Four-Year Gain in Bone Mineral in Girls With and Without Past Forearm Fractures: A DXA Study


  • The authors have no conflict of interest


We have previously shown that girls with a recent distal forearm fracture have weaker skeletons than girls who have never fractured. This could be a transient or persistent phenomenon. The present study was undertaken to determine whether the bone mineral content (BMC) of girls with previous distal forearm fractures remains lower 4 years postfracture or if catch-up gain has occurred. We report baseline and follow-up dual energy X-ray absorptiometry (DXA) results for 163 girls: 81 girls from the original control group who remained free of fracture (group 1) and 82 girls from the original group with distal forearm fractures (group 2). In data adjusted for bone area, height, weight, and pubertal status, group 2 girls had 3.5-8.5% less BMC at the total body, lumbar spine, ultradistal radius, and hip trochanter than group 1 at baseline, and 2.4-5.7% less BMC at these sites at follow-up. Even girls from group 2 who did not experience another fracture after baseline (n = 58) did not display greater BMC at follow-up compared with baseline values at any site, indicating that the decreased BMC at the time of fracture had persisted. In group 2, the relative gain in BMC after adjusting for the initial BMC and current bone area, height, weight, and pubertal stage was less than or similar to, but not greater than that of group 1 (ratio [95% CI]: total body, 0.985 [0.972-0.998]; lumbar spine, 0.961 [0.935-0.987]; ultradistal radius, 0.968 [0.939-0.998]; hip trochanter, 0.955 [0.923-0.988]; femoral neck, 0.981 [0.956-1.007]; and 33% radius 0.999 [0.977-1.021]). These findings indicate that girls with distal forearm fractures do not improve their gain of BMC. We conclude that girls who have sustained a distal forearm fracture maintain their lower BMC at most sites for at least 4 years.


MORE THAN one-half of adult skeletal mass is accrued during the pubertal years of rapid bone growth.(1) Maximizing the total bone mass acquired during childhood and adolescence is considered to be as important for osteoporosis prevention as slowing bone loss in older age.(2) Consequently, optimizing bone gain during childhood and adolescence is critically important. Others have shown that genetics, gender, hormonal status, nutrition, and amount of weight bearing exercise during growth may all affect bone acquisition.(3–6) However, recent work suggests that the size of bone and its mineral content may track from childhood to sexual maturity.(7)

Distal forearm fractures are the most common childhood fracture,(8) and approximately 7500 occur annually in New Zealand children ages 3-15 years.(9) We have previously shown cross-sectionally that girls ages 3-15 years with a distal forearm fracture had lower bone density at all sites measured and lower total body bone mineral content (BMC) than their age-matched fracture-free controls.(10) We subsequently showed in these girls that low total body bone density at baseline independently increased the risk of new fractures.(11) These results indicate that those girls with a poorer skeleton have a higher risk of fracture in the same manner as adults.

Low bone mass in girls with distal forearm fractures could be transient, leading to only short-term increase in fracture risk, or more persistent. We are not aware of any previous studies in childhood examining bone mass in children with a fracture history at a known interval postfracture. Thus, we do not know whether catch-up gain in BMC occurs during growth to strengthen the bones of girls who have sustained distal forearm fractures in the past. The aim of this longitudinal study was to determine whether, 4 years from baseline, the BMC of girls with a distal forearm fracture remains lower than that of fracture-free girls or if catch-up gain has occurred.


In 1994 and 1995 we recruited a consecutive series of girls discharged from the fracture clinic of our hospital with a recent distal forearm fracture (n = 100) and age-matched friend controls who had never fractured (n = 100).(10) The girls were aged between 3 and 15 years and all were white. Girls with any fracture of the distal forearm (radius, ulna, or both; displacement or greenstick; growth plate or shaft) and any type of orthopedic management were included in the 100 cases we recruited initially. The study children were seen in our clinic for assessment of anthropometry, bone development, body composition by dual energy X-ray absorptiometry (DXA; DPX-L scanner; GE Lunar, Madison, WI, USA), dietary calcium intake, and physical activity.(10) We assessed density at the ultradistal radius, 33% radius, total body, left hip, and lumbar spine (Pediatric software version 1.5h; GE Lunar). In case subjects, the nonfractured forearm was measured within 6 weeks of plaster removal, and in control subjects, we were careful to match for dominance with each index case.

