Association between sports type and overuse injuries of extremities in children and adolescents: a systematic review

A librarian-assisted search was conducted on November 2014 and a final search performed in June 2016 in PubMed, SportDiscus, PsycInfo and Web of Sciences using the search terms « overuse injuries OR cumulative trauma disorders OR musculoskeletal injuries » AND « extremity OR limb » AND « physical activity OR sport OR risk factor OR predictors OR exercises » AND « child OR adolescent OR young adults » in different combinations (MeSH terms and free text). An additional citation search of reference lists of the retrieved articles was performed. No restrictions were placed on date of publication. No attempts were made to search the grey literature.

Our approach was modelled on the Preferred Reporting Items for Systematic reviews and Meta-Analysis (PRISMA) flow-chart, as described below [10]. The first author applied the inclusion criteria to the title and abstract of the articles identified from the literature search. Blinded full-text screening was then done by two authors to determine which articles should be included in the final review. Inclusion criteria were: 1) a study design that was prospective, retrospective or cross-sectional, 2) age 19 years or below, 3) the OIE should be reported in relation to a particular sports type, and 4) sample size greater than 50. Only articles written in English, French or a Scandinavian language were considered. A list of articles excluded from the review on the basis of the full-text screening is available from the authors on request.

Two descriptive checklists, one quality checklist, and three tables of results were created by the authors for the purposes of this review. Information was grouped by sports type.

The first checklist presents information on type of sport, year of publication and country of study, as well as the observational setting (i.e. specific college, sports club or competition), the level of sport, the age and sex of the children, the number of participants and participation rate, the duration of data-collection, and the data source.

The second checklist was used to describe the definitions of injury and overuse injury used in the articles. Regarding criteria used for sports-related injury we chose those described by Fuller [6], whereas our overuse definition was inspired by criteria proposed by Bahr in 2009 [7] which are: repeated micro-trauma, no single identifiable cause, gradual onset, and activity exceeds tissue tolerance. Our classification of overuse injury was based on diagnosis, as described by Roos [8]. Moreover, because some articles did not use the criteria described above we added a column titled “other”, in which we quoted the definition used in the article.

In the third checklist, which related to the methodological quality, we were interested in whether: 1) participation rate was clearly stated, 2) the diagnosis of OIE was collected by medical personnel 3) the incidence of OIE was clearly reported in the article, 4) diagnoses and areas of overuse injury had been clearly and completely reported, and 5) the number of injuries could be reported in relation to the number of hours of exposure.

Checklists 2 and 3 were used to identify the strengths and weaknesses of this research area but were not taken into account during the interpretation of findings.

The findings were reported in three evidence tables. Table 4 shows the incidence of OIE and the proportion of OIE regarding the number of injuries per hour of exposure. Incidence was provided only if explicitly reported in the article. In addition, the proportion of OIE was calculated if information was provided on the number of injuries in relation to the number of hours of exposure.

Table 5 focuses on the site of overuse injuries (e.g. shoulder, elbow). Injuries clearly pertaining to the extremities were included in this table. Diagnoses that were clearly related to a particular site were recorded accordingly (e.g. ‘Osgood Schlatter’ = knee injury). Vague descriptions such as ‘tendinitis’ that could not be linked to a specific anatomical region were not recorded.

Table 6 reported on diagnosis related to injury site. If the diagnosis was provided but not the site, making it impossible to relate it to the limbs, the information was not used in the analysis. In one article on running, the authors reported stress fracture without specifying the localization. We decided to describe these injuries as extremity injuries and to include them in Tables 4, 5, 6 and 6 because we assumed that a stress fracture in running concerned the lower extremity. The two most frequently reported areas or diagnoses were emphasized in the tables.

A MeaSurement Tool for the Assessment of multiple systematic Review (AMSTAR) checklist [11], was used as a guide for this article. However, tests for homogeneity for control for likelihood of publication bias could not be carried out.

The checklists and tables were tested for relevance and user-friendliness, adjusted as needed, and thereafter used by two authors, who separately (and blind to each other’s findings) extracted information from the included articles. The data were later compared to detect extraction errors. If necessary, a third person arbitrated any disagreements between the two reviewers. Interpretation of data was done in collaboration between the first two authors. The review was registered in the PROSPERO database: CRD42014007079.

After primary selection we obtained 17 articles that fulfilled our initial inclusion criteria out of the 753 articles identified through the database and citation searches (Fig. 1). Most of the excluded studies either did not include children or did not deal with specific sports. However, in some of the 17 articles it was not possible to interpret the data correctly because it was not always clear if a stated condition was an OIE or not. We, therefore, realized that we had to add a new inclusion criterion: “the article must state clearly if reported cases were classified as traumatic or overuse injuries “. With this specific selection we were left with eight articles to which was added one obtained through hand search of reference lists, i.e. a total of nine articles (please see Fig. 1).

Seven studies were prospective [12‐18] and two were retrospective in design [19, 20]. As can be seen in Table 1, the nine articles included in the analyses reported on 6 different types of sport, i.e. soccer, gymnastics, handball, orienteering, running and dance and were published between 1986 and 2011. Of these, 5 were conducted in Europe, particularly in the Scandinavian countries.

As can be seen in Table 1, the number of children and adolescents observed in the various studies ranged from 79 to 837, with response rates varying from 63 to 100 %. Three of the articles included children and adolescents (ranging between 7 and 18 years), while six articles included only adolescents (between 12 and 19 years). Six studies included only girls, and the other three studies included both boys and girls. Duration of data collection varied from one season to eight seasons but was not always well described. Moreover, the data source was either the child, the trainer, or medical personnel or, usually, a combination thereof.

