The effect of physical activity on upper limb fractures was examined in this population-based case control study with 321 age- and gender-matched pairs. Sports participation increased fracture risk in boys and decreased risk in girls. Television viewing had a deleterious dose response association with wrist and forearm fractures while light physical activity was protective.
Introduction: The aim of this population-based case control study was to examine the association between television, computer, and video viewing; types and levels of physical activity; and upper limb fractures in children 9–16 years of age.
Materials and Methods: A total of 321 fracture cases and 321 randomly selected individually matched controls were studied. Television, computer, and video viewing and types and levels of physical activity were determined by interview-administered questionnaire. Bone strength was assessed by DXA and metacarpal morphometry.
Results: In general, sports participation increased total upper limb fracture risk in boys and decreased risk in girls. Gender-specific risk estimates were significantly different for total, contact, noncontact, and high-risk sports participation as well as four individual sports (soccer, cricket, surfing, and swimming). In multivariate analysis, time spent television, computer, and video viewing in both sexes was positively associated with wrist and forearm fracture risk (OR 1.6/category, 95% CI: 1.1–2.2), whereas days involved in light physical activity participation decreased fracture risk (OR 0.8/category, 95% CI: 0.7–1.0). Sports participation increased hand (OR 1.5/sport, 95% CI: 1.1–2.0) and upper arm (OR 29.8/sport, 95% CI: 1.7–535) fracture risk in boys only and decreased wrist and forearm fracture risk in girls only (OR 0.5/sport, 95% CI: 0.3–0.9). Adjustment for bone density and metacarpal morphometry did not alter these associations.
Conclusion: There is gender discordance with regard to sports participation and fracture risk in children, which may reflect different approaches to sport. Importantly, television, computer, and video viewing has a dose-dependent association with wrist and forearm fractures, whereas light physical activity is protective. The mechanism is unclear but may involve bone-independent factors, or less likely, changes in bone quality not detected by DXA or metacarpal morphometry.
Fracture incidence is bimodal,(1) with a peak in later life caused by osteoporosis and a less recognized peak between 10 and 15 years, especially for upper limb fractures.(2–6) The reasons for this increase in early life are poorly understood. A lag of bone mineralization compared with linear growth is hypothesized based on comparisons of bone density accrual and linear growth,(4,7) while uncontrolled studies attribute the peak to an increased participation in both organized and informal sports as well as the overall high level of physical activity during adolescence.(2,8–10) Several case control studies have indirectly supported the first hypothesis, suggesting that low bone mass is a risk factor for wrist and forearm fractures in children.(11–13) On face value, the latter hypothesis seems contradictory because accumulating evidence in children and adolescents strongly suggests that physical activity is beneficial with regard to bone mass.(14–16) Thus, one would expect physical activity to decrease fracture risk in children unless the increased risk of fall-related trauma outweighed the positive effects on bone density. So far, few controlled studies have been done in a young population to assess the overall impact of physical activity on different types of fractures and whether this is mediated by or is independent of bone density. The aim, therefore, of this population-based case control study was to investigate the association between television, computer, and video viewing; types and patterns of physical activity; and upper limb fractures in children 9–16 years of age.
MATERIALS AND METHODS
This study was conducted from 1998 to 2002 in Hobart, Tasmania, and included the southern Tasmania metropolitan council areas of Hobart, Clarence, Glenorchy, and Kingborough. Whites are predominant in this population. Its aims were to investigate the role of growth, bone strength, sports participation, risk taking, and coordination in the etiology of upper limb fractures in children ages 9–16 years old. Subjects who provided informed consent to take part (and/or consent was obtained from their parent/guardian where relevant) underwent an extensive protocol involving measurements of anthropometry, pubertal stage, bone density in all subjects by DXA at the spine, hip, and total body, metacarpal morphometry, bone age, clumsiness, risk-taking behavior, physical activity, sunlight exposure, habitual intake of dairy products, questionnaire assessment by parent/guardian of socioeconomic factors, and details of fracture. The current study relates to physical activity only. Ethical approval for this study was obtained from the University of Tasmania Ethics Committee (Human Experimentation).
Selection of cases and controls
From May 1998 to January 2002, subjects who sustained a single site upper limb fracture and were 9–16 years old were invited to take part in a pre-existing fracture registry.(6) In brief, the fracture registry received reports from all radiology providers containing the word “fracture” from southern Tasmania on a monthly basis. This region is larger than and fully encompasses the study area. Potential subjects were then screened from the registry by the authors and invited to take part (by a letter of invitation) in a follow-up telephone call for subject and parental consent. Fracture subjects were excluded if they had diseases, such as cerebral palsy and arthrogryposis, that would prevent them from completing the full protocol; had moved out of the study area; were not enrolled in school; or had a previous upper limb fracture between the ages of 9 and 16 years. Nonresponders who were more than 3 months past the date of fracture were also excluded.