In 1998 and 1999, 4 years from baseline, we invited all of the original subjects to return for a second assessment of their bone health. The local ethics committee approved this phase of our study, and all participants and their caregivers gave informed written consent. Six girls declined to take part in the second phase, and 24 girls had relocated to live outside Dunedin, leaving us with a 96.6% response rate of those who continued to live locally. Figure 1 shows a flow-chart of the follow-up groups used for this study. This shows that seven girls from the original control group were excluded from the present analysis. They are not included because they sustained fractures after baseline and thus could not be considered fracture-free; this group was too small for separate analysis.

Figure FIG. 1..

Study profile.

The girls were seen 4 years (±2 weeks) after the date of their first measurements. Height (Harpenden stadiometer; Holtain Ltd., Croswell, UK) and weight (electronic scale) were measured in duplicate without shoes and in light clothing. Body composition, BMC, and bone area were measured by DXA (software package 1.35; GE Lunar) at the same sites scanned at baseline.(10) To minimize operator-related variability, the same experienced operator analyzed both the baseline and repeat scans. The scanner performance was monitored throughout the study period, and no significant scanner drift was observed. CV for repeat in vivo measurements in adults of BMC (g) and projected bone area (g/cm2), respectively, for the different sites measured are as follows: total body, 0.84% and 0.94%; lumbar spine (L2-L4), 1.23% and 1.35%; ultradistal radius, 1.96% and 0.59%; 33% radius, 1.36% and 0.96%; femoral neck, 2.42% and 1.34%; and hip trochanter, 2.73% and 2.02%.

Stage of pubertal development at both baseline and follow-up was self-assessed by a validated technique.(12) Participants were shown pictures and written descriptions of Tanner stages of pubertal development and asked to select the stage that most resembled their current development. The caregivers helped younger girls when required. Information was collected by questionnaire concerning menstrual history, medical history over the last 4 years, estimated hours of television viewing per day, and physical activity. Physical activity was assessed in two ways. First, each girl self-assessed their physical activity on a scale of “much less, less, the same, more, much more” physical activity in comparison with other girls their own age. Second, the girls were asked to estimate the amount of time spent in vigorous activity (defined as activity that makes them puff and sweat) on an average day. Dietary calcium intake was assessed using a validated food frequency questionnaire.(13)

STATA was used for all statistical analysis (Stata Statistical Software Release 7.0; StataCorp., College Station, TX, USA). Raw data for differences with 95% CIs for the 4-year gain in all variables measured are shown. Log transformation was undertaken for the remainder of statistical analysis of all anthropometric and DXA scan data; this was done to stabilize the variance, which increased with age. BMC was adjusted for bone area, height, weight, and pubertal stage using multiple regression following the method of Prentice et al.(14) Differences between the groups at the beginning and the end of the study are shown as ratios with 95% CIs using the fracture-free group as the reference group. No adjustments were made for multiple tests.