Table 2 shows how the definition of “injury” differed between articles. In general, the definition was tied to complaints related to sport activities concerning their consequences (i.e. time loss or medical attention). The same table shows how the overuse aspect was not well defined and quite heterogeneous. None took into account whether the activity exceeded the tolerance of tissue, repeated micro trauma was only included in one article, and only two used the criterion of no single identifiable cause, whereas gradual onset was commonly used as the only criterion of overuse injury. Three articles did not use any of these criteria. In one article, only the title indicated the topic. In another one, an example of overuse injury was given, and in a third article, the diagnosis seemed to be one of exclusion (if not traumatic in origin then it had to be overuse).

In five of the articles, the study response rate was not reported and could not be calculated (Table 3). The diagnosis was established by health personnel in six of the nine studies.

Specific sites and diagnoses of OIE were often incompletely reported. Only two articles showed the incidence of some of the OIE. Nevertheless, proportions of injuries in terms of hours of exposure could be calculated in six articles.

Table 4 shows the results regarding incidence and proportions of OIE. Specific incidence estimates were available only in two articles on soccer, for the diagnoses periostitis, tendon pain, and tendinopathy, at <0.5 OIE for 1000 h of exposure.

The highest overall proportion of OIE was found for soccer: 1.8/1000 h of exposure, followed by 0.7 in one study on soccer, and in one study on gymnastics. A third study on soccer had an overall proportion of OIE of 0.4/1000 h of exposure, whereas handball and another study on gymnastics had lower estimates (<=0.1/1000 h of exposure).

In general, the lower limb was most often affected, and especially the knee, tibia, and thigh (Table 5). The most frequently provided diagnoses were tendinitis/bursitis and periostitis (Table 6).

For all sports covered, the lower limb was more often affected than the upper. It was not possible to compare the incidence rates because it was only rarely provided. In handball, elbow injuries were equally common and noted as the most common injury site equal to the lower leg injuries. The foot was the second most common site in gymnastics. No differences in diagnosis could be found between different types of sports, mainly because only few articles clearly provided the diagnosis.

To our knowledge, this is the first systematic review on this topic. Our aim was to improve the understanding of the relationship between type of sport and OIE in children and adolescents. The individuals included in this study were more often adolescents than children, but the articles included did not allow us to distinguish between these two groups. Whilst the reporting styles of the reviewed articles made it impossible to compare exact incidence rates, relative differences in occurrence could be studied in relation to the proportion of injuries per numbers of hours of exposure and injury site and diagnosis.

In a previous large study on 1259 school-children aged from 6 to 12 years old, in which data were collected with weekly text messages over 2.5 years, the lower extremities were generally affected more often than the upper extremities [9]. The results of our review largely confirm this finding. Specifically, our results indicate that the most commonly injured anatomical sites were the knee and lower leg.

They were two exceptions to this: handball and gymnastics. In handball the most commonly injured areas were elbow and the lower leg; they were equally frequent. In the two articles concerning gymnastics, the foot was the second most common affected area. In one article, the knee was in the first position, whereas in the other it was the lower leg.

Furthermore, our results indicated the most common diagnoses to be tendinitis/bursitis and periostitis and these were similar across all sports (when reported) but, again, no incidence estimates could be extracted.

In relation to the external validity of these results, it is not known how representative the study samples would be for children as a whole, as the articles reported on specific sport groups and there were no population-based studies. The main objective in most of the studies was to describe the musculoskeletal problems that occurred in a sports club or school, i.e. using convenience samples. A larger number of studies would probably have given a wider range of results. Low response rate could also affect the validity of data, and response rate information was provided in only five studies.

Injury was never clearly defined on its own. Instead, problems arising in relation to a sport activity would be reported if there was also a time loss for that sport, or if medical attention was required. The definition of “injury” was therefore a definition of consequences, e.g. time-loss and/or medical attention, rather than a painful condition.

Further, the overuse definition lacked a substantial pathological aspect relying mainly on the observation of gradual onset of symptoms. Bahr has already discussed this problem [7]. We propose that it would probably decrease the risk for misclassification, if at least three of the following criteria were fulfilled: repeated micro trauma, no single identifiable cause, activity exceeds tissue tolerance, and gradual onset.

Furthermore, the level of expertise differed between persons responsible for defining the injury, ranging from medical specialist to the injured individual. Depending on the definition and site of injury, it can be expected that self-reported problems might be less accurate than those obtained through a medical examination. The definition is important when comparing estimates between studies, because, as pointed out by Bahr, self-reported pain will result in higher estimates than pain with a consequence in terms of medical attention or time-loss [7].

The fact that not all diagnoses and sites of injuries were clearly reported in each article was an additional difficulty.

In sum, this area of research would benefit from a well-reasoned consensus approach to the various relevant definitions, making it possible to compare findings between studies.

Relevant articles written in other languages than those included in our review may exist and their inclusion could have changed the results.

Checklists and evidence tables used for data extraction were tested for user-friendliness and adjusted in a pilot study before being used in the main study. The review was carried out independently by two reviewers, one of whom was experienced in performing systematic reviews. Because the objectives of the articles were different from ours, it was frequently necessary to discuss and interpret their texts to correlate with our checklist but the third reviewer never had to be called in for arbitration.

In some sports, such as handball and gymnastics, the upper limb is subjected to more repetitive stress than in other sports, such as soccer and running. One could assume that it therefore should be affected at least as often as the lower limb but, generally even in these sports the lower limb was most commonly affected. Perhaps the explanation for this is that the lower limb always carries the weight of the body, which puts the lower limb in a constant stress situation. In comparison with this constant weight, other sport-associated demands on the upper body are probably much less important.

The results of our study indicate that the most frequent diagnoses are fairly similar across sports, mainly reported as tendinitis/bursitis, and periostitis. Not surprisingly, OIE seem to occur mainly where the bodily structures are most subjected to repeated stress.