Journal Logo

BASIC SCIENCES: Epidemiology

Sociodemographic Predictors of Sport Injury in Adolescents


Author Information
Medicine & Science in Sports & Exercise: March 2008 - Volume 40 - Issue 3 - p 444-450
doi: 10.1249/MSS.0b013e31815ce61a
  • Free


Adolescence has been identified as a time of life that predisposes the individual to a unique set of risk factors for injury in sport, which may be physical, psychological, or social in origin. Physical characteristics of the developing body place young athletes at high risk for fractures and overuse injuries (1,2,18). It has also been suggested that inappropriate training techniques and aggressive attitudes fostered by parental and coaching pressures contribute to the injury rate (2,18).

Risk factors for injury may be extrinsic to the individual (such as sport played or playing surface) or intrinsic. Intrinsic risk factors can be physical factors such as skeletal immaturity and developmental variability (e.g., size), psychological, or social. There are many factors that may predispose an adolescent to injury in sport, even prior to the specific mechanism of injury (i.e., pivot maneuver on the basketball court or body contact in hockey) (6). Some risk factors may be modifiable, such as exposure to the inciting event (i.e., games or practices), and some may be potentially modifiable (such as risk taking behaviors or use of protective equipment), but others factors are nonmodifiable (such as sociodemographic factors and previous injury). Identification of nonmodifiable risk factors can assist in determining high-risk populations, which should be targeted with appropriate injury-prevention strategies. Although sports and recreation injuries (SRI) have been identified as the leading cause of injury in adolescents, both on a population basis (16,19) and emergency department (ED) visits (2,12,24), few studies have focused solely on injury risk factors in the adolescent age group.

Previous research has suggested that risk factors for injury in children may be quite different to those for adolescents and that the risk factors for SRI may be quite different from risk factors for other injuries (25). A study of injury hospitalizations in children aged 0-19 in Manitoba shows that among children from rural areas, those from the lowest-income areas had rates almost three times higher than those from the highest-income areas (4). For a given income quintile, children in urban areas had almost twice the injury rate than those in rural areas (4). Unfortunately, in many injury studies, adolescents are often combined with young adults (e.g., aged 15-24 yr) or children (e.g., less than 19 yr), and participation or exposure hours are not considered (2,5,20). In addition, sports and recreation injuries are usually examined as a subgroup of all injuries, resulting in a lack of precision (19).

There has been a call for population-based SRI studies rather than hospital-based studies (ED presentation and admissions) or sport-specific studies (10). Injury data based on attendance at a hospital ED has the substantial limitation of capturing only a portion of the most severe SRI injuries and excludes the majority of medically treated injuries (24). Therefore, we designed a school-based survey to specifically examine the risk factors for SRI in adolescents (aged 14-19 yr). We estimated that in the Calgary area, 94% of high school students participated in sport and recreational activities in 1 yr. Sixty percent of these students reported at least one SRI during the year, and 40% sought medical attention for an injury (7). Participation in sports and recreational activities has been promoted as good for physical and mental health, but it is not without risk of injury (23). Injuries incurred in sports and recreation in adolescence may discourage further participation as a result of chronic problems, such as early development of osteoarthritis (28), or fear of further injury, which can subsequently lead to a sedentary adult life. It has been estimated that 8% of adolescents drop out of sport annually because of injury (13).

We present here, the results of a multivariate analysis of sociodemographic risk factors for sport injury to identify population subgroups at risk.


Study population.

A retrospective survey design was used to examine sport participation, associated injury, risk factors, and safety behaviors in Calgary area adolescents. Thestudy sample comprised high school students (grades 10-12, ages 14-19) from 24 Calgary and surrounding area high schools. These schools were randomly approached to participate from 38 Calgary and area high schools (including Calgary Public School Board, Calgary Separate School Board, Rockyview and Foothills Public School Boards, Christ the Redeemer Separate School Board high schools) (7). From each school, two mainstream classes were selected from each of grades 10, 11, and 12 (7). Informed consent was required to participate as per ethics approval from the office of medical bioethics, University of Calgary, and five participating school boards. High school students completed a 40-min written questionnaire in the classroom in the winter term of 2004. A similar questionnaire was developed previously in Australia and was customized to suit Canadian adolescent sports participation and seasonal differences (7). The variables covered in the questionnaire included subject demographics, sport participation, sport injury in the previous year, and injury-prevention practices (i.e., warm-up activities, use of protective equipment).