Controls were randomly selected from the same school class as the cases in a ratio of one control for every case (using available school lists and random numbers). They were also individually matched with cases by gender. Potential controls that had experienced an upper limb fracture between the ages of 9 and 16 were excluded. The response rate was 56% in both groups. There were no significant differences in mean age, male percentage, proportion of fracture type, or age distribution between responders and nonresponders for cases (all p > 0.05). although there was a trend toward higher nonresponse among males.(13)
A total of 32 subjects were excluded in the whole group because of prior fracture.
Weight was measured with light indoor clothing without shoes to the nearest 0.1 kg using electronic scales (calibrated at the beginning of the study by the manufacturer). Height was measured without shoes to the nearest 0.1 cm on a stadiometer. Body mass index (BMI) was calculated as weight divided by height squared (kg/m2). Overweight and obesity was defined using recent international age- and sex-specific standards.(17)
Types of trauma
Information of types of trauma associated with the fracture event was collected by questionnaire from the parent or guardian. The definition of degree of trauma was taken from previously published criteria.(2) Slight trauma was defined as an injury caused by forces exerted by falling to the ground from standing height or less. A fall from 0.5–3 m or similar velocity trauma was defined as moderate trauma, and greater than 3 m was defined as severe trauma. Where a fall was not involved in fracture etiology, a judgment was made on the equivalent degree of trauma.
Assessment of physical activity
Physical activity was retrospectively assessed in the year before study entry, using a questionnaire validated in U.S. adolescents,(18) which was modified after piloting to include popular Australian sports. The reference point was before the fracture date in both cases and controls. The test-retest Spearman correlation of overall leisure physical activity in hours per week over the last year was found to be 0.66. This questionnaire has demonstrated predictive validity in our hands.(19) This questionnaire has items on days of either vigorous activity or strenuous activity for greater than 20 minutes in the last 2 weeks (1, none; 2, 1–2 days; 3, 3–5 days; 4, 6–8 days; 5, 9 or more days); daily television, computer, and video viewing in last week (1, up to 1 h; 2, 2–3 h; 3, 4–5 h; 4, ≥6 h); number of competitive sports in the last 12 months (1, none; 2, one; 3, two; 4, three; 5, 4 or more); and activities done at least 10 times in the last 12 months. These sports were further defined as contact and noncontact sports. For the purposes of this study, the definition of contact sports is a sport that involves physical contact between players as part of normal play, such as Australian football league, rugby, and indoor/outdoor soccer. The definition of noncontact sports is a sport where contact with another player is not allowed or penalized. Only individual sports with more than 100 participants in total were analyzed because of sample size considerations.
Paired t-tests were used to compare the mean differences in age, anthropometry, and BMI between controls and cases. χ2 tests were used to compare the percentage of overweight/obesity and type of trauma between cases and controls for the different types of upper limb fractures. Univariate conditional logistic regression analyses were used to obtain the odds ratios for all of the physical activity variables. Test for trends of categorical variables were undertaken by replacing the binary variables used to estimate categorical effects with a single linear variable. In exploratory analyses, high-risk sports were defined as those sports with odds ratios greater than 1 with a p value <0.10 in boys only (as odds ratios were generally less than 1 in girls). These included Australian football league, rugby, soccer, cricket, and surfing. Any physical activity variable with a statistically significant association (p < 0.05) with that fracture type was examined in multivariate analysis. Where significant heterogeneity caused by gender was present, results are presented stratified by gender. For correlated statistically significant variables, backward stepwise regression was used to determine the best predictor for each type of fracture. This was done before and after adjustment for bone density (both DXA and metacarpal morphometry simultaneously). A p value less than 0.05 (two-tailed) or a 95% CI not including the null point were regarded as statistically significant. All statistical analyses were performed on SPSS version 10.0 for Windows (Cary, NC, USA).
A total of 642 subjects took part in this study (boys, n = 215 pairs; girls, n = 106 pairs). The number for different types of upper limb fractures was 91 for hand, 190 for wrist and forearm, and 40 for upper arm. The median time between fracture and interview was 105 days (IQR 77–115).
Table 1 documents the physical characteristics of the study population. Cases and controls were closely matched on age, weight, height, BMI, and percentage who were overweight or obese. However, the distribution of types of trauma was significantly different for the various upper limb fracture sites, with most trauma being observed for upper arm fractures and least trauma for hand fractures. Significant differences in number of sports were observed for high-risk and competitive sports for hand fractures; Tanner stage; light physical activity and television watching for wrist and forearm fractures; and contact and high-risk sports for upper arm fracture.
