| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Menzies Research Institute, Hobart, Tasmania 7000, Australia
Address all correspondence and requests for reprints to: Graeme Jones, Menzies Research Institute, GPO Box 252-23, Hobart, Tasmania 7000, Australia. E-mail: g.jones{at}utas.edu.au.
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Bone mass has been inconsistently associated with fractures in children (7, 8, 9, 10, 11, 12, 13). The strongest evidence reported a significant association between bone mass and distal forearm fractures in both boys and girls (10, 11). However, all studies have variations with regard to the age group, choice of controls, and fracture type studied. These factors, when combined with generally small sample sizes, are likely to explain some of the inconsistency. Furthermore, other bone strength-related factors such as cortical thickness, biomechanics, and microstructure may also be important in fracture etiology in children (12, 14) as has been shown in adults (15, 16). These factors are difficult to assess in children due to concerns about radiation exposure. Metacarpal morphometry is a technique that may lead to further understanding of the role of these factors, but there have been no studies to date.
The aim, therefore, of this population-based case-control study was to investigate the association between bone mass assessed by both dual energy x-ray absorptiometry (DXA) and metacarpal morphometry and upper limb fractures in boys and girls 9–16 yr of age.
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Selection of cases and controls
From May 1998 to January 2002, subjects aged 9–16 yr from a preexisting fracture registry (6), who sustained a single-site upper limb fracture, were invited to take part. In brief, the fracture registry received reports containing the word "fracture" from all radiology providers 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, with a follow-up telephone call for subject and parental consent. Fracture subjects were excluded if they had diseases that would prevent them from completing the full protocol, such as cerebral palsy and arthrogryposis; if they had moved out of the study area; if they were not enrolled in school; or if they had a previous upper limb fracture between the ages 9 and 16 yr. 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 the 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 yr were excluded.
Anthropometry measurements and Tanner stage assessment
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. Tanner stage was assessed by a validated self-administered instrument using drawings made from Tanner’s photographs, which illustrate the five stages of genital development for boys and breast development for girls (17).
Bone density measurements
Bone mass was measured using DXA at the total body, lumbar spine, and right femoral neck. The instrument used was a QDR2000 densitometer on array setting (Hologic, Inc., Waltham, MA). Bone mass was examined as bone mineral content (BMC), areal BMD (aBMD), and bone mineral apparent density (BMAD). BMAD (grams per cubic centimeter) is an approximation of the volumetric density of bone and is calculated at different regional sites as BMC/Ap3/2 (lumbar spine); BMC/Ap2 (femoral neck); and BMC/[Ap2/h] (total body; Ref. 18). Precision estimates in vivo are not available for our subjects but are 1–2% in adults. The longitudinal coefficient of variation for our instrument between 1998 and 2002 using daily measurements of a spine phantom was 0.50% for aBMD. Body composition estimates were also available from the DXA scan.
Metacarpal morphometry
Measurements were made from postero-anterior left hand radiographs taken at a uniform 1-m tube-to-film distance. Morphometric measurements are performed at the midshaft of the left second metacarpal by one examiner. The detailed method in defining the inner limit and length of second metacarpal was as previously described (19). Hand radiographs were placed flat on a lighted viewing box. The measurements of total outer width (W) and medullary inner width (w) were made halfway up the second metacarpal with a digital caliper, which was calibrated to the nearest 0.01 mm. The intraobserver reliability was assessed in the measurement of outer width and inner width by repeating the assessment of 31 hand radiographs at 1 month. The coefficients of variation were 0.81% for outer width and 3.73% for inner width. The combined cortical thickness was calculated as (W - w). Metacarpal index (MI) was calculated as the ratio of cortical width to total width (W - w)/W.
Statistics
A 
2 test was used for the comparison of the age distribution, percentage of males, and proportion of fracture types, and unpaired t tests were used for the age comparison between respondents and nonrespondents of cases. Due to the matched design, paired t tests were used to compare the mean differences of anthropometry, DXA, and metacarpal measures between cases and controls. To estimate odds ratios (ORs) for fracture risk conferred by a 1 SD reduction, Z-scores were calculated for all measures. Firstly, bone measures were log-transformed. Secondly, each log-transformed variable was regressed on age for the controls only. The standardized residuals from this linear regression were then saved as Z-scores for controls. The regression coefficient obtained for each variable was then used to calculate Z-scores for cases using standard formulae. All of the converted Z-scores were used as continuous variables in 1:1 matched conditional logistic regression to obtain ORs for all measures. Interaction terms (bone measures x gender) were created to test effect modification by gender. For further analysis of the association between low BMD and fractures, a low Z-score was defined as less than -1. The variables with the highest ORs were selected from DXA and metacarpal measures individually and then forced into a multivariate model to determine whether they were independent predictors of fracture. Discriminant function analysis was used to get an integrated discriminant score for each subject when taking both BMAD and MI into account in the function. The area under the receiver operating characteristic curve (AUC) was used to test statistical significance. A P value of 0.05 (two-tailed) or a 95% confidence interval (CI) not including the null point was regarded as statistically significant. All statistical calculations were performed on SPSS version 10.0 for Windows (SPSS, Inc., Chicago, IL).
