FormalPara Key Points

There is good evidence that exercise-based injury prevention programs can result in an injury reduction of around 46 % in organized youth sport

Jumping/plyometric exercises appear to be particularly relevant for injury reduction

The beneficial effects are independent of whether the program is implemented during the pre-season or in-season

1 Introduction

Physical inactivity is one of the leading causes of chronic diseases, and largely contributes to the burden of disease, death, and disability worldwide. Physical activity (PA) has proven to cause immediate positive effects on health risk factors, skeletal and psychological health, as well as on mental, cardiorespiratory, and neuromuscular fitness in children and adolescents [13]. In addition, participation in organized youth sport is positively associated with a higher level of adult PA [4, 5]. It can be stated that youth sport has important implications for long-term individual and public health benefits. Therefore, PA must be fostered starting at a young age [6, 7].

However, participating in sports bears a risk of sustaining an injury. Sport and recreational activities are the leading cause of injury in youth [810]. Injuries in young athletes can lead to a reduction in current and future involvement in PA [11]. This, in turn, may have considerable impact on future health as well as on quality of life [12]. The economic burden associated with injury involves medical, financial, and human resources at many levels. As such, it relates to the individual and society as a whole [1315].

In a Swiss survey, sport-related injuries (organized and non-organized sports) represented 55–60 % of all self-reported injuries in 9- to 19-year-olds [16]. Data from the US [17], Canada [8], France, The Netherlands, and the UK [18] show that organized sport is the main cause of injury in adolescents. This is further supported by recent data from Sweden showing that sport is the most common cause of injuries in 11- to 18-year-olds [19]. Prospectively assessed injury incidences range between 0.50 (95 % CI 0.29–0.71) per 1,000 h of physical education (PE) classes for 10- to 12-year-old children and 63.0 (95 % CI 57.5–69.1) injuries per 1,000 match hours in U18 male rugby union football players [20, 21].

In the US, children aged 6–12 years spend an average of 5–6.5 h per week doing sport [22]. It is estimated that each year more than one-third of school-age children sustain a PA-related injury that needs medical care. Based on 30 million children and adolescents participating in sports in the US, the costs of treating these injuries were estimated to be $1.8 billion per annum in 1997 [23]. In The Netherlands. direct medical costs of PA injuries in children were estimated at €170 million (plus indirect costs of €420 million) in 2003 [24]. In Australia. sport-related injuries in children younger than 15 years of age, accounted for 3.1 times the number of years lost to disability, 1.9 times the number of bed-days, and 2.6 times the direct hospital costs compared with road trauma. From 2002 to 2011, the number of sports injuries leading to hospitalization showed a significant yearly increase of 4.3 % (95 % CI 3.4–5.4) [25].

Injuries are an unfortunate consequence of participation in sport, and every effort must be made to prevent their occurrence. The strong need for PA on the one hand and the negative outcome of sport-related injuries on the other hand clearly demonstrate the importance and necessity of sports injury prevention in youth.

Few narrative [2629] or systematic [12, 3034] reviews on risk factors and/or injury prevention in children and adolescents have been published. To our best knowledge, no meta-analysis quantitatively investigated the effectiveness of exercise-based programs to reduce sport-related injuries in children and adolescents. Thus, the aim of this meta-analysis was to quantify the effectiveness of exercise-based injury prevention programs in organized sports in athletes under 19 years of age. The detailed objectives were:

  1. 1.

    To quantify the effect of exercise-based injury prevention in children and adolescents based on a meta-analysis of (cluster-) randomized controlled trials (RCTs) and controlled intervention studies in organized sports.

  2. 2.

    To describe the characteristics of the study population and the intervention.

  3. 3.

    To calculate cumulative effects and effects for specific subgroups.

  4. 4.

    To provide recommendations for future research.

2 Methods

2.1 Literature Search Strategy and Selection of Studies

The present meta-analysis was conducted without an open-access research protocol. Relevant studies were identified using an Internet-based search in six databases from inception until 14 October 2013 (electronic supplementary material [ESM] S1). The following search terms were used with Boolean conjunction: (sport injur* OR athletic injur* OR sport accident*) AND (prevent* OR prophylaxis OR avoidance) AND (child* OR adolescent OR youth). The search was conducted by two researchers (RR and TS) independently. Moreover, citation tracking and hand searching of key primary and review articles were carried out.

