Systematic literature search

A first search was performed in May 2015 and a final search in April 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 exercises » 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 and no attempts were made to search the grey literature.

Inclusion criteria

We used the Preferred Reporting Items for Systematics reviews and Meta-Analysis (PRISMA) guidelines in this review [6]. The first author applied the inclusion criteria to the title and abstract of the articles identified as possible relevant research articles from the literature search. Full-text screening was then done by two authors independently of each other to determine which articles should be included in the review. Inclusion criteria were: 1) a study design that was prospective or cross-sectional; 2) at least 2/3 of the study population should consist of ≥19 years olds or results should be reported specifically for different age groups. To determine this we looked for information on the range age, the mean age with the standard deviation, and the proportion of adults, when data were reported for age groups. In study samples consisting of “professionals” but no further information of age, we assumed that these would consist mainly of adults; 3) the article must state clearly if reported cases were classified as traumatic or overuse injuries in relation to a particular sports type; 4) a sample size greater than 50; and 5) the article must not deal with specific occupational subpopulations (such as military) nor with clinical populations. Only articles in English, French or a Scandinavian language were considered, as the authors could read these languages.

Data extraction

The checklists were extracted from a previous review on OIE and sports’ type on children and adolescents [5]. We used two descriptive checklists, one quality checklist and three tables of results [5].

The quality checklist can be seen in Table 3. It was reported in this table if 1) the participation rate was stated (or could be calculated), 2) the injury was diagnosed by a health professional, 3) the diagnosis and anatomical site were clearly and completely reported, 4) the incidence of OIE was reported, and 5) if the number of injuries could be reported in relation to number of hours of exposure and individuals.

N no

Y yes; when positive answers have been highlighted

In Table 5 the numbers of OIE were listed by anatomical area. We highlighted those two that were most commonly reported. Table 6 showed the same type of information but based on the diagnosis.

The two most common injury sites in each article are highlighted: Like this for the most common and like this for the second most common

“?”= Information not provided

OIE: Overuse injuries of extremities

Articles in which all OIE are described and in which all the sites of OIE are clearly described are framed, i.e. Author/Year

*: the number of injuries was not reported in this article but we have the incidence so we could rank the localisation. 1 means the most often reported, 2, the 2nd most often, and so on

OIE: Overuse injuries of extremity

The two most common diagnoses in each article are highlighted: Like this for the most common and like this for the second most common

Six articles have been excluded in this table because they did not mention any diagnosis:

Articles in which all OIE are described and in which all the diagnosis of OIE are clearly described are framed, i.e. Author/Year

*: The number of injuries was not reported in this article but we have the incidence so we could rank the diagnoses. "1" means the most often reported, "2" the 2nd most often, and so on

The AMSTAR checklist [9] was used as a guide for this review. However, tests for homogeneity and publication bias were not carried out because no such statistical information could be extracted to be used in this review. Furthermore, articles were not screened for conflict of interest statements, as this aspect was irrelevant for the current topic (no obvious financial gains).

Review process and interpretation of data

Two of the authors extracted the information separately and blind to each other’s findings. Their findings were compared to detect extraction errors. The third author was available for arbitration in case of disagreements between the two reviewers. The quality data were used for descriptive purpose only and to provide a basis for research recommendations.

The review was registered in the PROSPERO database: CRD42015032477.

Number of articles

Initially, on the basis of the database and citation searches, 1435 articles were identified, leaving 1080 articles after duplicates were removed. Of these, only 10 were retained after scrutiny of their title, abstract and full-text. The criteria of non-inclusion of the articles are presented in Fig. 1. Most of the excluded studies did not deal with specific sports or OIE. Although it often was difficult to extract some of the data, it was never necessary to use the arbitration process.

Study design, participants and method

Three sports were covered in the 10 articles included in this review: soccer [10–17], beach volleyball [18], and triathlon [19]. They were published from 1989 to 2013 and nine were conducted in Europe. The design was prospective for all studies except for one [18], which combined a prospective and retrospective study, but only results from the prospective study were used in this review. In all studies, the study samples were obtained from sports clubs or at competitions. The level of sport participation of the study participants varied from recreational to elite level, but for the majority of articles it was at an elite level.

The number of participants ranged from 123 to 1507 (Table 1). Four studies included only men, three studies only women, and three studies both sexes (Table 1). The age of participants was not clearly described in all articles, but when it was, it varied from 15 to 39 (Table 1), with the mean age of 23 to 38 years.

