In childhood, the most common site of fracture is the distal forearm. To determine whether young girls with these fractures have low bone density more commonly than fracture-free controls, we measured bone density at the radius, spine, hip, and whole body and total body bone mineral content, lean tissue mass, and fat mass by dual-energy X-ray absorptiometry in 100 Caucasian girls aged 3–15 years with recent distal forearm fractures and 100 age- and gender-matched controls. Bone density (age-adjusted ratios of all cases:controls with 95% confidence intervals) was lower in cases at the ultradistal radius 0.963 (0.930–0.996), 33% radius 0.972 (0.945–0.999), lumbar spine 0.945 (0.911–0.980), hip trochanter 0.952 (0.918–0.988), and total body 0.978 (0.961–0.995). Moreover, osteopenia (defined as Z score below −1), was more common in cases than controls (p < 0.05) in the forearm, spine, and hip, with one third of fracture cases having low spinal density. Odds ratios (95% confidence intervals) for low bone density were: ultradistal radius, 2.2 (1.1–4.6); lumbar spine, L2-L4, 2.6 (1.3–4.9); and femur trochanter, 2.0 (1.0–3.9). Fracture patients aged 8-10 years weighed more (mean ± SD) than age-matched controls (37.2 ± 8.0 kg vs. 32.5 ± 6.6 kg, p < 0.01) while older patients reported lower current and past calcium intakes than matched controls (p < 0.05). We conclude that low bone density is more common throughout the skeleton in girls with forearm fractures than in those who have never broken a bone, supporting the view that low bone density may contribute to fracture risk in childhood.
Accumulation of strong bone during childhood and adolescence is considered critical for the development of optimal peak bone mass.(1,2) In children, forearm fracture incidence peaks at the age of maximal growth spurt, and some have postulated that this may be a manifestation of poor bone mineral density (BMD).(3-6) Low BMD is known to increase fracture risk in adults,(7-9) but measurements of BMD in fracture children are sparse(10) and limited to single-photon absorptiometry.(11,12) Yet early identification of osteopenia in childhood could enable dietary and exercise changes to be instigated which would enhance BMD, help prevent fractures, and improve adult bone health.
The distal forearm is the most common site of fracture in childhood.(4) Our study objective was to determine whether children with a recent forearm fracture have lower BMD, either in the forearm or at other skeletal sites, than age-matched children who have never sustained a fracture.
MATERIALS AND METHODS
The study was approved by our hospital ethics committee. Participants gave informed consent.
All Dunedin girls aged 3–15 years (n = 123) with distal forearm fractures (radius, ulna, or both) confirmed on X-ray, who were discharged consecutively from our hospital between February 1994 and June 1995, were invited to enroll; 103 girls were studied (83.7% participation). Data from non-Caucasians (n = 3) were excluded. All children were treated as outpatients. They were given a light below-elbow cast which left fingers and thumbs free so that movement of the shoulder joints and elbows was not restricted. Most casts were removed after 3–4 weeks, though seven girls had casts for 6–7 weeks. No restrictions on physical activity, apart from advice not to participate in contact sports, was given by the orthopedic surgeons. No child underwent bedrest or immobilization after fracture, and all returned to school within a few days after fracture.
Each fracture case provided the names of three friends of their own age; the first friend who had never had a fracture and agreed to take part as a control was enrolled. We contacted 116 friends; 7 refused to participate and 7 were ineligible (past fractures). We studied 102 controls (participation rate 93.6%) and excluded non-Caucasians (n = 2).
The final study sample had a power of 80% to detect a difference of 0.05 g/cm2 in spinal BMD at the 5% level of significance.
Each child came to the hospital with a parent/guardian. Information was collected by questionnaire concerning medical history, dietary calcium intakes (current and past),(13) and physical activity (hours per week spent in vigorous exercise). Self-assessed body image was determined to the extent that older girls (6–15 years, n = 85 per group) were asked whether they wished to be the same weight, heavier, or lighter. Tanner stage of puberty was self-identified by a validated technique.(14) Parents were asked to rate height gain (centimeters in the preceeding year).