To assess the gain in BMC in our two groups, the initial BMC, final bone area, height, weight, and pubertal stage were included in the regression model for the final value of BMC when assessing relative gain in BMC. This was done because differences in BMC values between the beginning and end of the study are likely to be inversely correlated with the initial values, so those with the lowest bone mass at the beginning of the study seem to improve more than those with values closer to the center of the distribution, because of regression to the mean rather than because they have shown a greater improvement. If this effect is not taken into account, the comparisons are likely to favor the fracture group because their baseline values were lower than the fracture-free group. This removes the effect of the baseline value and provides valid estimates of differences in BMC gain between the groups.(15)


Tables 1 and 2 show the characteristics of our 163 subjects at baseline and after 4 years; 81 girls from the original control group who had never fractured at baseline and were still fracture free at time two (group 1), and 82 girls who were fracture cases at baseline (group 2). Twenty-four girls from group 2 had sustained 37 new fractures during follow-up as previously reported.(11) Sites affected were distal forearm (n = 23), fingers (n = 3), toes (n = 3), and thumb (n = 3), plus a single fracture at each of the scaphoid, patella, elbow, nose, and ankle. The girls were generally in good health at follow-up. In groups 1 and 2 respectively, four and six girls were taking inhaled steroids for asthma, and five and two girls were taking oral contraceptives. Three girls in each group reported regular cigarette use (>7 cigarettes per week). One girl from group 1 had insulin dependent diabetes, and one girl from group 2 reported she had chronic fatigue syndrome. There were no differences between the groups for medications or cigarette smoking.

Table Table 1.. Anthropometric Variables, Body Composition, and Pubertal Status at Baseline and After 4 Years in Girls Who Have Never Fractured (Group 1) and Those With Previous Fractures (Group 2) (Raw Data)
original image
Table Table 2.. Regional Bone Variables at Baseline and After 4 Years in Girls Who Have Never Fractured(Group 1) and Those With Previous Fractures (Group 2)
original image

The pubertal distribution in groups 1 and 2 (Table 1) did not differ at baseline (κ2 = 3.89, p = 0.27) or at study end (κ2 = 1.5, p = 0.68). The average age at which menarche began was calculated at follow-up for girls who had progressed to this stage. The two groups did not differ in age at menarche (mean ± SD): 12.8 ± 1.1 years for group 1 (n = 48) and 12.8 ± 0.9 years for group 2 (n = 54). By follow-up, 102 (63%) girls were postmenarchal by an average of 2.8 ± 1.8 years in group 1 and 2.5 ± 1.6 years in group 2 (not significant).

Average dietary calcium intake (mg/day) in our 163 girls decreased from baseline (1175 ± 382) to follow-up (910 ± 461; p < 0.001). Only six (4%) girls consumed less than two-thirds of their recommended daily calcium intake(16) at baseline, whereas 50 (31%) girls did this at follow-up. However, there were no differences in mean calcium intake between the groups at baseline (group 1, 1164 ± 326; group 2, 1187 ± 432) or at follow-up (group 1, 914 ± 358; group 2, 907 ± 547). Similarly, we found no group differences in reported hours of vigorous activity per day, hours of television viewing per day, or self-assessed physical activity score.

In contrast, in data adjusted for bone area, height, weight, and pubertal stage, there were marked group differences in most skeletal measures (Table 3). Girls in group 2 (previous fracture group) had significantly lower total body BMC, lumbar spine BMC, ultradistal radius BMC, and hip trochanter than the fracture-free girls (group 1) both at baseline and after 4 years. However, at the 33% radius and the femoral neck site, group 2 values were significantly lower than group 1 at baseline but not after 4 years. Although there seemed to be a trend toward improvement in the ratios relative to group 1 between baseline and follow-up, this did not attain statistical significance at any site.

Table Table 3.. Ratio of Bone Variables at Both Baseline and After 4 Years, Comparing Those With Previous Fractures (n = 82) With Those Who Have Never Fractured (n = 81), with Subgroup Analysis of Those Who did not Fracture Again after Baseline (n = 58) and Those Who Sustained New Fractures During Follow-up (n = 24) Compared with Those Who Hav Never Fractured (n = 81)
original image

A supplementary analysis of group 2 (Fig. 1) was undertaken to see whether BMC differed in girls who had not sustained further fractures and those who sustained new fractures in the 4 years of follow-up (Table 3). Low BMC persisted at the total body, lumbar spine, and ultradistal radius even in the 58 girls who had not broken any more bones. Although the 24 girls with new fractures tended to show even lower ratios at most sites, some improvement after 4 years was observed in this group at the total body and 33% radius only.

