The proportion of injury that is related to sport and recreation activities reported in geographically, culturally and climatically diverse countries clearly establishes sport and recreation participation as a leading cause of the paediatric injury burden.1 For example, epidemiological studies conducted in emergency departments and hospitals in Australia,2 3 Norway,4 5 New Zealand6 and France7 have reported that sport and recreational injuries account for 15–20% of injury presentations, and an injury surveillance system in Swansea, Wales reported that 36% of children’s fractures were due to sport and leisure activities.8

However, although the contribution of sport and recreation to the overall burden of injury is well documented, most studies have not quantified the incidence of sport and recreational injury in terms of exposure to risk. The studies that have quantified exposure-specific risks of injury are few and largely focused on a limited number of types of sport. No systematic reviews have been conducted that collate information from across the literature and synthesise the information to facilitate translation of the science to public health benefit. A summary of the available exposure-based injury risk estimates is essential to compare the injury risk for different types of activities across population distributions, and to support calculations of the balance of potential benefit versus harm for various activities.9

This subject is critical given recent efforts to encourage greater participation in physical activity by children to combat the growing obesity epidemic.10 11 A responsible public health policy requires that the potential risks as well as the benefits of increased physical activity through sports participation be quantified. Increased knowledge about the injury harms associated with specific physical activity exposures is an important part of an overall risk management strategy.

The aim of this systematic review was to summarise the literature reporting exposure-based injury rates for various forms of physical activity in children younger than 16 years of age.

METHOD

Inclusion criteria

Studies were included in the review if they reported injury rates in relation to a measure of exposure to a type of physical activity for children in the age range 5–15 years. The types of activities covered included both sports and individual activities such as athletics, martial arts and bicycle riding. However, studies on playground or general school time injuries were not included in the review. Studies were included if they measured either organised (community or school) or unorganised activity. Both prospective and retrospective studies were included, but the results are summarised separately.

The measures of exposure included were primarily hours of activity participation, but studies that used other exposure denominators such as athlete exposures, participant seasons, balls bowled and “player plays” were also included. Studies were excluded if they did not provide an objectively measured exposure denominator enabling the quantification of meaningful injury rates. Studies were also excluded if it was not possible to calculate injury rates for children exclusively in the 5–15 year age range. Hence, studies that included older children or adults were excluded if they did not report age breakdowns of injury occurrence and sample size.

Search strategy

A number of techniques were used to locate relevant studies for the review. These included electronic database searches (Ovid Medline, CINAHL, EMBASE), searching international government and university websites for published reports, scanning reference lists of relevant publications, and hand searching selected scientific journals including the American Journal of Sports Medicine, British Journal of Sports Medicine, International Journal of Sports Medicine, Journal of Science and Medicine in Sport, Pediatrics and Injury Prevention.

Data extraction

For each included study, all available information on the sample population (age, gender, source), injury definition, method of injury ascertainment and measurement was extracted. Injury rates per the nominated measure of exposure with 95% CI were also extracted for children aged 5–15 years. For some studies, both injury rates and 95% CI were provided, but for others these were calculated from information provided in the text, tables or graphs. For a limited number of studies, there was insufficient information available to calculate CIs.

Some studies provided information that allowed the breakdown of injury rates into gender, age, play mode (game or practice) and injury severity categories.

DESCRIPTION OF STUDIES

The literature search revealed a large number of studies on the topic of childhood injury and physical activity, but the vast majority of published articles were case series reports which did not define the population at risk and were limited to analysis of those who sought treatment only.

Forty eight studies that met the inclusion criteria were located. Twenty seven of them had prospective study designs and reported injury rates per hours of player exposure to the sport or activity. Table 1includes the description of these studies. The remaining 21 studies which reported injury rates in relation to some other exposure measure are described in table 2. The latter is divided into two sections: prospective (n = 14) and retrospective (n = 7) study designs.

Tables 1 and 2 display information for each study on the location, activity studied, sample size (or other size measure, eg, player hours), population characteristics, injury definition and method of data measurement. In table 1, injury rates are expressed per 1000 hours of player exposure, and in table 2, they are displayed according to the relevant exposure measure.

The included studies reported injury rates for 14 different activities: soccer (indoor and outdoor, 20 studies), ice hockey (nine studies), baseball/softball (six studies), American football (six studies), rugby (league and union, six studies), basketball (two studies), handball (two studies) and Australian Rules Football (AFL), netball, volleyball, bandy, bicycle riding, judo and karate (one study each).

Injury definitions

A wide range of injury definitions was used by the included studies, mostly dependent on time lost from activity participation or the necessity for medical treatment. Other injury definitions were limited to a particular type or body region of injury including upper extremity injuries,59 facial injuries58 and elbow injuries.54

Injury definitions had a big effect on the resulting injury incidence, mainly because of the differing threshold of severity necessary to constitute an injury event. The most inclusive definition of injury encompassed minor discomfort that did not restrict further play,30 and the most restrictive required a minimum of 7 days absence from the sport.41

Injury ascertainment and measurement

The studies used various methods to identify and record details of injury episodes including direct observation of matches by trained study personnel (five studies), injury surveillance implemented either in the general community or during sporting tournaments (11 studies), reporting by team coaches or other school or team personnel (26 studies), self or parent report (12 studies), and retrospective analysis of medical insurance claims (six studies). Nearly a quarter of the studies combined two12 16 18 20 21 33 34 36 39 41 or three40 of these methods to identify injury cases.

Once injuries were ascertained, the relevant information was obtained, normally by a combination of methods including direct examination or interview with the injured player or a parent or from data recorded by either the team coach or other team personnel, injury surveillance staff, a treating specialist or doctor or on medical insurance claims.

