Myelomeningocele (MMC) is the most serious and commonly occurring form of spina bifida, which is the most prevalent neural tube defect today. Most lesions occur in the lumbar or sacral regions, affecting the pelvis and lower extremities. During infancy, spontaneous leg movements are depressed and lower limb motor skills are delayed or fail to develop.1–4 In addition to poor motor skill development, the population presents poor bone growth, such as low bone mineral content (BMC), osteopenia/osteoporosis, and pathologic fractures.5–10 For example, by the end of the first year, infants born with MMC exhibit shorter leg lengths and smaller shank circumferences than those of age-matched infants with typical development (TD).9 Also, bone fracture rates ranging from 11.5% to 30% have been reported for children with MMC.6–8 What is not clear is how early BMC is affected and what effect upright supported stepping activity might have on the development of bones in these infants. Because bone tissue is dynamic, it is conceivable that early intervention to increase leg activity of this nature could have a preventive effect.
To date, only a few studies have reported on development of BMC (g) and bone mineral density (BMD [g/cm2]) in the population with MMC and focus has been on older children and adults. These studies used dual-energy x-ray absorptiometry (DXA), which is a standard measure of bone health status because the technology is accurate, reliable, and uses very low levels of radiation.10–14 Generally, children and young adults with MMC have lower BMD in both the upper and lower extremities compared with those of children and young adults with TD. Studies of specific regions of the lower extremities, such as the femoral neck, show that children with MMC who are able to ambulate have higher BMD values in this area than those who cannot ambulate.11,12,14 However, BMC and BMD data for infants with TD and infants with MMC have not been published; thus, we do not know how early and how rapidly these pathologies begin in children with MMC. To be more specific, no research has documented status of or changes in BMC in infants with MMC across the first year after birth, as an isolated population or in comparison with infants with TD.
Whereas the change in BMC and BMD over time is not known, functional activities have been shown to correlate with BMD levels. For example, as ambulatory status in children with MMC increases, so does BMD,10,11,13 but lesion level is also associated with BMD.11,12 We argue that although independent ambulation itself may or may not be possible outcomes for some, BMD might be improved therapeutically by increasing reasonable stresses and forces on the limbs via muscle activity in weight-bearing postures.15–17
Several factors contribute to why so little is known about BMD in babies, with and without MMC. One factor may be that reference values for infants do not yet exist and normalizing techniques have not been well established. In the pediatric population generally, interpretation of bone scan data is complicated by issues related to ongoing skeleton growth in length and shape and changing tissue characteristics. Thus, most bone studies accept raw BMC values to reflect bone development for infants18 but this can be misleading because an increase in bone length without changes in density could suggest improvement in bone “quality” where it does not exist.19–21
Our goals for this pilot study were to describe the developmental trajectory of BMC over the first 18 months of life in infants with MMC—those who received upright supported stepping practice (USSP) and those who did not—to compare BMC in infants with MMC with that of infants with TD and to determine the most appropriate way to normalize BMC values. Although the manufacturer of the technology we used has written software designed for research purposes to normalize the raw data (GE Lunar enCORE software), we believe it is prudent to compare the values obtained in multiple ways to determine the most appropriate approach for early intervention studies. Thus, we compared the raw BMC values, BMD values generated by GE Lunar enCORE software, and normalization of the raw BMC values as a function of the body segment size and mass (see the Appendix).
We hypothesized that
- Based on relatively TD in functional movements of the upper body in children with lumbar- or sacral-level MMC, the arm BMC would show a steady increase with age, and predicted no differences between the TD and MMC groups with age. Whole-body and leg BMC, however, would differ at birth, and the differences would increase with age.
- Based on evidence of weight-bearing exercise on children with TD, infants with MMC who received USSP would show higher BMC values than infants with MMC who did not receive this practice and their rate of increase in BMC at all measured regions would be more similar to that of infants with TD.
- Among the normalization techniques examined to compare BMC values across age in infancy, normalization to segment length or width (eg, leg) would be, overall, the most reasonable option because it related BMC to change in size reflecting relative changes in body tissues.