Interestingly, group 2 girls did not display greater gain in BMC over the 4 years of follow-up than the fracture-free group (Fig. 2). The adjusted relative gain at all sites in BMC was either less than (total body, lumbar spine, ultradistal radius, and hip trochanter) or the same as (femoral neck and 33% radius) that of the group that had never fractured. Supplementary analysis of group 2 showed similar patterns of relative gain in BMC in the 24 girls with and the 58 girls without new fractures after baseline. Values (ratio and 95%CI, respectively) were as follows: total body, 0.966 (0.948-0.984) and 0.993 (0.979-1.006); lumbar spine, 0.943 (0.906-0.982) and 0.967 (0.939-0.995); ultradistal radius, 0.968 (0.925-1.013) and 0.968 (0.936-1.000); hip trochanter, 0.963 (0.915-1.013) and 0.952 (0.916-0.988); femoral neck, 0.983 (0.945-1.022) and 0.981 (0.953-1.009); and 33% radius, 1.004 (0.970-1.038) and 0.997 (0.973-1.021).

Figure FIG. 2..

Ratios with 95% CIs of the relative 4-year gain of bone mineral content (BMC) at different regional sites in 82 girls with previous fractures compared with those who have never fractured (n = 81). Results were adjusted for baseline BMC and time two variables of bone area, body height, body weight, and Tanner stage of development.


This is the first study to investigate prospectively the relationship between distal forearm fracture and the gain of bone mineral in children. We show that in appropriately adjusted data, a group of 82 girls, ages 3-15 years at baseline, who experienced a distal forearm fracture during childhood or adolescence, maintain their lower BMC over 4 years and do not improve their relative bone gain compared with girls who have never fractured. This is remarkably strong evidence that girls who break their distal forearms show sustained low BMC compared with fracture-free girls. Even girls who had originally fractured but had not sustained new fractures during follow-up still exhibited lower BMC relative to fracture-free controls. Our results corroborate the view that BMC tracks.(7,17,18)

The 4-year persistence of lower BMC at most sites is especially striking given the large gains of bone demonstrated in our study participants. The majority of the girls were undergoing rapid pubertal growth. Around the 2 years of peak bone growth at puberty, approximately one-quarter of the adult skeleton is accrued.(1) Bone gain subsequently slows.(19) Some suggest that there may be a transient lag in areal bone mineral density (BMD) during early puberty as either the bone formation outstrips mineral deposition or as bone porosity increases due to the greater remodeling rate of the growing skeleton.(18,20,21) This may explain the rise in fracture incidence in early adolescence in both sexes.(8,9,22–24) Our results do not support this concept because, if the increased fracture rate was entirely due to a transient drop in areal BMD, we would expect the bone mineral values to have returned to normal or near normal after 4 years, particularly in the girls who had fractured at baseline but not again in the 4 years of follow-up. Rather, our results suggest tracking of low BMC for bone size, height, weight, and pubertal stage puts some girls at greater risk of fracture. During early adolescence this risk may be increased when turnover accelerates and rapid growth occurs. Girls with low BMC may reach the fracture threshold more easily at this time than girls with stronger bones.

No serial assessments of possible posttraumatic osteopenia following a distal forearm fracture have been undertaken in children. Therefore, we cannot rule out the possibility that posttraumatic osteopenia may have contributed to the lower BMC of girls with fractures. In adults, posttraumatic osteopenia develops rapidly after fracture and may persist.(25–27) The magnitude of this osteopenia at different sites varies.(28) Adults who sustained lower limb fractures as children show decreased BMD in the fractured limb compared with the uninjured limb.(29–31) However these fractures involve considerable periods of immobilization. In contrast, our children with fractures of the distal forearm remained physically active, making it unlikely that their low baseline measurements, which were seen throughout the skeleton within a short time of fracture, were due to low mobility. In our view, osteopenia is primary and not secondary to the fracture event itself.