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
), although there was a trend toward higher nonresponse among males. The number for different types of upper limb fractures was 91 in the hand, 190 in the wrist and forearm, and 40 in the upper arm.
|
documents the physical characteristics of the study population. Cases and controls were well matched in all factors in boys (all P > 0.05), whereas cases of girls had lower body mass index (BMI; P = 0.044) and less bone free lean mass (P = 0.049) than controls, but no significant differences in weight, fat mass, and fat percentage. The difference in BMI and lean mass between cases and controls in girls was no longer significant after adjustment for aBMD at most bone sites (all P > 0.05, except BMI difference after adjustment for total body BMAD). There was no difference in Tanner stage distribution between cases and controls in boys (P = 0.15) and girls (P = 0.66).
|
details the comparisons of DXA and metacarpal measures between cases and controls for different types of upper limb fractures. For total fractures, cases had lower aBMD and BMAD at all bone sites (1.1–3.3%; all P < 0.05). A larger reduction was observed in wrist and forearm fractures (1.2–4.5%; all P < 0.05 except total body BMAD), but neither clinically nor statistically significant differences were observed for hand (-1.6 to +1.2%; all P > 0.05), and only possible trends were observed for upper arm fractures (0.9–4.8%; all P > 0.05).
|
). The differences were statistically significant for wrist and forearm fractures, but not for hand or upper arm fractures. Results did not differ if fractures were broken down by body side (data not shown). Furthermore, MI did not differ between right-handed and left-handed controls [0.45(0.08) vs. 0.45(0.06); P = 0.12], and adjustment for handedness made little difference to the association between MI and fracture (data not shown).
Table 4 documents ORs (per SD reduction) for DXA and metacarpal measures for wrist and forearm fractures. As a whole group, the association between aBMD, BMAD, and fracture risk is similar to that of metacarpal measures. If the analysis is broken down by gender, aBMD and BMAD tended to be better predictors of fracture in girls, whereas metacarpal measures appear better in boys, but none of the differences were statistically significant (P > 0.05 for all interactions). Total body BMAD was the only bone measure that did not show any association with fracture risk in boys and girls. A low aBMD (Z-score < -1) was more common in cases at all sites (ORs, 1.8–2.1; all P < 0.05), but the percentage with low BMD only reached statistical significance at the lumbar spine in girls and at the femoral neck in the total group and in boys. Almost one third of cases with wrist and forearm fractures had low lumbar spine aBMD in girls (31% vs. 17%; P = 0.028) and low femoral neck aBMD in boys (29% vs. 16%; P = 0.019; Fig. 1).
|
|
). However, the best sensitivity and specificity remained modest.
|
|
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Previous fracture case-control studies in children are few, and the association of bone mass with fractures remains contradictory. Landin and Nilsson (7) and Chan et al. (8) reported a 7–8% reduction of forearm BMC in total fracture cases. Cook et al. (9), however, found that there was no difference in spine and femoral neck BMD between cases with traumatic fractures and controls. Recent studies from New Zealand (10, 11) using DXA measurement with a larger sample size reported that cases with distal forearm fractures had lower aBMD throughout all bone sites in both boys and girls. In contrast to this, a separate study using computerized tomography measurement of bone mass revealed no reduction in radius cancellous, integral, and cortical bone densities in cases with forearm fractures (12). Our results are broadly consistent with those of Goulding et al. (10, 11). Cases had 1–5% lower aBMD and BMAD throughout different bone sites, and wrist and forearm fracture risk increased approximately 1.5 times for each 1 SD reduction in BMD. This is only marginally lower than the 3–6% reduction in BMD and ORs of 1.6–2.0 in Goulding’s study (10). In comparison with adult studies, the strength of the association between BMD and fracture is somewhat weaker (20), suggesting that bone-independent factors may be more important in children as we have previously reported in a prepubertal sample (13). Indeed, there were no differences in all bone measures of either clinical or statistical significance in the prepubertal group within the current study (data not shown) or the New Zealand study (10, 13).