We tested 36 infants—14 with TD, 13 with MMC, and 9 with MMC who received USSP—across the ages of 1 to 18 months (Table). Of the 36 infants, the infants with TD and MMC without USSP were tested cross-sectionally; their ages at testing were distributed across the 18 months. The 9 infants with MMC who received USSP were tested longitudinally as participants in a larger study of the effects of USSP. These infants with MMC received home-based USSP; the majority of infants started at 1 month of age (mean age at pretest, first DXA scan = 53.2 days ± 24.5) and USSP continued for 12 months. Comparison groups were recruited after the larger study of USSP was initiated and thus random assignment of infants with MMC to subgroups was not possible. Both MMC subgroups received pediatric physical therapy, typically via the “Early On® Michigan program,” when they reached the age of 2 months or more. Inclusion criteria for all infants with MMC were only lumbar- or sacral-level lesions, gestation age at birth greater than 32 weeks, and no known physical problems beyond those associated with MMC. Infants with TD were born full-term, with no known cognitive or physical problems. We recruited infants with MMC via medical teams and support groups across Southeast Michigan and Northeast Ohio. Infants with TD were recruited via the University of Michigan Hospital System's Clinical Studies recruitment Web site and word of mouth. All procedures were approved by the University of Michigan institutional review board. Consent was obtained from parents prior to their infants' participation in the study. Families were given a monetary gift for participating.
Mean weight-for-length values (defined by the World Health Organization)22 for infants with MMC appeared to be higher than those for our infants with TD, although the differences did not reach statistical significance (see Table 1). Furthermore, mean weights for infants with MMC in this study did not reflect overweight status, which is 85th percentile or more by World Health Organization standards, and were similar to the samples of infants with MMC we have studied previously.9
All DXA scans took place at the Michigan Clinical Research Unit in the Cardiovascular Center at University of Michigan Hospital. Each infant with MMC in the USSP subgroup received a DXA scan prior to the start of practice, after 6 months, and after 12 months of practice. Infants in the other 2 subgroups received 1 scan. We conducted testing sessions at the time of day that parents indicated their infant was most likely to sleep. To ensure quality control of the system (GE Lunar Prodigy Advance Plus; GE Health Care, Diegem, Belgium), the University of Michigan clinic DXA technician calibrated the scanner daily. Prior to the scan, the baby's weight and length were measured and entered into the scanner data collection and processing program (GE Lunar enCORE software–Infant Mode; GE Health Care, Diegem, Belgium). We asked parent(s) to change their babies' diapers and dress their infants in clothing without metal parts (eg, snaps) and to swaddle their infants in a cotton blanket if preferred to encourage sleep. When infants slept, parents placed their baby in the supine position on the scan bed [262.3 cm (L) and 90 cm (W)], with arms and legs as close to anatomic position as possible. Once infants were in a deep sleep, the DXA scan began. When excessive movement occurred during the scan, a second scan was attempted.
For normalization and segment mass calculations, we measured infants' anthropometrics: body weight (digital scale, Tanita model #1583, Tanita Cooperation, Japan); lengths and widths (GPM anthropometer; Siber Hegner & Co, Zurich, Switzerland)—total body (crown-heel, measuring board), upper arm (acromion process-lateral epicondyle), forearm (lateral epicondyle to styloid process), thigh (greater trochanter-knee joint), shank (knee-lateral malleolus), and foot (heel-big toe); circumferences (tape measure)—upper arm, forearm, thigh, and shank; widths—malleolus and knee.
MMC With USSP Subgroup
This subset of infants with MMC was provided with a custom-engineered pediatric treadmill (Carlin's Creations, Sturgis, Michigan). Parents were asked to help their babies' practice being upright and stepping 10 continuous minutes per day, 5 days per week at a belt speed of 0.144 m/s. The protocol was progressive with variations in USSP adjusted to each baby's individual rate of stepping/neuromotor strength and control. The USSP continued for 12 months. At intervention onset, we encouraged parents to use a combination of 3 approaches to engaging their baby in upright activity: (a) newborn step elicitation, (b) bouncing on parent's lap, and (c) treadmill step elicitation. Because some very young infants do not step well on a treadmill, options a and b enabled parents an opportunity for encouraging infants to accept some weight when upright and to activate their extensor muscles in situations familiar to both parents and therapeutic. Parents started with at least 3 minutes of treadmill stepping; when infants' responses on the treadmill were calm and included some weight-bearing at least 50% of the 10-minute practice time, we asked them to increase the proportion of treadmill practice by another minute until all 10 minutes consisted of this practice. The rate at which this was achieved varied from approximately 3 to 6 months.