The inclusion criteria were:

  • full-text paper published in English in a peer-reviewed journal;

  • prospective controlled intervention study (randomized RCT, quasi-experimental, case control, or cohort design) with one group not receiving any intervention;

  • assessing the effect of an injury prevention program in organized sports;

  • intervention program based on/utilized physical exercises;

  • participants were younger than 19 years of age;

  • Outcome variables include number of injuries and exposure data and/or injury incidence.

The exclusion criteria were:

  • combined injury data from organized and unorganized sports (e.g. global injury incidence of high-school sports and leisure time PA) without specifying separated data;

  • study on (only) currently injured athletes or sample with a specific health problem (e.g. obesity, recurrent injuries).

According to the above-mentioned criteria, final inclusion/exclusion decision was made by two researchers (OF and RR).

2.2 Assessment of Methodological Quality

The methodological quality of eligible studies was rated using a study quality score developed by Abernethy and Bleakley for a review on the same topic [30]. The scale consists of a 9-item checklist whereby, for each item, 0, 1 or 2 points are attainable, enabling a maximum rating of 18 points (ESM S2).

To increase rating accuracy, two researchers (LD and RR) independently conducted the rating process. The raters were not blinded to study authors, place of publication, and results. In case of disagreement that arose between the first two raters, a third rater (OF) was consulted and consensus was achieved.

2.3 Data Extraction

Relevant study data were independently extracted by two researchers (RR and TS). These data comprised, amongst others, country, study design, number, age, and sex of the athletes in the intervention and the control group, type of sport and level of performance, content, duration and implementation of the prevention program, compliance, study duration, injury definition, number of injuries, and exposure measurement.

2.4 Statistical Analysis

We used the data of the primary outcome of each study. Whenever reported in the publication, the rate ratio (RR) adjusted for clustering was used. Otherwise raw data (number of injuries and exposure measure) were extracted and used to calculate the RR of the study. In some cases [3542], values had to be calculated (using the incidence rate and the number of injuries/the exposure measure). If necessary, injury incidence rates were calculated for each study arm (intervention and control group). These injury incidences represent a proportion of the injury frequency based on either a time component (e.g. per 1,000 player hours) or a countable number (e.g. per 1,000 athlete sessions). The RR was then calculated by dividing the injury rate of the intervention group by the injury rate of the control group.

A natural logarithm transformation of all RRs was conducted. According to the Cochrane Manual, the standard error of the natural-logarithm-transformed RRs was calculated [43]. The inconsistency statistic was used to measure the heterogeneity of the included studies. Because the observed value was moderate to high (71 %) within the group of eligible studies [44], the analysis was conducted using a random effects model [45]. The inverse-variance method according to Deeks and Higgins [46] was calculated by means of the Cochrane Review Manager Software (RevMan 5.1, Cochrane Collaboration, Oxford, UK; ESM S3). To assess the risk of a potential publication bias, a funnel plot was created (ESM S4).

Three risk-of-bias-related sensitivity analyses to detect potential influences of methodological differences between studies were conducted:

  • influence of study quality;

  • influence of randomization;

  • influence of type of exposure measurement.

Comparison of effects between the following subgroups was accomplished:

  • boys/girls;

  • elite level/sub-elite level;

  • football (soccer)/handball/basketball;

  • pre-season/in-season/pre-season and in-season;

  • balance exercises/jumping and plyometric exercises;

  • all/specific injuries (lower extremity, knee, and ankle injuries).

To test for a potential ‘shift in injury severity’ due to the intervention, three injury categories were compared—mild/moderate/severe injuries.

3 Results

3.1 Trial Flow

Of 1,835 potentially relevant articles, 94 full-texts were retrieved, of which 70 did not meet the inclusion criteria and 3 met the exclusion criteria (Fig. 1). The remaining 21 studies were included in the quantitative analysis.