The duration of data collection, when described, ranged from 26 weeks to 2 seasons and for one article it took place during five championships. The frequency of follow-ups within this duration, when reported, was often weekly (Table 1).

Definition of overuse sport injuries

The definition of ‘sports injury’ differed between articles (Table 2). Most commonly, a case was defined by time-loss and was nearly always depending on a link to the specific sport activity studied. The specific definition of overuse injury was most commonly based on the concept of the absence of a single, identifiable (traumatic) cause (N = 6), followed by gradual onset (N = 3), repeated micro-trauma (N = 2), and combinations thereof. Nobody stated explicitly that the activity had to exceed tissue tolerance, although this probably would have been taken into consideration during medical examination.

Quality of the studies

Although method sections in this type of studies often are very similar, specific information was sometimes difficult to obtain for our purposes. As can be seen in Table 3, response rates were often unreported, as well as incidence estimates. Therefore, it became necessary to calculate the proportion of cases based on exposure, which explains the last column in the quality checklist. In relation to outcomes, overuse injuries per anatomical area and diagnosis were often not systematically reported. On the positive side, health professionals were usually responsible for the data collection.

Incidence estimates of overuse injuries

The incidence estimates of OIE are shown in Table 4. These were reported in only four articles dealing with soccer but they all reported it differently.

Tegnander et al. [11] calculated the incidence distinguishing training from game exposure. Moreover, they reported the incidence for OIE in general with incidence estimates of OIE being 0.8 per 1000 h of game and 0.7 per 1000 h of training.

The other three articles reported the incidence based on 1000 h of sport participation. Jacobson et al. [12] provided the incidence for OIE based on the area injured, which varied from 0 for the hip, groin and thigh to 0.6 for the knee. Eirale et al. [15] provided the incidence for various diagnoses and localisations. Regarding the diagnoses, the incidence varied from 0.1 for fracture and synovitis/periostitis to 2.0 for muscle rupture/cramps. Regarding the localisation, the incidence estimates varied from 0.03 for shoulder, ankle and foot/toe to 1.7 for the thigh. Ekstrand et al. [16] reported the incidence for the most common OIE subtypes while combining the diagnosis and the localisation. The incidence varied from 0.03 for the ankle joint synovitis and calf muscle cramp/spasm to 0.5 for hamstring overuse/hypertension.

Proportion estimates of overuse injuries

Table 4 shows also the proportion of OIE based on exposure. It could be calculated in 8 articles and varied from 0.18 to 13.32 per 1000 h of exposure. The two studies that did not study soccer reported higher proportion of OIE than the others. Methodological differences could probably explain these results.

Injury site and diagnosis in general

The lower limb was most often affected (Table 5) and especially the knee, tibia, thigh and pelvis/hip/groin.

Only few articles described the diagnosis of overuse injury. For that reason, only 4 articles could be included in Table 6. The most frequently provided diagnoses were tendinitis/bursitis, and strain.

Differences in overuse injuries according to sports type

For all sports covered, the lower limb was more often affected than the upper limb. Again, it was impossible to compare the incidence rates between sports, because it was only reported in the articles on soccer. When considering the proportion of OIE per 1000 h of exposure, different results are found. For soccer, this proportion is <3.5 (and often around 1), 8 for triathlon, and 13 for beach volleyball. However, methodological considerations could well explain these differences.

In soccer, the pelvis/hip/groin appeared to be more commonly affected than in the two other sports.

We could not compare the diagnosis of OIE between sports because only articles on soccer reported the diagnosis.

Summary of findings

This appears to be the first systematic review on OIE in adults comparing the occurrence in various sports. We attempted to identify any differences between sports in relation to diagnoses and anatomical areas most likely to be injured. We were able to retrieve 10 studies on three different sports: soccer (N = 8), triathlon (N = 1) and beach volleyball (N = 1). Methodological differences between studies and a limited number of studies and sports studied made it difficult to provide clear answers. However, in relation to the proportion of OIE it varied between 0.2 to 13.3 per 1000 h of exposure, with soccer not having the highest estimates. This proportion is generally more important in adults than in youngsters, where results around 0.5 were found [5].

Injury site was, as for the youngsters [5], mainly the lower limb. However, in adults this was reported rather for the knee, tibia, thigh and pelvis/hip/groin whereas in youngsters it was the knee and the lower leg.