To monitor adherence to the intervention, we asked parents to complete a log of daily stepping practice: duration practiced per day, approach used, and infants' general response to the treadmill. A small gauge on the side of the treadmill also recorded the number of minutes the treadmill was used. Researchers visited each family's home biweekly to check compliance, answer questions from the family, and offer suggestions for optimizing practice sessions. Parents were most likely to provide exercise an average of 6 to 8 minutes daily, which was lower than the requested 10 minutes a day.
Post-DXA scan, we reviewed the image to ensure that the body segment template was aligned appropriately over the infant's body and no movement artifact occurred. As needed, we adjusted the template geometric lines via the software to fit variation in body posture for regions including legs, pelvis, trunk, arms, and head (Figure 1).
Normalization of Raw BMC
Starting with the raw BMC values, we normalized the data for whole body, legs, and arms separately, using the anthropometrics we measured directly and estimates of segmental mass, to establish BMC values expressed as a function of each individual baby's respective body size. See the Appendix for the specific formulas used.
All data were reviewed for distribution normality and homogeneity in error variance, ensuring that assumptions of multiple linear regression analyses were not violated. We used a multiple regression model to compare BMC at all sites measured across the groups over developmental time. In the multiple regression models, we allowed age to vary across groups so that we could apply regression analysis with varied distribution of age points among the groups. We chose an α level of 0.05 to determine the significance of all dependent variables (2-tailed).
Across raw data and all levels of normalization, whole-body BMC increased for all subgroups over the first 18 months. However, small but important differences were evident, among the normalization procedures used. Figure 2A shows that raw whole-body BMC values were very similar among the subgroups in the neonatal period. With time, a clear advantage for infants with TD emerged compared with infants with MMC, with infants with TD gaining bone mineral more rapidly than infants with MMC. Statistically, no significant differences emerged among the 3 subgroups for slope of their data regression lines (Wald χ2 = 2.734, P = .255). Also, no differences between MMC subgroups were observed.
These relations among subgroups varied, however, when normalization procedures were applied. When whole-body data were normalized to skeletal length and mass, a trend toward lower values for both MMC subgroups than for infants with TD was seen, although they still failed to reach statistical significance (Wald χ2 = 2.68, P = .262). Furthermore, when plotting the GE BMD values, subgroup relations inverted with age compared with what was observed in all other data treatments; infants with MMC without USSP achieved the highest BMD values of all 3 subgroups (Figure 2B). Infants with MMC in the USSP emerged as intermediate in the 3-subgroup comparison, and in infants with TD bone density scores were lowest. Again, the visually observed trends were not strong enough to reach statistical significance (Wald χ2 = 3.32, P = .19).
Figure 3 illustrates raw data and data normalized via several techniques. Infants with TD tended to show steady improvement with age and higher values than both MMC subgroups. Infants with MMC who received USSP showed values that tended to rise in parallel to their peers with TD, whereas infants with MMC without USSP tended to show the lowest values across age and sometimes a negligible rise in slope as age increased.
Raw BMC values for the legs was greater for infants with TD than for both subgroups of infants with MMC at all ages, with the difference increasing over developmental time (Figure 3A). Infants with MMC in the USSP subgroup showed lower raw leg BMC data than infants with TD, but the slope tended to increase at a rate similar to infants with TD across age. A comparison of raw leg BMC data regression lines demonstrated a significant difference among groups (Wald χ2 = 14.90, P = .00). Post hoc analysis produced statistical differences between MMC subgroups (Wald χ2 = 5.05, P = .03). In addition, a significant difference was found between infants with TD and infants with MMC who received USSP (Wald χ2 = 6.26, P = .01).