No particular DXA site stood out as a significantly better predictor of upper limb fractures in the current study, although total body BMAD was the weakest fracture predictor, possibly reflecting its somewhat artifactual mode of calculation. This is similar to the studies of Goulding et al. (10, 11), who reported the largest difference of bone density in lumbar spine and the smallest difference in total body, but the risk estimates for each site overlapped, indicating no site-specificity for distal forearm fracture prediction. On the other hand, this study has been the first to document different strengths of associations between bone mass and fractures at different upper limb sites in children. No significant association was found between BMD and either hand or upper arm fractures. This suggests that wrist and forearm fractures in children might be more bone mass dependent than other upper limb fractures, as has been shown for different adult fracture sites (20, 21). However, the sample size for hand and upper arm fractures was relatively small. This may result in lower power to detect an important difference in bone mass. The magnitude of the difference observed was small for hand fractures but larger for upper arm fractures, suggesting that some caution is necessary in concluding that bone density is not related to upper arm fractures.
In contrast to previous studies (10, 11, 12), we did not observe a higher weight and fat mass in cases compared with controls, suggesting that this association may not apply to all populations. Indeed, girls in the current study with upper limb fractures had a lower BMI and less lean mass than controls, which is directly opposite to previous reports (10, 12). This finding does seem more biologically plausible as bone mass increases proportionally to lean mass in boys and girls because most bone loading is due to muscle forces (22). Thus, a low bone mass may be contributed by low lean mass, consistent with our observation that the lean mass difference largely disappeared after adjustment for aBMD.
With respect to metacarpal measures, the results indicated that the strength of the association with wrist and forearm fractures was similar to DXA measures, and cases had a thinner cortical thickness and lower MI. MI has been reported to correlate modestly but significantly with DXA measures at all bone sites and particularly with the distal radius, and it is not as dependent on body weight and body surface area as aBMD (23), suggesting that MI may represent an inexpensive assessment of fracture risk in clinics where assessment of bone age is common. Although the measurement of MI requires training and a high-resolution caliper, another advantage is that this measure could be used as a parameter to reflect the structural properties of cortical bone, such as cortical thickness accrual in children and loss in adults (24, 25). Thus, the findings from metacarpal measures imply that both material and geometric properties of cortical bone are involved in the etiology of wrist and forearm fractures in children. This adds to a recent report that girls with forearm fractures had an 8% decrease in radial cross-sectional dimensions without a difference in cortical density (12). Moreover, a metacarpal measure is also able to assess both periosteal apposition and medullary contraction of cortical bone in children (25, 26). In the current study, we did not find a difference in outer width between cases and controls, but there was a significantly greater inner width in those who fractured. This implies that cases might have the same periosteal apposition as controls but have greater medullary expansion, probably due to a higher endosteal resorption or lower endocortical apposition. This concurs with observations regarding osteoporotic fractures in women where an enhanced endosteal resorption has been proposed as one of the principal mechanisms of osteoporosis (27, 28). The mechanisms of this in children are poorly understood. Low protein or calcium intake could lead to poor development of cortical thickness in children (26), whereas hormonal factors such as androgens, estrogens, and GH may also be important (29, 30). Further studies are desirable to understand the factors associated with medullary expansion or cortical thinning in children. Lastly, there was a nonsignificant trend for metacarpal measures to have a stronger association with fracture risk than DXA measures in boys and the reverse in girls. This possible gender difference merits further investigation in larger populations to explore potentially different fracture risk factors in boys and girls.
It is of marked interest that both BMAD and MI remained significant independent predictors of wrist and forearm fractures in multivariate analysis. This is likely to reflect a combined contribution of both trabecular and cortical bone to fracture risk in adolescence and suggests that measures incorporating multiple bone sites will lead to better prediction of upper limb fractures in children. Indeed, from the receiver operating characteristic curve analysis, we observed that the AUC was improved by approximately 25% after taking these two factors into account in comparison with BMAD only. Although the sensitivity and specificity for this combined model are modest, they are very similar to reports in adults (31) and probably reflect the role of bone-independent factors in fracture etiology in children as has been demonstrated in adults (32).