As for the diagnoses, they were most frequently (when at all provided) reported to be tendinitis/bursitis, and strain, whereas in children and adolescents the most commonly reported diagnoses were tendinitis/bursitis and periostitis [5].

Methodological aspects of the articles reviewed

A large body of literature on sports injuries of adults, as well as of children, is written by a group of researchers that uses the same methodological approach when surveying injuries in different sports. Typically, they study injuries in single sport clubs or during specific sports events with the ultimate goal to compare risk estimates for various sport activities. To record a sufficiently large number of injuries of specific sports in the general population is of course difficult, hence this approach. However, when choosing such a tactic, it would be relevant to collect similar data from several clubs/events, in order to even out any bias associated with single convenience samples of such type.

After having reviewed this literature on both children/adolescents and adults, it is clear that even when multiple studies are found for similar sports, data are often collected at different intervals, in different ways, using different definitions for injury, and for different specific types of injuries. Authors do not clearly report diagnosis and anatomical areas of injury, and if they do, they often leave out the one or the other. This, also, makes it difficult to make comparisons and to establish risk estimates. A simple example is the difference in estimates expected when the presence of an “injury” is reported as “complaint”, as “sought care”, or “time loss”. Further, in the case of “overuse”, absence of a traumatic etiology seems often automatically to result in a diagnosis of an “overuse” injury, merely because the person with the complaint was involved in a sporting activity. It is not logical that people involved in studies on sport injuries only have these two possible diagnoses, traumatic or overuse injury. Surely patients from the general population are diagnosed from a larger spectrum of possibilities. Clear criteria for this diagnostic label have been proposed [5] and discussed in the literature [20], but seem to be largely ignored, at least when reports are written up.

As for the definitions of “incidence” and “prevalence”, true incidence and prevalence estimates are usually not distinguished in studies within this area. The incidence is defined as number of injuries based on 1000 h of session (training, competition or both), in general without regards concerning the previous injury. In fact, this should not really be called incidence but prevalence. This issue has been previously discussed by Bahr [7]. Further, the numbers of potential and included study subjects are often not reported. Clearly, an injury rate (per 1000 h) would be more credible when obtained from many study subjects than from a few. It would therefore be useful for the reader to have access to both these denominators.

Admittedly, the objectives of our review were not the same as the objectives of the studies under review, which makes difficult the extraction of information in our review. Nevertheless, as we have already discussed in our previous review on children/adolescents [5] in our opinion, this research area would benefit from a well-reasoned consensus approach to the various definitions.

Methodological aspects of our review

Our review followed the current guidelines, using a transparent approach, searched several databases, and data were extracted blindly by two reviewers. However, it is possible that some articles could have been missed, as only texts written in English, French and Scandinavian languages were acceptable for inclusion. Checklists for data extraction have been previously tested and used in a previous review and were therefore known to be user-friendly and relevant.

Sometimes we had to make assumptions regarding the nature of injuries, when exact information regarding the site of injury was missing. Thus two diagnoses, tendinopathy and periostitis, were systematically considered as extremity injuries, whereas some diagnoses such as strain was not, because it could affect the spine.

Discussion of findings regarding the incidence of OIE

We did not find any information in the literature on OIE in the general population of adults. However, the incidence of OIE in general population of schoolchildren has been reported to be 2.3(1.6–3.0 95% CI) for the upper extremity and 3.7(3.5–4.0) for the lower extremity [4].

Discussion of findings regarding the anatomical site of OIE

As observed in the previous review on children and adolescents [5], the lower extremity is more often affected than the upper extremity in the sports studied. Only three sports could be considered in this review, so it is difficult to compare the localisation of OIE between sports. However, we noted that in soccer, in youngsters and in adults, the pelvis/hip/groin are more often affected than in the other sports. We assumed that this is due to the shearing force often imposed on the pelvis in soccer.

Discussion of findings regarding the diagnosis of OIE

Only four articles provided good information on the diagnosis of OIE and they all studied soccer, making it impossible to compare this finding with other sports. In childhood, 8 articles reported the diagnosis making a comparison relevant. However, for all sports covered, it was always the two same diagnoses that were reported.

Tendinis/bursitis is the most common diagnosis both in childhood and adulthood, followed in adults by synovitis, and in youngsters by periostitis. Probably because of the difference in bone skeletal maturity, osteochondral disorders, present in youngsters, did not appear in adults.