Fig. 1
figure 1

Flow diagram of the literature selection process

3.2 Characteristics of Study Population, Intervention and Outcome Variables

The included studies comprised 27,561 athletes, with a median sample size of 829 (range 50–5,703) per study (Table 1). The median age of those studies that reported athlete age was 16.7 years (range 10.7–17.8). Only one study focused on primary-school children under the age of 14 years [47], and some others included children younger than 14 years but did not report separate age-related data [4852]. Ten studies involved girls only [35, 36, 38, 4042, 50, 5254], four studies involved boys only [37, 39, 51, 55], and seven studies involved both sexes [4749, 5659]. In total, just 12.7 % of participants were boys. Four studies (10.5 %) investigated the elite level [39, 51, 53, 55, 56], 15 (82.2 %) investigated the sub-elite level [3538, 4042, 4850, 52, 54, 5759], and one study (7.3 %) investigated participants of PE classes at school [47].

Table 1 Overview of studies investigating exercise-based injury prevention programs (alphabetical order by first author)

Nine studies exclusively analyzed football players [35, 37, 39, 40, 48, 50, 5254], and four further studies included football together with other types of sport [36, 38, 41, 58]. Two studies focused on handball [42, 59] and three on basketball [49, 51, 56]. Ten studies aimed at prevention of all injuries [35, 37, 42, 4749, 51, 54, 55, 57], and 11 aimed at injuries in specific body parts or specific diagnosis (lower extremity, knee, or ankle injuries) [36, 3841, 50, 52, 53, 56, 58, 59]. Three studies investigated the effect of pre-season conditioning on injury incidence during the subsequent season [35, 36, 55], 15 analyzed the effects of an intervention that was conducted during the competitive season [3742, 4749, 5153, 56, 57, 59], and three did both [50, 54, 58]. Thirteen programs [35, 36, 38, 4042, 47, 48, 5053, 59] contained jumping/plyometric exercises, and eight [37, 39, 49, 5458] contained no such exercises. Fourteen programs [38, 39, 42, 4854, 5659] consisted of or included balance exercises, while seven studies [3537, 40, 41, 47, 55] did not. Eighteen studies used an exposure measure based on hours or the count of training sessions/games [3638, 4042, 4756, 58, 59], and three used the number of athletes or

athlete seasons [35, 39, 57]. Twelve studies reported injury severity data [37, 39, 42, 4851, 5355, 58, 59].

3.3 Quality of the Studies

Thirteen studies used a (cluster-) randomized design [35, 38, 42, 4749, 5154, 5759], and eight studies investigated effects compared to a control group in a non-randomized setting [36, 37, 3941, 50, 55, 56].

On average, the quality score of the studies was 11.8 (standard deviation [SD] 3.3), ranging from 6 to 16. The mean score of the 11 ‘high-quality’ studies was 14.6 (SD 1.2) [38, 4749, 5154, 5759], and of the 10 ‘poor-quality’ studies, mean score was 8.6 (SD 1.6) [3537, 3942, 50, 55, 56]. The most obvious differences between these studies were definition of inclusion and exclusion criteria; description of dropouts, including dropped-out participants, in the analysis; blinding of injury assessors; and randomization of participants. There were nearly no differences between ‘high-’ and ‘poor-quality’ studies concerning the definition of outcome measure, active surveillance/appropriate duration of the study period, and description of the applied intervention program (ESM S5).

3.4 Risk of Bias

The funnel plot (ESM S4) showed neither a perfect funnel-shape nor an obvious publication bias, although it seems that small-sized studies with indifferent effects are missing.

To determine whether the methodological quality of included studies affected the cumulative effect, ‘high-’ and ‘poor-quality’ studies were compared. The cumulated RR of ‘high-quality’ studies (0.59 [95 % CI 0.46–0.76]) and ‘poor-quality’ studies (0.47 [95 % CI 0.33–0.67]) was not significantly different (p = 0.29). Studies with a randomized design (0.54 [95 % CI 0.42–0.70]) did not significantly differ from non-randomized studies (0.54 [95 % CI 0.37–0.78], p = 0.99). Studies reporting exposure based on hours or number of sessions (RR 0.55 [95 % CI 0.44–0.68]) showed a similar (p = 0.83) effect as studies with an exposure measurement based on athlete seasons (RR 0.51 [95 % CI 0.30–0.89]).

3.5 Quantitative Data Synthesis

The cumulative analysis showed a significant overall effect of injury prevention programs in children and adolescents (RR 0.54 [95 % CI 0.45–0.67]; Fig. 2).