Depending on the normalization method used, regression line slopes varied. Leg BMC data normalized to knee, and malleolus widths showed significant differences among subgroups, with the lowest BMC in MMC subgroup without USSP over the first 18 months of age (for leg BMC normalized by knee width: Wald χ2 = 14.43, P = .01 and by malleolus width: Wald χ2 = 14.88, P = .01) (Figure 3C). Post hoc analysis indicated that differences occurred between MMC subgroups with and without USSP (Wald χ2 = 6.29, P = .01). In addition, a significant difference was found between infants with TD and infants with MMC in the USSP group (Wald χ2 = 4.68, P = .03). However, when raw leg BMC data were normalized to leg length and BMD, the slopes within each set of 3 regression lines were not significantly different (Wald χ2 = 4.54, P = .10, for BMC normalized by leg length; Wald χ2 = 0.58, P = .97 for leg BMD) (Figures 3B and 3D). Leg BMC normalized by mass produced the reverse pattern: regardless of subgroup, all demonstrated a decrease in leg BMC with age, but no statistically significant differences were seen among subgroups (Wald χ2 = 8.57, P = .65) (Figure 3E).
Figure 4 illustrates that across raw data and all normalized data, both MMC subgroups tended to show lower values than infants with TD. Infants with MMC who received USSP generated values that tended to rise in parallel to their peers with TD, whereas the MMC subgroup without USSP tended to show the lowest values with age and sometimes a small rise in slope across age.
Statistical analysis of raw arm BMC regression lines resulted in a significant difference among groups (Wald χ2 = 6.92, P = .03) (Figure 4A). For infants with MMC with USSP, raw BMC values were slightly lower than those of infants with TD; nevertheless, their values rose with age, indicating that the rate of accrual of bone mineralization was similar to that of infants with TD. Between the 2 MMC subgroups, a trend toward higher values for raw arm BMC values was seen in infants with USSP (Wald χ2 = 6.46, P = .01). None of the normalization procedures revealed statistically significant differences among subgroups for normalized arm BMC (see 1 example for BMD in Figure 4B). Similar to the values for leg BMC normalized to mass, arm BMC normalized to arm mass also showed strong negative correlations with age across all subgroups (Wald χ2 = 4.91, P = .78).
To the best of our knowledge, this pilot study is the first to describe the development of whole-body and body segment BMC in infants with MMC across the first 18 months postbirth. It is also the first to assess the effect of USSP on BMC in infants with MMC. As expected for babies with TD, and as we found for infants with MMC in both groups, whole-body BMC increased with age. We observed small, but not statistically significant, differences in whole-body BMC among these 3 subgroups; infants with TD consistently demonstrated slightly higher values than MMC groups, regardless of whether we examined the raw data or data normalized by skeletal length, BMD, or body mass. However, USSP seemed to improve the rate of bone mineralization in infants with MMC at all sites measured, but its effect was more statistically pronounced in the lower body.
Trajectory of Raw BMC Over the First 18 Months
Whole-body BMC increased steadily during infancy regardless of group, but the rate of change differed slightly among the groups. At birth, BMC was similar, but with age, differences emerged. Birth similarities may be due to similar levels of spontaneous activity reported for these populations during the fetal period.1,3,4 Activity increases healthy stresses on the skeleton that lead to replacing cartilage with bone mineral. The transition from pre-to postnatal life sets the stage for cascading effects. The uterine environment creates buoyancy and environmentally induced motions that may trigger motility. Postnatally, gravity increases the demands on energy and neural control to create movement. And, infants with MMC may require postnatal spinal surgery. These experiences may disrupt and reduce activity patterns.
BMC for legs and arms seemed to diverge more quickly postbirth than whole-body BMC for infants with MMC than for their peers with TD. Moreover, infants with MMC who did not receive USSP showed no change in leg or arm BMC. Lower raw BMC values observed for babies with MMC than for babies with TD aligned with previously reported data for children and adults, particularly in the lower body.10–14 Reported BMD values for all sites examined in the lower extremities, including femoral neck, trochanteric region of hip, mid-tibia, and metatarsals, were approximately 1 to 2 SDs below the mean of age- and gender-matched individuals with TD. Our results suggest that this lag and the lower BMC mean values start in infancy, setting up a potential for cascading effects including osteopenia, osteoporosis, and fractures.