METHODOLOGICAL QUALITY ASSESSMENT OF INCLUDED STUDIES

The most salient methodological limitations associated with the included studies were those related to the injury capture processes, and direct observation was used in only five of the studies. This was particularly a problem when the technique did not guarantee that all incidents that complied with the injury definition would be identified. For example, de Loes and Goldie15 used a definition for injury that required limitations to further participation, but injuries were identified by an injury surveillance system implemented by medical treatment facilities within the community. It is possible that participants may have sustained an injury that met the definition criteria, but did not seek professional treatment. These cases would therefore have been missed by the surveillance system.

Failure to identify all injury cases was possible even when the method used to capture cases was more coherent with the injury definition. Team coaches or parents may under-report injury occurrence, and insurance claims data may similarly be incomplete. To minimise the loss of potential injuries, some studies used more than one method for capturing data. Examples of this included cross-checking reports from team personnel against emergency department records,12 direct observation of randomly selected games, as well as reliance on coach reports40 and the collection of reports from both team coaches and parents of players.21 41

Methodological quality of the included studies was also affected by limited reporting. Only seven of studies included 95% CIs along with injury rates,18–20 27 45 47 56 although most studies provided sufficient information for their calculation to be possible. They could not be calculated for five studies because of a lack of either sample size or injury incidence reporting.28 34 38 53 55 Information was also generally lacking with regard to drop-outs and compliance with the study protocol.

Additional limitations were that the study populations were generally convenient samples with no random selection of participants, and exposure was mostly estimated collectively rather than on an individual basis.

RESULTS

Injury risk per exposure hours

Twenty seven studies reported injuries per hours of exposure. Risk per 1000 h for various sports ranged from 0.04 to 127.3. The highest injury risk was recorded for ice hockey, and the lowest was for soccer, although wide variations occurred between studies reporting the risk of soccer injuries. These variations in injury risk spanned a difference of over 1000-fold (ranging from 0.04 to 75.8 per 1000 h) depending on the threshold of injury definition, age of players, and type of soccer (indoor vs outdoor). CIs were very wide for two studies,12 25 but generally indicated the estimated injury rate to be reasonably close to the reported result.

Injury risk related to other measures of exposure

Other measures of exposure included per participant season (11 studies), per game or practice (six studies), per participant per year (three studies), per participant lifetime (two studies), per balls pitched (for baseball, one study), and per player plays (for American football, one study). The highest rates of injury per player season were reported for ice hockey for players aged 9–11 years (70 per 100 player seasons)40 and soccer for players aged 11–14 years (51.2 per 100 player seasons).52

Age and gender differences

In general, injury rates increased with age, but this was not always the case. The most notable exceptions were for ice hockey, for which two studies showed a trend toward higher injury rates in younger players.31 40

Gender differences were often non-significant, but girls were injured at a higher rate than boys in most studies presenting gender differences for soccer injury.

DISCUSSION

This systematic review found 48 studies that quantified the risk of injury from physical activity in children aged 5–15 years. Thirteen different sports and activities were represented, with soccer the game studied most often.

The magnitude of injury risk varied across sports: the highest injury rate per hours of exposure was reported for ice hockey, and the lowest was for soccer. However, it appeared that injury rates were greatly influenced by the definition of injury used and the age of participating players.

The highest injury rates expressed as injuries per player season were also reported for ice hockey. Direct comparison between injury rates per player seasons may be problematic, however, if the duration of the playing season varies between sports or organisations such that players are exposed to different numbers of games of practice sessions. This information was not provided in all of the studies.

Comparison between sports using a measure not based on hourly exposure is more difficult. However, some authors contend that such exposure measures may be inappropriate.49 For example, it has been argued that the constant interruptions that characterise a game of American football mean that players are inactive during a significant proportion of game time, rendering a time-based exposure measurement meaningless. Stuart and colleagues49 therefore expressed injury rates for American football in terms of player plays. A similar approach was adopted by Hale,55 who expressed injury rates in terms of balls pitched during football, and by Kujala and colleagues,56 who reported injuries per bicycle riders and bicycle uses.

Although a wide number of different sports were represented by the included studies, the current knowledge of physical activity-related injury risk in young children remains limited. Except for one study which reported cycling injury rates, all the studies were conducted in organised sports. There were no studies that reported injury rates for unorganised, daily activity.

In addition, very few children at the lower end of the age spectrum were included in the studies, with only seven including children aged less than 8 years.21 29 36 38 47 50 52 The study results indicate that injury incidence increases with age, but the magnitude of injury risk in very young children remains unknown.

Policy implications

The risk of physical activity-related injury must be considered within the context of the substantial benefits that children gain through participation in sports and activities. In general, the overall injury risk may be considered to be quite low compared with the opportunities for improving physical, psychological and mental health that many activities offer.

At some point, however, questions will be asked about sports and activities for which a higher injury risk is consistently demonstrated. Discouraging children from participation in such activities is undesirable. However, anxious parents may not permit children to participate if they believe their safety is compromised.

Modification of rules for younger players is one approach that has met with previous success in reducing injuries in some sports. Such approaches are being explored for many activities and coupled with improvements in coaching techniques, equipment safety and maintenance of playing fields may improve the safety of childhood sports and activity without reducing the associated enjoyment and health benefits.

CONCLUSION

The risk of injury per exposure time in children aged 15 years and younger has been quantified for a large number of sports and activities; however, direct comparison of injury risk between sports remains difficult because of varying definitions of injury and heterogeneity in study methods. The injury rates per hours of exposure varied between studies and sports by a magnitude of 10 000. Very few studies have ascertained sport-related injury risk for very young children (aged 8 years and younger).