Fig. 2
figure 2

Overall effect of exercise-based sport injury prevention programs (sorted by weight). SE standard error, IV inverse-variance, CI confidence interval, df degrees of freedom

Studies showed significant beneficial prevention effects (p < 0.001) for minor (RR 0.75 [95 % CI 0.63–0.88]), moderate (RR 0.58 [95 % CI 0.44–0.78]) and severe (RR 0.68 [95 % CI 0.51–0.90]) injuries, with no significant difference (p = 0.36) between the three degrees of injury severity (minor ≤1 week of absence, moderate 1–2/3/4 weeks of absence, severe ≥2/3/4 weeks of absence). There was similar effectiveness of programs in reducing ‘all’ (RR 0.62 [95 % CI 0.48–0.81]), ‘lower extremity’ (RR 0.57 [95 % CI 0.44–0.72]), and ‘ankle’ (RR 0.51 [95 % CI 0.31–0.81]) injuries (Fig. 3).

Fig. 3
figure 3

Effects of exercise-based sport injury prevention programs focusing on all (global), knee, ankle, and lower-extremity injuries. SE standard error, IV inverse-variance, CI confidence interval, df degrees of freedom

By trend (p = 0.10), injury prevention of ‘knee’ injuries (RR 0.32 [95 % CI 0.15–0.68]) was more effective compared with the subgroup of studies focusing on ‘all’ injuries.

Injury prevention programs were significantly more effective when exclusively girls were targeted (RR 0.44 [95 % CI 0.28–0.68]) than when only boys were included in the study (RR 0.71 [95 % CI 0.60–0.85], p = 0.05). Studies on the sub-elite level (RR 0.51 [95 % CI 0.39–0.67]) tended to show greater injury reduction than studies on elite athletes (RR 0.67 [95 % CI 0.55–0.80], p = 0.11). Studies on programs that included jumping/plyometric exercises showed a significantly greater injury preventive effect (RR 0.45 [95 % CI 0.35–0.57]) than studies without such exercises (RR 0.74 [95 % CI 0.61–0.90], p = 0.003).

No significant differences were observed between studies on football, handball, and basketball (Fig. 4); pre-season conditioning, programs during season and pre-season- plus in-season-conditioning (p = 0.93); and programs with and without balance exercises (p = 0.76).

Fig. 4
figure 4

Effects of exercise-based injury prevention programs in football (outdoor only), basketball, and handball (sorted by weight). SE standard error, IV inverse-variance, CI confidence interval, df degrees of freedom

4 Discussion

4.1 Comparison with Other (Systematic) Reviews

To date, no meta-analysis was available that specifically examines the effects of injury prevention programs in children and adolescents. Ten years ago, Emery wrote a systematic review on risk factors in child and adolescent sport [12]. Mainly based on the evidence presented in case-control and cross-sectional studies, she concluded that injury prevention programs targeting potentially modifiable risk factors are warranted and proprioceptive training is recommended. She noted that there is only limited evidence from high-quality studies, especially RCTs.

In 2007, a systematic review of Abernethy and Bleakley which included seven studies, reported beneficial effects of sports injury prevention programs in adolescent sport (without providing a quantitative synthesis) [30]. Currently, sports injury prevention is a trending topic, and within the last 6 years, since the review by Abernethy and Bleakley, ten studies have been published, of which eight were high-quality studies. Consequently, we included these in our systematic review and meta-analysis. In 2009, Frisch et al. [32] systematically reviewed the effects of exercise-based injury prevention programs in youth sports. Without providing a quantitative synthesis, they concluded that injury prevention is effective when compliance to the program is high. Our systematic review updates their findings as, since then, six new studies have been published, of which five are high-quality studies.

Ladenhauf et al. [29] reviewed anterior cruciate ligament (ACL) injury prevention programs in young athletes. They concluded that programs are effective in reducing injury risk and recommend age-appropriate strength and neuromuscular balance exercises. Gagnier et al. [60] also focused on ACL injuries. They conducted a systematic review and meta-analysis of ACL prevention programs in adolescents and adults and found a significant reduction of injuries (RR 0.49 [95 % CI 0.30–0.79]). Myer et al. [61] conducted a meta-analysis to investigate whether the effectiveness of ACL injury prevention programs in female athletes is age dependent. They found an age-related association between the application of injury prevention programs and reduction of ACL incidence, and recommended the implementation of ACL prevention during early adolescence. Herman et al. [62] systematically reviewed neuromuscular warm-up programs, which require no additional equipment, for preventing lower-limb injuries. They found beneficial effects in five different prevention programs. Six of the studies they included comprised youth athletes. Thus, we also considered these studies.