Outcome definition.

For this analysis, we defined sport injury as a positive response to the question, "In the past year, have you had at least one sports injury?" Students who responded affirmatively were asked to think of their most serious injury and respond to the question, "When you were injured, were you treated by a medical person (e.g., doctor, physiotherapist, certified athletic therapist, chiropractor)?" Thus, we had two categories of SRI; the first category covered all SRI, and the second covered medically treated SRI based on one reported "most serious" injury in the past year.

Risk factor definitions.

Students were asked to report their age, gender, and ethnicity and location of residence: in a city, small town (population > 1000), or rural area. As a measure of social economic status, students were asked to report the highest educational level achieved by both mother and father separately, on a seven-point scale ranging from less than grade 7 to postgraduate work at university. Finally, students were asked to report their height and weight, from which we calculated the body mass index (BMI) in kilograms per square meter. For this exploratory analysis, BMI was divided into six categories based on the 10th, 25th, 50th, 75th, and 90th percentiles. As a measure of exposure to sports and recreational activities, students were asked to report the number of hours per week and the total number of weeks that they engaged in their top three activities for participation in the past year. Exposure (average number of hours per week of participation) was also divided into six categories based on the 10th, 25th, 50th, 75th, and 90th percentiles.

Statistical methods.

Statistical analyses were performed using Stata Statistical Software, release 8.0. Incidence proportion (IP) based on the most serious injury reported in the past year and expressed as the proportion (%) of participants injured in the past year and odds ratios were estimated for different levels of the categorical variables. Standard errors, 95% confidence intervals, and P values were adjusted to take into consideration the potential for clustering effects within schools. A multivariate main effects logistic regression model (allowing for clustering within schools) was developed, with all of the previously described potential predictor variables as candidates for inclusion, but only those that were at the 5% level of significance were included in the final model. We used a GEE approach to logistic regression, which adjusts for the effects of clustering through employing robust variance estimators through consideration of intracluster correlation. All possible two-way interactions were examined.


Study participants.

Twenty-four of 38 possible high schools participated in this survey study. Of 3986 students who were requested to participate, 2873 students returned the survey, and 2850 completed at least part of the survey, resulting in an overall compliance of 71.5%. Compliance with completion of the survey ranged from 41 to 98% within each school (16). To be eligible for inclusion in this analysis (N = 2721), the students had to identify that he or she participated in sport and physical activity and to provide an answer to the question, "In the past year, have you had at least one sports injury?" Age and gender distributions of the sample are presented in Table 1. The majority of the students were white (85%). The next-largest group was Asian (110 Oriental and 35 Indian), and the remaining students were black (46), South American (31), or other non-Caucasian (152). Most participants lived in Calgary (45.6%) or a small town (29.7%), and 24.6% lived in rural Calgary. More than 66% of the fathers had postsecondary educations, as did 60% of the mothers.

Demographic risk characteristics of the sample (N = 2721) and incidence proportion (%).

Sport injury.

The overall IP rates of self-reported and medically treated sports injuries were 67.5% and 43.2% per year, respectively. The 95% confidence intervals, adjusting for the clustering effect of schools, were 64.2-71.1% and 40.4-46.3%, respectively. IP by explanatory variables are presented in Table 1, and the results of the univariate logistic regression models are presented in Table 2. All potential predictor variables were significantly related to injury, except for age and gender. In general, the coefficients for all SRI were very similar to those reported for the medically treated SRI.

Univariate logistic regression predicting injury.