Low BMC in the arms of both subgroups of infants with MMC was not predicted. However, both subgroups of infants with MMC may have demonstrated less progress in bone mineralization than infants with TD as a function of their parents' protective treatment of their medically fragile babies with regard to positioning. When these newborns undergo spinal surgery to place the cord in the vertebral column, the practice of placing infants “back to sleep” is achieved after spinal surgical wounds resolve sufficiently. Subsequently, parents may be more reluctant than parents of infants with TD to “push” their babies into tummy time, which facilitates greater muscle activity and weight-bearing in the arms. The USSP received by 1 of the MMC subgroups may have enhanced bone development by engaging them in holding onto parents' arms during practice and otherwise encouraging more spontaneous upper limb movements during and beyond actual practice time.
The arm BMC difference between infants with TD and infants with MMC may resolve with age, as greater upper limb strength and skill are acquired by those with MMC. By the time they are adolescents, many persons with MMC use assistive devices requiring greater than average weight support on the upper extremities, contributing to healthier bone density in the upper body.10 Nevertheless, at least 1 study found reduced BMD at the distal radius in children with MMC compared with children with TD.23 Clearly, more longitudinal studies with larger samples are needed to understand more clearly the developmental trajectory of arm BMD in persons with MMC and the factors that influence this beyond the neural defect.
A large database to which infant participants' raw data can be compared to establish a normative percentile does not exist. And, increases in the quantity of BMC raw values may or may not reflect proportional increases within the bones. Increases can occur if proportions remain the same and bones simply become larger. Thus, normalization techniques are needed to compare BMC across ages and subgroups. Among the normalization methods we used, our results suggest that, depending on the region of interest of the body, some were more sensible than others to reflect the underlying trajectory of BMC change among subgroups, across ages, and the effect of USSP. Although our data are preliminary, the method we found to work best, to be most consistently aligned with the literature and what might support hypotheses for differences among subgroups, was normalization to bone size: skeletal length and joint widths. We did not measure wrist or elbow width for normalization methods of arm BMC but suggest that future studies examine these as options. Unlike normalization to skeletal lengths or widths, normalization to mass and area produced less defensible patterns and sometimes patterns that were inverted in comparison with raw BMC or contrary to what extant literature would predict.
Normalization to segmental mass resulted in decreases in arm and leg BMC values as age increased. In retrospect, these trajectories make sense, given the relative changes during infancy in body weight and body length. From birth to 12 months of age, a 50% increase in infants' skeletal length occurs whereas their body mass is increased to values 3 times what they weighed at birth. Thus, while bones grow longer and denser, their relative increases are much smaller than growth in body mass. That infants with MMC show more negative slopes than their peers with TD may simply reflect their relatively greater increase in adipose tissue rather than bone and muscle. There are a few studies suggesting that body composition (ie, lean mass) can explain the changes of total body BMC minus the head values for children and adults.24 However, deconstructing body tissue mass into its components for children younger than 2 years is less reliable and less accurate due to their small body size than for older persons.25
BMD also showed a different rate of change compared with whole-body BMC or whole-body BMC normalized using our other methods. Previous researchers have argued that small, nonuniformly developing bones may distort the edge detection of body areas captured by DXA–infant software and thus BMD calculations for smaller areas than the whole may lose some accuracy as compared with the adult scan. For infants younger than 2 years, whole-body scans reflect bones with contours different from more mature bones, causing more complicated estimation of the area of individual bones. Therefore, even with well-derived infant software algorithms, interpretation of bone density results is challenging for patients who are very young. This may be why BMC is currently the preferred variable for describing bone mineral status in growing infants, rather than area BMD. Nonetheless, the disparate findings for BMD compared with raw and normalized BMC via other factors must be tested further with larger samples to draw more firm conclusions about validity.