Children were weighed and measured without shoes in light clothing. Metal objects (zippers, money, buckles, jewelery) were removed before scanning the lumbar spine, left hip, total body, and one radius (Lunar DPX-L dual-energy X-ray absorptiometer, Software packages 1.5e and 1.3z; Lunar Corp., Madison, WI, U.S.A.). Cases were scanned within 6 weeks of cast removal (average time 47 days postfracture), with density being measured in the nonfractured arm. In controls, scanning of the radius was matched for dominance with each index case. Our scanning precision (coefficients of variation) in adults is (%): 1.42 for neck of femur, 0.67 for lumbar spine L2–L4, 0.70 for total body, 1.61 for ultradistal radius, 1.52 for 33% radius, 2.52 for fat mass, 1.11 for lean tissue mass, and 1.52 for total body bone mineral content (BMC).
Results are presented as means (SD). Log transformation was undertaken for all anthropometric and scan data (SAS Institute Inc., Cary, NC, U.S.A.). Each of the transformed variables was regressed on age to derive the residual mean squares as a measure of the standard deviation using only the control group data. The regression coefficients were used to compute the predicted values for both groups, and individuals whose observed values fell below or above 1 SD of the predicted values were attributed Z scores below −1 or above +1. In the first analysis, Mantel–Haenszel odds ratios and odds ratios based on logistic analysis adjusting for weight as a categorical variable were calculated. In the second analysis (Fig. 1.), results were obtained using bone mineral measures and body weight as continuous variables. Because our samples were matched for age, age was regarded as a confounding factor and included in this analysis.
The children were in good health. Two girls (one case, one control) had insulin-dependent diabetes. Twenty cases and 22 controls took medication irregularly for mild asthma, but no girls were taking oral corticosteroids. One fracture child was a slow learner, and in each group four girls had consumed soy milk rather than cow's milk in early childhood. No child had any history of anorexia.
Pubertal development was similar in cases and controls: Tanner stage 1, n = 51 and 54; stage II, n = 23 and 22; stage III, n = 14 and 9; stage IV/V, n = l2 and l5. Cases achieved menarche at 12.7 (± 0.8) years and controls at 12.2 (± 1.2) years.
One child broke her arm on a playground roundabout; the others sustained fractures after falls, mostly occurring at play. Precipitating trauma, classified according to Landin,(4) was slight in 60 girls, moderate in 38 girls, and severe in 2 girls. Most girls (n = 71) fell directly on a hand or wrist, 38 falling forward, 31 backward, and 30 to the side. One child broke both arms at the same time falling backward from a swing. Thirty-three cases had previous fractures: 3 fractures (n = 3), 2 fractures (n = 12), and 1 fracture (n = 18). No fractures were from traffic accidents.
The number of cases in each 1-year age group from age 3–15 years, was: 2, 3, 10, 6, 6, 6, 16, 15, 13, 13, 2, 5, and 3, showing that fractures peaked at 9–12 years of age (57% of all fractures). Tables 1 and 2 show unadjusted data by age with ratios of the mean results for all cases relative to the mean results for all controls derived from analysis of covariance of the log-transformed data, adjusting for age. Cases and controls were well-matched for age and height, but fracture girls tended to have greater fat mass (Table 1), with those aged 8–10 years being 4.7 kg heavier (p < 0.01) with higher body mass index (p < 0.02), fat mass (p < 0.05), and lean tissue mass (p < 0.005) than age-matched controls. Despite extra weight bearing, total body BMC was not greater in these fracture cases than in controls. Furthermore, age-adjusted ratio data showed that the total body BMC of cases was significantly lower in cases, being 97.6% (CI 96.9–98.4) of the controls (Table 1). Bone values at the ultradistal radius, spine, and trochanter were significantly lower in fracture girls aged 11–15 years (Table 2). Age-adjusted ratio data showed that cases had bone densities that were 3–6% lower than those of controls at all sites except the neck of femur (Table 2). The largest significant difference was seen in the spine (94.5% [CI 91.1–98.0]) and the smallest in the total body (97.8% [CI 96.1–99.5]). Despite these differences, the groups did not differ in bone size, and no significant differences between cases and controls in the age-adjusted ratio data for cross-sectional bone area (cm2) were seen at any site. Thus, ratios with 95% confidence intervals for bone area cases:controls in our whole population sample were: ultradistal radius 1.011 (CI 0.972–1.052), 33% radius 0.994 (CI 0.969–1.019), L2–L4 0.971 (CI 0.941–1.003), and total body 1.006 (CI 0.973–1.040).