The results of the multivariate analysis are presented in Table 3. There were four simultaneous significant predictors of both all SRI and medically treated SRI: ethnicity, location of residence, body mass index, and weekly hours of exposure. The adjusted odd ratios were remarkably similar for both types of injury. Students who lived in small towns had a lower risk of both self-reported SRI and medically treated SRI than did those who lived in Calgary (OR = 0.72 (95% CI 0.58-0.88) for all injuries and OR = 0.78 (95% CI 0.65-0.93) for medically treated injuries) and also compared with those who live in rural Alberta. Non-Caucasian students had a lower risk of injury than did Caucasian students (estimated OR = 0.59 (95% CI 0.47-0.76) for all SRI and 0.54 (95% CI 0.44-0.68) for medically treated SRI). Students with a body mass index less than 20 (< 25th percentile) had the lowest risk of injury for both types of injuries, and the relationship between BMI and the risk of injury seemed to be curvilinear, with the highest risk of injury in adolescents with a BMI between 25 and 26 (75th-90th percentile) for SRI and between 23 and 26 (50th-90th percentile) for medically treated SRI. This relationship did not differ by gender. The risk of injury increases as the hours of participation per week increase. After including the exposure variable in the model, the variables representing parental education were no longer significant.

Multivariate main effect logistic regression model predicting medically attended sports injury.


Sport injury.

The SRI rates in this study may seem high at first glance, but it must be remembered that these rates encompass the entire pyramid of the incidence of SRI rates, which has seldom been described. In general, injury incidence rate pyramids have been viewed as three layers: emergency room visits as the base, hospital admissions in the middle, and fatality at the peak. We suggest that the pyramid describing SRI incidence rates actually consists of six levels, as in Figure 1. The lowest two levels of injury rates have not usually been reported, but in fact they account for the vast majority of the injuries, as we have seen in this study. The base of the pyramid consists of SRI self-treated injuries (such as rest, ice, compression, and elevation), and the next level consists of injuries that receive first aid by a nonmedically qualified person (such as a parent, teacher, or coach). The next level consists of medically treated SRI either by qualified medical or paramedical personnel other than physicians such as physiotherapists, certified athletic therapists, nurses, or paramedics. These three layers of injury incidence have seldom been reported. The traditional three levels of the injury incidence pyramid then top this broad base. It should be noted that the pyramid presented here depicts injury incidence only, not the severity of injury. Whereas emergency department data may suggest a gauging of the severity of SRI, it is extremely difficult to capture exposure data, and, as such, person-time incidence rates cannot be calculated from these data sources. In the context of sociodemographic predictors of sports injury, it should be noted that in Alberta, costs of treatment by physicians are covered by the provincial health care plan, but treatment by other medically trained personnel is not, although partial costs are recoverable through extended health care insurance. As such, capturing complete SRI and exposure data through these insurance data sources is impossible.

Triangle of sport and recreational injury incidence.

Because of the design of the questionnaire, only students who had participated in sports during the previous year were asked to provide details about their injuries. Thus, the estimated SRI IP rates of 67.5% and 43.2% for medically treated SRI are for adolescents who have participated in sports and recreation during the previous year. Among the respondents, 93.5% participated in some form of sporting activity during the previous year. Therefore, the population-based estimates of injury incidence can be obtained by multiplying our estimates by 0.935, which would result an estimated IP of medically treated SRI of 40.4%.

International comparisons of estimated injury incidence are extremely difficult to make. We provide a brief discussion of some of these issues here. There are different definitions of what constitutes a "medically" treated injury; some definitions are much more inclusive of "first aid"-type medical treatment. Also, there are different definitions of what constitutes a "sports and recreational activity." Biases are introduced by recall periods (14), which range from 6 wk to 12 months, and in some studies the respondent is a parent rather than the adolescent. In addition, different age groups within adolescence are surveyed, and response rates are highly variable.

In particular, we observed no significant difference in IP by age or gender. This is consistent with some surveys (21) but not with others (3). The lack of a difference by age is probably attributable to the fairly small range of age in our study. Some sport- and injury-specific studies (i.e., ACL injury) have found higher injury incidence in girls (26), whereas overall higher SRI incidence is more often reported in boys (27).