Effects of USSP on BMC
Despite the fact that the MMC population is at increased risk for low bone density as they age, the implementation of USSP to increase bone strength in nonambulatory populations has long been neglected, particularly during infancy. Our pilot results suggest that BMC in infants can be increased via USSP compared with infants with MMC who do not receive USSP. Interestingly, a slight difference was observed for whole-body BMC; the increase was focused more on the arms and legs. These findings are thus likely to be a result of the benefits of early and repeated USSP on bone mineralization during early infancy.
The reduced spontaneous leg activity characteristic of infants with MMC in general may contribute to the reduced rate of bone mineralization in our MMC subgroups of participants who did not receive frequent opportunities to be upright and active. The USSP, designed to elicit mainly muscle activity across the leg muscles and mechanical compression on joints of the lower body, influenced a systemic effect on the entire body, at least extending to the arms, instead of localized effects on the legs. That is, USSP could play an important role in reducing the risk of low BMC at all measured sites in infants with MMC. Our pilot findings enhance previous work showing that even passive range-of-motion exercise, provided daily to the upper and lower extremities, increased bone mineralization in premature infants.26–28 Similarly, investigators of at least 1 study reported that an increase in BMD at the lumbar vertebrae and femoral neck regions in adults with MMC who participated in sport activities in comparison with those who did not.10
Our work focused only on infants with spinal-level lesions in the lumbar and sacral regions, thus the prospects of standing or walking (with or without assistance) are good but cannot be predicted with certainty at 1 month postbirth. Nevertheless, our results support that early and repeated USSP may be expected to improve BMC in the bones of infants with MMC. Ultimately, although this practice may not be in service of standing or walking, it could contribute to the broader health goals and may present an option to reduce secondary consequences (eg, osteoporosis or fractures) in children with MMC that, ultimately, lead to decreased mobility and quality of life.
This pilot work suggests that early efforts to engage infants with MMC in USSP might be a valuable component of physical therapy. However, the work needs to be replicated and confidence in the importance of the results would increase with a larger sample with more infants across each age. Future studies with larger samples should compare the effects of different doses and types of physical activity on BMC in infants.
The authors thank the families who participated and clinic staff who helped them in recruitment. The authors also thank Arielle Bianco and Rachel Lee for their research assistance in data collection.
1. Rademacher N. Early spontaneous leg movements in infants born with and without myelomeningocele. Pediatr Phys Ther. 2008;20(2):137–145.
2. Findley TW, Agre JC, Habeck RV, Schmalz R, Birkebak RR, McNally MC. Ambulation in the adolescent with myelomeningocele, part I: early childhood predictors. Arch Phys Med Rehabil. 1987;68(8):518–522.
3. Sival D, van Weerden T, Vles JSH, et al. Neonatal loss of motor function in human spina bifida aperta. Pediatrics. 2004;114(2):427–434.
4. Smith BA, Teulier C, Sansom J, Stergiou N, Ulrich BD. Approximate entropy values demonstrate impaired neuromotor control of spontaneous leg activity in infants with myelomeningocele. Pediatr Phys Ther. 2011;23(3):241–247.
5. Anschuetz RH. Severe fracture complications in myelodysplasia. J Pediatr Orthop. 1984;4(1):22–24.
6. Akbar M, Bresch B, Raiss P, et al. Fractures in myelomeningocele. J Orthop Traumatol. 2010;11(3):175–182.
7. James CC. Fractures of the lower limbs in spina bifida cystica: a survey of 44 fractures in 122 children. Dev Med Child Neurol Suppl. 1970;22(suppl 22):88–93.
8. Lock TR, Aronson DD. Fractures in patients who have myelomeningocele. J Bone Joint Surg Am. 1989;71(8):1153–1157.
9. Teulier C, Smith BA, Kubo M, et al. Stepping responses of infants with myelomeningocele when supported on a motorized treadmill. Phys Ther. 2009;89(1):60–72.