Table Table 1. Anthropometric and Body Composition Data
Table Table 2. Bone Mineral Content (BMC) and Areal Bone Mineral Densities (BMD) at Different Skeletal Sites
There were more individuals with osteopenia (Z score below −1) and more with high fat mass (Z score above +1) among the cases than among the controls (Table 3). Odds ratios calculated using data unadjusted for weight indicated that osteopenia was significantly more common in cases than controls at the ultradistal radius, lumbar spine, and femoral trochanter. Adjusting data for weight further increased these odds ratios. Weight adjustment also strengthened odds ratios calculated by logistic regression for a 1 SD reduction in each variable (Fig. 1). These values were statistically significant at every measured bone site (content and density).
Table Table 3. Odds Ratios for Bone Measures and Weight Measures
We found no evidence that bone density differed in fracture children reporting slight trauma and more severe trauma (p > 0.1). In addition, including terms for asthma in the models did not add to the explanatory power of the models and changed the odds ratios for bone measures by very small amounts. No significant differences between cases and controls, respectively, were noted for use of medication, pubertal development, age at menarche, low total daily dietary calcium consumption (below Recommended Dietary Allowance,(15)n = 19 and 21), low levels of vigorous physical activity (<8 h/week, n = 35 and 26), or rapid growth (>7 cm height gain/year, n = 33 and 36). However, more cases than controls expressed a desire to be lighter (43/85 vs. 22/85; χ2 = 8.93, p < 0.05).
Average dietary intakes of calcium (mg/day) from all dairy products for 100 cases (858 ) and 100 controls were similar (829 ) but fracture girls aged 11–15 years reported lower current calcium intakes from dairy products (787 ) than age-matched controls (947 , p < 0.05). Cases aged 11–15 years had also consumed milk less frequently (χ2 = 4.70, p < 0.05) and in smaller amounts between 6–10 years of age than controls, calcium intakes from milk at this age (mg/day) being 364 (222) versus 499 (296), p < 0.05. Current calcium intakes from milk (mg/day) were also lower in cases than controls aged 3–7 years (372  vs. 509  p < 0.02), but similar in cases and controls 8–10 years of age (451  vs. 395 ), and 11–15 years of age (428  vs. 510 ).
The most important conclusion to emerge from our study is that low BMD is more common throughout the skeleton in young girls with recent forearm fractures than in controls who have never fractured. This does not appear to be due to smaller bone size in fracture cases. We show for the first time that girls with distal forearm fractures have lower spinal and hip densities, and confirm reductions in radial density.(11) It is worrying that one third of our cases exhibited overt spinal and trochanteric osteopenia, which could increase their risk of osteoporotic fractures later in life. Like others, we found that forearm fracture in children is associated with only slight trauma.(4,6) Fall direction at fracture was the same as in adults.(16) However, fracture patients aged 8–10 years weighed more than their fracture-free controls, and more individuals with high body fat mass and more with osteopenia were found among study cases than among controls. Indeed, weight-adjusted odds ratios demonstrated that both low BMD and low BMC were significantly more common at every measured skeletal site in cases. Our results indicate that poor BMD may contribute to the risk of forearm fracture in young girls, as in adults.(8,16) We believe that low BMD has a causative, rather than a consequential, role in children with distal forearm fracture. Although bone mass may change in the fractured arm in adults after Colles' fractures,(17,18) we do not consider that osteopenia in our cases was due to postfracture bone loss. Because we measured radial density only 47 days after fracture in the nonfractured radius, and our children had remained physically active after fracture, we consider that low bone density throughout the skeleton cannot be explained by sedentary behavior postfracture.