The relationship between BMI and SRI has not received much attention previously. The cut points as summarized in Tables 1-3 are 10th percentile (18 kg·m−2), 10th-25th percentile (19-20 kg·m−2), 25th-50th percentile (21-22 kg·m−2), 50th-75th percentile (23-24 kg·m−2), 75th-90th percentile (25-26 kg·m−2), and 90th percentile (≥ 27 kg·m−2). We observed an interesting curvilinear relationship, where the lowest risk of injury was observed in adolescents with a BMI < 25th percentile or > 90th percentile, after controlling for other factors. The highest risk was observed for adolescents with BMI in the range of 50th-90th percentile, which encompasses the high end of the normal range (18.5-24.9) and the low end of the overweight range (25.0-29.9). We included BMI in our analysis because many sports are played on gender- and age-based teams, and we expected adolescents who are small for their age to be more likely to be injured in contact sport particularly (6). Conversely, it has been hypothesized that adolescents who are overly heavy may be at high risk for injury (6,15). However, our results suggest the opposite. Michaud et al. (17) found, in a multivariate logistic regression model, that chronological age, gender, and body mass index were not significant predictors of injury (self-reported injury, half of which were SRI), but that adolescents (aged 9-19) with Tanner pubertal stage of 4/5 had an odds of injury that was 1.3 (95% CI 1.2-1.4) times that of those with stages 1/2/3. They conclude: "The risk of injury appears to be linked to biological development (pubertal stage) more than to actual size and weight, body mass index, or chronological age." Clearly, further research is needed in this area, because BMI may be a modifiable risk factor.

Parental education.

In the univariate analysis, we found a gradient of risk of injury associated with the level of parents' education, but after adjusting for exposure these variables were no longer significant predictors. It has been previously recognized that there is an association between SES and the occurrence of various diseases or conditions in children. In particular, an increased risk of injury mortality has been associated with lower SES (9). It is controversial whether this negative association pertains to injury morbidity (8). An analysis by Pickett et al. (24) suggests that contrary to expectation, that there may be an increased risk of SRI associated with increased SES. This was confirmed in a cross-national study done by the same authors (25). Results from a study in Scotland in 1994 found no evidence of a socioeconomic gradient in the overall incidence of medically treated SRI, but a positive relationship between both parental occupation and family affluence in terms of sports injuries (29).


We found that non-Caucasian adolescents had a much lower risk of SRI than Caucasians, both for all SRI and medically treated SRI, suggesting that the injury rate is not dependent on cultural difference in seeking medical care. In addition, ethnicity is less likely to be an economic factor in Canada than in the United States. Recently in the United States, Conn et al. (5) have estimated that the age adjusted rate of SRI among whites (28.8 per 1000 population) was nearly 1.5 times higher than that of black people (19.0 per 1000 population). Similarly, Ni et al. (20)have shown that the risk of recreational injuries for non-Hispanic white children (ages 6-17) was twice the risk for non-Hispanic black and Hispanic children (P < 0.001). For non-Hispanic whites, the rates for 11- to 17-yr-olds were 163.5 (11.5) for boys and 96.4 (8.7) for girls; for Hispanic boys it was 62.8 (10.9), and for Hispanic girls it was 48.2 (12.5). For non-Hispanic blacks, the rate for boys was 86.4 (13.50), and for girls it was 38.9 (9.8). The reasons for this disparity remain unknown, but they are likely multifactorial, and they may include physiological and psychosocial factors.

Location of residence.

There is little evidence of the effect of urbanicity of place of residence in the SRI literature; however, it has been shown consistently that sport team membership and the presence of sport and recreation facilities are associated with an elevated risk of injury (1,11,21,27). Increased risk of injury in the urban community in our study may relate to more opportunity to participate in organized sport setting in recreational facilities in these communities. In addition, there may be the additional risk that urban settings may have more hazardous facilities/places (i.e., skateboard parks) where these SRI may occur. Contrary to our findings, the annual prevalence of injuries, including all non-SRI, in Maryland adolescents (ages 11-17), was 65% in the rural areas and 53% in the urban areas (27). This may be related to an increased risk of farm-related injuries in these rural communities. On the other hand, consistent with our study, Overpeck et al. (22) found no significant effect of location of residence in an analysis of all injury risk (including non-SRI) using the 1988 NHIS, and Ni et al. (20) found no difference in the age- and sex-adjusted recreational injury rates in children (aged 6-17 yr) between rural and urban areas.

We included both all SRI in addition to the medically treated SRI, so that we could assess whether any differences in injury rates were attributable to differences in access to medical care for students in rural areas. We found that youth living in small towns had a significantly lower risk of both self-reported SRI and medically treated SRI than did youth living in the city or in rural areas. This suggests that the lower medically treated SRI risk is not simply attributable to access to medical care, even though the small towns included in this survey did not have hospitals.