Van Beijsterveldt et al. [63] conducted a systematic review on exercise-based injury prevention programs with a specific focus on football players. Although the focus of their review was not specifically set to youth athletes, five of six studies investigated youth football. Consequently, we also included these five studies in our review.

Nauta and colleagues [64] recently reviewed the effectiveness of school- and community-based injury prevention programs on risk behavior and injury risk in 8- to 12-year-old children. They concluded that the results with regard to active prevention were inconclusive. This is probably due to the small number of exercise-based injury prevention studies in the school- and community-based setting.

The present systematic review is the first meta-analysis that quantifies the effects of injury prevention programs in children and adolescents in organized sports. It updates and extends the systematic reviews of Frisch et al., as well as of Abernethy and Bleakley [30, 32], and provides cumulative and detailed analyses to clarify more specific questions in particular subgroups.

4.2 The Overall Effect of Injury Prevention Programs

The cumulated overall effect size indicates an injury reduction of 46 %. This value is slightly reduced to 41 % when only ‘high-quality’ studies are taken into account. However, even a moderate reduction of all sports injuries is of acute significance for young people’s health and could have a short- and long-term economic impact on healthcare costs [15, 65]. The sensitivity analyses did not reveal significant differences with respect to study quality and type of exposure measurement.

4.3 Sex

Most of the studies involved girls and, thus, boys were highly underrepresented, accounting for just one-eighth of all participants. This is in contrast to the higher PA participation rates observed in boys compared with girls [6668]. The risk of sports injuries is similar for both sexes, except for some specific types of injuries (e.g. ACL injuries, concussions) [6972].

The present meta-analysis revealed that girls profited significantly more from injury prevention than boys. Based on the present data, it is speculative to assume that girls have a greater potential to respond to exercise-based injury prevention. As data for boys are underrepresented, further research is required to clarify underlying reasons.

4.4 Level of Competition

Both elite and sub-elite athletes profited significantly from prevention programs. The slightly lower effect in elite than sub-elite athletes could be due to a ceiling effect, meaning that better trained athletes have less potential for further improvements (e.g. neuromuscular performance). To minimize the probability of ceiling effects, programs should enable the possibility of variation and progression [32].

4.5 Timing of Implementation

The comparison between programs that implemented ‘pre-season only’, ‘in-season only’ or ‘pre-season and in-season’ revealed very similar effects. Based on this finding, injury prevention programs can be recommended regardless of timing of their implementation.

4.6 Type of Sport

No statistically significant difference was found between studies on football, handball, and basketball. All three subgroups showed significant preventive effects and, thus, at least in team sports, the injury reduction effect seems to be independent of the sports performed.

4.7 Balance Exercises/Jumping or Plyometric Exercises

Basically, balance exercises and plyometric/jumping exercises are two different approaches as one focuses on proprioception and the other one on lower-leg strength/power. Therefore, these two concepts are compared with regard to their effects on injury reduction [12, 36]. While programs that incorporated balance exercises did not result in an increased injury reduction, programs including plyometric and jumping exercises showed a significantly greater preventive effect than programs that did not apply such exercises. A possible explanation could be the fact that injuries are often related to high-impact situations (landing, change in moving direction, opponent contact) [53, 73], and that the neuromuscular system is best prepared to resist these influences through high-intensity exercises such as jumps and landings [36].

4.8 Global Versus Specific Prevention Programs

Although not significant (p = 0.10), a tendency towards greater preventive effects of programs focusing on knee injuries was observed. However, it has to be considered that a certain amount of injuries is not preventable through exercise-based programs (e.g. head injuries as a result of a collision). This basic amount of injuries is not considered in studies with a ‘specific’ focus, whereas studies with a ‘global’ focus include these non-preventable injuries in their analysis. Thus, greater preventive effects are to be expected in studies with a ‘specific’ focus.