Table Table 1. Characteristics of Study Sample
Table 2 details the crude odds ratios for all physical activity variables by gender as well as the number of participants in each category. Contact, high-risk, and competitive sports were associated with a significantly increased fracture risk in boys, whereas total sports, noncontact sports, and high-risk sports were associated with a significantly decreased fracture risk in girls. A similar trend was observed for individual sports. Boys had an increased risk of fractures, although these only reached statistical significance for Australian football league/rugby, soccer, and surfing, whereas girls had a decreased risk that only reached statistical significance for soccer. Consistent gender discordant associations were observed for total, contact, noncontact, and high-risk sports participation and four individual sports (all p < 0.05). No gender discordance was observed for patterns of physical activity. However, statistically significant linear trends with total fracture risk were only observed in girls for light physical activity, strenuous physical activity, and television, computer, and video viewing, with the former two being protective and the latter deleterious.
Table Table 2. Comparison of Crude Odds Ratios for Types and Patterns of Physical Activity by Gender
Table 3 documents the crude odds ratios for all physical activity variables for wrist and forearm fractures stratified by gender. Light physical activity and competitive, contact, and high-risk sport participation was associated with reduced fracture risk in girls only, with no apparent trend in boys, whereas television watching increased fracture risk in boys, with a similar trend in girls. The television watching association was also significant if treated as a continuous variable in boys and both sexes combined (Table 3; Fig. 1), indicating a dose-response association. The association between television watching and fracture risk was very similar in those with and without a sports associated fracture (OR for non-sport-caused fractures and television watching: 1.4 [95% CI 1.0, 1.9]). High-risk and competitive sports were associated with increased hand fracture risk in boys only. The number of contact sports was associated with increased upper arm fracture risk in boys only, whereas light physical activity was protective (Table 4). No trends, either deleterious or beneficial, were observed for girls for these fracture types, probably because of an insufficient sample size (data not shown). Being overweight or obese was not significantly associated with any type of upper limb fracture.
Table Table 3. Odds Ratios of Physical Activity and Sports Participation for Wrist and Forearm Fractures in Boys and Girls
Table Table 4. Multivariate Model of Selected Factors for Different Types of Upper Limb Fractures
Table 4 documents the results of multivariate analysis. All associations persisted after adjustment for other physical activity variables and bone mass with the exception of contact sports and wrist and forearm fracture risk in girls and total and competitive sports and hand fracture risk in boys. All of the above associations did not alter after further adjustment for body composition (data not shown).
This population-based case control study is the first report comprehensively describing the associations between television, computer, and video viewing; physical activity, and upper limb fractures in children. Importantly, it adds to the existing literature on adverse effects of television viewing in children. Furthermore, the results suggest that an assessment of the impact of physical activity on the risk of upper limb fractures in children needs to consider types and patterns of physical activity as well as the effect of gender.
The increased time spent by children watching television/video or playing computer games is of increasing public health concern. It has been linked to physical inactivity, obesity, aggressive behavior, and short-term growth delay.(20–24) In this study, we found a deleterious dose-response association between time spent television, computer, and video viewing, and wrist and forearm fracture in both sexes, but especially in boys. This was not observed for hand and upper arm fractures. Somewhat surprisingly, this effect was independent of bone mass and patterns of physical activity in our population. In addition, it was also independent of coordination measures and risk taking (data not shown), which does not provide a clear mechanism for this association. One possible explanation is that there might be an adverse effect on bone quality that is not detected by DXA or metacarpal morphometry. Alternatively, it is more likely to be related to behavioral abnormalities such as social problems, delinquent and aggressive behavior, or externalization, which have been linked with excessive television, computer, and video viewing,(21,22) and closely parallel a report documenting a high prevalence of social competence and behavioral difficulties in children with fractures.(25) This implies that television, computer, and video viewing may be acting either as a cause or a marker of behavioral and psychosocial disturbance in children.