Strengths and limitations.

Because the survey response rate was variable between schools, our results should be interpreted with caution. The survey response rate was higher (almost 100%) for adolescents from the small town and rural areas, because the school boards required only "assumed consent" or "adolescent assent" following parental review of an informational letter, rather than "written parental consent." This suggests that the estimation of the effect of small town versus adolescents living in rural areas is more accurate than comparisons involving students who live in the city. It is possible that higher rates of injury reported in the city were related to selection bias based on parents being more likely to complete and return consent forms if their child both participated in sport and had sustained a sport injury.

Based on cross-sectional data, we were able to estimate exposure to risk (person-time) based on three sports for each participant. While we included this estimated exposure variable in our model, a true incidence rate based on person-time was not estimated. A limitation in this study includes that the estimation of IP based on the proportion of participants injured in the previous year was used to determine risk, which does not consider person-time.

It is difficult to measure SES using a survey of adolescents. Often, adolescents do not know enough about their parent's occupation to report it accurately enough for coding purposes, and they are also unable to report family income accurately. Ni et al. (20) have suggested that at least two measures of SES be used. In their study, SES was defined by two indicators: family income level, and highest educational level attainment of all adults in the family. We found that injury rates increased as parental education increased, which agrees with some recent results. However, this was primarily attributable to the relationship between parental education and exposure. Adolescents whose mothers had attended university had a mean of 9.1 h·wk−1 (SD = 5.9) compared with a mean of 5.9 (SD = 5.5) in those whose mothers had less than a high school education.

The strengths of this study are that it is population based, it includes a large random sample of adolescents, it focuses on adolescents aged 14-19, and it examines SRI only. Limitations include potential bias from the nature of the self-report survey techniques used. Other studies demonstrate underreporting of injury in similar studies requiring 1-yr recall (3). It seems that injury rates were higher when the adolescent rather than the parent responds to the questionnaire.


High rates of sport participation and sport injury in adolescents lead to a significant public health impact. Preventionof sport injury in adolescents is critical to maintain high rates of participation in physical activity and to reduce the risk of early osteoarthritis related to injury. This study identifies specific sociodemographic risk factors for injury in adolescents, which will help target appropriate populations for future research examining risk factors and prevention strategies in adolescent sport. Risk factors including ethnicity, home location, and BMI must be considered in future research in injury prevention in adolescent sport. This study also highlights the importance of considering both exposure to risk (i.e., hours of participation) and level of treatment following injury in future research.

This research was supported by the Canadian Institutes of Health Research, Institute of Musculosketoskeletal Health and Arthritis. Carolyn Emery is supported by the Alberta Heritage Foundation for Medical Research and the Canadian Institutes of Health Research. We acknowledge the school boards, high school principals, and participating teachers. Without their support, this research would not have been possible. We are especially grateful to the many high school students who consented to participate in this study.


