and Engebretsen et al18 defined an injury as a musculoskeletal condition that had occurred during the Olympic Games and that had required medical attention.

In weightlifting and powerlifting the body is exposed to great forces since the athlete's aim is to lift the maximum weight possible. Since heavy lifting is a well-known injury risk in the general population3 coupled with the complex demands on balance and coordination, one can assume that the risk of injury is high. However, the results of the studies included in the present systematic review showed that the incidence of injuries were similar compared to other sports with comparable characteristics to weightlifting and powerlifting, that is, non-contact sports which require strength/power. In weightlifting there were 2.4–3.3 injuries/1000 hours of training17 ,21 and in powerlifting the incidence was 1.0–4.4 injuries/1000 hours of training.16 ,20–22 In comparison, a study of injuries in track and field by Jacobsson et al30 showed an incidence of 3.57 injuries/1000 hours of training and a study of injuries in alpine skiing by Westin et al31 showed an incidence of 1.7 injuries/1000 hours of training. When comparing the results to the incidence of injury in popular contact sports which also requires strength/power, for example, American football and wrestling (9.6 injuries/1000 hours of training and 5.7 injuries/1000 hours of training respectively32), the incidence of injuries in weightlifting and powerlifting could be considered low. The relatively low injury incidence in weightlifting and powerlifting should be interpreted with the background that most of the included studies were of a retrospective design. In studies of injuries in more popular sports, a prospective design is more common30–32 and could therefore contribute to reflect a more accurate injury risk by reducing the risk for recall bias. Therefore, a suggestion for further research in this area is to conduct prospective, rather than retrospective, studies regarding injury incidence. A further recommendation for future research might be to ask the athletes to report exposure to particular exercises included in the sports instead of hours of training. By this approach, the relative risk of particular activities to the injury can be examined. An example of this is the study by Winwood et al33 where they were able to compare the absolute and relative injury rates to strongman events and traditional weight training exercises by an estimate of the time the athletes allocated to specific exercises.

Several studies describe injuries in weightlifting and powerlifting. The relationships between specific exercises included in the sports and the injuries are, however, still unclear. Raske and Norlin21 assessed whether there were any relationships between shoulder injuries and certain exercises; however, they did not find any significant relationships. In the study by Keogh et al,20 52% of the injuries were caused by the three lifts in powerlifting; however, the authors did not see any indications that a specific exercise would lead to more injuries than the others. The other articles included in the present study did not assess risk factors which resulted in that we could not describe any risk factors for the injuries that occur in weightlifting or powerlifting. However, a study that was excluded due to its study design34 had collected 60 cases of pectoralis major ruptures and 80% of the ruptures had occurred during the bench press exercise. This suggests that there might in fact be a relationship between the bench press exercise and injuries to the pectoralis muscle, and strengthens the above suggestion that future studies should include reports of exposure to certain exercises rather than hours of training.

An important factor that might have influence on the relationships between exposure to training and injury is the use of anabolic androgenic steroids. So far, only case studies35 ,36 have suggested that there might be a relationship between use of anabolic androgenic steroids and injuries: Sollender et al36 described four cases of distal ruptures of the triceps brachii muscle where three of the patients had their injury while performing bench press. In a study by de Castro Pochini et al,34 96% of the participants who suffered an injury due to bench press, had also used anabolic androgenic steroids. With this in mind, the use of steroids might be an important mediator in the process of developing injuries and future studies should always enquire about steroid use. In the study by Keogh et al20 the authors state that they did not know whether the athletes had used steroids or not, and the other included articles do not mention the issue at all.

The included studies in this systematic review are, according to the assessment tool, generally of ‘poor’ or ‘fair’ methodological quality. The assessment tool we used is made both for cohort studies and cross-sectional studies and two questions: number 6 ‘For the analyses in this paper, were the exposure(s) of interest measured prior to the outcome(s) being measured?’ and number 7 ‘Was the time frame sufficient so that one could reasonably expect to see an association between exposure and outcome if it existed?’ can only be answered for cohort studies. In accordance with the assessment tool these questions have been answered ‘no’ since the included studies are cross-sectional studies. Two other questions were not possible to answer for most of the included articles: number 12 ‘Were the outcome assessors blinded to the exposure status of participants?’ and number 13 ‘Was loss to follow-up after baseline 20% or less?’ and they have therefore been answered with ‘not applicable’ for most articles. The questions answered ‘no’ may contribute to lowering the methodological quality even though they were not relevant. The questions answered ‘not applicable’ may contribute to risk of bias since blinding and follow-up was not made. However, in cross-sectional studies blinding and follow-up cannot be made since the exposure has already been made and therefore the answers to these questions have been ignored.

The problem with the questions answered ‘no’ have been corrected by classifying question 1 ‘Was the research question or objective in this paper clearly stated?’ and question 2 ‘Was the study population clearly specified and defined?’ as less valuable. Question 8 ‘For exposures that can vary in amount or level, did the study examine different levels of the exposure as related to the outcome (eg, categories of exposure, or exposure measured as continuous variable)?’, question 9 ‘Were the exposure measures (independent variables) clearly defined, valid, reliable and implemented consistently across all study participants?’ and question 11 ‘Were the outcome measures (dependent variables) clearly defined, valid, reliable and implemented consistently across all study participants?’ were considered more valuable since they were harder to fulfil. This is the reason why the study by Keogh et al20 has been classified as ‘good’ even though it has only answered ‘yes’ to 6 of the 14 questions.

It is not easy to compare the results of the included studies since they used different methods and different ways of presenting their data. For example, regarding time periods, Jonasson et al19 assessed if and where athletes had pain during the previous year whereas the study by Junge et al10 assessed acute injuries during the Olympic Games 2008. These data-collection periods probably differed in terms of exercise intensity and it makes the results from Junge et al10 difficult to compare with the studies assessing injury prevalence/incidence during an ordinary training period.

The fact that the included studies used different methods for data collection also contributed to difficulties in making fair comparisons between studies. Six studies16 ,19–23 collected data by self-report from the athletes. The three other included studies10 ,17 ,18 collected their data through medical records. In studies with self-report there is a risk of underestimating the true number of injuries since the athletes may have forgotten injuries that were not of great severity and therefore have not reported them. This can be the reason why the majority of injuries in the study by Raske and Norlin21 were rated severe. In Calhoon's and Fry's study17 minor injuries are over-represented which can be explained that the athletes were in living in a training facility with constant access to medical staff that could examine them every time they felt pain. These examinations would then have been recorded as an injury.

In weightlifting and powerlifting injuries to the shoulder, knee and low back were common. Knee injuries were more common in weightlifting and Keogh et al20 states that a reason for this might be that the weightlifters perform squats with the barbell resting on the upper parts of the trapezius muscle with a vertical back position which results in greater torque around the knees joints.

In this review, shoulder injuries were more common among powerlifters than weightlifters. A possible explanation may be the bench press exercise. Green and Comfort37 describes bench press with a wide grip as risky because the shoulder joint is in an abducted and externally rotated position. Since the aim is to lift the heaviest weight possible, one can postulate that the athlete often use a wide grip allowing them to make the lift as short as possible, thus perhaps increasing injury risk to the shoulder. What contradicts this is that weightlifters also are exposed to this extreme position of the shoulder joint when performing the snatch. However, the weights used in bench press are often much higher than the weights used in snatch, for example, the world record in bench press is 275 kg2 and the world record in snatch is 216 kg1 in respective heavyweight divisions. It should, however, also be said that the relevance of these possible explanations in relation to shoulder pain in weightlifting or powerlifting has not, to the best of our knowledge, been investigated.