The effect of physical activity on fracture risk is determined by its net effect on fall-related trauma and bone strength. Based on observational studies, leisure time physical activity is associated with decreased fracture risk in older adults, and this reduction is larger than would be expected from an effect on bone density alone, suggesting it also reduces the risk of falling.(26,27) However, most uncontrolled studies in children have linked higher physical activity level and sports participation with increased rates of injury and fracture.(2,8–10,28) The results from two available case control studies are contradictory, with one reporting an inverse association between activity level and fracture risk and the other a positive association, although the former may have been biased by assessing physical activity with a reference point after the fracture event.(29,30) The current study favors a protective effect of strenuous activity in girls for total fractures and light activity in both sexes, but especially girls for wrist and forearm fracture. These associations were also largely independent of bone strength measured by both DXA and metacarpal morphometry, which we have previously reported as a fracture risk factor in this same sample.(13) This implies that the beneficial effect of physical activity is mostly caused by a reduction in the risk of fall-related trauma. Furthermore, the observation that physical activity but not obesity or overweight was associated with fracture risk does not support a primary role for obesity per se in the etiology of these fractures in our population. This is directly contradictory to studies in New Zealand children and suggests geographic variation in fracture risk with body composition.
Another novel finding was that sports participation was associated with fracture risk, and this association is dependent both on type of sport and type of fractures. Sports participation increased the risk of total, hand, and upper arm fractures in boys and decreased total and wrist and forearm fracture risk in girls. In this study, high-risk sports were Australian football league, rugby, soccer, cricket, and surfing. All of these sports except surfing are on the list of five most common sports associated with sport injuries in Australian children.(31) The similar trait is that most are contact and/or competitive team sports or have the potential for substantial trauma, suggesting participation in such sports may reflect a higher risk-seeking behavior as proposed by Patel.(32) This is also consistent with our previous report documenting sports participation as a risk factor for total fracture risk in prepubertal children independent of bone mass.(33) In addition, our recent report in this sample has shown low bone mass is associated with wrist and forearm fracture but not hand and upper arm fractures,(13) which suggests that a higher degree of trauma might be required for hand and upper arm fractures compared with wrist and forearm fracture. Our data partially support this with a higher percentage of moderate to severe trauma for upper arm fracture but less trauma for hand fracture. The latter finding could be explained by hand bones being smaller in size and thus requiring relatively less trauma to fracture.
A gender discordant effect of sports participation was also observed. There was a consistent trend for all types and levels of physical activity and sports participation to be protective for upper limb fractures in girls, although not all were statistically significant. Conversely in boys, participation in most sports seemed to increase fracture risk. This gender discordance was significant for four individual sports and all sport categories with the exception of competitive sports. The discordance was also present for all fracture types, but particularly hand and upper arm fractures. This is consistent with a previous uncontrolled report indicating a higher proportion of fractures caused by sport accidents in boys compared with girls.(2) These effects were independent of bone mass and suggest different approaches to sports in boys and girls. They imply that, specifically for fracture prevention, all sports should be encouraged in girls, but that boys should be steered toward sports with a lower potential for trauma.
This study has a number of potential limitations. First, the response rate was less than ideal. However, based on available information in nonrespondents, there were no significant differences with participants in mean age, male percentage, fracture types, and age distribution, suggesting that these associations may be generalizable to other populations. However, it is also possible that the nonparticipants differed in other factors that we did not assess, such as severity of fracture and patterns of physical activity. Excluding subjects with a previous fracture may have also decreased generalizability. However, relatively few were excluded, probably because of the young mean age of our sample. Second, the measurement of physical activity in children is difficult. In this study, physical activity level was assessed based on a 2-week recall period before the fracture event to avoid any alterations in physical activity induced by the fracture itself. Because this was measured within 4 months of the fracture event in most cases, there is a small potential for misclassification of activity based on poor recall. Prospective studies will be required to more fully address this. This questionnaire will reasonably assess current physical activity but may only approximate habitual physical activity. However, it correlates moderately well with past year physical activity, suggesting that this is not a major problem,(18) and the overall effect of measurement error will be an attenuation of the association between physical activity and fracture risk. Last, there was a high overall level of physical activity, with a high average number of sports and over 85% of controls and 83% of cases reporting >5 days per fortnight involved in strenuous physical activity in the last 2 weeks. This is consistent with a previous Australian study(34) but may make it more difficult to observe associations between strenuous activity and fracture risk
In conclusion, there is gender discordance with regard to sports participation and fracture risk in children, which may reflect different approaches to sport. Importantly, television, computer, and video viewing has a dose-dependent association with wrist and forearm fractures, whereas light physical activity is protective. The mechanism is unclear but may involve bone-independent factors, or less likely, changes in bone quality not detected by DXA or metacarpal morphometry.
We thank research assistants Fiona Wilson, Anitra Wilson, Lesley Oliver, and Val Walsh, and biostatisticians Leigh Blizzard and Jim Stankovich as well as the staff of the Medical Imaging Department at Royal Hobart Hospital. This work was supported by the National Health and Medical Research Council of Australia and Clifford Craig.