1. Alexander CS, Somerfield MR, Ensminger ME, et al. Genderdifferences in injuries among rural youth. Inj Prev. 1995;1:15-20.
2. Bienefeld M, Pickett W, Carr PA. A descriptive study of childhood injuries in Kingston, Ontario, using data from a computerized injury surveillance system. Chronic Dis Can. 17 1996:21-7.
3. Bijur PE, Trumble A, Harel Y, Overpeck MD, Jones D, Scheidy C. Sports and recreation injuries in US children and adolescents. Arch Pediatr Adolesc Med. 1995;149:1009-16.
4. Brownell M, Mayer T, Martens PJ, et al. Using a population-based health information system to study child health. Can J Public Health. 2002;93(Suppl. 2):S9-14.
5. Conn JM, Annest JL, Gilchrist J. Sports and recreation related injury episodes in the US population, 1997-99. Inj Prev. 2003;9:117-23.
6. Emery CA. Risk factors for injury in child and adolescent sport: a systematic review of the literature. Clin J Sport Med. 2003;13:256-68.
7. Emery CA, Meeuwisse WH, McAllister JR. Survey of sport participation, sport injury and sport safety practices in Calgary and area adolescents. Clin J Sport Med. 2006;16:20-6.
8. Engstrom K, Diderichsen F, Laflamme L. Socioeconomic differences in injury risks in childhood and adolescence: a nationwide study of intentional and intentional injuries in Sweden. Inj Prev. 2002;8:137-42.
9. Faelker T, Pickett W, Brison RJ. Socioeconomic differences in childhood injury: a population based epidemiologic study in Ontario, Canada. Inj Prev. 2000;6:203-8.
10. Finch C, Valuri G, Ozanne-Smith J. Sport and active recreation injuries in Australia: evidence from emergency department presentations. Br J Sports Med. 1998;32:220-5.
11. Gerberich SG, Gibson RW, French LR, et al. Injuries among child and youth in farm households: a regional rural study. Inj Prev. 2001;7:117-22.
12. Gotsch K, Peal AJH. Nonfatal sports- and recreation-related injuries treated in emergency departments-United States, July 2000-June 2001. Morb Mortal Wkly Rep. 2002;51:736-40.
13. Grimmer KA, Jones D, Williams J. Prevalence of adolescent injury from recreational exercise: an Australian perspective. J Adolesc Health. 2000;27:266-72.
14. Harel Y, Overpeck MD, Anderson J, Scheidt PC, Bijur PE, Jones DH. Effect of recall on estimating annual nonfatal injury rates for children and youth. Am J Public Health. 1994;84:565-99.
15. Kaplan TA, Digel SL, Scavo VA, Arellana SB. Effect of obesity on injury risk in high school football players. Clin J Sport Med. 1995;5:43-7.
16. King M, Pickett W, King A. Injury in Canadian youth: a secondary analysis of the 1993-94 Health Behaviour in School Aged Children Survey. Can J Public Health. 1998;89:397-401.
17. Michaud PA, Renaud A, Narring F. Sports activities related to injuries? A survey among 9-19 year olds in Switzerland. Inj Prev. 2001;7:41-5.
18. Micheli LJ. Sports injuries in children and adolescents. Br J Sports Med. 1991;25:6-9.
19. Mummery WK, Spence JC, Vincenten JA, Voaklander DC. A descriptive epidemiology of sport and recreation injuries in a population-based sample: results from the Alberta Sport and Recreation Injury Survey (ASRIS). Can J Public Health. 1998;89:53-6.
20. Ni H, Barnes P, Hardy AM. Recreational injury and its relation to socioeconomic status among school aged children in the US. Inj Prev. 2002;8:60-5.
21. Nordstrom DL, Zwerling C, Stromquist AM, Burmeister LF, Merchant JA. Identification of risk factors for non-fatal child injury in a rural area: Keokuk County Rural Health Study. Inj Prev. 2003;9:235-40.
22. Overpeck MD, Jones DH, Trumble AC, Scheidt PC, Bijur PE. Socioeconomic and racial/ethnic factors affecting non-fatal medicallyattended injury rates in US children. Inj Prev. 1997;3:272-6.
23. Paffenbarger RS, Kamput JB, Lee IM, et al. Changes in physical fitness and other lifestyle patterns influence longevity. Med Sci Sports Exerc. 1994;26(7):857-65.
24. Pickett W, Garner MJ, Boyce WF, King MA. Gradients in risk for youth injury associated with multiple-risk behaviours: a study of 11,329 Canadian adolescents. Soc Sci Med. 2002;55:1055-68.
25. Pickett W, Molcho M, Simpson K, et al. Cross national study of injury and social determinants in adolescents. Inj Prev. 2005;11:213-8.
26. Powell JW, Barber-Foss KD. Sex-related injury patterns among selected high school sports. Am J Sports Med. 2000;28:385-91.
27. Riley AW, Harris SK, Ensminger ME, et al. Behavior and injury in urban and rural adolescents. Inj Prev. 1996;2:266-73.
28. Roos EW. Joint injury causes osteoarthritis in young adults. Curr Opin Rheumatol. 2005;17:195-200.
29. Williams JM, Wright P, Currie CE, Beattie TF. Sports related injuries in Scottish adolescents aged 11-15. Br J Sports Med. 1998;32:291-6.


©2008The American College of Sports Medicine