4.9 Recommendations

While it is of special importance to prevent severe injuries such as ACL ruptures or severe ankle sprains, it can also be argued that prevention programs should focus on the most frequent injuries. It is therefore recommended that injury prevention targets the reduction of injuries in the broadest possible way without losing its specificity to tackle the most severe injuries. It would seem reasonable to call for multimodal approaches that consist of different exercises, each one of which has a specific aim. We also have to be aware of the fact that some injuries will not be preventable through a modification of intrinsic risk factors.

4.10 Strengths and Limitations

This systematic review was conducted according to the PRISMA statement [74]. To the best of our knowledge, it is the first meta-analysis which cumulates the effects of injury prevention programs in organized child and adolescent sport. It gives a comprehensive overview of current scientific evidence. As recommended by Impellizzeri and Bizzini [75], no cut-off in quality score was used, first to avoid an influence of subjective study rating and, second, to get the broadest possible perspective. All subgroup analyses, except two, were planned a priori. The analyses which compared mild, moderate, and severe injuries, and the analysis that focused on elite and sub-elite level, were defined a posteriori. Therefore, the findings of these two analyses are exploratory and hence preliminary in nature, and should be carefully interpreted. A sensitivity analysis was undertaken to check for a potential bias due to study quality. The 21 studies included in the meta-analysis provided a large enough data pool for specific analysis of subgroups with different characteristics in relation to study population, characteristics of the injury prevention program, and outcome variables. The available studies on the topic vary considerably in characteristics of the study population, type of intervention (content, dose, and duration), injury definition, severity classification (e.g. ‘severe’ is defined as ‘more than 2 weeks’ or ‘more than 4 weeks’ of absence), exposure measurements, and research design. However, it can be argued that although different in nature, all programs do seem to have beneficial effects regardless of the specific setting in which preventive measures are applied. As the development of various different prevention programs seems to not be efficient, and current programs show similar effects, the current evidence may be used to establish a blueprint for effective injury prevention in children and adolescents.

Injury prevention trials in children under the age of 14 years are almost completely missing to date. Only one study focused solely on primary-school children, and a few others included children younger than 14 years of age. Thus, an analysis of the effectiveness of injury prevention programs in different age groups was not possible.

4.11 Directions for Future Research

This meta-analysis shows promising beneficial effects of injury prevention programs in organized child and adolescent sport, but more high-quality studies are required to clarify the effect of specific exercises and the influence of compliance. Studies on sports injury prevention in children under the age of 14 years, and in individual sports athletes, are desirable for the future.

Consistency with regard to injury definition and severity classification are key features to consolidate the evidence in the future. The success of an injury prevention program is not only based on a quantitative reduction of injuries but also on a reduction in severity of injury. Therefore, an intervention can be beneficial, even without an absolute reduction of injury incidence, if the severity of injuries is reduced. Kiani et al. [50] explicitly reported such an effect; however, this needs to be substantiated by further research.

To increase the quality of future studies, authors should report the definition of inclusion and exclusion criteria, use an intention-to-treat analysis, and assure blinding of injury assessors. As recently shown, the success of a sports injury prevention program depends essentially on compliance [76]. A dose-response relationship of adherence to the program and injury reduction effect was found [52, 76, 77]. Therefore, it is of particular importance to assess and report compliance with the intervention. To clarify the net effect, compliance and dose-response analyses are recommended for all future injury prevention studies. The development and application of a consensus statement on how to conduct studies on injury prevention programs in child and adolescent sports would be warranted, since homogeneity with respect to study design will enable a clearer interpretation of results.

The prevention of severe sports injuries is a major challenge of the future. Thereby, the 6- to 18-year-old age group is of particular interest as the proportion of sport-related injuries of all life-threatening injuries is much higher in children and adolescents (32 %) compared with adults (9 %) [78].

Exercise-based injury prevention should become an integral part of regular training sessions as it can improve physical fitness and technical performance [79]. There may be voices who claim a loss of practice time due to the application of injury prevention programs; however, considering injury reduction and performance enhancement effects, children, parents, coaches, sport institutions, and society in general can benefit from exercise-based injury prevention.

5 Conclusions

The present systematic review and meta-analysis reveals good evidence that exercise-based injury prevention programs can result in an injury reduction of around 46 % in youth sports. Multimodal programs including jumping/plyometric exercises can be recommended. There is a considerable lack of data for children (under 14 years), boys (representing only 12.7 % of the overall study population), and for individual sports. More high-quality studies are needed to clarify the effect of specific exercises and compliance.