Our present results also do not rule out the possibility that young fracture subjects with low bone mass during growth may accrue their skeleton more slowly but continue to gain bone over a longer period of time. Indeed their reduced gain in BMC relative to the fracture-free group at most sites may reflect this possibility. Further research is required to establish whether this occurs. Unless catch-up improvement in mineral accrual occurs in the future, these girls may arrive at maturity with lower peak bone mass than is desirable. This could place them at risk of osteoporotic fractures later in life.(2) It is encouraging to note that the girls who had experienced new fractures during follow-up showed some evidence of improved BMC relative to fracture-free girls at the second measurement.

Even short-term improvement in bone mass has considerable potential to reduce the risk of childhood fracture that we have shown both cross-sectionally(10,32) and longitudinally(11) increases when bone mass is low. Ours was not an intervention study. Nevertheless, other work suggests the ability to improve BMC during growth does exist. More physically active children have better bone status than their less active peers, even when young.(33) Intervention studies have shown that increasing weight-bearing physical activity augments bone mass.(34–39) Benefits seem to be most consistently observed at the hip, total body, and lumbar spine.(4) Some consider the beneficial bone effects of weight-bearing exercise persist,(40) although this view is contentious.(41) Variations in dietary calcium intake during growth may account for between 5% and 10% of the range in peak bone mass in adults.(5) Although the influence of calcium intake on bone in girls is controversial,(42) most short-term intervention trials(43–48) have demonstrated increased bone mass in children receiving additional calcium, particularly in the lumbar spine, total body, and radius.(4) However, some studies suggest that this is of most advantage to those with low spontaneous calcium intakes,(4,46) and other research suggests that the benefits do not last beyond the supplementation period.(48,49) Nevertheless it is encouraging to note that a recent report has documented sustained benefits of dietary calcium supplementation in young girls.(50) This suggests there is potential for improvement in the bone accrual of girls with past distal forearm fractures, particularly those with low calcium intakes. Combining increased weight-bearing physical activity with increased calcium intake may have an additive effect.(51)

Our study is novel and has a number of strengths because of its prospective design. At baseline, we recruited 84% of a consecutive series of girls who presented at the only fracture clinic for our district with a distal forearm fracture, and then we recruited friend controls of the same age who had never fractured. After 4 years, we followed up a high proportion of this original sample, and only the girls we saw twice are included in our baseline values. We were careful to only include radiologically proven fractures both at baseline and over the ensuing 4 years. The same DXA scanner and operator were used on both occasions. Although we do not have hormonal measurements, self-assessed pubertal staging at baseline and pubertal progression during the 4 years did not differ between the two groups. Moreover, their average age of menarche was also similar. Thus, the bone differences seen between those girls who have never fractured and those girls with previous fractures were not explained by differences in pubertal development. Although oral contraceptives(52–54) and low-dose inhaled steroid medication(55,56) may have minor effects on bone mass, the proportion of girls from each group using these medications were similar.

Our study has important implications. We have shown in appropriately adjusted data that girls who have sustained a distal forearm fracture maintain their lower BMC at most sites and do not improve their relative bone gain compared with girls who have never fractured, for at least 4 years. Unless these girls continue to accrue bone for a longer period than fracture-free subjects, it seems possible that girls with a distal forearm fracture may reach peak bone mass with less bone than is desirable. Consequently, we suggest that interventions to improve bone accrual in girls who sustain distal forearm fractures may be important, first for the treatment of the existing childhood skeletal fragility and second because they have the potential to improve peak bone mass.


We are grateful to the study participants and their families for their willing participation in this research. The Health Research Council of New Zealand funded this research.