10. Valtonen KM, Goksor LA, Jonsson O, Mellstrom D, Alaranta HT, Viikari-Juntura ER. Osteoporosis in adults with meningomyelocele: an unrecognized problem at rehabilitation clinics. Arch Phys Med Rehabil. 2006;87(3):376–382.
11. Apkon S, Fenton L, Coll J. Bone mineral density in children with myelomeningocele. Dev Med Child Neurol. 2009;51(1):63–67.
12. Ausili E, Focarelli B, Tabacco F, et al. Bone mineral density and body composition in a myelomeningocele children population: effects of walking ability and sport activity. Eur Rev Med Pharmacol Sci. 2008;12(6):349–354.
13. Rosenstein BD, Greene WB, Herrington RT, Blum AS. Bone density in myelomeningocele: the effects of ambulatory status and other factors. Dev Med Child Neurol. 1987;29(4):486–494.
14. Szalay EA, Cheema A. Children with spina bifida are at risk for low bone density. Clin Orthop Relat Res. 2011;469(5):1253–1257.
15. Eliakim A, Raisz LG, Brasel JA, Cooper DM. Evidence for increased bone formation following a brief endurance-type training intervention in adolescent males. J Bone Miner Res. 1997;12(10):1708–1713.
16. Slemenda CW, Miller JZ, Hui SL, Reister TK, Johnston CC Jr. Role of physical activity in the development of skeletal mass in children. J Bone Miner Res. 1991;6(11):1227–1233.
17. Karlsson MK, Ahlborg H, Obrant KJ, Nyquist F, Lindberg H, Karlsson C. Exercise during growth and young adulthood is associated with reduced fracture risk in old ages. J Bone Miner Res. 2002;17(suppl 1):S297.
18. Carter DR, Bouxsein ML, Marcus R. New approaches for interpreting projected bone densitometry data. J Bone Miner Res. 1992;7(2):137–145.
19. Koo WWK, Bush AJ, Walters J, Carlson SE. Postnatal development of bone mineral status during infancy. J Am Coll Nutr. 1998;17(1):65–70.
20. Koo WWK, Walters J, Bush AJ, Chesney RW, Carlson SE. Dual-energy x-ray absorptiometry studies of bone mineral status in newborn infants. J Bone Miner Res. 1996;11(7):997–1002.
21. Salle BL, Braillon P, Glorieux FH, Brunet J, Cavero E, Meunier PJ. Lumbar bone mineral content measured by dual energy x-ray absorptiometry in newborns and infants. Acta Paediatr. 1992;81(12):953–958.
22. WHO Multicentre Growth Reference Study Group. WHO Child Growth Standards: Length/Height-for-Age, Weight-for-Age, Weight-for-Length, Weight-for-Height and Body Mass Index-for-Age: Methods and Development. Geneva, Switzerland: World Health Organization; 2006.
23. Quan A, Adams R, Ekmark E, Baum M. Bone mineral density in children with myelomeningocele. Pediatrics. 1998;102(3):E34.
24. Crabtree NJ, Kibirige MS, Fordham JN, et al. The relationship between lean body mass and bone mineral content in paediatric health and disease. Bone. 2004;35(4):965–972.
25. Hammami M, Koo W, Hockman E. Technical considerations for fan-beam dual-energy x-ray absorptiometry body composition measurements in pediatric studies. JPEN J Parenter Enteral Nutr. 2004;28(5):328–333.
26. Litmanovitz I, Dolfin T, Friedland O, et al. Early physical activity intervention prevents decrease of bone strength in very low birth weight infants. Pediatrics. 2003;112(1, pt 1):15–19.
27. Moyer-Mileur L, Luetkemeier M, Boomer L, Chan GM. Effect of physical-activity on bone mineralization in premature-infants. J Pediatr. 1995;127(4):620–625.
28. Moyer-Mileur LJ, Brunstetter V, McNaught TP, Gill G, Chan GM. Daily physical activity program increases bone mineralization and growth in preterm very low birth weight infants. Pediatrics. 2000;106(5):1088–1092.
29. Schneider K, Zernicke RF. Mass, center of mass, and moment of inertia estimates for infant limb segments. J Biomech. 1992;25(2):145–148.