This study has a number of potential limitations. Firstly, the response rate was 56%, which is less than ideal. However, based on available information in nonrespondents, there were no significant differences between participants in mean age, male percentage, fracture types, and age distribution, suggesting that these associations may be generalizable to other populations. Secondly, the manual measurement of metacarpal morphometry may be affected by measurement error. However, this was measured by one observer, and reproducibility compares favorably to other studies (33). Lastly, bone density was measured after the occurrence of the fracture, and thus the actual difference in bone mass may have been magnified by post fracture bone loss. We attempted to minimize this bias by only studying upper limb fractures, which rarely require bed rest, and by measuring all subjects within 3 months of the fracture event. Prospective studies may appear ideal to resolve this conflict. However, they will also lead to potential measurement error from growth-related effects on bone density if there is a significant interval between measurement of bone density and occurrence of fracture.
In conclusion, a combination of DXA and metacarpal measures improves wrist and forearm fracture prediction in children. This suggests that low bone mass (both cortical and trabecular) is a risk factor for wrist and forearm fractures in children but not other upper limb fractures. The magnitude of this association is somewhat weaker than in adults but suggests that optimizing age-appropriate bone mass will lessen the risk of fracture in children.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Abbreviations: aBMD, Areal BMD; AUC, area under the receiver operating characteristic curve; BMAD, bone mineral apparent density; BMC, bone mineral content; BMD, bone mineral density; BMI, body mass index; CI, confidence interval; DXA, dual energy absorptiometry; MI, metacarpal index; OR, odds ratio.
Received October 28, 2002.
Accepted December 19, 2002.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  G. Ducher, S. L Bass, G. A Naughton, P. Eser, R. D Telford, and R. M Daly Overweight children have a greater proportion of fat mass relative to muscle mass in the upper limbs than in the lower limbs: implications for bone strength at the distal forearm Am. J. Clinical Nutrition, October 1, 2009; 90(4): 1104 - 1111. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. L Lambert, R. Eastell, K. Karnik, J. M Russell, and M. E Barker Calcium supplementation and bone mineral accretion in adolescent girls: an 18-mo randomized controlled trial with 2-y follow-up Am. J. Clinical Nutrition, February 1, 2008; 87(2): 455 - 462. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Winzenberg and G. Jones Recommended Calcium Intakes in Children: Have We Set the Bar Too High? IBMS BoneKEy, February 1, 2008; 5(2): 59 - 68. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Powell, D. Ma, and G. Jones Determinants of Skeletal Age Deviation in a Cross-Sectional Study J. Clin. Endocrinol. Metab., February 1, 2008; 93(2): 521 - 526. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Winzenberg, K. Shaw, J. Fryer, and G. Jones Effects of calcium supplementation on bone density in healthy children: meta-analysis of randomised controlled trials BMJ, October 14, 2006; 333(7572): 775. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E.M. Clark, J.H. Tobias, and A.R. Ness Association Between Bone Density and Fractures in Children: A Systematic Review and Meta-analysis Pediatrics, February 1, 2006; 117(2): e291 - e297. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. Eastell Role of oestrogen in the regulation of bone turnover at the menarche J. Endocrinol., May 1, 2005; 185(2): 223 - 234. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J H Davies, B A J Evans, and J W Gregory Bone mass acquisition in healthy children Arch. Dis. Child., April 1, 2005; 90(4): 373 - 378. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
V. Matkovic, P. K Goel, N. E Badenhop-Stevens, J. D Landoll, B. Li, J. Z Ilich, M. Skugor, L. A Nagode, S. L Mobley, E.-J. Ha, et al. Calcium supplementation and bone mineral density in females from childhood to young adulthood: a randomized controlled trial Am. J. Clinical Nutrition, January 1, 2005; 81(1): 175 - 188. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. B Leonard, J. Shults, B. A Wilson, A. M Tershakovec, and B. S Zemel Obesity during childhood and adolescence augments bone mass and bone dimensions Am. J. Clinical Nutrition, August 1, 2004; 80(2): 514 - 523. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. Barnes, P. Wong, B. Egan, T. Speller, F. Cameron, G. Jones, H. Ekert, and P. Monagle Reduced Bone Density Among Children With Severe Hemophilia Pediatrics, August 1, 2004; 114(2): e177 - e181. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 I. E. Jones, S. M. Williams, and A. Goulding Associations of Birth Weight and Length, Childhood Size, and Smoking with Bone Fractures during Growth: Evidence from a Birth Cohort Study Am. J. Epidemiol., February 15, 2004; 159(4): 343 - 350. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