Cases and controls had similar general health.(4) Although mild asthma was common, this was equally distributed in cases and controls. Moderate asthma medication does not alter BMD or bone growth in children.(19,20) Women rarely experience forearm fractures between 15 years of age and the menopause, when distal forearm density is high. By contrast, forearm fractures are common in childhood when radial density is low.(3,6) A low bone density, inadequate to take existing body weight during minor falls, seems a biologically plausible reason for fracture in children. In both genders, peak fracture incidence coincides with the growth spurt,(6,21) when the metaphyseal/diaphyseal density ratio is lowest and dissociation between longitudinal growth and mineral accrual increases bone fragility and alters bone quality and microarchitecture.(6) Parfitt(21) suggests that cortical bone porosity may develop during rapid growth to supply calcium needed by the growing ends of the long bones. In our cases, average bone densities were 3–6% less at different skeletal sites than in controls. Landin(11) reported similar forearm density reductions. In adults with wrist fractures, decreases in density of 4–8% occur at the hip and spine,(22) and these patients have an increased risk of sustaining fractures at other sites.(8,23-26)
Poor bone density in cases could arise from genetic or environmental factors, or both.(27) We did not examine genetic markers of bone, such as the vitamin D receptor genotype. Interestingly, recent work shows that in prepubertal girls, bone density differs according to vitamin D receptor genotype without there being differences in bone size.(28,29) Modifiable risk factors affecting density that we assessed included medication affecting bone, inadequate diet with low current or past calcium intakes, a less active lifestyle, a higher degree of obesity, delayed puberty or low estrogen status, and an excessive growth spurt. Of these factors, inadequate dietary calcium and excessive weight relative to bone development seemed the most likely factors related to fracture. Seeman(1) postulates that poor nutrition, inadequate (or excessive) weight-bearing, or poor hormonal status, during critical stages of growth, may reduce peak bone mass and induce site-specific osteopenia. Although delayed puberty or low circulating sex hormones could have contributed to reduced bone accrual, especially in the spine,(2,30) our cases and controls had similar pubertal development.
Since calcium intake influences mineral accrual in growing children,(10,31-34) inadequate calcium consumption for bone needs may have helped to generate low bone density. Avoidance of dairy products, which adolescents often perceive to be fattening, may have contributed to low calcium intakes since cases reported an increased desire to be thin. Older cases reported lower present and past calcium intakes than controls, and approximately one fifth of our girls were consuming less total calcium than recommended.(15)
In youth, physical activity with intermittent overloading of bone is strongly osteogenic, and regular exercise is critical for maximizing bone mass.(30,34-36) Inadequate bone development to support body weight would increase vulnerability to fracture. Obese girls have less bone for a given body weight than girls of normal weight.(37,38) Girls with fractures tended to be heavier than controls, the difference being significant in 8- to 10-year-olds. These girls had an increased fat mass which could indicate lower physical activity, although the vigorous exercise scores of cases and controls were similar. The excess adiposity of fracture children was unexpected, since adults with Colles' fractures tend to weigh less than controls.(22,39)
In summary, our study indicates that low bone density is found more commonly in girls with forearm fracture than in those who have never had fractures. Physicians should be alert to the possibility that children with forearm fractures may have poor bone density, which makes them vulnerable to fracture. Osteopenia could be transient or persistent. Future investigations should be undertaken to determine whether the osteopenia in young girls with forearm fractures persists into adulthood and the extent to which low BMD is due to genetic, hormonal, or environmental influences.
We are indebted to Prof. A.K. Jeffery, staff of the Fracture Clinic of Dunedin Hospital, and the children and their families who took part in the study. We thank Prof. G.O. Barbezat and Prof. J. Mann for critical advice. This work was supported in part by grants from the Otago Medical Research Foundation, the New Zealand Lotteries Grant Board, and the Health Research Council of New Zealand.